Enhanced toughness and thermal conductivity for epoxy resin with a core–shell structured polyacrylic modifier and modified boron nitride

A new epoxy-based composite with higher toughness and thermal conductivity was developed. First, a poly(n-butyl acrylate)/poly(methyl methacrylate-co-glycidyl methacrylate) (PBMG) core–shell structured latex was prepared by seeded emulsion polymerization to toughen the epoxy resin (EP). Second, boron nitride particles were modified into nano-scale sheets and added to the epoxy/PBMG blend to improve the thermal conductivity of the resulting composite material. The properties of the constituent materials were determined prior to fabrication and testing of the composite product. The monomer conversion in the emulsion polymerization process of the PBMG was checked by determining the solid particle content. The PBMG particle size was characterized by dynamic laser scattering, and the morphology of the particles was characterized by scanning and transmission electron microscopy. The exfoliation of the modified boron nitride (MBN) flakes was verified by TEM and Raman microscopy. The mechanical properties and the thermal conductivity of the EP/PBMG/MBN composite were determined at various constituent contents. Results showed that the unnotched impact strength of the composite increased by 147%, the flexural strength increased by 49.1%, and the thermal conductivity increased by 98% compared with pristine EP at a PBMG content of 5 wt% and MBN content of 7 wt%. With the enhanced properties and ease of fabrication, the developed composite has good potential for application in high-end industries such as microelectronics packaging.


Introduction
Epoxy resin (EP) has been widely used due to its many desirable properties such as: good electrical insulation, heat resistance, ease of processing, and low cost. It has been used in composite laminate, adhesives, semiconductor packaging, aerospace, and other applications. [1][2][3][4][5] However, the cured resin has high crosslinking density and is very brittle with low resistance to the initiation and growth of cracks, which make the epoxy matrix undesirable in many applications. Researchers have been working to toughen epoxy and broaden its use in more demanding applications. To achieve this purpose, a variety of reinforcements, including liquid rubber tougheners, 6-8 coreshell rubber (CSR) particles, [9][10][11][12][13][14][15][16][17][18][19] thermoplastics, 20,21 and inorganic ller [22][23][24] have been used. However, sometimes these led to a decrease in glass transition temperature, modulus and tensile strength of the epoxy matrix. To overcome this conict, hybrid epoxy composites reinforced by both rigid particles and rubber particles have been developed by incorporating the two reinforcing agents, [25][26][27][28][29] which are shown to possess a synergetic toughening effect. Acrylic core-shell polymers (CSP) have been used in toughening various resins because of their good weather resistance, low shrinkage, and low cost of preparation. Moreover, the composition and size of the particles can be controlled to obtain desired properties. These particles consist of a so rubbery core inside a harder polymer shell. The particle size can be controlled as they are formed by emulsion polymerization before they are dispersed into the epoxy resin. [30][31][32][33][34] With the rapid development in electronic technology, demand for high performance and small size electronic devices has been on the rise. Such high demand causes these devices to generate and accumulate a great amount of heat that should be effectively dissipated to prolong the service life of the devices. 35 Commonly suggested llers to dissipate the heat include: graphite, 36 carbon nanotubes 37,38 and graphene, [39][40][41] however these llers are also electrically conductive and cannot be applied in composite materials that are used for electrical insulation. To effectively solve the thermal dissipation problem and obtain high performance, different llers have been introduced into polymers to provide thermally conductive but electrically insulative polymer-matrix composites. These llers include oxides (Al 2 O 3 , 42 SiO 2 , 43 and ZnO 44 ), diamond, 45 y ash 46 and nitride (AlN and BN) 47,48 which are thermally conductive but electrically insulative. Hexagonal boron nitride (h-BN) represents a ller with a thermal conductivity of up to 600 W (m À1 K À1 ) and high electrical insulation 49 that can be used as a material to dissipate heat in such applications. [50][51][52][53][54][55][56][57] Researchers have shown that functionalized h-BN has high thermal conductivity, excellent thermal properties, and that it can be incorporated as a conductive ller in polymeric matrices. [58][59][60][61][62][63][64] Gu et al. used NH 2 -POSS functionalized nBN llers to improve the thermal conductivities for PPS dielectric nanocomposites. 58 However, research on improving both the thermal conductivity and toughness of epoxy matrix is scarce and not widely reported. A novel high performance composite material consisting of epoxy, core-shell polymers, and modied boron nitride developed to improve both of these properties, is presented in this paper. The composite has been successfully prepared by sequentially adding moderate amounts of poly(nbutyl acrylate)/poly(methyl methacrylate-co-glycidyl methacrylate) core-shell structured particles (PBMG) and modied boron nitride (MBN) to the epoxy (EP) resin.

