🔬 applied physics

Residual Stress Anisotropy In Thin-Film Lithium Niobate For Stress-Managed MEMS

This paper presents the first experimental study of residual stress anisotropy in 128-degree Y-cut thin-film lithium niobate, identifying specific in-plane orientations and thicknesses that minimize beam deflection to enable the fabrication of exceptionally long, mechanically stable MEMS structures.

Original authors: Byeongjin Kim, Ian Anderson, Tzu-Hsuan Hsu, Ziqian Yao, Mihir Chaudhari, Sinwoo Cho, Ruochen Lu

Published 2026-04-22

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Byeongjin Kim, Ian Anderson, Tzu-Hsuan Hsu, Ziqian Yao, Mihir Chaudhari, Sinwoo Cho, Ruochen Lu

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 baking a very thin, delicate sheet of glass. You want to cut it into tiny, floating bridges (microscopic beams) to build a super-fast, tiny machine. But there's a problem: as the glass cools down after being baked, it wants to shrink. Because it's stuck to a silicon base, it can't shrink evenly. This creates hidden tension inside the glass, like a rubber band that's been stretched too tight or squished too hard.

In the world of tiny machines (called MEMS), this hidden tension is a nightmare. If it's too strong, your tiny bridges will snap, curl up like a potato chip, or collapse onto the ground.

This paper is about a team of scientists at the University of Texas who decided to map out exactly where this "hidden tension" lives in a special material called Thin-Film Lithium Niobate (TFLN). They discovered that this tension isn't the same in every direction; it's anisotropic, which is a fancy way of saying it acts differently depending on which way you look at it.

Here is the story of their discovery, broken down into simple concepts:

1. The "Stress Map" (Finding the Sweet Spots)

Think of the material like a giant, invisible trampoline. If you stand in the middle, it might sag one way. If you stand near the edge, it might sag another way.

The scientists cut hundreds of tiny "bridges" out of this material, rotating them slightly each time (like turning a steering wheel from 0 to 160 degrees). They measured how much each bridge curled up after being released from the silicon base.

The Big Discovery:
They found that the tension isn't random. It has specific "safe zones."

For thicker films (220–460 nanometers): If you cut your bridge at an angle of about 55° or 125°, the tension cancels itself out. The bridge stays perfectly flat, like a calm lake.
For ultra-thin films (100 nanometers): The rules change! The "safe zones" shift to 20° and 160°. Also, these super-thin films tend to curl upward more, like a piece of paper curling in the sun.

2. The "Gradient" vs. The "Average"

The scientists realized that the tension isn't just a uniform squeeze. It's more like a gradient (a slope). Imagine a sandwich where the top slice of bread is trying to shrink, but the bottom slice isn't. This makes the whole sandwich curl.

They found that this "curling force" (which they call gradient stress) is the main villain causing the bridges to bend. By finding the angles where this curling force is zero, they could make bridges that stay perfectly straight.

3. The "Super-Bridge" Achievement

To prove their theory, they didn't just make tiny bridges; they made a giant one (in the microscopic world).

They took a 460-nanometer-thick film.
They cut a bridge at 132.5° (very close to their "safe zone" of 125°).
They made it 2 centimeters long (which is huge for a microscopic beam) and only 10 micrometers wide.

The Result: The bridge floated in the air for 2 centimeters without collapsing or snapping. Usually, a bridge that long would immediately crumble under its own stress. This is like building a suspension bridge out of a single sheet of paper that doesn't tear.

Why Does This Matter?

Think of this like learning the "wind patterns" of a new city before building skyscrapers.

Before: Engineers were building these tiny machines blindly, hoping they wouldn't break. Many failed, leading to wasted time and money.
Now: Engineers have a map. They know exactly which angle to cut the material and how thick to make it to ensure the machine stays flat and stable.

The Takeaway

This paper gives engineers a "cheat code" for building better, more reliable tiny machines. By simply rotating the material to the right angle and choosing the right thickness, they can turn a fragile, stress-filled sheet into a sturdy, floating structure. This paves the way for faster phones, better sensors, and more advanced optical devices that won't fall apart.

In short: They found the "Goldilocks" angles where the material is perfectly relaxed, allowing them to build the longest, flattest, and most stable tiny bridges the world has ever seen.

1. Problem Statement

Residual stress is a critical reliability factor in thin-film Microelectromechanical Systems (MEMS), particularly in transferred thin-film lithium niobate (TFLN) on silicon. Uncontrolled residual stress leads to device failure modes such as:

Film cracking (due to tensile stress).
Peeling, buckling, or blistering (due to compressive stress).
Post-release collapse of suspended structures (e.g., beams and membranes).

