3.1. Structure and morphology of 3DBT/EP-GO composites
Figure S1 illustrates representative SEM images of cleanroom wiper and the obtained 3DBT network structure. The cleanroom wiper exhibits a loose interwoven structure (Fig. S1a), which provides a 3D skeleton for the formation of 3DBT network (Fig. S1b). The porous 3DBT network and large specific surface area enables the penetration of epoxy resin and facilely forms continuous polarization pathway in the composite accordingly. The microstructure and morphology of GEBT and EBT systems are shown in Fig. 2. Figure 2a exhibits a representative low magnification SEM image of GEBT-3 composite. It is seen from the cross section of GEBT-3 that epoxy resin is successfully permeated into the 3DBT network. The thickness of GEBT is about 140 µm as determined from the SEM image, a little larger than that of pristine cleanroom wiper (120µm) template, which may be ascribed to the slight expansion of BT gelation during calcining. The upper right of Fig. 2a shows an enlarged view of the interface between BT and polymer, from which it can be observed that epoxy has good affinity with 3DBT. No obvious pores are observed with the aid of suction filtration. From Fig. 2b, it is evident that the 3DBT still maintains its continuous network structure, as marked by the green riverbed-like curves. In contrast, island structure of traditional BT powder filled epoxy composite (EBT-1) can be observed in Fig. 2c, including aggregated BT clusters. Figures 2d and 2e are representative optical microscope images of GEBT-3 and EBT-1, respectively. Compared with SEM images, the optical microscope images provide much intuitive observation of the interlacing 3DBT structure within polymer matrix. Such continuous 3DBT network is expected to construct perforative pathway of polarization within the composite [14], thus leading to the improved dielectric and energy storage properties thereafter.
Figure 3a illustrates representative XRD patterns of epoxy, EBT-1, GEBT-3 and standard PDF card of BaTiO3. The peaks located at 22.2o, 31.5o, 38.9o, 45.5o, 50.8o, 56.3o, and 65.8o correspond to the (100), (110), (111), (200), (210), (210), and (220) characteristic diffractions of cubic BT, respectively [22]. These characteristic peaks can be traced in the GEBT and EBT composites. Figure 3b displays the FTIR results of pure epoxy, EBT-1, and GEBT-3. The absorption peaks located at 826 cm− 1 and 1230 cm− 1 correspond to the asymmetric and symmetric vibrations of epoxide group, respectively. The peaks at 1034 cm− 1 and 1180 cm− 1 represent C-O stretching of epoxide group and C-C stretching of molecular backbone [23]. The characteristic peaks of benzene ring stretching vibrations can be found at 1506 cm− 1 and 1608 cm− 1. In addition, with the introduction of BT and 3DBT, the infrared curves of EBT-1 and GEBT-3 composites exhibit a strong wide peak around 553 cm− 1, which belongs to typical barium ion stretching vibration of barium titanate [24]. The small addition of GO can be traced form the new C = O stretching vibration of carboxyl group at 1717 cm− 1. Other characteristic peaks of GO, such as epoxide (1034 cm− 1) and hydroxide (3400 cm− 1) groups, are weak and somewhat overlapped with that of neat epoxy.
The TGA and DSC curves of representative epoxy, EBT-1, GEBT-1 and GEBT-3 systems are shown in Fig. 3c. It can be concluded from TGA curves in Fig. 3c and Table S1 that the content of 3DBT in all GEBT systems is in the vicinity of 15.0 wt%, endowing the brittle 3DBT with excellent flexibility and mechanical strength. Furthermore, form the initial decomposition temperature (Tdi, at 5 wt% weight loss), the GEBT composites show better thermal stability than that of EBT-1 system, which may be ascribed to the higher cross-linkage of epoxy in GEBT than that of EBT systems. In addition, the thermal stability of GEBT composites can also be verified by the final decomposition temperature. The introduction of small amount of GO further improves the thermal stability of GEBT composites as revealed by the DSC curves. The final decomposition temperature (Tdf) of GBET-3 composite with only 0.5 wt% GO reaches 543.4 oC, 29.6 oC and 14.4 oC higher than that of EBT-1 and GEBT-1 systems, implying that GO is an effective stabilizer for GEBT composites due to its various functional groups.