Emulsion polymerization process of PBMG
Poly(n-butyl acrylate)/poly(methyl methacrylate-co-glycidyl methacrylate) (PBMG) latex was synthesized in a 500 ml fournecked ask equipped with a mechanical stirrer and reux condenser. The temperature of the water bath was maintained at 78 C, and the reaction was carried out in an argon atmosphere. First, the surfactant (0.25 g) and deionized water (140 g) were added to a four-necked ask and stirred at a 180 rpm for 30 minutes. Then a mixture of seed monomer (BA, 10 g) and a crosslinking agent (ALMA, 0.054 g) were added to the above dispersed system. Aer stirring for 10 minutes, a KPS aqueous solution (KPS, 0.43 g per DIW, 20 g) was added to start the polymerization reaction. More KPS aqueous solution (0.11 g/10 g) was added aer 55 minutes and stirred in for 5 minutes to complete the reaction of the seed stage. The second stage was the growth stage, which involved two layers of pre-emulsied monomers. The rst layer was the rubber core, containing BA (130 g), surfactant (1.625 g) and ALMA (2.275 g). The second layer was the rigid shell, consisting of MMA (60 g), surfactant (1.2 g), and functional monomer GMA (1.5 g). The pre-emulsied mixture was added into the reaction ask at a rate of 2 ml min À1 during the growth stage. The initiator, KPS aqueous solution (0.043 g/10 g), was added into the reaction ask at 60 min intervals aer the beginning of the second stage.
The nal mixture was allowed to remain for another 60 minutes to ensure that the monomer reaction was complete. Then the latex was cooled to room temperature and ltered through a 53 mm sieve to obtain the coagulated content. Finally, the PBMG particles were obtained aer freezing the ltrate for 12 h and freeze-drying it for another 12 h. The product was collected and ltered through a 100 mesh screen to obtain the nal product. A schematic diagram of the preparation of PBMG is shown in Fig. 1.

Characterization of the PBMG latex particles
At 30 min intervals, 5 ml samples of the latex were transferred into pre-weighed vials containing 1 ml hydroquinone solution (2 wt%) and then cooled down in an ice batch to terminate the polymerization. Gravimetric analysis was used to determine the monomer conversion. The particle size and polydispersity index (PDI) were measured with dynamic light scattering (DLS) using a Malvern Zetasizer NANO-ZS90 (Worcestershire, UK). The cell temperature was held at 25 AE 0.1 C, and the z-average diameter was calculated.
The growth of the latex particle size in the seed emulsion polymerization was investigated. The particles diameters determined by DLS were compared with theoretical particle diameters which were calculated according to the following equation: 65 where d t is the diameter of the particle at time t, M t is the total mass of the monomers added at time t, I t is the instantaneous conversion at time t, M s is mass of monomer added in the seed stage, and d s is the seed particle diameter as measured by DLS. Instantaneous and overall conversions were calculated using the following equations: 66 Instantaneous conversion% ¼ [mass of polymer formed/mass of monomer added] Â 100 (2) Fig. 1 Preparation procedure of PBMG.
where the mass of monomer added is the sum of the monomer at the seed stage and any monomer that has been added at each sampling time during the growth stage.
Overall conversion% ¼ [mass of polymer formed/total mass of monomer] Â 100 (3) where the total mass of monomers added is the sum of monomers at the seed stage and all monomers added during the growth stage.

Morphology of the PBMG core-shell particles
The morphology of the PBMG latex was characterized by a JEOL JEM-2100 transmission electron microscope (TEM). The latex was rst dispersed in deionized water with ultrasonic waves for 3 min, then single drops of diluted solution were deposited onto a carbon-coated copper grid, and dried with infrared light to prepare samples for TEM analysis. The external morphology of the latex particles was characterized by an FEI Nova NanoSEM 450 eld emission scanning electron microscope (SEM). The sample was diluted with distilled water to achieve the desired concentration, followed by ultrasound treatment for 5 min, then single drops of diluted solution were deposited on a silicon wafer which was then placed in a vacuum and dried at room temperature for 24 h. The surface morphology of the core-shell particles was observed and characterized.

