3.1. XRD of graphite and carbon materials
X-ray diffraction (XRD) analysis was performed to investigate the phase composition and structural characteristics of the various graphite and carbon powders. The results are presented and discussed below, categorized based on the observed XRD patterns.
Group 1: Technical Carbon and Graphite Powders (K-354, P-701, T-900). The XRD patterns for technical carbon and graphite powders, namely trademarks K-354, P-701, and T-900, exhibited similar diffraction features (Fig. 1a). The observed peaks corresponded to graphite (COD 96-110-0004), with a hexagonal space group of R-3m (166) and lattice parameters of a = 2.456 Å and c = 10.041 Å. The degree of crystallinity for these samples was approximately 69.5%.
Group 2: Graphite from Ozersk, Trademarks GE-1, GK-1, and Colloidal Carbon S-1.
This group of samples exhibited distinct XRD patterns. The graphite from Ozersk was identified as a single-phase material, corresponding to graphite with a hexagonal space group of P63/mmc (194) and lattice parameters of a = 2.464 Å and c = 6.711 Å (COD 96-901-1578). The GK-1 sample displayed a mixture of the two graphite phases mentioned above, R-3m and P63/mmc. Samples GE-1 and S-1 shared a common feature: an oxidized surface, evidenced by the presence of carbon oxides in their XRD patterns (Fig. 1b). The degree of crystallinity for these samples was approximately 100%.
Group 3: Carbon Black and Biocarbon. The biocarbon produced from orange peel exhibited an almost amorphous structure with weak graphite peaks (Fig. 1c). The XRD pattern of carbon black was difficult to interpret definitively, with peaks potentially attributable to monoclinic silicon oxide (coesite) or calcium carbonate (calcite) within an amorphous carbon matrix. However, the absence of characteristic Raman shifts at 525 cm⁻¹ and 1085 cm⁻¹, corresponding to these impurities, suggests that these peaks likely arise from carbon. Moreover, similar but sharper peaks were observed in the biocarbon S XRD pattern (Fig. 1c). Consequently, the XRD pattern was attributed to carbon (ICDD 00-46-0943). The biocarbon S sample was identified as a two-phase material: the first phase, as described above, was carbon as in carbon black, and the second was an orthorhombic carbon with a space group of Cmc21 (36) and lattice parameters of a = 2.460 Å, b = 4.260 Å, and c = 28.960 Å. All samples within this group displayed low crystallinity.
The Scherrer equation was employed to calculate the crystallite size of the graphite and carbon samples (Table 3). The results indicated that partially amorphous carbons exhibited very small crystallite sizes, ranging from 1 to 1.74 nm. In contrast, graphite materials exhibited significantly larger crystallite sizes, ranging from 18.13 to 45.33 nm. Considering the potential applications, the roughness of coupling surfaces can vary from 12.5 to 0.4 µm. Therefore, the primary particles of all materials under investigation could effectively fill these gaps. However, it is important to note that primary particles rarely exist as isolated and well-distributed entities in powders. They commonly form aggregates or agglomerates from nanoparticles, and their size and morphology will be examined in detail in subsequent sections.
Table 3
Graphite and carbon materials crystallite size
Powder | 2 Theta ± 0.03, deg. | Crystallite size, nm |
Carbon black | 29.35 | 1.74 |
Technical carbon | 25.24 | 1.52 |
Т-900 | 27.97 | 1.60 |
Ozersk | 26.44 | 34.00 |
K-354 | 24.19 | 1.15 |
P-701 | 24.49 | 1.35 |
GK-1 | 26.53 | 34.01 |
GE-1 | 24.88 | 5.42 |
S-1 | 26.44 | 18.13 |
Biocarbon O (Carbon 2H) | 26.32 | 45.33 |
Biocarbon O (Carbon O) | 24.16 | 1.0 |
Biocarbon S (Carbon O) | 24.97 | 1.48 |
Biocarbon S (Carbon C) | 28.57 | 8.28 |
From the perspective of XRD phase analysis, samples with a high degree of crystallinity are considered more promising due to their higher thermal conductivity compared to partially amorphous materials.
3.2. Raman spectra of graphite and carbon
Raman spectroscopy was employed to investigate the structural characteristics and graphitization degree of the graphite and carbon powders. The Raman spectra were obtained by irradiating the samples with a 532 nm laser, and the spectral range analyzed was from 1000 to 3400 cm− 1 (Fig. 2a).
The Raman spectra of all carbon samples exhibited the characteristic G and D modes of graphite, as reported in previous studies [21–25]. The G mode, located around 1580 cm⁻¹, originates from the in-plane bond-stretching motion of pairs of sp² hybridized carbon atoms. The D mode, typically located around 1350 cm⁻¹, represents a breathing mode of A1g symmetry involving phonons near the K zone boundary, indicative of the presence of sp³ hybridized carbon. The D mode is negligible for ideal graphite and becomes more pronounced with increasing disorder [26].
