🔬 materials science

Thermal response of an in-situ STEM MEMS chip under rapid pulse heating

This paper characterizes the thermal response of an uncoated Protochips Fusion MEMS chip under rapid pulse heating in an in-situ STEM setup, revealing a cooling time constant of 1.80 ms and an average cooling rate of approximately $7.9\times 10^{4}$ K/s to establish practical thermal limits for high-speed solidification experiments.

Original authors: Phillip Dumitraschkewitz, Thomas Kremmer

Published 2026-09-09

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Phillip Dumitraschkewitz, Thomas Kremmer

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 trying to understand how a metal hardens as it cools from a liquid. The speed at which it loses heat determines the tiny crystal structure that forms inside, which in turn dictates whether the final material is strong, brittle, or flexible. Scientists have long wanted to watch this process happen in real time, but the cooling often occurs so quickly that standard tools cannot keep up. To solve this, researchers use a special kind of microscope that can see individual atoms, combined with a tiny, heated chip that holds the sample. This chip acts like a miniature furnace, capable of heating a speck of metal to extreme temperatures and then letting it cool down in a fraction of a second. The challenge has always been knowing exactly how fast that cooling happens, because the instruments used to measure the temperature often react too slowly to capture the true speed of the event.

In a recent study, a team of researchers set out to measure just how fast this specific type of microscopic chip can cool down. They focused on a commercial chip designed for use inside a scanning transmission electron microscope, a powerful tool that allows scientists to peer deep into the structure of materials. The chip itself is a tiny membrane made of silicon carbide, a material that conducts electricity well and can withstand high heat. When electricity flows through this membrane, it heats up, much like the filament in an old-fashioned light bulb. To study rapid solidification, the researchers needed to turn this heat off and watch the temperature plummet. However, they found that simply looking at the electrical current flowing into the chip was not enough to tell the whole story. While the electricity could be cut off almost instantly, the physical temperature of the chip took a little longer to respond, creating a gap between what the power source did and what the material actually felt.

To get a precise reading, the team built a custom measurement system using a device that can generate very specific electrical signals and a high-speed camera for voltage. They programmed the chip to heat up to temperatures as high as 750 degrees Celsius and then drop the power to let it cool. By watching how the electrical resistance of the chip changed during this process, they could calculate the temperature with incredible precision. They discovered that while the electrical current dropped in less than one-thousandth of a second, the chip's temperature followed a slightly slower path. The temperature did not drop in a straight line; instead, it relaxed exponentially, meaning it cooled very fast at first and then gradually slowed down as it approached the lower temperature.

The researchers measured the time it took for the chip to cool down from its peak temperature to a lower, stable point. They found that the chip cooled with a time constant of about 1.8 milliseconds. To put that in perspective, a single blink of an eye takes roughly 300 milliseconds, meaning this chip cooled down nearly two hundred times faster than a human blink. Based on this rapid drop, they calculated an average cooling rate of approximately 79,000 degrees Celsius per second. This is an incredibly fast pace, comparable to the rates seen in specialized industrial processes designed to create unique metal alloys. The team also checked their work carefully, running the experiment multiple times and testing the equipment with known resistors to ensure their numbers were accurate. They found that while the chip's electrical properties shifted slightly over time due to repeated heating cycles, this drift did not change the fundamental speed of the cooling event.

The study confirms that this type of microscope chip is a powerful tool for studying how materials solidify at extreme speeds. By proving that the chip can cool at rates of nearly 80,000 degrees per second, the researchers have established a reliable benchmark for future experiments. This means scientists can now design experiments to create new materials with confidence, knowing exactly how fast the heat is being removed. The findings also highlight a crucial detail for anyone using this technology: the temperature of the material lags slightly behind the electrical signal used to control it. Understanding this delay is essential for interpreting the results correctly, ensuring that the tiny structures seen under the microscope are truly the result of the intended thermal history. With these measurements in hand, the door is open for more precise investigations into how metals and other materials behave when they are heated and cooled faster than ever before.