Modication of the BN powder
The preparation of boron nitride nanosheets was based on the reported article with minor modications, including three steps. 67 First, 0.5 g of h-BN powder and 20 ml of benzyl benzoate were added into four sealed steel milling vials with 25 g steel balls that are 12.7 mm in diameter. The vials were lled with pure argon gas at a pressure of 200 kPa to avoid environmental contamination. The rotation speed of the four-station horizontal planetary mill was set at 350 rpm to generate enough shearing force to effectively exfoliate the h-BN powder. The milling time was set at 24 h. Secondly, the milled product was diluted with benzyl benzoate and sonicated at 100 W for 12 h to peel off the h-BN akes more thoroughly. Finally, the modied h-BN was centrifuged at 3000 rpm for 20 min to remove the agglomerated large particles. The supernatant was collected and washed three times with absolute ethanol. The nal product was ltrated and dried at 80 C under vacuum for 12 h. The resulting nano-scale akes were designated as modied boron nitride or MBN.
The surface morphology of the boron nitride before and aer modication was prepared in the same way as the PBMG. It was diluted with distilled water to achieve the desired concentration, single drops of the diluted solution were deposited on a silicon wafer and placed in a vacuum for drying at room temperature for 24 h then examined with the SEM. The BN powder before and aer modication was analyzed by Bruker AXS D8 Advance X-ray diffractometer at a scan range of 10-90 , and a scan rate of 6 min À1 . The h-BN and MBN powder were also analyzed with a DXR Raman Microscope from Thermal Scientic Corporation. The laser wavelength was 532 nm.

Preparation of the composite samples
Initially the PBMG and MBN powder were dried in vacuum ovens at 40 C and 80 C, respectively, for 24 h. The weighed amount of PBMG and MBN were sequentially added to 80 g of epoxy, and the mixture was fairly dispersed for 15 min using a three-roll grinder to ensure that the llers were evenly distributed in the epoxy. Then a hardener (68 g) and accelerator (1.6 g) were added to the mixture and stirred with a glass rod. Finally, the mixtures were degassed in a centrifugal deaerator for 10 min at a revolution speed of 3000 rpm, and a rotation speed of 900 rpm. The mixtures were poured into stainless steel molds slowly aer degassing and cured at 90 C for 1 h, then post-cured at 150 C for 4 h to ensure complete cure of the composites. Samples were of EP/5 wt% PBMG with MBN contents of 1%, 3%, 5%, 7%, and 10% by weight. The size of the composite specimens in each sample was 80 mm long, 10 mm wide, and 4 mm thick.

Mechanical properties and failure analysis
Impact and exure properties were obtained using China National Standards Testing Methods. The composite samples were conditioned at 23 AE 1 C and 50 AE 5% relative humidity for 48 h before testing. The unnotched impact strength was determined according to the National Standard Testing Methods GB/ T2571-1995 using a SANS ZBC-4 impact testing machine. The exure properties were obtained according to Standard GB/ T2570-1995 using a SANS CMT-6104 universal testing machine. Six replicate specimens of dimensions 80 mm Â 10 mm Â 4 mm were used in each of the impact and exure tests. The fracture surface of failed specimens of the EP/PBMG/ MBN composites samples were analyzed using SEM.

Thermal conductivity measurement of the composites
The thermal conductivities of the EP/PBMG/MBN composites were evaluated by TC 3000 Series Thermal Conductivity Apparatus (Xi'an Xiaxi Co. Ltd., China), based on transient hot-wire technique at room temperature according to ASTM C1113-99 (2004). Every specimen was measured for ve times at room temperature and an average value was calculated.

