Highly improved dielectric properties of polymer/ a -Fe 2 O 3 composites at elevated temperatures

a -Fe 2 O 3 particles were incorporated into ten kinds of polymer matrices, and the dielectric properties of the resultant polymer/ a -Fe 2 O 3 composites were investigated between 40 (cid:1) C and 160 (cid:1) C. We found that the dielectric properties strongly depended not only on the temperatures but also on the kind of polymer matrices. For engineering plastics such as polyetherimide (PEI), the dielectric constant 3 0 r was highly enhanced at around 1 kHz by incorporation of the a -Fe 2 O 3 particles owing to Maxwell – Wagner polarization of free electrons in the a -Fe 2 O 3 particles. This is probably because the p electrons in the aromatic structures of the engineering plastics strongly interact with the electrons in the a -Fe 2 O 3 particles. Furthermore, the dielectric loss factor 3 00 r for the engineering plastics became small at elevated temperatures because the conductivity of the a -Fe 2 O 3 particle was enhanced and therefore the relaxation frequency of Maxwell – Wagner polarization was shifted to higher frequency. PEI/ a -Fe 2 O 3 composites exhibited highly improved dielectric properties at around 1 kHz, the high 3 0 r and very low 3 00 r at the elevated temperatures above 120 (cid:1) C. It was demonstrated that the PEI/ a -Fe 2 O 3 composite was comparable to the PEI/BaTiO 3 composite in dielectric performance at 160 (cid:1) C. Because the cost of a Fe 2 O 3 is much lower than that of BaTiO 3 , the PEI/ a -Fe 2 O 3 composite might be promising as a low-cost dielectric material for high-temperature applications.


Introduction
Dielectric materials are mainly used in capacitors to store electrical energy. [1][2][3] In comparison to ceramic materials, polymer ones have the advantages of low cost, lightweight and high processability. However, the very low dielectric constant 3 0 r of the polymer material strongly prevents improvement of the capacitors. Therefore, polymer/ceramic composite materials have been much studied over the last few decades. Among the ceramics, BaTiO 3 has been most widely used for enhancement in the 3 0 r of polymer materials because of the very high 3 0 r of BaTiO 3 . 4-21 However, the high cost of BaTiO 3 is a serious problem for industrial applications.
On the other hand, hematite (a-Fe 2 O 3 ) is a very low cost ceramic material although a-Fe 2 O 3 has not been paid attention as a high 3 0 r ller for polymer materials owing to the low 3 0 r of a-Fe 2 O 3 ; according to the reported literature, the 3 0 r of a-Fe 2 O 3 is as low as 30. [22][23][24][25][26][27] Therefore, few studies are available on the dielectric properties of polymer/a-Fe 2 O 3 composites. [28][29][30] Djoković and coworkers reported dielectric properties of a composites epoxy resin and a-Fe 2 O 3 nanorods. 28 Also, a-Fe 2 O 3 nanorods were dispersed in a poly(vinyl alcohol)/poly(ethylene glycol) blend by Sayed et al. 29 There are only small improvements of the 3 0 r s of the polymer materials in these studies.
In this paper, we focus on the conductivity s of a-Fe 2 O 3 at elevated temperatures. Typically, a-Fe 2 O 3 is known to have s of about 10 À5 to 10 À6 S cm À1 at room temperature. 26 The s of semiconductors such as a-Fe 2 O 3 is enhanced as temperature is raised because electrons are excited from a valence band to a conductive band. 31 It is expected that the s of a-Fe 2 O 3 is especially increased owing to its narrow band gap of 2.1 eV. 32 At elevated temperatures, the enhanced s of the a-Fe 2 O 3 particles would bring a high 3 0 r to a polymer/a-Fe 2 O 3 composite owing to Maxwell-Wagner polarization 33 of the resultant free electrons in the a-Fe 2 O 3 particles. The polymer/a-Fe 2 O 3 composite might be promising as a low-cost dielectric material for high-temperature applications such as hybrid and electric vehicles. 34,35 In this study, a-Fe 2 O 3 particles were incorporated into ten kinds of polymer matrices, and the dielectric properties of the polymer/a-Fe 2 O 3 composites were investigated between 40 C and 160 C. We found that the dielectric properties strongly depended not only on the temperatures but also on the kind of polymer matrices. Furthermore, the dielectric performance of one of the polymer/a-Fe 2 O 3 composite was compared with that of a corresponding polymer/BaTiO 3 composite at the elevated temperatures.

