Fabrication of Fe 3 O 4 @SiO 2 Nanoﬂuids with High Breakdown Voltage and Low Dielectric Loss

: Insulating oil modiﬁed by nanoparticle (often called nanoﬂuids) has recently drawn considerable attention, especially concerning the improvement of electrical breakdown and thermal conductivity of the nanoﬂuids. However, traditional insulating nanoﬂuid often tends to high dielectric loss, which accelerates the ageing of nanoﬂuids and limits its application in electrical equipment. In this paper, three core-shell Fe 3 O 4 @SiO 2 nanoparticles with di ﬀ erent SiO 2 shell thickness were prepared and subsequently dispersed into insulating oil to achieve nanoﬂuids. The dispersion stability, breakdown voltages and dielectric properties of these nanoﬂuids were comparatively investigated. Experimental results show the alternating current (AC) and positive lightning breakdown voltage of nanoﬂuids increased by 30.5% and 61%, respectively. Moreover, the SiO 2 shell thickness of Fe 3 O 4 @SiO 2 nanoparticle had signiﬁcant e ﬀ ects on the dielectric loss of nanoﬂuids. stretching vibration absorption at 1100 cm − 1 was found in this spectrum, and the wide absorption peak at 3485 cm − 1 represents the –OH group. The above results indicate that the oleic acid molecules were successfully bonded onto surfaces of the Fe 3 O 4 @SiO 2 nanoparticles. The surface modiﬁcation of Fe 3 O 4 @SiO 2 nanoparticles by oleic acid can improve the dispersion stability of nanoparticle in insulating oil.


Introduction
Adding well-dispersed nanoparticles into insulating oil can effectively improve the insulation properties and thermal conductivity of insulating oil [1,2]. The Fe 3 O 4 nanoparticles have been proven to improve the AC and lightning breakdown voltage of insulating oil [3], and the insulation properties of nanofluids are determined by nanoparticle size, surfactant and structure [4]. The mechanism of Fe 3 O 4 nanoparticle effect on the breakdown voltage was studied at a certain level. It was found that nanoparticles can inhibit space charge accumulation and uniform electric field in insulating oil [5]. Several studies have shown that nanoparticles increase the trapping density and depth, and reduce the velocity of streamer propagation in nanofluids [6,7].
Dielectric loss factor is a key parameter for insulating nanofluids. In fact, many high breakdown voltage nanofluids have been prepared but they tend to have high dielectric loss [8], especially for nanoparticles with high relative permittivity. For instance, the relative permittivity of Fe 3 O 4 nanoparticles is 80 [9], much higher than that of insulating oil, and this will lead to a significant increase in polarization loss of insulating oil. Therefore, how to prepare high breakdown voltage nanofluids with low dielectric loss factor is a very interesting topic.
At present, a lot of researches have been carried out on the synthesis of nanoparticles with different sizes and morphologies [10,11]. Grzelczak et al. shows the preparation methods of gold nanoparticles with different sizes [12], and it has been proven that the nanoparticle size and morphology have an important effect on the dielectric properties of materials [13,14]. However, it is difficult for a single nanoparticle to meet the requirements of increasing breakdown voltage and reducing dielectric loss of insulating oil. Core-shell structure nanoparticles show great potential for fabricating nanocomposites, because of their unique properties, such as high thermal conductivity, large surface area and special dielectric properties. Li et al. prepared insulating rPANI@rGO nanocomposites by an in situ polymerization method with high dielectric constant and low dielectric loss [15]. Grumezescu et al. reported Fe 3 O 4 -oleic acid-usnic acid core-shell-extra-shell nanofluids could have application for different medical devices [16]. However, most of the works ignored the influences of core-shell nanoparticle size on the dielectric properties of nanofluids, and the method of prepared nanofluids with high breakdown voltage and low dielectric loss still has never been reported.
This paper researches the effect of Fe 3 O 4 @SiO 2 nanoparticles on the insulation properties and dielectric loss of insulating oil. The insulating nanofluids were prepared by adding three different sizes of Fe 3 O 4 @SiO 2 nanoparticles; thereafter the breakdown voltages and dielectric properties of nanofluids were presented and discussed. The Fe 3 O 4 @SiO 2 nanofluids showed high insulation performance as well as low dielectric loss factor, which indicates significant application in the power industry.

