As it turns out, nothing is evaporating into thin air. According to a two-year electron microscope investigation by hardware enthusiast, professional overclocker, and Thermal Grizzly CEO Roman "der8auer" Hartung—published on his YouTube channel under the title How Liquid Metal Really Reacts! Analysis & a 2-Year Test and detailed by Tom's Hardware—liquid metal TIM does not dry out at all. Instead, its primary ingredient chemically reacts with your cooler's baseplate to form a solid intermetallic compound.
For Windows power users, custom loop builders, and laptop owners pushing thermal envelopes to the ragged edge, understanding what happens at that microscopic border between silicon and cold plate is the difference between a pristine cooling setup and an hours-long scraping session.
The Two-Year Electron Microscope Experiment
To figure out what actually happens over time, Hartung tracked a liquid metal application across a 24-month window using an electron microscope. Because liquid metal alloys—such as Thermal Grizzly's Conductonaut—rely on a blend of gallium, indium, and often tin, they remain liquid well below room temperature. While pure gallium melts at roughly 29.8°C (about 85.6°F), alloying it with indium and tin in a eutectic mixture drops the melting point to around 11°C. Moreover, peer-reviewed materials research published in Advanced Functional Materials notes that gallium-based liquid metals have effectively zero vapor pressure at room temperature, meaning they simply cannot evaporate the way solvent-based silicone thermal pastes do.
In Hartung's long-term observation, the timeline of physical change was remarkably front-loaded:
- At the 1-hour mark: The TIM remained visibly liquid under examination.
- At the 2-month mark: All visible liquid had disappeared, replaced by a field of solid crystals across the contact surface.
- At the 2-year mark: The crystalline structure looked identical to the two-month inspection. As Hartung noted, after two full years, the material honestly did not change any further.
It is worth noting an important experimental caveat here: between electron microscope sessions, Hartung stored the test apparatus in a vacuum or a nitrogen-filled container to prevent ambient oxidation. While that controlled environment differs from a daily-driver Windows gaming PC cycling through heat loads in open air, it isolated the purely metallurgical reaction between the TIM and the metal contact surface.
Why Interface Bottlenecks Matter More Than Ever
Why bother with volatile liquid metals in the first place? Because modern silicon is becoming a miniature furnace. As transistor density climbs, the heat generated per square millimeter—power density—has skyrocketed. Hartung pointed to Nvidia's RTX 5080 and RTX 5090 GPUs as prime examples, noting that they push roughly two to three times the power density of a classic GeForce 8800 GTX. When you are trying to evacuate hundreds of watts through a postage-stamp-sized area, the microscopic air gaps and thermal resistance at the die-to-cooler interface become the primary cooling bottleneck.
Section Summary: Liquid metal TIM has virtually zero vapor pressure and cannot evaporate. Hartung's two-year electron microscope test revealed that the liquid completely transforms into solid crystals within two months by reacting chemically with the cooler's metal, remaining unchanged through the two-year mark.
A Tale of Three Metals: Aluminum, Copper, and Nickel
What happens after you apply a gallium-based TIM depends entirely on the metallurgy of your heatsink or water block. If you treat all shiny metals the same, you are setting your hardware up for either cosmetic staining or catastrophic structural failure.
| Cooler Baseplate Material | Chemical Reaction with Gallium | Reaction Layer Thickness / Speed | Impact on Hardware & Cooling |
|---|---|---|---|
| Aluminum | Penetrates grain boundaries once oxide layer breaks | Swells in ~5 mins; brittle within 30 mins (2 mm sheet demo) | Catastrophic: Destroys structural integrity (liquid metal embrittlement) |
| Bare Copper | Bonds to form solid $\text{CuGa}_2$ (leaves indium/tin behind) | ~1 µm/day up to 50–100 µm (or ~4 µm in 96 hrs at 70°C) | Moderate: Permanent solid layer (~100 W/m·K vs. copper's ~400 W/m·K) |
| Nickel-Plated Copper | Forms solid $\text{NiGa}_5$ barrier layer | ~1 to 5 µm (5 to 10 µm on presentation slide) | Minimal / Best Practice: ~10x thinner layer protects underlying copper |
1. Aluminum: The Five-Minute Disaster
Every tube of liquid metal comes with a bold warning never to use it on aluminum heatsinks, and Hartung's demonstration illustrated exactly why. Once the thin natural oxide skin on aluminum is scratched or breached, liquid gallium aggressively migrates along the aluminum's internal grain boundaries.