Experimental
Sample preparation a-Fe 2 O 3 particles were obtained from Kanto Chemical (Japan). The density and specic area of the a-Fe 2 O 3 particles were determined to be 4.94 g cm À3 and 8.71 m 2 g À1 using helium pycnometry and a standard BET method, respectively. The average diameter of the a-Fe 2 O 3 particles was calculated to be 139 nm from the obtained density and specic area data. In the same manner, the density, specic area and average diameter of BaTiO 3 particles obtained from Wako Pure Chemical Industries (Japan) were determined to be 5.73 g cm À3 , 9.99 m 2 g À1 and 105 nm, respectively.
A series of polymer/a-Fe 2 O 3 composites were prepared by simply blending the a-Fe 2 O 3 particles with the polymer matrices. First, the a-Fe 2 O 3 particles and the polymer matrix except PES were homogeneously dispersed in 1,4-dioxane by sonication. The dispersion was quickly frozen by liquid nitrogen, and then freeze-dried under vacuum. For preparation of PES/a-Fe 2 O 3 composites, dichloromethane was used instead of 1,4-dioxane because PES was insoluble in 1,4-dioxane. Subsequently, all the pre-mixtures thus-obtained were further kneaded in molten states. Also, PEI/BaTiO 3 composites were prepared in the same manner. The resultant composites were hot-pressed into disklike specimens with a diameter of 33 mm and a thickness of $0.52 mm. For electrical measurements, two gold electrodes with a diameter of 27 mm were deposited on the top and bottom of the specimens.

Measurements
Complex permittivity was obtained in the frequency f range of 10 2 to 10 6 Hz at 2 V using an LCR meter (E4980A, Agilent) coupled with a rheometer (ARES G2, TA instruments). The temperature of the specimen was controlled in the oven of the rheometer. Direct current (DC) resistance of the specimen was measured at room temperature under a nitrogen atmosphere using a megohmmeter (SM-8220, Hioki), and the value was recorded aer 10 minute from application of a voltage of 500 V. The cross-section of the specimen that had been attened with argon ion beam 38 was observed using an SEM (S-4300, Hitachi) operated at an accelerating voltage of 2 kV.  Table 1. There is a weak tendency for the 3 0 r of the composites prepared using the engineering plastics to be high. This is an interesting phenomenon that has never been reported. However, it is difficult to correctly evaluate the 3 0 r enhancement effect by direct comparing of the 3 0 r values of the composites because each 3 0 r of the polymer matrix 3 0 r polymer is different to some extent as shown in Table 1. Therefore, we suggest that apparent 3 0 r of a-Fe 2 O 3 , 3 0 r Fe 2 O 3 is calculated using Lichteneker's logarithmic mixing rule and then compared in order to separate a contribution of the 3 0 r polymer itself from the 3 0 r of the composite. It is known that the 3 0 r of a polymer/ller composite oen obeys Lichteneker's logarithmic mixing rule as follows: 4,7,9,15-17,39

Results and discussion
where 3 0 r ller and F are the 3 0 r and volume fraction of the ller. Notice that the 3 0 r ller is not the 3 0 r of the ller itself but an apparent 3 0 r of the ller particles in the polymer matrix.