3.2. Dielectric properties
Figures 4a-4c display the frequency dependence spectra of dielectric constant, dielectric loss, and conductivity of epoxy, GEBT and EBT composites, respectively. It can be seen from Fig. 4a that the permittivity of all systems gradually decreases with increasing frequency, indicating a phenomenon of polarization hysteresis to external frequencies [25]. However, the dielectric constant of composites increases with increasing of both 3DBT and BT nanoparticles, implying enhanced interfacial polarizations after incorporating 3DBT and BT nanoparticles. The GEBT systems with continuous 3DBT network exhibits much better dielectric performance than that of EBT composites. The permittivity of GEBT-1 (15.0 wt% 3DBT) reaches 9.2 at 100 Hz, which is almost double times higher than that of EBT-1 (ε′ = 5.2) with the same BT loading and even larger than EBT-2 with 25 wt% BT loading. The advantage of construction 3DBT network is evident. What’s more, the dielectric constant of GBET composite is further improved with small addition of GO (also shown in the inset of Fig. 4d). The permittivity of GEBT-4 system with only 0.75wt% GO achieves 15.6, almost 1.7 and 3.5 times higher than that of GEBT-1 and epoxy, respectively. The distinct enhancement in permittivity with small addition of GO is attributed to the good dispersion of multi-functional GO nanoplatelets, which for one thing forms numbers of micro-capacitors [26], and for another, brings in strong interfacial polarizations [27]. Figure 4b illustrates the variation of loss tangent (tanδ) with external frequencies. The loss tangents of these composites on the whole increase with 3DBT and BT ceramic contents, however, it still remails a relatively low level (tanδ = 0.013 ~ 0.024 at 100 Hz) in the same orders of magnitude (Table S1). The addition of GO in GEBT has little effect on the dielectric loss of composites, implying excellent insulating behavior of GO. As expected, the conductivity of GEBT and EBT composites shown in Fig. 4c increases almost linearly in log-log plots within the tested frequency range [28]. The conductivity of representative GEBT-4 system is only 1.8×10− 9 S·m− 1 at 100 Hz, which is a little larger than neat epoxy (5.5×10− 10 S·m− 1), exhibiting good insulating behavior as well.
The variation of dielectric constant for GEBT and EBT systems with different 3DBT or BT contents under 100 Hz is displayed in Fig. 4d. Accordingly, Maxwell-Garnett and Logarithmic mixing rule are used to theoretically predict the variation of permittivity for the studied GEBT and EBT systems [29], provided that εepoxy = 4.4, εBT = 2000. It can be seen from Fig. 4d that the dielectric constant of EBT system obeys the mixing rule well, indicating uniform dispersion of BT nanoparticles in epoxy matrix, while the permittivity of GEBT composite far exceeds theoretical predictions from the above two theories with the same BT content. It is known that Maxwell-Garnett theory and logarithmic mixing rule are good models to simulate polymer dielectric composites filled with spherical nanofillers [29]. Therefore, the far cry from theoretical values can be due to the interwoven structure of 3DBT network in epoxy matrix that constructs continuous polarization channels capable of promoting the polarization ability of GEBT composites. The effect of 3DBT to the interfacial polarization can be clearly observed from the ε′-ε″ and M′-M′′ plots as shown in Figs. S2a-S2b, where large difference can be seen in the low frequency region, indicating strong interfacial polarization effect with increasing of BT and GO contents [30]. The relaxation behavior of composites is also displayed in Figs. S2c-S2d in an expression of frequency-dependent dielectric modulus, indicating no big relaxation changes after incorporating 3DBT or BT nanofillers in the studied frequency range.
3.3. Breakdown strength
The breakdown strength of composites was determined by a two-parameter Weibull statistical distribution method [31], which can be written as Eq. (1):
$$P=\exp \left[ { - {{\left( {{\raise0.7ex\hbox{${{E_b}}$} \!\mathord{\left/ {\vphantom {{{E_b}} {{E_0}}}}\right.\kern-0pt}\!\lower0.7ex\hbox{${{E_0}}$}}} \right)}^\beta }} \right]$$
1
where Eb is the experimental breakdown strength, P is the cumulative probability of electrical failure, β is the shape parameter related to the scatter of the data, E0 is the scale parameter representing the breakdown strength at cumulative failure probability of 63.2%. In addition, Eq. (2) is a linear expression derived from Eq. (1), form which the breakdown strength E0 can be derived.
