EFFECTS OF GRAPHITE SELECTION ON THERMALLY CONDUCTIVE COMPOUNDS FOR LED LAMP HEAT SINKS Daniele Bonacchi IMERYS Graphite & Carbon, Bodio Switzerland Abstract Thermally conductive compounds are viewed as potential replacements of metal based heat sinks in automotive and non-automotive LED lamp applications. Graphite is certainly the main candidate for thermally conductive applications that tolerate electrical conductivity for their high efficiency and reduced costs. In this article we demonstrate that the introduction of graphite increases substantially the thermal conductivity especially along the plastic flow (in plane) direction. We have tested several commercially available graphite grades in polyolefin model polymers and have seen that the crystallinity, the average particle size and the aspect ratio are the three main factors that promote thermal conductivity. In this comparative study we have also tested special high aspect ratio graphite that delivers high thermal conductivity at low loadings giving an advantage in terms of weight reduction. Introduction Metal replacement by polymer compounds is increasing the demand of thermally conductive plastic materials for applications that requires heat dissipation or heat transfer [1]. The use of polymers offers substantial advantages compared to standard metal-based technologies in terms of weight reduction, corrosion resistance, design flexibility and manufacturing cost reduction. Since polymers are inherently thermally insulating (<0.5 W/m.K), conductive fillers have to be added in order to provide the requested thermal conductivity. Metallic powders are known to be difficult to handle and are potentially explosive. Ceramic materials, on the other hand, have low efficiency, could be abrasive and can be extremely expensive although they have the advantage of being electrically insulating. Carbon-based fillers are promising thermally conductive additives. There are many allotropes of carbon: graphite, carbon nanotubes (CNT), carbon black and graphene are some of them, and added to a polymer they also give electrical conductivity above a certain concentration (percolation threshold) to the material, meaning that at high loadings they cannot be used in applications that require electrical insulation [2]. LED lamps need to dissipate thermal energy in order to reduce the operating temperature that is known to reduce the lamp lifetime. It has been shown that the limiting factor of the cooling power of an aluminum based heat sink is the air convection. In other words the thermal conductivity of aluminum (200 W/mK) exceeds the requirement for medium size LED lamps [3]. Some studies [3-4] indicate that a thermal conductivity above 10 W/mK is an acceptable value to dissipate the thermal heat generated by most of the indoor LED lamps. For these reasons carbon blacks and carbon nanotubes cannot be used for these applications as they are unable to deliver such high thermal conductivity to plastic compounds [5]. Graphite on the other hand is known to increase considerably the thermal conductivity of plastic compounds, especially at high loading [6]. In this article we have screened various graphite types looking for the highest performing graphite grades for heat sink applications. It must be noted that, although polyolefin polymers are rarely seen as polymers for heat sink applications, it allowed us to draw general conclusion that can be used with engineered polymers that are used more often. Page 1 SPE ANTEC Anaheim 2017 / 285
Experimental section Materials High density polypropylene (HDPE) Hostalen GF4750 (MFI=0.4 g/10min at 190 C/2.16Kg) and polypropylene Moplen HP400R (MFI=25g/10min at 230 C/2.16Kg) from LyndellBasell have been used in this study. Primary synthetic graphite (KS44, SFG44, KS150-600SP at >99.9% purity) and natural graphites (PP44, 80x150, 50x80, 20x50 at >96% purity) and special graphite (C-Therm001 at >99.7% purity) from IMERYS Graphite and Carbon have been used as thermally conductive additives. KS44, SFG44 and PP44 have a D 90 of 44 micron while KS150-600SP has a size distribution between 150 micron and 600 micron. The numbers of natural graphites 80x150, 50x80 and 20x50 indicate the mesh size. C-THERM001 has been produced starting from graphite with a D 90 of 81 micron. Sample preparation In the case of Polypropylene, Haake Polylab OS equipped with internal mixer unit Rheomix 600 has been used to mix the polymer and the thermal additive. The polymer has been melted in the internal mixer for two minutes at 200 C at 100 rpm. Graphite has then been added and mixed at the same speed for five minutes. The polymer compound has then been taken out from the chamber and compression molded at 200 C for five minutes and cooled down to room temperature with cooling plates. For high density polyethylene compounds, a twin screw extruder Haake Polylab OS equipped with PTW 16/40 (16mm screw diameter, 40L/D) has been used. The graphite has been introduced after the first polymer melting zone to increase dispersion and avoid agglomeration of the graphite. The granules with the desired graphite concentration (concentrations are always expresses as weigh percentage) were successively injection molded with an Engel ES330/80HL CC90 in the shape of 2mm thick sheets or dog bones (ISO:3167) for mechanical property testing (mold temperature was set at 40 C) or compression molded in a similar way to polypropylene compounds. Instruments Instron universal machine 5500R has been used for tensile tests (ISO:527) while Instron/CEAST 9050 impact tester has been used for Charpy