The ratio of the intensities of the G and D modes (IG/ID) is commonly used to characterize the degree of graphitization in carbon materials. A higher IG/ID value signifies a higher degree of graphitization. As observed in the presented Raman data (Fig. 2), the D peak intensity for well-crystallized graphite was significantly lower than that of the G peak. Figure 3 depicts the ratio of band intensities and their full width at half maximum (FWHM). Notably, the GK-1 graphite sample exhibited a peak area ratio[27] of 4.87, while other samples ranged from 0.3 to 1.94.
Raman spectroscopy reveals a correlation between peak shifts and molecular bond length. A decrease in bond length results in a shift of the Raman peak towards higher wavenumbers (characteristic of graphite containing materials), while an increase in bond length corresponds to a shift towards lower wavenumbers (typical of carbon-based materials).
The presence of a 2D band in the Raman spectrum indicates the presence of graphene layers [27]. The second-order Raman spectrum exhibited weak bands at 2452 cm⁻¹ and 3244 cm⁻¹, corresponding to the D + D” and 2D’ harmonics, respectively. The D” band is attributed to an in-plane longitudinal acoustic (LA) branch near the K point, while the D’ band corresponds to a phonon of the in-plane longitudinal optical (LO) branch near the zone center (Γ point) [28].
In the Raman spectra of almost all samples (Fig. 2b), peaks were detected at approximately 250 cm⁻¹ and 360 cm⁻¹, likely associated with a high density of states of disorder-activated acoustic phonons (DAAP) [29, 30]. Upon Lorentzian decomposition, a gentle peak around 2000 cm⁻¹ was observed in the biocarbon O sample, potentially corresponding to the vibrational symmetric mode of CH₂ [31] or the presence of adsorbed CO [32].
The spectra of biocarbon, carbon black, and technical carbons exhibited a broad D-band (Fig. 3) in the range from 1350 to 1379 cm⁻¹. In contrast, samples with high crystallinity displayed a narrower D and G band FWHM (6.7–40 cm⁻¹). The smallest D-band shift was observed for carbon K-354 and P-702 (1350–1354 cm⁻¹), while the largest shift occurred for biocarbon (1379 cm⁻¹). Additionally, a prominent peak in the range of 2800–3000 cm⁻¹ was observed for the biocarbon O sample, a common mode attributed to ν(CH₂) as a surface impurity after synthesis.
Based on their Raman spectra (Fig. 4), the carbon samples could be broadly classified into two groups:
-
Graphite (G peak position ~ 1580 cm⁻¹, ID/IG ratio ~ 0.5)
-
Nanocrystalline Graphite (G peak shifted to ~ 1600 cm⁻¹, ID/IG ratio increased to 2.0) [26].
3.3. Morphology of graphite and carbon
Scanning electron microscopy (SEM) was employed to investigate the morphology and particle size distribution of the different graphite and carbon powders. The SEM images and resulting observations are discussed below, categorized by sample type:
Biocarbon and Colloidal Carbon: samples of biocarbon and colloidal carbons (Fig. 5a-c) exhibited a well-developed surface morphology characterized by large agglomerates surrounded by smaller, irregularly shaped particles.
Well-Crystallized Graphite: powders of well-crystallized graphite (GE-1, GK-1, graphite from Ozersk) were observed as large agglomerates of irregular shapes (Fig. 5d-f).
Carbon Black and Technical Carbons: samples of carbon black and technical carbons were represented by small, spherical graphite particles (Fig. 5g-k). From a morphological perspective, technical carbons appear most promising for achieving densely packed structures.
Particle size measurements were obtained from SEM images and calculated from surface area data using Eq. 1, assuming spherical particle morphology. These results are presented in Table 4. Due to the complex morphology of biocarbon O particles, size measurements for this sample were inconclusive.
Several samples, including carbon black, T-900, and P-701, exhibited similar particle sizes according to both SEM analysis and calculations from surface area data (Eq. 1), indicating a well-distributed particle distribution as observed in the SEM images (Fig. 5). In contrast, other samples displayed an aggregate structure. While particle size could be calculated from surface area data, the SEM images revealed aggregates, and the measured size corresponded to the aggregate rather than the individual particles.
$$\:\begin{array}{c}a=\:\sqrt{\frac{{S}_{a}}{4\pi\:}}\#\left(1\right)\end{array}$$
Where a – calculated spherical particle size, Sa – surface area.
Combining particle size data with the presence of agglomerates and aggregates suggests that well-distributed particles are the most promising fillers for thermal interface materials.
Table 4
Graphite and carbon powder surface area and particle size
Powder | Surface area, m2/g | Calc. particle size, µm | Particle size (SEM), µm |
Carbon black | 10 | 0.89 | 0.57 |
Technical carbon | 10 | 0.89 | 24.8 |
Т-900 | 15 | 1.09 | 0.66 |
Ozersk | 5 | 0.63 | 23.9 |
K-354 | 112 | 2.99 | 12.0 |
P-701 | 24 | 1.38 | 0.59 |
GK-1 | 63 | 2.24 | 19.9 |
GE-1 | 28 | 1.49 | 118.4 |
S-1 | 110 | 2.96 | 10.5 |
3.4 Thermal conductivity and operational bench test
The thermal conductivity of the prepared thermal greases was investigated, and the results are discussed below, focusing on the correlation between material characteristics and thermal performance.
