Think of thermal paste as grout between two tiles: the CPU lid and the cooler base. Both surfaces look flat but have tiny gaps full of air, which blocks heat. Paste fills those gaps. The thinner and more even the grout line, the better heat crosses. Conductivity is how good the grout material is, but the thickness of the final layer and how long it stays in place matter just as much. Easy, safe application beats a headline number you cannot reproduce.
The physical type: paste, liquid metal, thermal pad, or phase-change pad.
When it matters: Paste suits almost everyone. Pads are for gaps on graphics card memory and SSDs, or for fuss-free CPU mounting. Liquid metal is for experienced users only.
The main filler: ceramic, metal oxide, carbon, silver-based, or liquid metal (a gallium alloy).
When it matters: Carbon and ceramic pastes are safe around electronics. Silver and liquid metal types can conduct electricity. Trap: liquid metal attacks aluminum coolers.
What the maker designed it for, such as CPUs, graphics cards, laptops, or extreme overclocking.
When it matters: A quick filter. GPU and laptop use needs pump-out resistance; desktop CPUs just need easy, reliable paste.
How well the material itself carries heat, in W/mK (watts per meter-kelvin). Higher is better on paper.
When it matters: Reality check: brands measure differently, so numbers rarely compare across brands. Real temperatures depend more on layer thickness and mounting pressure. Differences between good pastes are small.
Where the conductivity figure comes from: maker claim or independent test, and which lab method was used.
When it matters: Figures from different test methods cannot be compared. An unnamed method is a warning sign of an inflated number.
Skip it if: You judge paste by real temperature results.
How much the applied layer resists heat flow, in C-cm2/W. Lower is better.
When it matters: Closer to real-world results than conductivity, because it includes layer thickness. Few brands publish it, so it is rarely useful for comparison.
Skip it if: The brands you compare do not both list it.
How thin the paste layer can get when squeezed between CPU and cooler, in mm.
When it matters: A thinner layer passes heat better. Pastes with large filler particles cannot squeeze as thin, whatever their conductivity rating.
Skip it if: You use a well-known mainstream paste.
Whether the material conducts electricity.
When it matters: Conductive pastes and liquid metal can short parts around the CPU if they spill or spread. Non-conductive paste is the safe choice for beginners and laptops.
How strongly the material resists electric current, in ohm-cm. Higher means safer if it touches circuits.
When it matters: A high value confirms a paste is electrically safe. It matters most around exposed parts next to the CPU or GPU die.
Skip it if: The paste is already listed as non-conductive.
How much the material stores electric charge, measured at 1 MHz. Lower is better near sensitive circuits.
When it matters: Mostly an engineering spec for pads near high-speed circuits. It has no effect on cooling.
Skip it if: You are applying paste to a desktop CPU.
Whether the material is safe on aluminum surfaces.
When it matters: Trap: gallium-based liquid metal eats into aluminum and ruins the cooler. Check the cooler base material before using liquid metal.
What happens if the material lands on the board or components: harmless, needs cleanup, or risks a short circuit.
When it matters: Useful for first-time builders and for laptops and graphics cards, where parts sit close to the die.
Skip it if: You use non-conductive paste.
How thick the paste is, in Pa-s (pascal-seconds). Thick pastes hold shape; thin ones spread easily.
When it matters: Thin paste spreads itself under pressure. Very thick paste needs warming or careful spreading. Too thick can leave a gap and higher temperatures.
Skip it if: You apply a pea-sized dot and let the cooler spread it.
How easy the material is to apply correctly.
When it matters: Easy pastes give near-best results with a simple dot. Hard ones, including liquid metal, punish mistakes with poor contact or spills.
Whether a spreader, stencil or cleaning wipe comes in the box.
When it matters: Helpful for thick paste and liquid metal. Most pastes do not need a spreader.
Skip it if: You use the dot or line method.
Whether the paste stays the same after application or firms up over time and heat cycles.
When it matters: Non-curing paste performs fully from day one. Some pastes improve slightly after curing. Trap: curing pads and adhesives are hard to remove.
Skip it if: You use a non-curing paste.
How many hours of use the maker says the paste needs to reach full performance.
When it matters: Explains why temperatures can drop slightly over the first days. Test cooling after burn-in, not right after building.
Skip it if: The paste lists zero burn-in time.
How long the maker says the applied paste keeps performing, in years.
When it matters: Paste can dry out or thin over years. Replacing it every four or five years is sensible, and any time you remove the cooler.
Skip it if: You repaste whenever you clean the cooler.
How well the paste stays put as the chip heats and cools. Pump-out slowly squeezes paste from the center.
When it matters: Matters most on graphics cards and laptops with bare dies and big temperature swings. Poor resistance shows up as creeping temperatures months later.
Skip it if: You use it on a desktop CPU with a metal lid.
The lowest and highest temperatures the material is rated to work at, in degrees Celsius.
When it matters: Any reputable paste covers normal PC temperatures. The range matters for extreme cooling with dry ice or liquid nitrogen.
Skip it if: You run a normal PC at room temperature.
The temperature at which a phase-change pad softens and flows like paste, in degrees Celsius.
When it matters: Below it the pad stays solid and cools poorly. Run the PC under load once after mounting so it melts in.
Skip it if: You are using ordinary paste, not a phase-change pad.
How much material is in the tube (grams or ml), and roughly how much surface area it covers.
When it matters: Trap: tubes with the same weight hold different volumes because densities differ. One CPU application uses little, so a small tube covers several builds.
Skip it if: You only need one or two applications.
How thick a thermal pad is, in mm.
When it matters: Trap: the wrong thickness causes poor contact or stops the cooler seating properly. Replace pads with the same thickness as the originals.
Skip it if: You are buying paste, not pads.
The dimensions of the pad sheet, in mm.
When it matters: Make sure one sheet covers every chip you need to pad, such as all memory chips on a graphics card.
Skip it if: You are buying paste.
How soft the pad is, on the Shore 00 scale. Lower numbers are softer.
When it matters: Softer pads squeeze into uneven gaps and put less pressure on parts. Harder pads can hold the cooler off the main chip.
Skip it if: You are buying paste.