Development of the production technology of a new highly effective thermal grease

A rapid increase in the power of microelectronic devices, along with a reduction in their size, leads to a rapid growth in the density of dissipated heat flows. As a result, thermal management becomes a crucial factor for maintaining the stable uninterrupted operation of microelectronic devices. Stricter requirements for thermal interface materials (TIMs) make it necessary to optimize their production technology. A solution-based technology for obtaining grease with enhanced thermophysical properties is proposed. It has been shown that heat treatment of a mechanical mixture of aluminum nitride (AlN) and graphite (C) (1:1 by weight) in a vacuum at temperatures of 250–350 °C makes it possible to clean the surface of the particles from moisture and organic impurities, which leads to an increase in the thermal conductivity of the materials obtained. It was revealed that the best solvent at the processing stage is AlN:C silane is ethanol due to the high chemical similarity with silane surfactant. In contrast, when introducing particles into polydimethylsiloxane (PDMS), the highest thermal conductivity results were achieved with acetone as a solvent. The use of ultrasonic treatment of the filler, when introduced into the polymer matrix, was considered. It was shown that the optimal duration is 10 to 15 min, depending on the surfactant. The resulting thermal pastes have sufficient thermal conductivity (up to 2.25 W/(m·K)) and high thermal stability up to the flash point of PDMS (340 °C).


Introduction
Electronic devices' thermal management plays a crucial role in safety and effectiveness. Therefore, the dramatically increased density of heat-generated flows leads to strict requirements for productive dissipation. Due to the surface roughness of the coupled devices and heat sinks, the major issue is to provide tight contact. Thermal interface materials (TIMs), such as thermal greases, can fill the gaps between the coupled interfaces and squeeze out the air. Thus, an effective heat transfer is obtained [4,9,31].
It is known that pure polymers used as a TIMs matrix have a low thermal conductivity, which needs to be remarkably improved [12,31]. The common way for enhancement is to add highly thermal conductive inorganic materials (fillers) to the silicone oil. Straight application of a chemical or morphological uniform filler does not allowing achieve a high thermal conductivity value [12,28]. As an example, the highest thermal conductivity values from single fillers Al, ZnO, AlN, graphite, and SiC were obtained 1.54, 1.15, 1.45, 0.95, and 0.47 W/(m·K) correspondingly [12]. The similar results were represented in the previous work [28], where the best results achieved for graphite (2.85 W/(m·K), SiC (1.37), Al (1.2), and AlN (1.09). Thus, for the following improvement of TIM's thermal conductivity, the plenty of parameters have to be varied such as interfacial interaction, synergy effect, and thermal conduction pathway tuning are of great interest to researchers nowadays [13].
The most common approach for the improvement of thermal conductivity is surface functionalization and processing due to the trend of agglomeration and poor affinity to silicone. The first way to ennoble thermal conductivity is a surfactant application [14,19]. Prof. Chung [22] achieved thermal conductivity enhancement for epoxy-AlN and epoxy-BN composites from 9.99 and 5.27 W/(m·K) to 10.98 and 10.31 W/(m·K) after raw powder treatment by silane. For liquid greases with polydimethylsiloxane (PDMS) binder, no such high values being reported even after silane treatment (2.01-2.32 W/(m·K) [12,23]. The second is a processing technique effect. The well-known fabrication technologies are solvent casting, melt extruding, compression, injection molding, and others [13].
Also carefully studied approach is a synergy effect [5,12,15]. There are a lot of powder particle parameters such as morphology, size, surface area, and porosity which in various combinations make such hybrid fillers substantial contribution to the TIM's thermal conductivity. It was found that the combination of particles with different size and morphology leads to the growth of thermal path quantity hence increase of thermal conductivity. Al-SiC and Al-AlN hybrid fillers allowing to achieve grease with thermal conductivity value 2.07 and 1.98 W/(m·K), consequently [12]. PDMS matrix composite with 12 vol.% BN nanoparticles + 70 vol.% spherical Al 2 O 3 has a 3.6 W/(m·K) thermal conductivity value [7]. As one of the most promising methods could be considered an application of 0, 1, or 2D nanofillers. Recently, plenty of works deal with graphene or graphene oxide 0-20 wt% addition to the hybrid filler with thermal conductivity values varying from 0.497 to 3.0 W/(m·K) [9,10,16,20,27,30].
The last but not the slightest approach is to diminish the percolation threshold to achieve high thermal conductivity. The recent challenge is to implement a high thermal conductivity at a low volume fraction of the filler. Such techniques as 3D structures, particle orientation, and others are applied to produce a strongly anisotropic structure with a high in-plane thermal conductivity (25 W/(m·K) and even higher) [3,6,21,24,26]. Therefore, such an approach could be used in cured polymer composite materials and greases.
A recent review of TIMs fabrication methods is presented in the work [1], but the solution-based methods are firstly focused on the last approach (orientation and techniques for its implementation). The second technique is also being observed and studied in several works [14,17,19]. Considering the above-mentioned a major issue is raised regarding a thorough investigation of the interfacial interaction effect. Consequently, the present work is consecrated to assessing diverse treatments (heat, surface, and ultrasonic) on the grease thermal conductivity.
Material selection for effective thermal grease production should meet two main requirements: high thermal conductivity and a low raw materials cost. Such wise graphene and carbon nanotube materials hardly could be considered as prospective fillers for industrial-scale application at the present. Graphite particles are suggested as a promising material for thermal conductive polymer composites [6,8]. Based on the data being reported and previous work [17,18,28], it is hybrid materials should be investigated. Different binary combinations of fillers Al, AlN, SiC, and graphite (C) were thoroughly considered by our group earlier [17], and it was found that the best results 2.84 and 3.02 W/(m·K) were achieved for AlN:graphite ratio 1:1 and 1:2 correspondingly. Although, the poor PC benchmark test results due to large graphite particles force to significantly reduce the graphite particle size and choose AlN:C 1:1 ratio because of the better fluidity of grease. Based on the AlN:C compound, the present investigation aims to develop a solution-based technology for a new highly effective thermal grease with a hybrid filler production.