Preparation of Nanofluids
The process of preparation of Fe 3 O 4 @SiO 2 nanoparticles is shown in Figure 1

Nanoparticle Characterization
The morphologies of the three different sized nanoparticles prepared by adding different amounts of TEOS were characterized by transmission electron microscopy (TEM, JEM-2100F, Japan Electronics Ltd, Tokyo, Japan), as shown from Figure 2

Nanoparticle Characterization
The morphologies of the three different sized nanoparticles prepared by adding different amounts of TEOS were characterized by transmission electron microscopy (TEM, JEM-2100F, Japan Electronics Ltd., Tokyo, Japan), as shown from Figure 2(A1-C1). The polydispersity of the three different sized nanoparticle were tested by Zeta potential size analyzer (MS-2000, Malvern Panalytical Ltd., Melvin, UK), and the results are shown in Figure 2(A2-C2). It is seen that the prepared nanoparticles are monodispersed spherical particles, and each nanoparticle is composed of two distinct regions. The darker central-core is the Fe 3 O 4 crystal, and the surrounding layer is low density shell of SiO 2 . The Fe 3 O 4 nanoparticles coated with SiO 2 can effectively avoid the agglomeration of the nanoparticle. It also can be seen that the size distributions of the three nanoparticles are narrow, indicating the three Fe 3 O 4 @SiO 2 are uniform nanoparticles. Meanwhile, the nanoparticle sizes tested by Zeta potential analyzer are basically the same as those observed by TEM.

Nanoparticle Characterization
The morphologies of the three different sized nanoparticles prepared by adding different amounts of TEOS were characterized by transmission electron microscopy (TEM, JEM-2100F, Japan Electronics Ltd, Tokyo, Japan), as shown from Figure 2(A1-C1). The polydispersity of the three different sized nanoparticle were tested by Zeta potential size analyzer (MS-2000, Malvern Panalytical Ltd, Melvin, UK), and the results are shown in Figure 2(A2-C2). It is seen that the prepared nanoparticles are monodispersed spherical particles, and each nanoparticle is composed of two distinct regions. The darker central-core is the Fe3O4 crystal, and the surrounding layer is low density shell of SiO2. The Fe3O4 nanoparticles coated with SiO2 can effectively avoid the agglomeration of the nanoparticle. It also can be seen that the size distributions of the three nanoparticles are narrow, indicating the three Fe3O4@SiO2 are uniform nanoparticles. Meanwhile, the nanoparticle sizes tested by Zeta potential analyzer are basically the same as those observed by TEM.
As the amounts of TEOS increase, the thickness of the SiO2 shells increase continuously. The SiO2 shells vary in thickness from ~7.5 nm (Figure 2(A1)) to ~50 nm ( Figure 2(C1)) via ~24 nm ( Figure  2(B1)). However, the sizes of the Fe3O4 cores do not change significantly, which are always ~60 nm. However, the sizes of the Fe 3 O 4 cores do not change significantly, which are always~60 nm. Figure 3 depicts the Fourier-transform infrared spectroscopy (FTIR, Nicolet, Thermo Electron Corporation, Franklin, TN, USA) spectra of the obtained nanoparticle samples. In the spectrum, the stretching vibration absorption of Si-O bands appears at 466, 801 cm −1 and the stretching vibration absorption of Fe-O emerges at 568 cm −1 in the spectrum. These absorption peaks prove that the Fe 3 O 4 @SiO 2 nanoparticle was synthesized. Furthermore, C-O stretching vibration absorption at 1100 cm −1 was found in this spectrum, and the wide absorption peak at 3485 cm −1 represents the -OH group. The above results indicate that the oleic acid molecules were successfully bonded onto surfaces of the Fe 3 O 4 @SiO 2 nanoparticles. The surface modification of Fe 3 O 4 @SiO 2 nanoparticles by oleic acid can improve the dispersion stability of nanoparticle in insulating oil.
absorption of Fe-O emerges at 568 cm −1 in the spectrum. These absorption peaks prove that the Fe3O4@SiO2 nanoparticle was synthesized. Furthermore, C-O stretching vibration absorption at 1100 cm −1 was found in this spectrum, and the wide absorption peak at 3485 cm −1 represents the -OH group. The above results indicate that the oleic acid molecules were successfully bonded onto surfaces of the Fe3O4@SiO2 nanoparticles. The surface modification of Fe3O4@SiO2 nanoparticles by oleic acid can improve the dispersion stability of nanoparticle in insulating oil.