In a demonstration using a scrap 2 mm aluminum sheet, visible surface swelling appeared after just five minutes, and the metal became completely brittle within half an hour. This phenomenon—known in materials science as liquid metal embrittlement and documented in Physical Review Letters—causes polycrystalline aluminum to suffer intergranular fracture under minuscule mechanical stress. Put simply: put gallium on an aluminum cooler, and the baseplate will literally crumble.
2. Bare Copper: The CuGa2 Alloy Trap
Copper is the gold standard for PC water blocks and heatpipes, boasting a stellar thermal conductivity of roughly 400 W/m·K. However, bare copper and gallium are chemically eager partners. In Hartung's testing on the bare copper cold plate of a Thermal Grizzly stainless steel water block, liquid metal could no longer be wiped away with acetone or isopropyl alcohol after less than five minutes of contact.
Instead of sitting passively on the surface, one copper atom bonds with two gallium atoms to form an intermetallic compound called $\text{CuGa}_2$. As the gallium gets locked into this solid crystal lattice, the leftover indium and tin from the TIM alloy segregate into their own solid pockets, making the entire application look like it has dried up.
What does that mean for your thermals and maintenance?
- Thermal Conductivity Drop: The $\text{CuGa}_2$ reaction layer has a thermal conductivity of about 100 W/m·K—roughly one-quarter of bare copper's 400 W/m·K rating. Because the layer is microscopic, it does not ruin the entire cooler's performance, but it does alter the immediate contact interface.
- Growth Rate: Depending on operating temperatures and how much liquid gallium is pooled on the surface, the $\text{CuGa}_2$ layer grows at roughly 1 micrometer per day until it plateaus between 50 and 100 micrometers thick. In a controlled test pressing pure gallium between copper plates at 70°C, the reaction layer reached about 4 micrometers over 96 hours.
- Permanent Solidification: Once $\text{CuGa}_2$ forms, it remains solid all the way up to approximately 254°C. Your CPU or GPU will hit thermal shutdown long before approaching a third of that temperature, meaning the compound will never melt back into a liquid inside a PC.
3. Nickel Plating: The Thin Protective Shield
If bare copper absorbs gallium like a metallurgical sponge, how do you stop the reaction from eating up your TIM? The answer is nickel plating.
When gallium meets a nickel-plated cold plate, it still reacts, forming a solid intermetallic compound that Hartung identifies as $\text{NiGa}_5$ (one nickel atom to five gallium atoms). While independent high-temperature metallurgical studies published in Materials note that gallium-nickel interfaces can form varying phases such as $\text{Ga}_7\text{Ni}_3$ alongside gallium-rich $\text{Ga}_5\text{Ni}$ layers depending on temperature, the practical takeaway for PC cooling is how thin the barrier stays.
According to Hartung, nickel plating restricts the solid reaction layer to roughly one-tenth the thickness seen on bare copper—measuring just 1 to 5 micrometers in his commentary, or 5 to 10 micrometers on his accompanying presentation slide. Even though $\text{NiGa}_5$ conducts heat less efficiently than $\text{CuGa}_2$, the layer is so vanishingly thin that its thermal penalty is negligible. Better yet, because the nickel plating stops the copper underneath from drinking up the gallium, your liquid metal pool stays intact rather than turning into a dry-looking crust requiring a re-application.
Section Summary: Aluminum is destroyed by gallium within minutes via grain-boundary embrittlement. Bare copper absorbs gallium to form a solid $\text{CuGa}_2$ layer up to 100 micrometers thick that conducts heat at a quarter of copper's rate. Nickel-plated copper remains the superior choice because it limits the reaction layer to roughly one-tenth the thickness, preventing gallium depletion.