Materials and composite material production
Aluminum nitride powder (99.9%, spherical particles with an average diameter of 2 µm), graphite powder (98%, flake particles with an average dimension of 20 µm), and polydimethyl silicone oil (PDMS, polymer number 1000) were purchased from a local supplier. AlN to graphite with the equal mass mechanical mixture was used as a starting material. The raw materials were selected from the results of the previous work [17].
The schematic flow chart of the composite material production is represented in Fig. 1. The technology was separated into 5 major stages (mixing, heat, surface, and ultrasonic treatments) and each was carefully studied and discussed in the further sections in the consequence represented on the right side of Fig. 1.
Mixing of materials and surface treatment was carried out in the solvent at the magnetic stirrer with a constant speed of 1000 rpm. Excess of the solvent was evaporated in the vacuum drying oven equipped PTFE membrane vacuum pump.
Heat treatment of the AlN:C mechanical mixture was carried out in the tube furnace equipped with Value VRD-16 vacuum pump at temperatures 25-900 °C for 2 h with a heating rate of 5 °C/min. After complete cooling in the vacuum, the resulting thermal treated mixture was immediately stirred with a PDMS and solvent. Excess solvent after mixing was distilled in the vacuum chamber at 100 °C for 2 h.
TechPan ultrasonic (US) disintegrator type UD-20 automatic was used for the treatment of AlN:C mechanical mixture in a solvent. The US treatment was carried out in a water bath.

Investigation methods
Thermal conductivity was measured at the custom-made apparatus IT-λ-400 described in detail earlier [17].
The powder mixture before and after the heat and surface treatment was cold pressed in the KBr matrix. The obtained specimen was studied by IR-spectroscopy at Bruker Vertex 80.
TGA for dynamic assessment of heat treatment was carried out at Setaram 92 with a low vacuum pump at 25-900 °C with steps at 250, 350, 500, 650, and 900 °C. Hold time at each step was set at 2 h.
The thermal stability of the produced grease was studied at DSK instrument-Sentsys EVO in the temperature range 25-900 °C. A heating rate in all experiments was set to 5 °C/min.