Results and discussion
3.1 Preparation of PBMG toughening particles and modication of h-BN powder 3.1.1 Preparation of PBMG toughening particles. A semicontinuous starvation-feeding method was used to prepare the toughening rubber-core PMMA-shell particles designated here as PBMG. The method has been widely used in the preparation of core-shell polymers. 65 First, the deposition rate of monomers must be slightly lower than the reaction rate of the monomer to prevent secondary nucleation or heterostructure of the core-shell particles formed during the polymerization process. Second, the reaction activity of the monomers should be relatively high to ensure the effectiveness of the starvation-feeding method.
The monomer conversion vs. reaction time of the emulsion polymerization is shown in Fig. 2(a). It can be seen in this gure that the monomers instantaneous conversion rates are higher than 90%, and the nal conversion rate is 97.45%. The total monomer conversion with the reaction time exhibited a linear growth mode, indicating that the polymerization process was relatively stable, and that the deposition rate of the monomers was appropriate in this reaction. During the stabilization stage (240-300 min), the instantaneous conversion of monomers coincided with the total conversion and the total conversion rate basically remained unchanged signaling that no more monomers were added to the reaction. Fig. 2(b) shows the measured and theoretical particle sizes vs. the reaction time. The gure presents that the particle diameter was 130 nm at the end of the seed stage, and the nal particle diameter was 358 nm. The measured particle size of latex particles was basically consistent with the theoretical particle size, which indicates that all the added monomers were polymerized on the surface of the original latex particle without secondary nucleation. Fig. 2(c) shows the distribution of the nal particle size of the PBMG latex. The particle distribution index was 0.044, which means that the nal particles size of the latex was uniform with a very narrow distribution range.
The PBMG latex was further analyzed by TEM and SEM. As shown in Fig. 3(a), the particles consisted of a dark core of poly(BA) and a brighter shell of poly(MMA), which clearly indicated that a core-shell structure of PBMG latex had been successfully constructed. The SEM image in Fig. 3(b) shows that the PBMG latex had a uniform distribution and consistent size.  This journal is © The Royal Society of Chemistry 2019 3.1.2 Exfoliation and modication of h-BN. Boron nitride powder was used to improve the thermal conductivity of EP/ PBMG matrix. To maximize its efficiency, the h-BN was modi-ed by peeling the akes into nanosheets and ensuring the exfoliation and well dispersion of the sheets in the composite. SEM images of h-BN before and aer modication are shown in Fig. 4. Fig. 4(a) and (b) show that the bulk h-BN akes have lateral dimensions of 3-5 mm and thickness of $300 nm. Fig. 4(c) and (d) show that the modied boron nitride (MBN) sheets have lateral dimensions of 1-2 mm and thickness of 30 nm indicating that the ball milling and sonication process were effective in peeling off thin layers of h-BN from bulk akes and/or breaking the akes into smaller pieces. It is worth noting that the edge of the MBN is curly and rough which could only be seen in the nanosheet samples. All observations demonstrated that exfoliated layers of boron nitride nanosheets were successfully obtained.
Thin specimens of MBN were prepared and analyzed under the TEM. The thin and transparent MBN nanosheets shown in The crystal structure and exfoliation of the MBN were characterized by X-ray diffraction. The XRD patterns shown in Fig. 6 indicate that almost all the peak positions of h-BN and MBN are matching. The diffraction peaks located at 26.72 , 41.63 and 55.14 of 2q angle corresponded to the (002), (100) and (004) planes according to the standard spectrum of hexagonal boron nitride. 70 Both X-ray diffraction patterns of h-BN and MBN were narrow and sharp, without new peaks. These results indicate that MBN nanosheets prepared by ball milling and sonication could well maintain the original crystal structure.
Based on these results, the (002) crystal plane along the C axis was the optimal orientation of the BN layered material. 10,71 Therefore, the better the exfoliation of boron nitride, the more (002) crystal planes are exposed in the BN. Comparing the two diffraction patterns, the intensities of the (002) and (004) planes of the MBN are signicantly enhanced and the intensities of diffraction peaks of other crystal planes are weakened, which indicates that more (002) planes were exposed during the modication process. Furthermore, according to the calculated result, the value of I (004) /I (100) of MBN was 5 times higher than the original h-BN which fully proved that the MBN had been sufficiently exfoliated. 72 The exfoliation of the MBN nanosheets was further veried using Raman microscopy, which was an effective method to characterize the structure and interlayer relationship of boron nitride. [73][74][75] Fig. 7 shows Raman spectra of h-BN and MBN. As can be seen from this gure, the characteristic peak position of h-BN and MBN were at 1365.28 cm À1 and 1363.35 cm À1 , respectively, which attributed to the E 2g vibration mode of boron nitride material. 76,77 Compared with h-BN, the Raman peak of the MBN produced a red shi of 1.93 cm À1 aer exfoliation and the full width at half-maximum (FWHM) of the MBN nanosheets increased. This was because the stretching vibration between the boron nitride nanosheets and silicon substrate in the Raman test affected the E 2g vibration of MBN, and the nanolayer structure of MBN had produced photon limitations. These results indicated that the exfoliation could effectively reduce the thickness of BN powder. 78 A similar phenomenon had been reported by Hou. 79 In addition, the Raman peak intensity of MBN was signicantly lower than that of h-BN, which may be explained by the widening of the spacing between the layers and the weakening of the interlaminar forces of boron nitride nanosheets. 80 In short, Raman spectroscopy proved that the wet ball milling combined with ultrasonic method efficiently exfoliated the h-BN, as indicated by the TEM images and the XRD analysis above.

