🔬 applied physics

Reduction of intrinsic losses in nanomechanical silicon nitride resonators through thermal treatment in ultrahigh vacuum

This paper demonstrates that thermal treatment in ultrahigh vacuum significantly reduces surface loss and enhances the intrinsic quality factor of silicon nitride nanomechanical resonators by converting surface hydroxyl groups into siloxane bridges, thereby establishing surface chemistry as a tunable parameter for achieving ultracoherent mechanical systems.

Original authors: Nicola Cavalleri, Ariane Giesriegl, Robert G. West, Kostas Kanellopulos, Saeed Rasouli, Daniele Nazzari, Pedram Sadeghi, Sebastian Alberti, Antonius Armanious, Silvan Schmid

Published 2026-08-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Nicola Cavalleri, Ariane Giesriegl, Robert G. West, Kostas Kanellopulos, Saeed Rasouli, Daniele Nazzari, Pedram Sadeghi, Sebastian Alberti, Antonius Armanious, Silvan Schmid

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 a tiny, invisible trampoline made of silicon nitride, so small you'd need a microscope to see it. Scientists have spent years trying to make these microscopic trampolines vibrate as perfectly as possible. When they vibrate, they store energy, but over time, that energy leaks away, and the vibration stops. This "leaking" is called dissipation. The goal is to make the vibration last as long as possible, measured by something called the "quality factor" (or Q). Think of it like a bell: a cheap bell rings for a split second, but a perfect crystal bell can ring for minutes.

Why does this matter? If we can make these tiny trampolines vibrate for a really long time without losing energy, they become incredibly sensitive. They could detect the tiniest forces in the universe, like a single atom landing on them, or even help us build quantum computers that work at room temperature instead of needing giant, expensive freezers. For a long time, scientists thought the main reason these trampolines stopped vibrating was because of their shape or the material inside. But a new study suggests the real culprit is hiding on the very surface, like a layer of sticky dust that slows everything down.

The Sticky Surface Problem

For the last two decades, engineers have been building these silicon nitride trampolines with a special trick: they stretch the material tight, like a drum skin. This stretching helps cancel out some of the internal friction, allowing the trampolines to ring for a long time. However, even with this trick, there was a stubborn limit. No matter how thin or perfect the trampoline was made, the vibration would eventually die out. Scientists suspected the problem was on the surface, but they weren't sure exactly what was causing the friction. Was it a layer of oxide? Was it dirt? Or was it something else entirely?

The Vacuum Oven Solution

In this new study, researchers at TU Wien in Vienna decided to treat these tiny trampolines like a piece of pottery in a kiln, but with a twist. They placed the resonators in an ultra-high vacuum chamber—a space so empty it's almost a perfect void—and baked them at scorching temperatures, up to 1000°C.

The results were dramatic. After this "baking" process, the quality factor of the thinnest trampolines (just 10 nanometers thick) jumped by a factor of 20. To put that in perspective, if the trampoline was previously ringing for one second, it now rings for twenty seconds. This wasn't just a small improvement; it was a massive leap. The researchers also noticed that the tension in the material increased, making the "drum skin" even tighter.

The "Silanol" Secret

So, what actually happened inside the oven? The team used special tools to look at the chemical makeup of the surface before and after baking. They found that the heat didn't change the bulk of the material or remove the entire oxide layer. Instead, it triggered a specific chemical reaction on the surface.

Imagine the surface of the trampoline is covered in tiny, wet hands (hydroxyl groups, or -OH) reaching out and grabbing onto each other. These "wet hands" are sticky and cause friction. When the researchers heated the trampolines in the vacuum, these hands let go of the water and grabbed onto each other instead, forming a strong, dry bridge (a siloxane bond). This process is called "silanol condensation."

The evidence for this was clear:

The Chemical Shift: Spectroscopy showed that the "wet hands" (N-H/O-H groups) disappeared, and the "dry bridges" (Si-O-Si bonds) appeared.
The Reversibility Test: This is the most convincing part. When the researchers took the baked, super-efficient trampolines and exposed them to humid air, the "dry bridges" broke apart, and the "wet hands" returned. The trampoline immediately lost its superpowers and went back to its old, sluggish self. This proved that the improvement wasn't a permanent change to the material's structure, but a reversible chemical state of the surface.

What It Is NOT

The researchers were careful to rule out other possibilities. They showed that the improvement wasn't just because the vacuum removed loose water molecules (which happens at much lower temperatures). It wasn't because the oxide layer grew thicker or thinner. It wasn't because the material inside the trampoline changed. The key was specifically the chemical termination of the surface—the "hands" reaching out.

The Bigger Picture

This discovery changes how we think about these devices. For a long time, surface loss was treated like a fixed, unchangeable property of the material, like the weight of a rock. This paper shows that it's actually a tunable chemical setting. By controlling the surface chemistry, we can make these tiny mechanical systems much more efficient.