Results and discussion
As the first point of the flow chart ( Fig. 1) mixing technique of inorganic filler (AlN:C) and PDMS was examined. Three samples were obtained: by direct mechanical mixing in the mortar and by magnetic stirring with a solvent. The thermal conductivity of the obtained samples are collected in Table 1.
It was found that direct mechanical mixing allows for obtaining the highest values of thermal conductivity due to the higher fillers content. Also, it should be noticed that the application of graphite with a finer particle size considerably reduces the thermal conductivity value due to a higher quantity of highly thermal resistant borders of inorganic filler-PDMS. The same trend was observed for Ag particles [29]. However, operational bench test results show the inability to achieve effective heat dissipation with a large particle grease [17], which forces it to significantly reduce average particle size for cooling improvement. According to the results, acetone is a more effective solvent for PDMS, both have CH 3 − radical groups enhancing solubility. Besides worse results mixing in the solvent has a higher potential to influence the resulting structure hence mechanical mixing was eliminated. The acetone was chosen as a solvent for the final mixing of AlN and C powders with a PDMS.
The thermal resistivity of inorganic filler and polymer matrix is the most common limitation issue of composite material thermal conductivity. It could be highly related to the affinity of the coupling materials. A treatment for surfactant application could significantly affect both structure and thermal conductivity. The phenomena will be discussed in the following paragraphs.

Thermal treatment
One of the most crucial powder parameters is the chemical purity of a surface. As the moisture and different organic compounds absorb at the surface, those result in enormous thermal resistivity growth. Hence, as the second step, heat treatment of the aluminum nitride and graphite mechanical mixture in a vacuum was studied to remove any organic or moisture impurity traces from the surface. The results of  thermal grease thermal conductivity with thermally treated filler are presented in Table 2. Thermal treatment at low temperatures (250 and 350 °C) results in light mass lost within calcination and sufficient thermal conductivity growth due to absorbed water removal. PDMS and water have completely different chemical bonds: polar for the water and nonpolar for the polymer; hence, water is insoluble in polymer and v.v. Thermal resistance at the border absorbed water-PDMS is extremely high.
Further temperature increase (T > 500 °C) results in both aluminum nitride and graphite partial oxidation. The alumina and aluminum oxynitride impurities at the AlN grain surface remarkably decrease the heat transfer from the bulk particle to the border particle/polymer. As a result, the thermal conductivity of the obtained grease drops. As well oxidized particle surface increases contact thermal resistivity.
The dynamic mass change observation in the TGA apparatus was carried out to analyze the thermal treatment effect. The gathered data of AlN and graphite mechanical mixture TGA in a low vacuum is illustrated in Fig. 2.
At low temperatures, the mass loss of the sample is close to the previously obtained values. At 500 °C, the difference appears (0.67 vs 1.06%) explains by the worse vacuum pump (lower vacuum values) in the case of the TGA experiment. The further data approves the assumption; after 550 °C, a dramatic weight change was observed as a consequence of oxygen ingress into the chamber and the graphite combustion. However, the low-temperature 0-500 °C data demonstrate sequential steps of the material weight. Hence, the balance between complete absorbed impurities removal and prevention of material oxidation should be chosen.

Infrared microscopy
The surface composition transformation investigation was made by IR spectroscopy. The gathered data from the thermal treated at different temperatures samples is illustrated in Fig. 3.
The presence of graphite in the mixture could be observed by vibration energy values 2930 and 2840 cm −1 which are matching to C-H bond. Impurity C = O is found in all samples at 2360 and 2341 cm −1 . Peaks connected to aluminum nitride are found at 1329 and 746 cm −1 . It should be noted that peak 669 cm −1 connected to partial graphite oxidation appears to stay until heat treatment at 650 °C and disappears with graphite combustion that approves the previous experimental data. Hence, CO 2 impurity removal dramatically  increases the thermal resistivity of C-PDMS which results in a reduction of both the volume ratio and thermal conductivity of the greases. The obtained experimental data shows the requirable temperature of the aluminum nitride and graphite mechanical mixture in a vacuum is about 350 °C.