Technical Summary: Thermal Response of an In-Situ STEM MEMS Chip Under Rapid Pulse Heating

Problem Statement
Advancements in additive manufacturing have increased the demand for studying rapid solidification and the correlation between thermal history and microstructure at small scales. While Fast Scanning Calorimetry (FSC) offers high-accuracy temperature measurements via thermocouples, in-situ Scanning Transmission Electron Microscopy (STEM) experiments utilizing MEMS heating chips lack independent temperature sensors decoupled from the power source. Instead, these systems rely on measuring resistance or current to infer temperature. Furthermore, standard commercial measurement setups for in-situ STEM often have integration times (e.g., 100 ms) that are too slow to track rapid heating pulses, and open-loop software limits may not reflect the physical thermal response of the chip. Consequently, there is a need to quantify the actual cooling response and thermal path limitations of commercial in-situ STEM chips (specifically the Protochips Fusion platform) under rapid pulse heating conditions to ensure accurate interpretation of microstructural evolution.

Methodology
The authors developed a customized measurement setup to quantify the thermal response of an uncoated commercial Protochips Fusion MEMS chip. The system replaces standard commercial control loops with:

Arbitrary Waveform Generator (AWG): A Keithley 6221 used to drive user-defined temperature programs via current control.
Digital Oscilloscope: A Tektronix MSO 44 used to record voltage drops across the chip and a series shunt resistor with sub-millisecond resolution.
Data Reconstruction: The setup measures total voltage ( $U_1$ ) and shunt voltage ( $U_2$ ) to reconstruct chip resistance ( $R_{chip}$ ) and subsequently temperature ( $T$ ) using pre-calibrated $R(T)$ and $i(T)$ relationships derived from the chip's SiC-based heated membrane.

The experimental protocol involved running parameterized temperature programs (heating to a maximum temperature, holding, and rapid cooling to a nucleation temperature) on an empty chip. The study focused on the cooling segment to determine the relaxation time and cooling rates. Robustness checks included repeated $R(T)$ measurements to assess drift, testing with a 10 kΩ load, and characterizing AWG leakage/offsets.

Key Results

Current vs. Temperature Response: The current drop time was verified to be less than 1 ms. However, the temperature derived from resistance ( $T(R)$ ), which reflects the physical thermal state of the membrane, evolves more slowly than the electrical signal ( $T(i)$ ).
Relaxation Time: Analysis of the maximum cooling step revealed an exponential-like relaxation behavior. The average time constant ( $\tau$ ) was determined to be 1.80 ± 0.04 ms. This value is consistent with thermal lag constants reported for FSC chips (e.g., 2.2 ± 0.6 ms).
Cooling Rates: Based on the time required to reach 95% of the temperature difference ( $\Delta T$ ), the average cooling rate was measured at approximately 7.9 × 10⁴ K/s (specifically ~79,000 K/s) in the temperature range between ~700°C and 100°C.
Measurement Robustness:
- Drift: Repeated measurements showed a modest downward drift in the $R(T)$ curve approaching an asymptote, likely due to prolonged air storage or annealing effects. However, sensitivity analysis demonstrated that linear transformations of the temperature scale (simulating drift) did not alter the calculated relaxation time $\tau$ , as the fitting equation relies on temperature ratios.
- Beam Heating: Calculations indicated that electron beam heating under the experimental conditions was negligible (< 1 mK).
- System Offsets: The study identified small leakage currents and zero-point offsets in the AWG and oscilloscope, which were accounted for in the analysis to ensure current accuracy within the instrument's specified tolerance.

Significance and Claims
The paper establishes practical bounds on achievable thermal-path rates for in-situ electron microscopy experiments using the Protochips Fusion platform. The authors claim that while the electrical control can switch rapidly (< 1 ms), the physical thermal response of the chip is limited by in-plane conduction, resulting in a relaxation time of ~1.8 ms. This finding is critical for researchers planning rapid solidification experiments, as it defines the temporal resolution limits for correlating thermal history with microstructure in this specific chip environment. The study confirms that despite the lack of a dedicated thermocouple, the resistance-based method, when coupled with high-speed data acquisition, can effectively characterize the thermal lag of the system.

Future Outlook
The authors suggest that future work should utilize closed-loop mode for initial $R(T)$ measurements and validate these critical cooling rates using metallic glass-forming alloys within the identified cooling rate window. Additionally, they propose that a coupled electrothermal finite-element model would further enhance the understanding of the chip's heat-flow behavior.

More like this