Introduction

The increasing integration of electronic devices into our daily lives has brought significant benefits. However, it has also posed new challenges, particularly concerning the issue of heat management within these devices. As electronic equipment becomes more compact and powerful, the heat generated within them increases, leading to potential malfunctions and failures of electronic parts and semiconductors [1, 2]. A thermally conductive material with efficient heat dissipation capabilities is needed to address this challenge. Simultaneously, the material must be insulated, especially when developing electric devices. This combination of characteristics would allow for effective heat management while ensuring the safety and functionality of the electrical components [3]. Polymer nanocomposites are an excellent example of effective heat management because they have thermal conductivity and electrical insulation properties [4].

Polymer nanocomposites are composite materials with a polymer matrix (a continuous phase) and nanoscale fillers (nanofillers or nanoparticles) dispersed within the matrix. The bulk polymer has very low thermal conductivity (0.01–0.1 W/(m·K)), but adding a certain amount of inorganic filler will increase the thermal conductivity of the polymeric composite [5, 6]. Metals have high thermal conductivity (100–2000 W/(m·K)) and excellent mechanical strength [7], but they also possess electric conductive properties. Some inorganic materials combine electrical insulation with thermal conductivity, such as alumina [8], beryllium oxide (BeO) [9, 10], magnesium oxide (MgO) [11], silicon nitride (Si3N4) [12, 13], hexagonal boron nitride (h-BN) [14,15,16], silicon carbide (SiC) [17], aluminum nitride (AlN) [18, 19] and so on.

AlN is an attractive inorganic filler for composites due to its exceptional combination of properties, such as thermal conductivity, electrical insulation properties, high melting point, wide bandgap, transparency, piezoelectric properties, low thermal expansion coefficient, and so on [20, 21]. Many research reports have been on compounding resins with insulating and highly thermally conductive AlN particles as fillers.

Choi and Kim designed alumina and AlN hybrid filler composites mixing large-sized (10 µm) AlN with small-sized alumina (0.5 µm) particles and large alumina (10 µm) with small-sized AlN (0.5 µm). The maximum thermal conductivities in the two systems were 3.402 W/(m·K) and 2.842 W/(m·K), respectively, at a total filler content of 58.4 vol.% when the volume ratio of large particles to small particles was 7:3 [22]. Guanglei et al. produced a composite sheet by blending polystyrene (PS) with AlN particles using a powder processing technique. Their investigation revealed that the thermal conductivity of the AlN/PS composite (0.489 W/(m·K)) surpassed that of pure PS (0.189 W/(m·K)) when the AlN content was 25 wt% [23]. Wang et al. produced a nanocapsule phase change material/octadecane/porous aluminum nitride particle (NPCM/ODE/PAlNP) hybrid composite by adsorbing ODE into the PAlNP. The composite NPCM/ODE/PAlNP showed that the AlN efficiently enhanced thermal conductivity by 3.5 (W/(m·K)), which was around 17.5 times of pure ODE [24]. Wei et al. fabricated epoxy (EP)/aluminum nitride honeycomb (AlN-H) composite materials by EP with AlN-H reinforcements. The process involved freeze-casting the AlN-H reinforcements and then infiltrating them with EP. The resulting composites demonstrated good thermal conductivity, reaching a maximum value of 9.48 W/(m·K). This remarkable thermal conductivity was explicitly observed when measuring in the directions aligned with the AlN-H channels. The composite achieved this high thermal conductivity at an AlN loading of 47.26 vol%, signifying that nearly half of its volume comprised the AlN-H structures [25]. Yunsheng et al. created a composite sheet combining polyvinylidene fluoride with AlN whiskers and particles. They researched this composite and achieved a notable enhancement in thermal conductivity, reaching 11.50 W/(m·K). This improvement was observed when the filler volume fraction was about 60%, and the ratio of whiskers to particles was 1:25.7 [6]. Additionally, about 50 vol% of particulate thermally conductive filler must be compounded to form a thermally conductive path in the resin. At a filler content of 50 vol%, resin's advantages, such as flexibility, light weight, flexibility, and adhesiveness, are lost [7, 26]. Additionally, About 50 vol% filler contents make an interface between the particles and create an obstacle to heat conduction even when the paths of the thermally conductive particles are formed in the composite [27].

We propose using several hundred nanometers in diameter inorganic nanofibers as a filler in the composites instead of inorganic particle fillers. By doing so, we considered that the heat conduction pass would be quickly introduced into the composites with a low filler contents because the ratio of length and diameter of nanofiber is, in theory, infinity. Our previous studies clarified that nanofiber shape helps create high thermal conductive composites [28, 29]. Inorganic nanofibers were formed using electrospinning. For example, the polyurethane sheet containing aligned α-alumina nanofibers, 33 vol%, showed a thermal conductivity of 14.6 W/(m·K) for the aligned nanofiber direction in the sheet. The inherent thermal conductivity of α-alumina is about 30 W/(m·K). This value is lower than that of previously made AlN nanoparticles at 50 vol% [30,31,32]. To further increase the thermal conductivity of the sheet containing inorganic nanofibers, we have tried to form AlN nanofibers and use them as fillers. The inherent thermal conductivity of AlN is high, 70 to 270 W/(m·K), but the thermal conductivity of the composites containing AlN particles has not fully used the property of AlN due to the interface between them as mentioned above.