Dispersion Stability of Nanofluids
The long-term dispersion stability of Fe3O4@SiO2 nanoparticles in insulating oil was characterized by natural deposition method. In Figure 4, three nanofluids with a much larger nanoparticle/oil weight ratio (0.1%) were set for 90 days in ambient condition to examine their storage-time dependent dispersion stability. The FR3 oil was also measured for comparison. It can be seen that there was no agglomeration and precipitation in the nanofluids, indicating that the nanofluids have good dispersion stability.

Dispersion Stability of Nanofluids
The long-term dispersion stability of Fe 3 O 4 @SiO 2 nanoparticles in insulating oil was characterized by natural deposition method. In Figure 4, three nanofluids with a much larger nanoparticle/oil weight ratio (0.1%) were set for 90 days in ambient condition to examine their storage-time dependent dispersion stability. The FR3 oil was also measured for comparison. It can be seen that there was no agglomeration and precipitation in the nanofluids, indicating that the nanofluids have good dispersion stability.

Breakdown Voltage of Nanofluids
The AC and lightning breakdown voltages of FR3 oil and three nanofluids were measured in accordance with IEC 60,156 and IEC 60,897 [17,18]. A ball-plate steel electrode was adopted for the AC breakdown voltage test and the electrode gap was 2.5 nm. The device for testing the lightning breakdown voltage is shown in Figure 5. This device was composed of a steel electrode and a container. The high voltage electrode was a steel needle, and the ground electrode was a 13 mmdiameter ball. The gap distance between high voltage electrode and grounding electrode was 15 mm. Standard lightning impulse of 1.2 (±30%)/50 µ s (±20%) with both negative and positive polarities were applied to all samples.

Breakdown Voltage of Nanofluids
The AC and lightning breakdown voltages of FR3 oil and three nanofluids were measured in accordance with IEC 60,156 and IEC 60,897 [17,18]. A ball-plate steel electrode was adopted for the AC breakdown voltage test and the electrode gap was 2.5 nm. The device for testing the lightning breakdown voltage is shown in Figure 5. This device was composed of a steel electrode and a container. The high voltage electrode was a steel needle, and the ground electrode was a 13 mm-diameter ball. The gap distance between high voltage electrode and grounding electrode was 15 mm. Standard lightning impulse of 1.2 (±30%)/50 µs (±20%) with both negative and positive polarities were applied to all samples.
FR3 oil was added to 0.1 wt % Fe3O4@SiO2 nanoparticles, and the nanoparticles were prepared with 2 mL of TEOS; (C) Sample B: 4 mL of TEOS; (D) Sample C: 8 mL of TEOS.