While the effect is reversible in humid air, the study suggests that if we can find a way to "lock" these dry bridges in place (perhaps through chemical passivation), we could build next-generation sensors and quantum devices that work incredibly well, even without needing to be kept in a deep-freeze. The path forward involves figuring out how to protect these super-efficient surfaces from the humidity of our everyday world, but the recipe for making them is now in our hands.

Technical Summary: Reduction of Intrinsic Losses in Nanomechanical Silicon Nitride Resonators

Problem Statement
Silicon nitride (SiN) nanomechanical resonators have established the benchmark for ultracoherent mechanical systems, largely due to the discovery of dissipation dilution, where induced tensile stress adds a nearly lossless restoring force. Despite geometry and strain engineering achieving quality factors ( $Q$ ) approaching $8 \times 10^9$ , surface loss remains the dominant and largely unaddressed dissipation channel. As film thickness ( $h$ ) enters the sub-micron regime, the intrinsic quality factor ( $Q_{int}$ ) scales linearly with thickness ( $Q_{int} \approx \beta \cdot h$ ), indicating that surface effects limit performance regardless of resonator geometry. While thermal treatment has been shown to improve $Q$ in crystalline silicon and cantilevers, no systematic study had previously addressed dissipation-diluted SiN membrane resonators. The microscopic origin of surface loss—whether native oxide, hydroxyl groups, or adsorbed hydrocarbons—remained unclear.

Methodology
The authors investigated stoichiometric and silicon-rich low-pressure chemical vapor-deposited (LPCVD) SiN square membranes with thicknesses ranging from 10 to 300 nm and varying pre-stress levels (200 MPa and 1100 MPa). The study employed the following experimental protocols:

Thermal Treatment: Samples were subjected to thermal cycles in an ultrahigh vacuum (UHV, $p < 5 \cdot 10^{-10}$ mbar) environment using a resistive heating stage, reaching temperatures up to 1200°C.
Mechanical Characterization: Intrinsic quality factors ( $Q_{int}$ ) were extracted by measuring high-order mode frequencies and ring-down times. The data was fitted to a dissipation model accounting for bulk loss, surface loss, and an excess volume loss term specific to ultrathin films. Tensile stress was derived from the frequency dispersion relation.
Surface Analysis: The chemical composition of the membrane surfaces was analyzed using nanomechanical photothermal Fourier-transform infrared spectroscopy (FTIR) and in situ X-ray photoelectron spectroscopy (XPS) before and after treatment.
Reversibility Testing: Treated samples were stored under three conditions: high vacuum ( $10^{-7}$ mbar), dry air (<10% relative humidity), and humid air (>95% relative humidity) to observe the stability of the improvements.

Key Results

Enhancement of Intrinsic Quality Factor: Thermal treatment at 1000°C in UHV increased the intrinsic quality factor of dissipation-diluted SiN membranes by up to a factor of 20. For 10 nm membranes, $Q_{int}$ rose from approximately $1 \times 10^6$ to over $30 \times 10^6$ .
Reduction of Surface Loss: The treatment improved the surface-loss coefficient ( $\beta$ ) by a factor of 8 (from $\approx 110$ nm $^{-1}$ to $\approx 910$ nm $^{-1}$ ) and reduced the excess loss in ultrathin films by a factor of 50.
Stress Modification: The treatment concurrently increased tensile stress, with the magnitude of the increase scaling approximately as $1/h$ , confirming a surface-localized origin.
Chemical Mechanism: FTIR and XPS data identified the underlying mechanism as thermally activated silanol condensation. The treatment converted surface hydroxyl terminations (Si-OH/N-H) into siloxane bridges (Si-O-Si), evidenced by a reduction in the O-H/N-H stretching band (3300 cm $^{-1}$ ) and a growth in the Si-O-Si band (1010 cm $^{-1}$ ). The bulk hydrogen content and the overall oxide layer thickness remained largely unchanged.
Reversibility: The improvements in $Q_{int}$ and stress were fully reversible upon exposure to ambient humidity. Samples stored in humid air reverted to their pre-treatment values within days, while those in vacuum or dry air retained the enhancements. This confirmed that the hydroxyl termination, rather than the oxide volume or adsorbed hydrocarbons, is the primary source of dissipation.

Significance and Claims
The paper establishes that surface loss in SiN resonators is not a fixed material property but a chemically tunable parameter. By targeting the surface chemistry—specifically the conversion of hydroxyl groups to siloxane bridges via thermal treatment in UHV—the authors demonstrate a geometry-independent method to significantly enhance mechanical coherence. The findings resolve the ambiguity regarding the origin of surface loss, identifying surface hydroxyls as the dominant lossy element rather than the oxide-nitride interface or the oxide volume itself.

The authors note that while the enhancement is substantial, practical application in non-vacuum environments requires chemical passivation of the condensed surface to prevent rehydroxylation. Furthermore, they suggest that while the current heating method (chuck heating) effectively treats the clamping regions, future applications to defect-mode resonators may require targeted photothermal treatment to heat specific localized regions where dissipation occurs.

Technical Summary: Reduction of Intrinsic Losses in Nanomechanical Silicon Nitride Resonators

More like this