Generally, the 3 0 r ller is much smaller than the 3 0 r of the corresponding bulk compound, mainly because the ller particles do not form a continuous phase in the polymer/ller composite system; the 3 0 r ller is signicantly reduced by the interface between the ller particles. Furthermore, the 3 0 r ller is affected not only by the size and form of the ller particle but also by the dispersivity of the ller particles in the matrix. 16 For example, the apparent 3 0 r of spherical BaTiO 3 nanoparticles with a diameter of less than 100 nm is reported to be around 100 according to literatures. 8,39 The calculated 3 0 r Fe 2 O 3 is shown in Table 1. There is a strong dependency of the 3 0 r Fe 2 O 3 on the polymer matrix; the 3 0 r Fe 2 O 3 is about 30 for PS, PMMA and P2VP whereas over 100 for PAR, PEI, PES and PPE. In other word, the 3 0 r for engineering plastics is more enhanced by the a-Fe 2 O 3 particles than that for general-purpose plastics such as PS and PMMA. Fig. 2 shows the frequency f dependences of the 3 0 r and 3 00 r of four kinds of composites for PEI, PSF, PPE and PS with F ¼ 0.2 obtained at 40 C. Every 3 0 r is decreased as f is increased. This is because the dielectric relaxation of a polarization occurs in the f range, which results in a 3 00 r peak as shown in Fig. 2b. This polarization would be attributed to the interfacial polarization of free electrons in the a-Fe 2 O 3 particles, that is Maxwell-Wagner polarization. 33 Because the relaxation frequency becomes high as the s of the a-Fe 2 O 3 particles is high for Maxwell-Wagner polarization, it is suggested that the s of the a-Fe 2 O 3 particles in the PEI/a-Fe 2 O 3 and PPE/a-Fe 2 O 3 composites is higher than that in the PS/a-Fe 2 O 3 composite, resulting in the higher 3 0     (Table 1). Under an electric eld E, the power loss of a dielectric material per unit volume, P is expressed as, 16 where 3 0 is the permittivity of vacuum. Because P is proportional to 3 00 r , the 3 00 r of the material must be low for industrial applications such as capacitors.
The large 3 00 r values of the polymer/a-Fe 2 O 3 composites are drastically reduced at 160 C. Dielectric characteristics (1 kHz) obtained at 160 C were listed in Table 2 for six kinds of polymer/a-Fe 2 O 3 composites with F ¼ 0.2. At 1 kHz, all the 3 00 r values are less than 0.05. Furthermore, for PEI, PSF, PAR and PES, the 3 0 r Fe 2 O 3 is more than 150 at 160 C. For the four kinds of composites, the f dependences of the 3 0 r and 3 00 r are shown in Fig. 3. Because the dielectric relaxations of Maxwell-Wagner polarization are not started at around 1 kHz as shown in Fig. 3b, the 3 00 r values are small in the lower f ranges than 1 kHz. This result indicates the s of the a-Fe 2 O 3 particle is higher at 160 C than 40 C. Fig. 4 shows a variation in the f dependences of the 3 0 r and 3 00 r of the PEI/a-Fe 2 O 3 composites with F ¼ 0.2 between 40 and 160 C. The 3 00 r peak is shied to the higher f until the temperature reaches 120 C.
For the four polymer matrices, we also prepared polymer/a-Fe 2 O 3 composites with F ¼ 0.4. Fig. 5 shows the f dependences of the 3 0 r and 3 00 r of polymer/a-Fe 2 O 3 composites with F ¼ 0.4 at 160 C. The 3 00 r of the PEI/a-Fe 2 O 3 composite remains small whereas for PAR and PES, the 3 00 r becomes very large over all the f range. In order to elucidate the large 3 00 r , the volume resistivity r DC of the composites was measured. At F ¼ 0.4, the r DC for PAR and PES is much lower than that for PEI and PSF as shown in Fig. 6. 3 00 r is the sum of a loss from s, 3 00 r(s) and losses from dielectric polarizations 3 00 r(DP) : 40 Furthermore, 3 00 r(s) is described using r DC as follows: It is suggested that because 3 00 r(s) is inversely proportion to f and r DC , the 3 00 r is highly increased at the low f range for PAR and PES. The low r DC for PAR and PES could be attributed to a leak current passing through the conductive networks of the a-Fe 2 O 3 particles, that is, percolation of the a-Fe 2 O 3 particles. According to the percolation theory with a random packing of spherical particles, the percolation threshold is calculated to be about 0.30. 41,42 Therefore, it was reasonable that the percolation of the a-Fe 2 O 3 particles occurred in the polymer/a-Fe 2 O 3 composites with F ¼ 0.4. To the contrary, it was unusual that the PEI/a-Fe 2 O 3 composite had a high r DC even at F ¼ 0.4. We suspected that the dispersivity of the a-Fe 2 O 3 particles in the polymer matrix was different between  the PEI/a-Fe 2 O 3 and PAR/a-Fe 2 O 3 composites, and observed it using SEM. Fig. 7 shows SEM images of the PEI/a-Fe 2 O 3 and PAR/a-Fe 2 O 3 composites with F ¼ 0.20 and F ¼ 0.4. There is no great difference in the dispersivity of the a-Fe 2 O 3 particles, which means that the signicant difference in r DC is not due to the dispersivity of the a-Fe 2 O 3 particles. In the PEI/a-Fe 2 O 3 composite, the a-Fe 2 O 3 particles would be covered with a PEI layer at a nanometer scale, resulting in the high r DC for PEI owing to inhibition of the tunneling conduction of the free electron 39,43 between the a-Fe 2 O 3 particles.
