ARCTIC MX-7 Thermal Compound Review – Will the price breaker also become an icebreaker in the top 10?

1 - Introduction and technical data 2 - Shear behavior, microscopy, and composition 3 - Usability, laboratory measurements, and bulk values 4 - Performance on CPUs, notebook chips and GPUs 5 - Durability (pump-out), summary, and conclusion

Tear-off behavior and shear structure in comparison MX-7 vs. MX-6 New Formula

Looking at the tear-off images of the MX-7, the first thing that stands out is the clearly pronounced island formation. The overview with a 1,000 µm scale shows a dense field of individual, differently sized paste islands in the lower part of the image, with relatively narrow, largely exposed metal areas in between. The shear paths run predominantly parallel, but are broken up several times locally, so that the paste does not remain as a continuous film, but breaks up into packets and drops. At the higher magnification of 250 µm, the surface of these islands appears grainy and slightly “glittery”, which indicates a visually rather coarser, clearly visible filler structure. The transitions between the paste packets and the free substrate surfaces appear irregular, sometimes with thread and droplet structures, as known from a rather viscoplastic, highly shear-loaded matrix. From a purely visual point of view, this indicates that the cohesive forces within the paste are high enough to hold larger clusters together, but that the bond to the substrate is locally exceeded and then breaks off in individual areas. The fragmentation of the layer into many islands indicates quite pronounced shear behavior and a matrix that does not “slide off” evenly at high elongation, but breaks up in cracks.

In contrast, the tear-off images of the MX-6 New Formula show a much more homogeneous picture. On the 1,000 µm image, a largely closed film can be seen on the left-hand side, which only tears into individual areas towards the right-hand edge of the image. The shear front appears more clearly defined and less “frayed”. The paste appears to shear off as a continuous sheet over long distances before it breaks up. In the detailed view at 250 µm, the surface of the MX-6 is more finely textured, the filler particles appear significantly smaller and more densely packed, and the layer appears smoother overall. The edges of the tear-off zone are less irregular, more like a stepped edge than a splattered tear-off. Visually, the impression is of a material that is displaced in a more ductile-plastic manner, forming longer shear paths without strong fragmentation.

If you compare the two pastes directly, the MX-7 shows a much stronger micro-segmentation and a more pronounced droplet formation in the tear-off, while the MX-6 tends to have longer, closed shear plates. Based purely on image interpretation, this suggests that the MX-7 fails in a more localized manner under load and, to a certain extent, withdraws into highly cross-linked clusters. During disassembly, this could lead to more individual paste islands remaining on both contact surfaces and less detachment as a continuous film. In contrast, the MX-6 acts like a paste that tends to break off as larger areas when the joint is opened and shifts as a whole before it breaks. I cannot verify whether this is primarily due to the matrix chemistry, the degree of filling or the particle morphology, but the observed structures are consistent with a finer formulated, more slippery MX-6 and a somewhat more viscous MX-7.

In terms of shear structure, it can be deduced from the images that the MX-7 tends to allow higher local shear stresses before fragmenting. The many small islands and threads indicate that the paste is mechanically “pulled” when the contact surface is opened, breaking up into individual strands and droplets instead of flowing evenly. This can be positive in operation when it comes to limiting “migration” of the paste under cyclical loads, as the matrix becomes more entangled in the microstructures of the surfaces and acts less as a sliding film. The MX-6, on the other hand, acts as if it absorbs shear more by shifting and thinning out the surface, which leads to a more even surface, but could also favor a certain tendency to pump-out or edge migration in the long term. Again, I cannot directly verify the causes, but the differences in the tear-off pattern are clearly recognizable.

To summarize, the MX-7 shows a coarser structured, more segmented shear zone with clearly visible filler clusters and thread formation in the tear-off image, while the MX-6 New Formula forms a more closed, homogeneous film with a finer surface texture and smoother shear paths. From a purely visual point of view, this is consistent with a paste that appears somewhat tougher and more cross-linked in MX-7, while MX-6 appears more finely dispersed, slippery and flowable.

Particle morphology and dispersion in comparison between MX-7 and MX-6

The relatively broad particle size distribution is already noticeable in the first MX-7 image. The measured values range between just under 5 µm and over 10 µm, with many particles grouped in the range between 7 and 10 µm. The surface appears significantly more heterogeneous than in the MX-6 and shows pronounced micro-elevations. The particles are clearly delineated, sometimes angular, sometimes roundish, and they lie in a matrix that holds the grains but does not embed them completely. This creates a relief that appears visibly rougher.

In the second MX-7 image, taken after laser ablation, the particles emerge as shiny, irregularly reflective structures. The larger particles of 15 to over 20 µm form clearly recognizable clusters, while the smaller particles only partially fill the spaces between them. This leads to a texture with a stronger topography, which could be reminiscent of a bimodal filler mixture in which larger grains provide structural stability and smaller particles provide surface filling. Although the dispersion is uniform enough to avoid local agglomeration, it is still visible that the matrix is not completely leveled. There are microscopic valleys between individual elevations, which indicates that the MX-7 relies more on a viscoplastic, slightly pasty matrix in which particle movement is limited under shear. The particles appear to lie relatively freely on the surface, which matches the segmentation observed in the tear-off image.