Improvement of impact toughness and exure strength
Before determining the change in thermal conductivity by the addition of MBN into the EP/PBMG blends, it was necessary to ensure that the addition of MBN did not signicantly reduce the mechanical properties of the blended epoxy achieved with 5 wt% PBMG content. A study of the effect of the core-shell PBMG on the mechanical properties of epoxy showed that the impact and exure strengths of the EP/PBMG increased with the PBMG content up to a certain weight fraction aer which these properties started to decrease. The optimum PBMG content to obtain the maximum toughness and exural strength was 5 wt%. 81 At this content the impact and exural strengths reached a maximum of 37.25 kJ m À2 and 168.4 MPa, respectively. The PBMG structure provided a PMMA shell which was compatible with the EP as the PBMG was functionalized with GMA on the outer shell. The epoxy groups of GMA on the surface of PBMG reacted with the epoxy resin during the curing process. 10,82 In addition, at 5 wt% the PBMG was well-dispersed and evenly distributed in the EP matrix. The increased plastic deformation and energy absorption of the EP/PBMG were   Paper mainly due to the rubber core in the PBMG particles. The deformation mechanism is most likely due to stress-activated shear yielding initiated in the matrix, in the regions of high stress concentration close to the equators of the rubber particles. The deformation leads to the formation of cavitation within the rubber particles which in turn promotes shearyielding and higher ductility. 81 A similar study was conducted for the EP/PBMG/MBN composite in which the PBMG was kept at its optimum value of 5 wt% and the MBN content was increased from 0 to 10% by weight. Fig. 8 shows the variation of the impact strength and exural strength of EP/PBMG/MBN composites with various MBN contents. It is shown in this gure that the impact and exural strengths increase up to a certain amount of MBN content, then start to gradually decrease with additional content of MBN. The ability of the EP/ PBMG to maintain its toughness and strength at low MBN content could be attributed to the high specic surface area and high surface activity of MBN, and a strong interaction between the MBN and epoxy matrix. In addition, the MBN may function as a layered reinforcement, hence increasing or maintaining the strength and toughness of the blended matrix. The highest impact and exural strengths were achieved at 3 wt% MBN content. The impact strength at 3 wt% MBN was 43.0 kJ m À2 and the exural strength was 179 MPa as shown in Fig. 8. The mechanical properties of the EP/PBMG/MBN composites started to gradually decrease aer the MBN content exceeded 3 wt%. This is most likely due to the agglomeration of MBN at higher concentrations that lead to the formation of brittle regions with high stress concentrations. As shown in the next section, the thermal conductivity increased linearly with MBN content. The optimum strengths at a relatively high thermal conductivity were achieved with a composite consisting of 7 wt% MBN by weight. At this MBN content the impact strength of the EP/PBMG/MBN composites was 29.6 kJ m À2 and the exural strength was 161 MPa. These values are lower than the strengths obtained at 3 wt% MBN content, however they are still 147% and 49.1%, respectively, higher than those of the neat epoxy. The major toughening mechanisms involve a suitable structured PBMG particle cavitation and/or debonding between the exfoliated BN layer and the EP, followed by massive shear yielding of the matrix.
In order to evaluate the interfacial interaction of h-BN and MBN with PBMG and EP matrix, the fractured surface of EP/ PBMG/BN composites were investigated by SEM. Fig. 9(a) shows that the neat epoxy resin exhibits a relatively smooth fracture surface. The river pattern indicates a typical cleavage fracture, accounting for the low fracture toughness of the unlled epoxy. In contrast, the EP/PBMG/BN composites show much rougher fracture surfaces indicating plastic deformation, as shown in Fig. 9(b) and (c). The increased surface roughness implies that the path of the crack tip is either arrested or deected by the ller, making crack propagation more difficult. 83 Large amount of micro cavities, 'toughness dimples' and macroscopically sized stress yielded zones can be observed, which indicate the development of localized shear-yielding plastic deformation under impact. Fig. 9(b) and (c) show that the composites exhibited plastic deformation and ductile failure at 3 wt% content of both h-BN and MBN. That could be due to the fact that the nanoscale BN may have stronger interfacial bond which can inhibit crack propagation or result in crack deection, leading to higher energy absorption and tougher matrix. 84 However, as shown in Fig. 9(d), with the increase of MBN content, agglomeration occurs gradually within the composite due to the van der Waals interactions between the adjacent MBN nanosheets, leading to some defects that may compromise the mechanical properties of the composites. In summary, the SEM images of fracture surfaces conrm that the incorporation of PBMG and exfoliated BN effectively improve the strength and toughness of EP/PBMG/ MBN composites and illustrate the reasoning behind the relative decrease in these properties with higher MBN content.