Comparing of dielectric performance for PEI/a-Fe 2 O 3 and PEI/ BaTiO 3 composites As mentioned above, the PEI/a-Fe 2 O 3 composite exhibits the high 3 0 r and very low 3 00 r at the elevated temperature above 120 C, which might be promising as a low-cost dielectric material for high-temperature applications; PEI has a maximum operating temperature of 200 C. 35 Therefore, the novel PEI/a-Fe 2 O 3 composite was compared with a conventional PEI/BaTiO 3   composite in dielectric performance. As shown in Fig. 6, the r DC values of both the composites are comparable at room temperature when F is less than 0.4, whereas the r DC of PEI/a-Fe 2 O 3 is much lower than that of PEI/BaTiO 3 at the higher F. This is because the interparticle distance becomes small enough for tunneling conduction to occur owing to high loading of particles. Fig. 8 shows the f dependences of the 3 0 r and 3 00 r of the two types of composites obtained at 40 C and 160 C. At 40 C, although the 3 0 r of PEI/a-Fe 2 O 3 is higher than that of PEI/BaTiO 3 in the lower f range, the 3 00 r of PEI/a-Fe 2 O 3 is much larger than that of PEI/BaTiO 3 . At 160 C, however, 3 0 r has comparable 3 00 r values to PEI/BaTiO 3 except with F ¼ 0.5, and higher 3 0 r than PEI/BaTiO 3 . The dielectric characteristics at 1 kHz obtained at 160 C are plotted as a function of F in Fig. 9. These results show that the PEI/a-Fe 2 O 3 composite was comparable to the PEI/BaTiO 3 composite in dielectric performance at 160 C. The tting curves for the 3 0 r of the two types of composites are drawn in Fig. 9a in solid lines using Lichteneker's logarithmic mixing rule. These theoretical curves well t to the experimental results and give 3 0   consistent with some reported results; 3 0 r BaTiO 3 is around 100. 8,39

Conclusion
The dielectric properties of polymer/a-Fe 2 O 3 composites strongly depended not only on the temperatures but also on the kind of polymer matrices. For engineering plastics such as PEI, PSF, PAR and PES, the 3 0 r was highly enhanced at around 1 kHz by incorporation of a-Fe 2 O 3 particles owing to Maxwell-Wagner polarization of free electrons in the a-Fe 2 O 3 particles. This is probably because the p electrons in the aromatic structures of the engineering plastics strongly interact with the electrons in the a-Fe 2 O 3 particles. Furthermore, the 3 00 r for the engineering plastics became small at the elevated temperatures because the s of the a-Fe 2 O 3 particle was enhanced and therefore the relaxation frequency of Maxwell-Wagner polarization was shifted to higher f. PEI/a-Fe 2 O 3 composites exhibited highly improved dielectric properties at around 1 kHz, the high 3 0 r and very low 3 00 r at the elevated temperature above 120 C. It was demonstrated that the PEI/a-Fe 2 O 3 composite was comparable to the PEI/BaTiO 3 composite in dielectric performance at 160 C. Because the cost of a-Fe 2 O 3 is much lower than that of BaTiO 3 , the PEI/a-Fe 2 O 3 composites might be promising as a low-cost dielectric material for high-temperature applications.