impact test at room temperature (ISO179). Netzsch Laserflash LFA447 has been used to measure the thermal conductivity according to ASTM E1461. In the case of through plane direction the samples were cut in the form of disks while for the in plane thermal conductivity the procedure described in reference 7 was used. Results Graphite KS44 has been used as a reference graphite, a broad range of concentrations has been explored ranging from 0 to 60%w/w. As can be seen in figure 1, in injection molded samples, there is a large difference in the thermal conductivity in the through plane and in the in plane directions. The origin of this anisotropy can be explained with the planar crystalline structure of the graphite. Due to the specific arrangement of carbon atoms in graphene planes, the transmission rates of phonons 1, that are responsible of the transport of the thermal energy in graphite, are different in the in plane and the through plane directions [8]. This effect should be averaged when the graphite particles are randomly oriented, but due to their flake morphology, they always tend to orient along the plastic flow originating the strong anisotropy observed. 1 Phonon is a quantum of sound or vibratory elastic energy Page 2 SPE ANTEC Anaheim 2017 / 286
7 Thermal conductivity (W/mK) 6 5 4 3 2 1 0 0 10 20 30 40 50 60 70 Graphite loading (%w/w) Figure 1: In plane (blue rhombus) and through plane (orange squares) thermal conductivity of injection molded HDPE 2mm sheet loaded with KS44 graphite. 25 20 Polypropylene (density =0.9g/cm3) Polycarbonate (density =1.2g/cm3) Volume fraction (%v/v) 15 10 5 0 0 5 10 15 20 25 30 35 Weight fraction (%w/w) Figure 2: Volume fraction versus weight fraction for two compounds made with polymers of different density and graphite. Page 3 SPE ANTEC Anaheim 2017 / 287
Although it is still unclear the specific role of two components of the thermal conductivity on the cooling power, which is also very dependent on the heat sink design, it is widely accepted that the in plane component is the most important one [3-4]. Since LED heat sinks require high in plane thermal conductivity, we can narrow the range of investigation to above 40%. Indeed, the typical loading for LED heat sink applications ranges between 40 and 70% depending on the polymer density, as can be seen in figure 2, two generic polymers with different density with the same volume fraction of graphite have very different weight fractions. Thermal conductivity (W/mK) 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 Through plane In plane 0.5 0.0 KS44 40% SFG44 40% PP44 40% Figure 3: Through plane (blue) and in plane (orange) thermal conductivity of injection molded HDPE sheets loaded at 40%w/w with different graphite grades. In figure 3 the in plane and the through plane components of the thermal conductivity of HDPE compounds made with different graphite grades are reported. The three graphite grades are of similar size (D90 of 44 micron). KS44 and SFG44 are synthetic graphite while PP44 is a natural graphite. It s evident that the natural graphite is better performing than synthetic graphite although the compound made with SFG graphite reaches a thermal conductivity close to the compound made with PP44 graphite. The difference in performance of the compounds made with the three different graphite grades can be ascribed to a mix of different crystallinity and particle shape. KS44 graphite is known to be less flaky than SFG or PP grades, for this reason KS44 gives higher through plane thermal conductivity than in plane conductivity. In figure 4 we present the in plane thermal conductivity of polypropylene compounds made with natural graphite grades of different average size and similar aspect ratio in comparison with KS44 compound. KS44 compound at 70% is not shown as we were unable to compound it properly due to the very high torque. The result shows that the thermal conductivity increases by increasing average particle size. This can be ascribed to the fewer particle-polymer or particleparticle interfaces that a phonon must travel per unit length in the in plane direction. It must be remembered that although larger particles are better performing, particles size must be selected Page 4 SPE ANTEC Anaheim 2017 / 288
according to the design. It s easy to understand that large particles cannot be used when the plastic parts to be formed have thin layers as they would easily block the plastic flow inside the mold. Figure 4 also tells us that by increasing the graphite loading in the matrix, the thermal conductivity reaches and surpasses 10 W/mK that is required for most LED lamp applications. It must be noted that above 60% loading, the material becomes extremely brittle and difficult to process. Although natural graphite seems to be the best solution for most heat sink applications, there are other cases that require extremely high purity for regulatory issues (for example drinking water applications) or to prevent polymer degradation during processing and service. For those applications, synthetic graphite is known to be a preferred solution due to its high purity. In plane Tc (W/mK) 20 15 10 60% 70% 5 0 KS44 80x150 50x80 20x50 Figure 4: In plane thermal conductivity at 60%w/w (blue) and 70%w/w (orange) in compression molded PP sheets loaded with different graphite grades. Other applications in which synthetic graphite can be useful are all the applications that require high through plane thermal conductivity (for example cooling pipe applications). Special KS graphite grades are known to be less flaky than other