Influence of Crystallinity and Particle Size: the highest thermal conductivity values were obtained for thermal greases containing trademarks GE-1, GC-1, and graphite from Ozersk. These materials share the common characteristics of high crystallinity (Fig. 1) and large particle size or agglomerates (Table 4). The high crystallinity preserves the inherently high thermal conductivity of crystalline graphite, while the larger particles/agglomerates reduce the number of thermally resistive interfaces between the graphite and the liquid polydimethylsiloxane (PDMS) matrix along the height of the test sample. This combination contributes to the observed high thermal conductivity values (Fig. 6).
Impact of Amorphization and Particle Size: despite having comparable particle sizes to GK-1 graphite, carbon black and biocarbon S, due to their high amorphization, their thermal conductivity values did not exceed 0.65 W·m⁻¹·K⁻¹. Technical carbons, with their smaller particle sizes and lower degree of crystallinity, also exhibited lower thermal conductivity values.
Correlation with Mass Fraction and Surface Area: similar mass fractions of graphite were observed for technical carbons (T-900, P-701) and carbon black (Table 2). This similarity can be attributed to their comparable particle size (Fig. 5, Table 4) and specific surface area (Table 4). K-354 carbon, which possesses a higher specific surface area (Table 4), demonstrated a lower mass fraction. This observation is explained by the need for a larger amount of polymer to effectively bind the particles with a more developed surface in order to maintain the desired viscous consistency of the thermal grease.
While commercially available thermal greases such as Arctic MX-4 (declared thermal conductivity of 8.5 W·m⁻¹·K⁻¹) and RGeek RG-5 (declared thermal conductivity of 15.7 W·m⁻¹·K⁻¹) claim high thermal conductivity values, the experimentally obtained results (Fig. 6) indicate a significant overestimation. This discrepancy is likely attributed to the measurement techniques employed by the manufacturers. Specifically, the thermal conductivity is typically determined by examining a thin layer of the composite material. This methodology allows for a wide range of possible thermal conductivity values to correspond to the same temperature gradient at the sample ends, considering the measurement error. This observation aligns with previously reported calculated data [3], suggesting that the declared values might be overestimated due to the limitations of the measurement technique.
This study investigated the potential of graphite and carbon-based thermal greases for enhancing heat dissipation in microelectronics devices. The performance of the developed materials was compared to commercially available thermal pastes, utilizing both benchtop thermal conductivity measurements and operational bench testing on a CPU.
A wide range of thermal greases with high thermal conductivity are commercially available for high-performance devices. Comparative evaluation of the developed materials against these existing solutions provides insights into their potential applications.
Operational bench testing, which simulates real-world operating conditions, is a critical approach for assessing thermal interface materials (TIMs). This study confirmed the significant influence of TIM layer thickness on heat dissipation from microelectronic devices [3]. Thinner TIM layers require lower thermal conductivity to achieve optimal device temperature control.
An operational bench test was conducted on a CPU, comparing the developed thermal greases with the commercially available KPT-8 thermal paste under 100% CPU load. The resulting temperature profiles of the processor cores (Fig. 7) revealed varying performance across the tested samples.
Sample P-701: This sample demonstrated the poorest performance, attributed to a lower mass fraction of graphite (43.4 wt%) compared to T-900 (49.6 wt%). This discrepancy likely arises from a greater tendency of P-701 particles to agglomerate during mechanical mixing, leading to non-uniform particle distribution within the polymer matrix. While this effect is minimized in bulk thermal conductivity measurements due to sample thickness, it becomes significant when applying a thin layer to the CPU.
Samples GK-1 and S-1: These well-crystallized samples reduced the maximum CPU temperature from 54°C to 49°C. This improvement is attributed to their small, well-crystallized graphite particles, enabling the application of a thin TIM layer with higher thermal conductivity (1.85 and 1.21 W·m⁻¹·K⁻¹, respectively) compared to KPT-8.
Samples GE-1 and T-900: Both samples achieved the lowest CPU temperature (47°C), demonstrating the effectiveness of contrasting approaches. GE-1's large graphite grains resulted in high thermal conductivity (2.19 W·m⁻¹·K⁻¹), surpassing KPT-8. Conversely, T-900's fine carbon particles allowed for a thin TIM layer, minimizing thermal resistance at the CPU-radiator interface despite lower thermal conductivity (0.6 W·m⁻¹·K⁻¹).
The thermal conductivity values achieved with the developed graphite-based thermal greases and their promising performance in the CPU operational bench test highlight their potential for developing new, highly efficient composite materials for microelectronics applications.