$$\ln \left[ { - \ln \left( {1 - P} \right)} \right]=\beta \ln {E_b} - \beta \ln {E_0}$$
2
Figure 5a illustrates the Weibull distribution of breakdown strengths of GEBT and EBT composites. As expected, the breakdown strength of composites decreases with increasing ceramic contents in both GEBT and EBT systems. The derived E0 for EBT-1 and EBT-2 system are 137.6 kV·mm− 1 and 121.0 kV·mm− 1, which are 98.1% and 86.4% of pure epoxy (140.2 kV·mm− 1), respectively. The breakdown strength of GEBT-1 composite reaches 128.8 kV·mm− 1, which is a little lower than that of EBT-1 system with the same filler loading. However, there is a significant increase in breakdown strength of GEBT composites with small addition of GO, which may be ascribed to the reduced current leakage caused by the deep carrier traps of GO [20]. The breakdown strength of GEBT-4 reaches 239.8 kV·mm− 1, which is 1.9 and 1.7 times higher than that of GEBT-1 and epoxy resin, respectively. Figure 5b presents a schematic illustration of the hindrance of electrical-treeing growth in the presence of GO nanoplatelets [25]. The depth of electrical breakdown is distinctly reduced, thus leading to the improved electrical breakdown performance. The increased dielectric constant and breakdown strength of GEBT composites in the presence of GO contribute greatly to the overall energy storage capacity of composites. The energy density of GEBT composites obtained from linear equation of Ul = 1/2ε0εrEb2 is summarized in the last column in Table S1. It is seen that the GEBT-4 possesses the highest energy density of 3.97 J·cm− 3, which is 10.4 and 9.0 times higher than that of pure epoxy and EBT-1, respectively.
To understand the underlying mechanism and to characterize the effect of 3DBT network structure on polarization and breakdown strength, the electric field distribution and polarization direction of representative GEBT-1 composite was simulated by finite element analysis [32]. Figure 5c is an optical microscope image of GEBT-1 and used as the simulation model. Figure 5d is the finite-element simulation of GEBT-1 composite, where the permittivities of epoxy and 3DBT are defined as 4.4 and 2000, respectively. It is observed that high electric field is concentrated on the interfaces of composite when an external electric field of 1 Volt is applied to the above model as evidenced from the color change of Fig. 5d. The simulated electric field intensity on the ceramic side is only EBT−max = 0.33×10− 3 kV·mm− 1, which is far less than that of on the epoxy side (Eep−max = 111.21 ×10− 3 kV·mm− 1). At the gap of two unconnected ceramic networks (the dotted area in Fig. 5d), a high accumulation of electric field (the red color) can be observed, where equipotential lines are densely distributed in the gaps of unconnected ceramics, implying less resistance to breakdown at interfaces and gaps. The direction and intensity of polarization can be seen from the yellow arrows in Fig. 5d. It shows that the polarization mainly occurs in the interior of ceramic network, and the direction is consistent with the trend of external field. The existence of 3D ceramic network effectively provides continuous polarization pathway, thus leading to the improved permittivity of composite.
3.4. Polarization-electric field loop
D-E loop is preferred to determine the energy storage performance of nonlinear dielectrics. Figure 6a shows the D-E loops of epoxy, EBT and GEEBT composites under 40 kV·mm− 1. The GEBT system exhibits greater polarization than that of EBT system and pure epoxy, and the loops enlarge with further addition of GO. The results have the same tendency with the variation of permittivity of composites. The GEBT-4 composite exhibits the maximum electric displacements as high as 5.07 µC·cm− 2 at 40 kV·mm− 1. However, its energy dissipation (Uloss) is also the largest (0.103 J·cm− 3) as compared to other specimens. Therefore, the discharge energy density (Ue) is only 0.36 J·cm− 3 (energy efficiency η = 25.7%, Fig. 6b). Large energy dissipation leads to the decreased energy storage density. In contrast, the GEBT-3 system with 0.5 wt% GO possesses the highest storage density of 0.47 J·cm− 3 at 40 kV·mm− 1, nearly double times higher than that of neat epoxy. The enhanced energy storage is as a result of the combination of 3DBT network and 2D GO. The dielectric properties and energy storage density of each composite are summarized in Table S1, showing excellent synergistic effect between 3DBT and GO. In comparison, the dielectric and energy storage performances of polymer-ceramic dielectrics based on structure design and hybridization are summarized in Table 1. The synergistic effect of constructing 3DBT network and the introduction of GO deep traps in our work is clear, showing potential applications in energy storage dielectrics.