Improvement of thermal conductivity
The thermal conductivity of the neat epoxy, the EP/PBMG blend, and the EP/PBMG/MBN composites with various MBN contents were investigated. In addition, the thermal conductivity of the EP/PBMG/h-BN composites were also investigated to clarify the importance of modifying the original h-BN. The thermal conductivity of the neat epoxy was determined as 0.191 W (m À1 K À1 ), and that of the EP/PBMG was measured at 0.192 W (m À1 K À1 ), indicating that the addition of core-shell particles did not affect the thermal conductivity of the epoxy, as shown in Fig. 10 and Table 1 below. The addition of BN leads to an increase in thermal conductivity and the percent increase in thermal conductivity, determined as the thermal conductivity improvement ratio in eqn (4) below, is shown in Fig. 10.
Thermal conductivity improvement ratio ¼ [(thermal conductivity of the composite À thermal conductivity of the epoxy)/thermal conductivity of the epoxy] Â 100 It is shown in Fig. 10 that the thermal conductivity of the EP/ PBMG/BN composites increases with the addition of both h-BN and MBN. More importantly, the graphs show that the addition of MBN is more effective in increasing the thermal conductivity of the composites than the h-BN. The high aspect ratio of the MBN nanosheets is critical in increasing the thermal conductivity improvement ratio shown in Fig. 10. The high aspect ratio of the MBN nanosheets leads to the formation of effective networks of conductive pathways that in turn lead to higher thermal conductivity. 76 In addition, the high surface activity of the MBN nanosheets and their strong bond with the epoxy matrix provide stable and highly efficient conductive networks, further enhancing the thermal conductivity of EP/PBMG/MBN composites. Table 1 shows that at 10 wt% the thermal conductivity of the MBN-lled composite is 0.452 W (m À1 K À1 ), while that of the EP/PBMG/h-BN composite is 0.372 W (m À1 K À1 ). These values are 1.37 and 0.95 times higher, respectively, than that of the neat epoxy matrix. In other words, at 10 wt% MBN, the thermal conductivity of the composite increased by 137%.

Conclusions
Poly(n-butyl acrylate)/poly(methyl methacrylate-co-glycidyl methacrylate) (PBMG) core-shell particles were synthesized by seed emulsion polymerization and added to epoxy resin to improve its toughness and strength properties. Modied boron nitride (MBN) nano-akes were prepared by ball milling and sonication and added to the EP/PBMG blend to improve the thermal conductivity of the composite. The thermal conductivity of the EP/PBMG/MBN composite increased linearly with MBN content and reached a maximum value of 0.452 W (m À1 K À1 ), at MBN content of 10% by weight. The highest impact and exural strengths were achieved at 5 wt% PBMG and 3 wt% MBN, but at 3 wt% MBN content the thermal conductivity was only 0.271 W (m À1 K À1 ). Therefore to get a good balance between the mechanical properties and the thermal conductivity, a 7 wt% MBN content is recommended. At this content the thermal conductivity is increased by 98% compared with pristine EP, the impact strength of the EP/PBMG matrix is increased by 147% from 12.0 kJ m À2 to 29.6 kJ m À2 , and the exure strength is increased by 49.1% from 108 MPa to 161 MPa. Material scientists and engineers can determine the appropriate MBN content within these limits to achieve desired   toughness and thermal conductivity for specic applications. With the improved toughness and thermal conductivity, the EP/ PBMG/MBN composite has high potential for applications in industries in which these properties are critical.

Conflicts of interest
There are no conicts to declare.