Surface treatment
The third stage of the suggested technology is a direct mixing with a surfactant. The application of surfactants such as TEOS or AMEO could considerably affect the properties of the greases due to decreasing the surface to PDMS thermal resistivity and agglomeration prevention. Because silanes lead to PDMS hardening, surface treatment and mixing with silicone oil were divided into different technological stages. TEOS and AMEO replace groups such as OH-at the particle surface (Fig. 4) by an O-Si-(OR) 3 chain, where R = C 2 H 5 − . The high affinity of the ethyl group to PDMS methyl ones reduces the thermal resistivity of the surface contact particle-silicon oil.
Surface treatment by TEOS and AMEO (Fig. 5) results in the appearance of new vibrations at 1241 and 1050 cm −1 that corresponds to SiO-CH 2 and Si-O-Si bonds. Heating at 200 °C in a vacuum of AlN:C sample treated by TEOS to evaporate surfactant residues shows no significant changes to the untreated sample. Such a result is connected with a remaining quantity of TEOS lower than the IR-spectroscopy detection limit.
Several technological parameters were varied in the series: surfactant (TEOS or AMEO), its quantity, and the solvent (acetone or ethanol). Heat treated at 350 °C mechanical mixture of aluminum nitride and graphite was blended with a surfactant in the solvent, and excess after treatment was evaporated at 100° in a vacuum. The gathered data of the obtained grease's thermal conductivity is presented in Table 3, where * is the sample without heat treatment of the initial powder mixture.
From the first two series, it was found that acetone is not the best choice as a solvent for silanes. Only the sample treated with 2.5% of AMEO does not reduce its thermal conductivity. In the case of PDMS, the same CH 3 − groups show better affinity, and acetone could be considered a better solvent, but TEOS and AMEO have C 2 H 5 − groups that are similar to the ethanol ones. As a result, better values of thermal conductivity were achieved for 2.5% of TEOS. The excess of surfactant in all cases leads to a thermal conductivity decrease which could be related to surface thermal resistivity raise as a result of partial pyrolysis of unreacted TEOS or AMEOS to amorph SiO 2 within the distillation of unreacted silane.

Ultra-sonic treatment
US disintegration of powder particles agglomerates is a well-known way to improve the distribution of the filler in the polymer matrix that allows for obtaining higher thermal conductivity values. In the fourth stage, US treatment by introducing the mixture into the PDMS was studied. The several series with AMEO and TEOS as a surfactant was investigated and the gathered data in comparison with untreated samples is presented in Table 4, where *-samples without heat treatment.
It could be found that pure US treatment of a raw mechanical mixture of aluminum nitride and graphite fails to achieve Dispersion of aluminum nitride and graphite mechanical mixture in PDMS with solvent is greatly heated during the US treatment. Water bath allows dissipation of the majority of the generated heat, nevertheless, local particle overheating results in enforced particle agglomeration. Surfactant greatly prevents this process until a temperature close to boiling point is achieved. The such wise higher boiling point for AMEO in comparison with TEOS 217 and 169 °C consequently results in a long time of US treatment without reverse agglomeration that considerably conforms with the experimental data.
Last but not least, the fifth stage of the offered technology was to study the effect of US treatment within surfactant addition to the mechanical mixture of AlN:C. It was suggested that deagglomeration of the inorganic particles within treatment allows for the distribution of TEOS more effectively over the entire surface. And therefore, preventing the following agglomeration and minimizing the thermal  resistivity at the border particle-polymer matrix. The additional sample (*) with the only US treatment within TEOS injection was produced to consider the necessity of two deagglomeration stages. The obtained data are presented in Table 5.
It should be noted that the same time dependence as for the previous stage is obtained. Hence, the best combination of such technological parameters as heat dissipation (water bath temperature), US power and time, and type and quantity of surfactant was discovered for the equipment discussed earlier. It should be noticed that the thermal conductivity growth with the application second US treatment is lower than the error rate of the investigation method. Hence, the fifth stage could be excluded from the principal technological scheme represented in Fig. 1. Moreover, US treatment within TEOS addition has an appreciably lower impact on the thermal grease properties than treatment with PDMS. Due to high viscosity, even particles with surfactant fails to distribute uniformly in the bulk matrix and the US application is required.