The following two images of the MX-6 New Formula show a significantly finer and more homogeneous particle structure in direct comparison. The majority of the particles measured are between 6 and 10 µm, with a few slightly larger. The distribution is much narrower, the surface appears smoother and shows less pronounced macrostructures. The particles appear more embedded and optically almost merge with the matrix, which suggests a finer formulation or a higher proportion of small particles.

In the last image, the uniform grain size is particularly striking. The surface appears almost “sanded”, without dominant large clusters. The dispersion is fine and uniform, the matrix seems to flow around the fillers much better and thus creates a practically flat microstructure. It is noticeable that there are hardly any height differences in the MX-6, resulting in a closed film, which was confirmed by smooth shear paths in the tear-off images. The finer and more evenly dispersed structure indicates lower internal stresses during curing or under shear stress and could therefore explain why the MX-6 rarely breaks into large islands during tear-off and tends to shift over a larger area.

The MX-7 uses visibly larger and more widely distributed particles, which also protrude more strongly from the matrix. This design indicates a formulation that is mechanically more robust and may have lower pump-out tendencies. The picture is characterized by topographical height differences, which occur with differently rigid filler packages. The matrix of the MX-7 thus differs significantly from the finer, smoother structure of the MX-6. The MX-6 shows a considerably more homogeneous particle distribution with narrower size dispersion and better embedding. As a result, the paste appears more flowable and more slip-resistant, which matches the observed smooth shear structure. However, the homogeneous microstructure can also lead to greater lateral migration of the matrix when cyclic temperature loads occur. Overall, the MX-7 appears coarser-grained and more segmented. The surface shows a texture that indicates a higher internal cohesion of the matrix. The MX-6, on the other hand, appears finer, more uniform and smoother, which is reflected in a more closed tear-off pattern.

Chemical composition (LIBS analysis)

I analyzed the sample with my LIBS system, which is based on a short-pulsed Nd:YAG laser and is used in my laboratory for the elemental analysis of thermally conductive pastes. The surface of the sample is ablated with short, high-energy laser pulses, creating a plasma that emits characteristic spectral lines. These are recorded by an echelle spectrometer with high resolution and then spectrally analyzed. The method allows a spatially resolved, low-destruction analysis without chemical preparation of the sample and provides a reliable quantitative distribution of the elements. The method is ideal for determining the inorganic fillers and the composition of the binder matrix, whereby organic components are only recorded indirectly via silicon, carbon and hydrogen content.

The LIBS measurement of the MX-7 shows a clearly structured and consistent picture, which is typical for a strongly oxide-filled thermal compound. The dominant elements are aluminum, oxygen and zinc, supplemented by a clearly measurable silicon content and small amounts of hydrogen. Carbon is not detectable, although this does not mean that the paste is carbon-free. Rather, the metal oxide fillers overlay the organic matrix in the plasma to such an extent that their signature is barely visible. The measurement therefore primarily reveals how high the inorganic content actually is.

Aluminum is clearly at the top with around 44 percent and, together with oxygen, forms a characteristic aluminum oxide profile. The lines are clearly separated and without extraneous peaks, which indicates a homogeneous filler. Zinc occurs at around 15 percent and forms a typical zinc oxide signature together with oxygen. This combination of aluminum oxide and zinc oxide is desired in many TIM formulations, as it reduces thermal resistance and creates a defined mechanical stability. Aluminum oxide provides structure and heat conduction, while zinc oxide provides better processability, lower hardness and more uniform spreading properties.

The silicon content of around 10 percent acts as an additional but clearly recognizable component in the LIBS spectrum. This silicon most likely comes from the polymer matrix, i.e. a silicone-based binder that coats the oxides and keeps the paste supple. The fact that carbon is not visible in the spectrum can be explained by the fact that the filler content is extremely high and the laser ablation essentially hits oxide particles. In such cases, the organic phase quickly falls below the detection limit, even though it is actually present. I can therefore not confirm that the matrix is carbon-free, but the measurement result is absolutely consistent with a silicone-containing organic base that is optically and analytically dominated by the fillers.

The visible data thus fits exactly with the previously observed microstructure of MX-7. The paste shows a relatively coarse, topographically strong particle distribution under the microscope, which corresponds well with a high proportion of solid oxides. The larger aluminum oxide grains provide mechanical stability and the distinctive island structures in the tear-off pattern. The softer zinc oxide particles and the silicone matrix, on the other hand, are used for plastic deformability, but remain restrained in the LIBS signal.

The hydrogen peak is not critical and can originate from the matrix itself. It has no influence on the functional interpretation of the paste. Carbon remains invisible, which is normal for highly filled pastes and has no analytical significance regarding the absence of organic substances. All in all, the MX-7 shows a filler system based on aluminum oxide and zinc oxide and held together by a silicone-containing matrix. The LIBS measurement thus confirms the mechanical and optical observations of the paste: a robust, highly filled and thermally stable formulation that relies on proven oxide systems and does not reveal any exotic or conductive additives.