CAYOM CA-4 vs. BOLIGO Z980 Thermal Pastes Review – More than just similar bargains?

1 - Introduction and overview 2 - Shear behavior, microscopy, and chemical composition 3 - Lab measurements, interface resistance and bulk thermal conductivity 4 - Real performance on CPUs, notebook chips, and GPUs 5 - Durability (pump-out), summary, and conclusion

Classification of the real measured values in the simulation

The simulations below show very clearly how close the field of pastes tested here is and how strongly the resulting temperature depends primarily on the actual layer thickness. As usual, the colors correspond to the previous classification, i.e. red for the CAYOM CA-4, green for the Boligo Z980 and blue for the Arctic MX-6 (new formula). The temperature curves increase almost proportionally with increasing BLT, which corresponds exactly to the previously measured thermal resistance behavior. The Arctic MX-6 achieves the lowest temperatures for all tested layer thicknesses, confirming the lower Rth gradient determined in the laboratory. The Z980 is consistently in between, while the CA-4 is slightly higher due to its slightly higher Rth. Although the differences remain in the range of a few tenths of a Kelvin, they are consistent and therefore reproducible.

The influence of the minimum compression thickness is particularly noticeable. As the CA-4 was able to achieve the lowest BLT in the laboratory, it compensates for part of its higher material resistance and thus comes close to the results of the Z980 and MX-6. The Z980 is only just behind at 15 µm. The MX-6 requires slightly more space between the surfaces and starts at around 16 µm. These differences may seem small at first glance, but they cause immediately measurable temperature deviations in the series measurement because the total thermal resistance is directly proportional to the coating thickness. Each reduction in BLT acts as a direct deduction from the total resistance and can therefore result in noticeable advantages even with a slightly less conductive paste.

However, it is important to note that a low BLT is only automatically advantageous on perfectly flat contact surfaces. As soon as the radiator base is slightly convex or the heatspreader relief is unevenly distributed, the paste is displaced more strongly locally or compressed differently. In such cases, a paste with the technically lowest BLT can even be disadvantageous if it no longer wets areas cleanly or runs out too thin. A slightly thicker, smoother-flowing paste can then improve the actual contact and thus lead to lower temperatures despite higher nominal Rth values.

With increasing BLT, the balance of forces clearly shifts in favor of material resistance. The thicker the layer, the more the slope of the Rth curve influences the result. From around 50 to 75 µm, the bulk resistance begins to play the main role and differences in the interface resistance increasingly fade into the background. From this point onwards, the better bulk conductivity of the Arctic MX-6 comes into its own and explains why it clearly pulls away from the other two products at higher BLT. All in all, the graphs show very clearly that real application scenarios are always an interplay of wetting behavior, minimum achievable layer thickness, mechanical boundary conditions of the cooler and the pure material performance of the paste.

CPUs with headspreader

CPUs without headspreader /Direct Die)

GPUs