Comparison of the results
A number of commercially available thermal grease have a huge declared value of thermal conductivity > 8.5 W/(m·K). The several thermal greases were purchased such as Maxtor Thermal Paste 8301A (11.2 W/(m·K), LK-17 thermal grease (17.0), Cooler Master Mastergel Maker (11.0), RGeek RG-5 (15.7), Arctic MX-4 (8.5), and KPT-8 (0.8). Despite the lowest thermal conductivity, it is the KPT-8 which is applied in industrial scale manufacturing because of its available cost, while the others are high price products for high-end performance microelectronics. The gathered data (Fig. 6) indicates excessive overstatement of the product's thermal conductivity. This is probably due to thermal conductivity measurement technique: due to the low value of the temperature gradient at the ends of the sample, the error in determining the temperature introduces a huge inaccuracy in the value of the thermal conductivity of the samples. As it was shown in the work [17], the thinner the layer of the thermal interface between the heat-generating and heat-dissipating devices, the lower the value of thermal conductivity at which the lowest value of the CPU temperature is achieved. Thus, when measuring thin layers of thermal pastes, there is a significant increase in the values of thermal conductivity, the declaration of which is not correct. This means that the values given by the manufacturers are only a marketing move, and not a reflection of the real properties of the greases. The required thickness of the sample obtained for the device used in this work (IT-λ-400) (with a thermal conductivity of specimen 1-5 W/(m·K) is at least 4 mm, which corresponds to its passport data.
The presented data in the articles [11,12,27,29] are very close to the achieved values, meanwhile, no such expensive fillers as graphene, graphene oxide, or silver were added. It was shown that the similar effect (increase thermal pathways quantity along the sample) to the graphene addition could be obtained with a graphite flake that remarkably reduces the cost of producing composite material.

Thermal stability of grease
One of the most crucial thermal grease properties thermal stability is illustrated by DSC/TGA curves which are shown in Fig. 7. The decrease in mass by 0.8% during heating from 94 to 342 °C is associated with removing the absorbed moisture and gases during the storage and production of the samples. Within further temperature increase to 342 °C, an  [11,12,27,29] exothermic peak is observed, which can be attributed to the flashpoint of polydimethylsiloxane. Thus, composite materials with a PDMS binder are completely thermally stable in the operating temperature range (339 °C [2]). At 451 °C, a peak is observed in good agreement with the literature data and is related to the self-ignition of PDMS [2]. The endothermic peak at 620 °C can be attributed to the partial burnout of graphite accompanied by according to the IR spectroscopy results CO 2 removal (mass loss 0.3%). At 795 °C, partial oxidation of powdered aluminum nitride [25] and graphite combustion begins. The weight loss before autoignition (451 °C) was 5.20 and after 19.52%. The total weight loss was 24.72%, which pretty well corresponds to the mass fraction of the grease.

Conclusion
Chemical purity of the surface is tremendously important for polymer-based composite materials with high thermal conductivity production. It was found that heat treatment of aluminum nitride and graphite mechanical mixture in a vacuum at temperatures (250-350 °C) allowing to improve thermal grease properties. But the additional raise of the temperature over 550 °C leads to graphite burnout and oxidation of AlN which has a negative impact on grease properties. Additional treatment by TEOS in acetone media and US application allowing to achieve deagglomeration of the particles in a short period of time. Period of US treatment closely connected with a surfactant boiling point. The highest thermal conductivity value of the thermal greases being obtained is 2.25 W/(m·K) that together with the commercial availability of raw materials makes them prospective as a high-end TIMs. Moreover, such greases have a high thermal stability up to polymer flash temperature (about 340 °C), which is significantly higher than the working temperatures of TIMs in microelectronics.
Author contribution Conceptualization, methodology, validation, formal analysis, investigation, resources, data curation, writing-original draft preparation, writing-review and editing, visualization, supervision, and project administration R.A.S. All authors have read and agreed to the published version of the manuscript.

Funding
The author appreciates the support of this work by the Russian Science Foundation under grant № 21-79-00123.