Breakdown Voltage of Nanofluids
The AC and lightning breakdown voltages of FR3 oil and three nanofluids were measured in accordance with IEC 60,156 and IEC 60,897 [17,18]. A ball-plate steel electrode was adopted for the AC breakdown voltage test and the electrode gap was 2.5 nm. The device for testing the lightning breakdown voltage is shown in Figure 5. This device was composed of a steel electrode and a container. The high voltage electrode was a steel needle, and the ground electrode was a 13 mmdiameter ball. The gap distance between high voltage electrode and grounding electrode was 15 mm. Standard lightning impulse of 1.2 (±30%)/50 µ s (±20%) with both negative and positive polarities were applied to all samples.    Figure 6 shows the measurement results of AC breakdown voltages of FR3 oil and three sizes of nanofluids at different nanoparticle contents. The FR3 oil was marked as the sample with 0 ppm Fe 3 O 4 @SiO 2 nanoparticles added. All the measurements were made on five oil samples. It was seen that the AC breakdown voltages of each nanofluid increases to a top value and decreases afterwards with higher nanoparticle content. For example, the AC breakdown voltage enhanced by 30.5% from 52.1 kV for FR3 to 68.0 kV for nanofluids C that was added with 100 ppm Fe 3 O 4 @SiO 2 nanoparticle. Fe3O4@SiO2 nanoparticles added. All the measurements were made on five oil samples. It was seen that the AC breakdown voltages of each nanofluid increases to a top value and decreases afterwards with higher nanoparticle content. For example, the AC breakdown voltage enhanced by 30.5% from 52.1 kV for FR3 to 68.0 kV for nanofluids C that was added with 100 ppm Fe3O4@SiO2 nanoparticle. The positive and negative lightning breakdown voltages of oil samples are summarized in Figures 7 and 8. The positive lightning breakdown voltage of nanofluids provides significant effects, but for negative lightning breakdown voltage the attained improvement is insignificant. Here the positive lightning leads to strongly increased breakdown voltages. For example, nanofluids A, which contained 300 ppm nanoparticles, shows the highest breakdown voltage of 68.5 kV, which is significantly higher voltage compared to the 42.4 kV for FR3 oil, improved by about 61%. However,  The positive and negative lightning breakdown voltages of oil samples are summarized in Figures 7  and 8. The positive lightning breakdown voltage of nanofluids provides significant effects, but for negative lightning breakdown voltage the attained improvement is insignificant. Here the positive lightning leads to strongly increased breakdown voltages. For example, nanofluids A, which contained 300 ppm nanoparticles, shows the highest breakdown voltage of 68.5 kV, which is significantly higher voltage compared to the 42.4 kV for FR3 oil, improved by about 61%. However, the negative breakdown voltage of nanofluids was not significantly increased. As the Fe 3 O 4 @SiO 2 nanoparticle content was 200 ppm, the breakdown voltage of nanofluids A was 6.9% higher than that of FR3 oil. This is mainly due to the different nanoparticle activities in oil under different polarity electric fields, leading to different impacts of the positive and negative lightning breakdown voltage of nanofluids [19]. The positive and negative lightning breakdown voltages of oil samples are summarized in Figures 7 and 8. The positive lightning breakdown voltage of nanofluids provides significant effects, but for negative lightning breakdown voltage the attained improvement is insignificant. Here the positive lightning leads to strongly increased breakdown voltages. For example, nanofluids A, which contained 300 ppm nanoparticles, shows the highest breakdown voltage of 68.5 kV, which is significantly higher voltage compared to the 42.4 kV for FR3 oil, improved by about 61%. However, the negative breakdown voltage of nanofluids was not significantly increased. As the Fe3O4@SiO2 nanoparticle content was 200 ppm, the breakdown voltage of nanofluids A was 6.9% higher than that of FR3 oil. This is mainly due to the different nanoparticle activities in oil under different polarity electric fields, leading to different impacts of the positive and negative lightning breakdown voltage of nanofluids [19].   Figure 9 shows the curves of relative permittivity of FR3 oil and three nanofluids between 10 −2 and 10 6 Hz. There is no visible difference among FR3 and three sizes of nanofluids at a frequency above 10 Hz. With frequency below 10 Hz, the relative permittivity of oil samples follows the sequence A > B > C > FR3. It is obvious that different thickness shells of Fe3O4@SiO2 nanoparticle endow nanofluids with varied relative permittivity values. As is well known, the relative permittivity of Fe3O4 is about 80 [9], which is much greater than that of 3.9 for SiO2 nanoparticle. According to the Maxwell-Garnett model [20], the relative permittivity of the Fe3O4@SiO2 nanoparticle, εn, can be calculated by following formula:  Figure 9 shows the curves of relative permittivity of FR3 oil and three nanofluids between 10 −2 and 10 6 Hz. There is no visible difference among FR3 and three sizes of nanofluids at a frequency above 10 Hz. With frequency below 10 Hz, the relative permittivity of oil samples follows the sequence A > B > C > FR3. It is obvious that different thickness shells of Fe 3 O 4 @SiO 2 nanoparticle endow nanofluids with varied relative permittivity values. As is well known, the relative permittivity of Fe 3 O 4 is about 80 [9], which is much greater than that of 3.9 for SiO 2 nanoparticle. According to the Maxwell-Garnett model [20], the relative permittivity of the Fe 3 O 4 @SiO 2 nanoparticle, ε n , can be calculated by following formula:

Dielectric Properties of Nanofluids
where ε 1 = 3.9 and ε 2 = 80, the relative dielectric constant of SiO 2 and Fe 3 O 4 nanoparticle, respectively, and ϕ v is the Fe 3 O 4 core volumetric concentration in the Fe 3 O 4 @SiO 2 nanoparticle. According to formula (1), the relative permittivity of Fe 3 O 4 @SiO 2 nanoparticles decreases rapidly with the increase of SiO 2 shell thickness. Figure 9 shows the curves of relative permittivity of FR3 oil and three nanofluids between 10 −2 and 10 6 Hz. There is no visible difference among FR3 and three sizes of nanofluids at a frequency above 10 Hz. With frequency below 10 Hz, the relative permittivity of oil samples follows the sequence A > B > C > FR3. It is obvious that different thickness shells of Fe3O4@SiO2 nanoparticle endow nanofluids with varied relative permittivity values. As is well known, the relative permittivity of Fe3O4 is about 80 [9], which is much greater than that of 3.9 for SiO2 nanoparticle. According to the Maxwell-Garnett model [20], the relative permittivity of the Fe3O4@SiO2 nanoparticle, εn, can be calculated by following formula:

Dielectric Properties of Nanofluids
where ε1 = 3.9 and ε2 = 80, the relative dielectric constant of SiO2 and Fe3O4 nanoparticle, respectively, and φv is the Fe3O4 core volumetric concentration in the Fe3O4@SiO2 nanoparticle. According to formula (1), the relative permittivity of Fe3O4@SiO2 nanoparticles decreases rapidly with the increase of SiO2 shell thickness.   Figure 10 gives the frequency dependence of dielectric loss factor for FR3 oil and three sizes of nanofluids. The curves of the dielectric loss factor show a decrease with increasing frequency and behave almost identically in the range of 1-10 7 Hz. However, the results present a difference between 10 −2 Hz and 1 Hz. It can also be seen that the dielectric loss factor of nanofluids decreases with the increase of SiO 2 shell thickness. At a frequency of 0.1 Hz, the dielectric loss factor of 3% for nanofluids A is significantly higher than 0.5% for FR3 oil. However, the dielectric loss factor of the nanofluids C was only slightly increased.  Figure 10 gives the frequency dependence of dielectric loss factor for FR3 oil and three sizes of nanofluids. The curves of the dielectric loss factor show a decrease with increasing frequency and behave almost identically in the range of 1-10 7 Hz. However, the results present a difference between 10 −2 Hz and 1 Hz. It can also be seen that the dielectric loss factor of nanofluids decreases with the increase of SiO2 shell thickness. At a frequency of 0.1 Hz, the dielectric loss factor of 3% for nanofluids A is significantly higher than 0.5% for FR3 oil. However, the dielectric loss factor of the nanofluids C was only slightly increased.

Conclusions
• In this work, core-shell Fe3O4@SiO2 nanoparticles were synthesized. The TEM test showed the SiO2 shell thickness increased with the increase of TEOS concentration. Zeta potential size test showed the three Fe3O4@SiO2 were uniform nanoparticles.

Conclusions
• In this work, core-shell Fe 3 O 4 @SiO 2 nanoparticles were synthesized. The TEM test showed the SiO 2 shell thickness increased with the increase of TEOS concentration. Zeta potential size test showed the three Fe 3 O 4 @SiO 2 were uniform nanoparticles.

•
The nanofluids with high breakdown voltage and low dielectric loss were developed by dispersing Fe 3 O 4 @SiO 2 nanoparticles with different thicknesses of SiO 2 shell. The long-term dispersion stability of Fe 3 O 4 @SiO 2 nanoparticles in insulating oil was characterized by natural deposition method. The AC and positive lightning breakdown voltage of nanofluids were significantly improved compared with that of FR3 oil, but the improvement is not obvious for the negative lightning breakdown voltage.

•
As the thickness of the SiO 2 shell increases, the relative dielectric constant and the dielectric loss factor decrease. When the SiO 2 shell is 50 nm, the dielectric loss factor of nanofluids is basically the same as that of FR3 oil. This work will be beneficial to the application of nanofluids in transformers.