On October 15, 2026, Roman Hartung, known as der8auer, published microscope observations of liquid metal applied to copper and tracked for about two years. The surface was covered in crystals within the first few months, but it looked largely the same at two months and at two years. The content explains part of the phenomenon known as liquid metal "drying out" as a chemical reaction with copper. However, a surface that has settled does not mean cooling performance has been maintained. To judge whether liquid metal holds up over long-term use, you need to look at what it reacts with and how heat crosses the interface.
Crystallization seen over two years
In the long-term test section of the video, Hartung showed a small copper sample coated with Thermal Grizzly's liquid metal "Conductonaut Extreme." He used a scanning electron microscope to observe the surface and follow how it changed over time. He explained that between observations, the sample was stored in a vacuum or in a box filled with nitrogen to avoid the effects of oxidation.
About an hour after application, the surface showed smooth, liquid-looking areas. After two weeks, crystals had appeared, and after one month, fine crystals had multiplied further. In the two-month image, the liquid-looking areas had almost disappeared from the observed surface, replaced by spreading crystals. Yet even after sitting for up to two years, the appearance reportedly did not change much from the two-month state.
The changes were concentrated early, after which a similar surface remained. This difference in timing is what makes the observation interesting. The idea that liquid metal is lost endlessly the longer it is used cannot fully explain the later period.
That said, the conditions differ from running a CPU with a cooler installed for two years. What was presented is an image of a copper sample's surface, not a performance curve of CPU temperatures or thermal resistance measured over two years. Hartung is also the head of Thermal Grizzly, the seller of the material in question. It is better read as an observation for understanding how things look after long-term use than as a guarantee of product lifespan.
The chemical reaction on the copper surface
Liquid metal is a thermal interface material placed between a CPU or GPU and its cooler, filling the microscopic gaps at the contact surface. Even when surfaces look flat, places where they do not fully touch hinder heat transfer.
Gallium reacts with copper to form an intermetallic compound, CuGa₂. It combines one copper atom with two gallium atoms, and its properties differ from both the original copper and the liquid alloy as applied. Because components of the liquid metal are taken up into a compound with copper, the surface gradually turns into solid crystals.
According to Thermal Grizzly's Conductonaut product description, the liquid metal contains indium, tin, and other metals in addition to gallium. Copper, the contact partner, joins the reaction in a liquid made by combining several metals. In other words, if you look at the residue after long-term use and conclude that the original alloy simply evaporated or cooled and solidified, you misread what happened.
The chemical reaction also changes what cleaning means. The usual intuition that wiping the surface restores its original state does not carry over. The manufacturer also explains that traces derived from the metal alloy may remain, and warns that if the liquid metal spreads onto the heat spreader, markings may become unreadable, which could affect the warranty. The change in appearance is also the result of the material interacting with the contact surface.
Why nickel plating is recommended
Thermal Grizzly recommends nickel-plated copper contact surfaces as a combination that emphasizes long-term stability. On the other hand, it explicitly says not to use the product with aluminum coolers. When handling liquid metal, you need to check the surface the liquid metal actually touches, not just the cooler's general material.
| Surface touched by liquid metal | Reaction and usage position |
|---|---|
| Aluminum | A material the Conductonaut manufacturer prohibits from contact |
| Unplated copper | Reacts with gallium to form CuGa₂ |
| Nickel-plated copper | The plating also reacts, but it suppresses reaction with the underlying copper, and the manufacturer recommends it for long-term use |
Nickel is not a protective film that reacts with nothing. According to Hartung's explanation, contact with gallium forms a different intermetallic compound, NiGa₅. However, the reaction layer is thinner than in the case of copper, and it suppresses gallium from being taken up into the underlying copper. Rather than eliminating the reaction entirely, its role is to reduce the changes occurring at the contact surface.
Also, having aluminum heat sink fins is not the same as having liquid metal touch aluminum. What needs to be checked is the base facing the CPU or GPU. And the recommendation of nickel-plated copper does not mean it is fine for liquid metal to flow out onto circuitry. The product itself is conductive, so material compatibility and leak prevention must be considered separately.
Does the reaction layer worsen cooling?
The fact that an intermetallic compound has formed does not by itself determine how cooling performance changes. Heat passing through the inside of a material and heat crossing the boundary between different materials are governed by different factors. In the former, the material's thermal conductivity and the layer's thickness matter; in the latter, the state of contact and the way the surfaces are bonded also matter.
In a 2025 experiment at Tsinghua University, forming CuGa₂ on a copper surface by hydrochloric acid treatment reduced the interfacial thermal resistance between copper and liquid gallium to about one-sixteenth of its untreated value.
This is the result of a study that Xudong Zhang and colleagues published in the peer-reviewed journal Advanced Materials Interfaces. The measured values in the results section were 1.11×10⁻⁷ m²K/W before treatment and 6.94×10⁻⁹ m²K/W after. Dividing the pre-treatment value by the post-treatment value gives about 16. Using time-domain thermoreflectance, a method that examines heat flow from changes in reflectance of a laser-heated sample, they measured the resistance to heat crossing the boundary.
The research team explained that the change from an interface with an oxide sandwiched in between to one in which the metals are bonded through CuGa₂ improved heat transfer. However, the experiment used pure gallium and copper, with the interface created by hydrochloric acid treatment. It was not a test of commercial liquid metal used in a PC over a long period, and the figure does not mean CPU cooling performance improved sixteenfold.
What can be read from this research is that the reaction can improve contact in some cases. On the other hand, how much a grown compound layer impedes heat also depends on its thickness. Even when the compound has the same name, the conditions for evaluation differ between research that deliberately creates a thin interface and the phenomenon of a reaction layer growing during use. From the appearance of a solidified surface alone, one cannot judge cooling to be improving or worsening in either direction.
Judging long-term use requires temperature records
This two-year observation cannot yield a rule such as "replace it after two months," nor can it support "performance does not change for two years." The former treats the moment crystals form as a failure; the latter substitutes stability of appearance for stability of performance. Both go beyond what the sample images show.
To judge long-term cooling, you need to track how temperature and thermal resistance change under the same load and cooling conditions. Surface observation offers clues for thinking about what is happening, but it is no substitute for temperature records. Decisions such as removing the reaction layer because of residue seen at disassembly, or reapplying after a fixed number of months, also cannot be justified by this observation alone.
When choosing liquid metal, compare the temperature drop right after application with the effort of keeping that state. The prerequisites are that the contact surface material is compatible, leaks are controlled, and the condition can be inspected when needed. To confirm its value as a long-lasting coolant, temperature changes through repeated heating and cooling while installed, in addition to photos of crystals, are what should inform the judgment.