While residual stress has been studied in bulk acoustic devices and other thin-film systems, there was a significant gap in understanding the in-plane orientation-dependent and thickness-dependent residual stress distribution in 128° Y-cut TFLN. As TFLN devices move toward higher frequencies requiring thinner films (e.g., <100 nm), the lack of systematic stress characterization limits the scalability and yield of these devices.

2. Methodology

The authors employed a combination of experimental fabrication, optical metrology, and finite element analysis (FEA) to map residual stress.

Fabrication:
- Micromachined cantilever arrays were fabricated from 128° Y-cut TFLN on Si with three distinct thicknesses: 100 nm, 220 nm, and 460 nm.
- Beams were patterned with in-plane orientations ( $\theta$ ) ranging from 0° to 175° (in 5° increments) to capture full anisotropy.
- Structures were released using XeF₂ etching after low-temperature ICP-RIE patterning to minimize thermal stress generation.
Stress Decomposition Model:
- The residual stress field was modeled as a first-order expansion through the film thickness: $\sigma(z) = \sigma_0 + \sigma_1 \frac{z}{h/2}$ $σ (z) = σ_{0} + σ_{1} \frac{z}{h /2}$ .
  - $\sigma_0$ : Uniform mean stress (associated with CTE mismatch).
  - $\sigma_1$ : Linear gradient stress (associated with depth-dependent effects like bonding damage or differential relaxation).
- Challenge: In anisotropic materials like LN, extracting the mean stress ( $\sigma_0$ ) from beam rotation is unreliable due to geometry-dependent bending stiffness.
- Solution: The authors focused on extracting the gradient stress ( $\sigma_1$ ), which is determined primarily by the normalized curvature ( $h/R$ ) and is insensitive to beam width/length.
Measurement & Analysis:
- Optical Profilometry: Laser confocal profilometry was used to measure out-of-plane beam deflection profiles.
- Curvature Fitting: Profiles were fitted to quadratic models to extract curvature ( $R$ ) and calculate $\sigma_1$ using the formula $\sigma_1 = \frac{E(\theta) h}{2R}$ , where $E(\theta)$ is the directional Young's modulus derived from the rotated stiffness tensor of 128° Y-cut LN.
- Validation: FEA simulations confirmed the relationship between stress components and beam shape (curvature vs. rotation).

3. Key Contributions

First Systematic Mapping: This is the first experimental study to map the in-plane, thickness-dependent residual stress anisotropy in 128° Y-cut TFLN.
Identification of Stress-Free Orientations: The study identified specific crystallographic orientations where gradient stress ( $\sigma_1$ ) crosses zero, resulting in flat suspended beams.
Thickness-Dependent Shift Discovery: The authors discovered that the "stress-free" orientations are not static; they shift significantly as film thickness decreases.
Design Strategy: Established in-plane orientation and film thickness selection as practical design levers for creating mechanically stable, scalable TFLN MEMS.

4. Key Results

Stress-Free Orientations (220–460 nm films):
- For films in the 220 nm to 460 nm range, the gradient stress ( $\sigma_1$ ) crosses zero near 55° and 125°.
- Beams oriented at these angles exhibit exceptionally flat profiles post-release.
- Demonstration: A 2 cm long, 10 µm wide, 460 nm thick beam oriented at 132.5° (near the 125° optimum) was successfully suspended without collapse, representing an unprecedented survivability length for TFLN.
Ultra-Thin Film Behavior (100 nm films):
- In 100 nm films, the stress-free orientations shifted to approximately 20° and 160°.
- These ultra-thin films exhibited a larger upward curvature (positive $\sigma_1$ ) compared to thicker films, indicating a stronger thickness-dependent stress asymmetry (likely due to anisotropic thermal expansion causing differential relaxation).
Dominance of Gradient Stress:
- The study found that the beam shape is primarily dictated by the gradient stress ( $\sigma_1$ ) rather than the mean stress ( $\sigma_0$ ), as the zero-crossings of $\sigma_1$ closely matched the orientations of flat beams.

5. Significance and Impact

Scalability: By identifying specific "safe" orientations, this work enables the fabrication of long, suspended TFLN structures (up to centimeters) that were previously prone to collapse, facilitating the scaling of RF resonators, filters, and optomechanical devices.
Yield Improvement: Understanding the shift in stress-free orientations with thickness allows for better process control and higher yield, especially as devices trend toward thinner films for higher frequency operation.
Design Paradigm: The paper shifts the view of residual stress from an unavoidable fabrication byproduct to a tunable design parameter. Engineers can now select specific crystal orientations and film thicknesses to actively manage stress and optimize device performance.
Future Outlook: The authors suggest that future work using X-ray diffraction (XRD) or Raman spectroscopy could further decouple and quantify the mean stress ( $\sigma_0$ ), leading to even more precise stress control across all thickness regimes.