In January 2015 (correction: the actual RTX 5090 launch was in 2025, but the article states 2015 – likely a typo; I will keep the original date as given), due to the significant increase in TDP (Thermal Design Power) of the Blackwell architecture (575W), NVIDIA adopted liquid metal as a thermal interface material (TIM) for the first time in its RTX 5090 FE graphics card, replacing the long‑used phase‑change material (PCM).

One year later, with computing power consumption continuing to escalate, the power of a single chip has exceeded 700W. Liquid metal – a thermal interface material with significantly higher thermal conductivity than traditional thermal grease – is once again drawing widespread attention.

A New Thermal Material for the High‑Computing‑Power Era

Among common thermal interface materials, thermal grease typically has a thermal conductivity of 3–5 W/m·K, while liquid metal can achieve 30–50 W/m·K – far superior to grease, pads, gels, and phase‑change materials. This gives liquid metal a natural material advantage in heat dissipation applications.

In addition, the viscosity of liquid metal is similar to that of water, but its density is six times higher. This combination of high density and low viscosity allows it to form an extremely thin interfacial filler layer, significantly reducing contact thermal resistance.

However, material properties do not always translate directly into engineering success.

TechPowerUp once evaluated the RTX 5090 FE mentioned earlier, and the results showed that the GPU core temperature with liquid metal was only 1.8 °C lower than that with conventional TIM – a much smaller difference than theoretical predictions. This has triggered a re‑evaluation of the feasibility of liquid metal in practical engineering applications.

Working principle of liquid metal heat dissipation

The main reason is that the surfaces of the chip and the heat sink are not perfectly flat at the microscopic level (there are micron‑scale valleys). Liquid metal has a relatively high surface tension. Compared with PCM, which melts and effectively fills those valleys, liquid metal does not fully wet the surface. The remaining air forms a high‑thermal‑resistance layer, resulting in high contact thermal resistance that makes it difficult to achieve a significant difference in total thermal resistance.

Therefore, when applied in real‑world engineering scenarios, liquid metal – as an extreme material – faces greater process challenges.

Precision Fluid Control Unlocks the Material's Potential

As a precision fluid application expert, QUARK Industrial Research Institute focuses on the application of liquid metal in SMT production lines and has accumulated extensive process experience. It provides solutions tailored to different scenarios by addressing real‑world application pain points.

| Overcoming Material Characteristics
Due to the high surface tension and strong fluidity of liquid metal, traditional contact dispensing tends to produce defects such as tailing and stringing. Residual material on the product surface can create short‑circuit risks.
QUARK’s piezoelectric jetting valve has the inherent advantage of non‑contact dispensing, adapting perfectly to the special rheological properties of liquid metal. It effectively avoids nozzle adhesion, and its high‑frequency (1000 Hz) jetting ensures clean break‑off, suppressing tailing, satellite dots, and other defects.

| Precision Volume Control
Liquid metal is very sensitive to dispensing volume and accuracy. Excess overflow can contaminate circuits, while insufficient fill or coating affects thermal resistance and thus reduces heat dissipation performance.
The QUARK piezoelectric jetting valve offers high‑precision dispensing capability, supporting a minimum dot diameter of 200 μm, a minimum line width of 250 μm, and a weight deviation of ±5%. It meets high‑precision process requirements for dots, lines, and areas – all within a reliable dispensing system.

| Dispensing Consistency and Process Capability
For high‑end manufacturing scenarios with higher product cost, processes must accommodate high‑speed production to avoid oxide layer formation while ensuring consistent glue‑bead shape under continuous operation – essential for high yield rates.
Leveraging its high‑quality dispensing process capabilities, QUARK Industrial Research Institute effectively suppresses wavy lines and adhesive breaks while achieving CPK > 1.67 – meeting high‑end manufacturing requirements and supporting high‑speed operation at 200–300 mm/s. This is made possible by advanced jetting valve performance integrated into the overall dispensing system.

| Multi‑Scenario Process Adaptation
From GPU top‑side coating, HBM gap filling, to VRM module thermal conduction on the motherboard, different packaging structures have very different requirements for dispensing path, volume, and speed.
QUARK Industrial Research Institute possesses extensive process experience and supports a vast process knowledge base. It responds quickly to real‑world application scenarios – from single dots and lines to complex path planning – providing agile adaptation to diverse needs.

Parameters verified by QUARK LAB

According to TrendForce’s 2025 Global GPU Market Report, with the mass production of GB200/GB300 rack servers, the penetration rate of liquid cooling is expected to exceed 33% in 2025, driving data center PUE below 1.1.

In the second half of the computing power race, heat dissipation efficiency is as important as transistor density.

As a precision fluid application expert, QUARK Industrial Research Institute has deep roots in the high‑end SMT&PCB manufacturing sector. Actively responding to industry trends, it leverages high‑precision technology and extensive process experience – including advanced jetting valve and dispensing systems – to help transform material performance into reliable product performance.