graphites (see above), for this reason the through plane thermal conductivity is higher (see figure 3 and figure 6). Also in this case, by increasing the average particle size, the through plane thermal conductivity increases. Unfortunately, the mechanical properties of highly loaded graphite polymers (table one) are extremely poor. Upon the addition of graphite, the material becomes brittle increasing the stiffness but decreasing the elongation at break and the impact properties. To elucidate the effect of processing on thermal conductivity we decided to produce a polypropylene compound loaded with 80x150 at 60%w/w with the twin screw extruder and to process the same compound via injection molding machine or the compression molding machine. The results are presented in figure 5 and has can be seen, the in plane thermal conductivity increases while the through plane component is reduced probably as a consequence of the higher orientation of graphite particles in the injection molded material. Page 5 SPE ANTEC Anaheim 2017 / 289
Thermal conductivity (W/mK) 12 10 8 6 4 2 Compression molding Injection molding 0 Through plane In plane Figure 5: Through plane (left) and in plane (right) thermal conductivity of compression molded (gray) and injection molded (yellow) polypropylene loaded with 60%w/w of 80x150 natural graphite. Through-plane Tc (W/mK) 4.0 3.5 3.0 2.5 2.0 1.5 1.0 KS44 80x150 KS150-600SP 0.5 0.0 0 10 20 30 40 50 60 70 Graphite loading (%) Figure 6: Through plane thermal conductivity of polypropylene compression molded sheets containing 80x150 (orange squares), KS44 (blue rhombus) and KS150-600SP (pale blue circles). Page 6 SPE ANTEC Anaheim 2017 / 290
Table I: Thermal and mechanical properties of HDPE compound loaded with graphite In plane thermal conductivity Tensile strength Elongation at break Charpy notched Density W/mK MPa % kj/m 2 g/cm 3 SFG44@40% 3.6 34.3 5.9 7.6 1.19 C-Therm001@20% 3.7 32.1 7.4 5.1 1.03 Another type of graphite of interest is represented by high aspect ratio graphite, commercially known as C-THERM (in this case C-THERM001 has been used). It is known that the aspect ratio determines the average distance between the particles [9] thus influencing thermal conductivity. For this reason this graphite performs extremely well in delivering high thermal conductivity at low loading as evident from figure 7. As a rule of thumb, this high aspect ratio graphite, delivers the same thermal conductivity at half loading of the other graphites. Surprisingly the mechanical properties of this high aspect ratio graphite compound are similar to the ones of SFG44 compound at the same in plane thermal conductivity (see table 1) despite the large difference in loading. Beside the advantage in weight saving, the much lower loading of the high aspect ratio graphite can be an advantage when used in combination with other mineral reinforcing additives like glass or carbon fibers. We report also a much better surface state of the compound made with high aspect ratio graphite that can be helpful in applications that require better esthetics. Thermal conductivity (W/mK) 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 Through plane In plane 0.5 0.0 KS44@40% SFG44@40% PP44@40% C-Therm001@20% Figure 7: Through plane (blue) and in plane (orange) thermal conductivity of injection molded HDPE sheets with different graphite grades. Page 7 SPE ANTEC Anaheim 2017 / 291
Conclusions Thermal conductivity of graphite loaded compounds for LED lamp heat sink applications was considered in this article. The addition of graphite to the polymer matrix increases considerably the thermal conductivity of the compound. Due to the particle orientation, the thermal conductivity has different values in the different directions. In plane thermal conductivity reached the required values for LED lamp heat sink applications at high graphite loadings. By screening different graphite grades we could conclude that natural graphite is very performing and only a specific synthetic graphite grade reaches similar performance. We have seen also that larger average particle size of similar aspect ratio increases the thermal conductivity of the compound. We could demonstrate that graphite selection depends on the specific requirements and for example that KS synthetic graphite grades are the most performing grades when through plane thermal conductivity is needed. Finally we examined high aspect ratio graphite properties and we discovered that high aspect ratio graphite is able to increase compound thermal conductivity at roughly half loading of the natural and synthetic graphite with advantages in weight reduction and possibly to the addition of other mineral reinforcing additives. Bibliography 1. Thermally Conductive Plastics Global Trends and Forecast Till 2020, Markets&Markets December 2015 2. D. Bonacchi, The carbon approach to thermal conductivity Compounding World, February 2016, pag.57-60 3. SABIC presentation by Dr. F. Mercx at Conductive Plastics 2015, Düsseldorf 4. R.H.C. Janssen, IR. K. Douven, H.K. van Dijk, Thermally-conductive plastics: balancing material properties with applications needs, Compounding World February 2010, p.38-42 5. Z. Han, A. Fina, Progress in Polymer Science (2011) vol 26, 7, p.914-944 6. I. Krupa, I. Chodàk, European Polymer Journal 37 (2001) 2159-2168 7. C. Raman, High Performance Plastics 2011 Proceedings, (2011) Smithers Rapra Technology 8. A. A. Balandin, Nature Materials (2011) vol-10, pag.569-581. 9. M. O. Saar, M. Manga, Physical Review E (2002), vol. 65, 056131 Page 8 SPE ANTEC Anaheim 2017 / 292