Cleaning Up the Crust and Market Realities
Because the dull gray patch left on a copper cold plate is a solid intermetallic alloy bonded right into the surface—and because gallium oxide skins stubbornly cling to metals—wiping it down with a paper towel and 99% isopropyl alcohol does next to nothing. Previously, enthusiasts resorted to lapping their heatsinks with fine-grit sandpaper to grind away the $\text{CuGa}_2$ stain.
Naturally, as the head of a cooling company, Hartung used part of the video to highlight his firm's own solution: Thermal Grizzly's TG Metal Polish & Restore, priced at $22.11 direct for a 30-gram tube, a polishing pen, and lint-free microfiber cloths. According to Thermal Grizzly's product documentation, the paste combines controlled abrasives with a chemical surfactant that lowers gallium's surface tension and binds freed gallium particles so they cannot re-adhere to the copper, brass, chrome, or nickel while you scrub. While Hartung showed it wiping away a fresh film on bare copper quickly, long-term buildup still requires serious elbow grease: Tom's Hardware highlighted a PC Games Hardware forum member who used the paste on an Aqua Computer water block that had been running liquid metal on a delidded Intel Skylake-X processor since 2018, and scrubbing that multi-year crust off still took about an hour.
The Soaring Cost of Gallium and Better Alternatives
Even if you do not mind the cleanup, geopolitics and supply chains are making liquid metal an increasingly pricey hobby. Hartung revealed that Thermal Grizzly's raw material cost for gallium has surged nearly tenfold over the past five years—jumping from €300 per kilogram to between €2,500 and €3,000 per kilogram, with prices more than doubling since early 2025 alone. Because China controls roughly 99% of global gallium production and its suspension of U.S. gallium export bans is only slated to last through November 27, 2026 (with strict export licensing still enforced), those prices are unlikely to cool down anytime soon.
Looking ahead, Thermal Grizzly hopes to introduce new cold-plate coatings next year to replace standard nickel plating, alongside an updated KryoSheet graphene thermal pad tailored for bare silicon dies. While Hartung estimates the graphene pad runs about 3°C to 4°C warmer than liquid metal, it avoids electrical conductivity hazards, intermetallic alloying, and gallium price spikes altogether.
Practical Guidance for PC Builders
So, should you put liquid metal on your next processor or graphics card? Unless you are delidding a flagship CPU for manual overclocking or servicing a laptop factory-equipped with liquid metal, the smartest move for most builders is to leave gallium in the syringe.
If you do decide to enter what Hartung calls the "enthusiast bubble," keep these ground rules in mind:
- Verify your cold plate material: Never let gallium near aluminum. Whenever possible, choose a nickel-plated copper cooler or water block over bare copper to keep the intermetallic reaction layer ~10 times thinner.
- Expect initial gallium absorption on bare copper: If you apply liquid metal to bare copper, understand that the copper will consume a portion of the gallium to build a 50–100 µm $\text{CuGa}_2$ layer. If temperatures creep up after the first month or two, the TIM did not evaporate—the copper absorbed the gallium, and a light top-up over the newly formed intermetallic layer will usually stabilize permanently.
- Consider graphene pads for bare-die peace of mind: If a 3°C to 4°C difference does not make or break your overclock, a solid graphene sheet gives you maintenance-free longevity without ever having to scrub intermetallic crystals off your hardware.
Have you ever cracked open a water block or delidded CPU after a couple of years with liquid metal? Drop by our cooling and overclocking threads here on WindowsForum.com and let us know whether you stuck with gallium or made the jump to phase-change and graphene pads!
References
- Experiment shows liquid metal TIM doesn’t dry out; it turns solid Tom's Hardware · 2026-10-16T11:00:00+00:00
- Why isn't there a liquid metal TIM that stays liquid? | Overclock.net overclock.net
- Formation and Growth of Intermetallic Compounds during Reactions between Liquid Gallium and Solid Nickel - PMC pmc.ncbi.nlm.nih.gov