Experimental

Preparation of aluminum nitride nanofibers through polyvinyl alcohol/boehmite composite precursor nanofibers using electrospinning technique

Boehmite (AlO(OH)·H2O) nanoparticles (Disperal P-2, dispersed particle size: 25 nm) were kindly gifted from Sasol Japan KK. Polyvinyl alcohol (PVA) (degree of polymerization: 1500, Wako Pure Chemicals Ind., Ltd., Japan) 10 wt% aqueous solution was prepared. Boehmite nanoparticles were dispersed into the PVA solution (spinning solution). The spinning solution was loaded into a plastic syringe (2 ml) with a needle. The solution extrusion rate was 1.0 ml/h. The metal rotating drum (diameter: 150 mm, spinning speed: 4000 rpm) was used as a collector, which was grounded, and the distance between the tip of the needle and the collector was 150 mm. A voltage of 20.0 kV was applied to the needle, and then the PVA/boehmite (PVA/alumina = 70/30 wt%) composite nanofibers were deposited on a collector as a nonwoven fabric. The PVA/boehmite composite nanofibers were used as a precursor of AlN nanofibers. The precursors obtained were heated up to a given temperature (1200 to 1500°C) in an alumina tube furnace (AHRF-30KC-9P, Asahi-rika Co. Ltd., Japan) under nitrogen gas flow, and they transformed into AlN nanofiber. The precursors were sandwiched with alumina plates to keep the flat shape of the precursor mat after heating when they were set in the tube furnace. The heating speed was 10°C/min, and the heating time at the maximum temperature was 2 to 20 h. The flow rate of nitrogen gas was 3.5 L/min. To remove excess amorphous carbon from the resultants, they were heated at 600°C for 3 h in the air during the cooling process. AlN did not change to alumina by this heating in the air.

Formation of polyvinyl butyral sheets containing aluminum nitride nanofibers

Polyvinyl butyral (PVB, Mowital B68/1SF) with the degree of polymerization = 750, degree of acetalization = ca. 70 mol%, and plasticizer [triethylene glycol bis (2-ethyl hexanoate)] = ca. 50 phr was kindly gifted from Kuraray Co. Ltd, Japan. PVB sheets (ca. 100 µm in thickness) containing AlN nanofibers were prepared as below. PVB was dissolved in ethanol as the content of the polymer varied from 5 wt.%. The PVB solutions were dropped on the AlN nanofibers mat and placed under a vacuum for 10 h. Silica (ethyl silicate) was used as a sintering additive to enhance AlN crystallinity. Ethyl silicate (Silicate 40, Tama Chemicals Co., Ltd., Japan) was used as a starting silica reagent. Ethyl silicate was added to the spinning solution to the ratio of alumina/silica = 95/5 wt%. The whole specific steps were illustrated schematically in Fig. 1.

Figure 1
figure 1

Experimental scheme for the preparation of PVB/AlN nanofibers heat-dissipating sheet.

Full size image

Characterization

The structure of the nanofibers was observed with a scanning electron microscope (SEM) (Keyence VE-9800, Japan). The average fiber diameters of each nanofiber (N = 100) were calaulated by using Adobe Photoshop CS3 extension software from the SEM images. Thermogravimetric (TG) analysis was performed in the air to estimate the AlN content in the PVB sheet at a heating rate of 10°C/min (Shimadzu DTG-60, Japan). X-ray diffraction (XRD) curves were taken using the Ni-filtered CuKα radiation (30 kV, 15 mA) (Rigaku Miniflex II, Japan). The Scherrer's equation also compared the average crystallite size (D). The Eq. (1) can be denoted as:

$$D \, = k \, \lambda /\left( {\beta \cos \theta } \right)$$
(1)

where K is a constant dimensionless shape factor (0.94). λ is the X-ray wavelength (0.15418 nm) used in the experiment. β is the full-width at half-maximum (FWHM) of a diffraction peak, measured in radians. θ is the Bragg angle at which the diffraction peak (°) is observed. The specific surface areas of the nanofibers were calculated from the nitrogen adsorption isotherms (−196°C) measured with Belsorp-mini II, BEL Japan Inc., Japan. Thermal diffusivity (m2/s) was measured in the planar and thickness directions with a Thermowave Analyzer TA35 (Bethel Co., Ltd., Japan) at room temperature. This system uses temperature wave analysis to measure thermal diffusivity. The average and relative standard deviations of 4 different measurements were calculated. The sample size was about 100 mm2. The thermal conductivity of the composite was calculated by using the density and heat capacity of AlN 3300 kg/m3 and 725 J/(kg·K), respectively, and for PVB 1100 kg/m3 and 1960 J/(kg·K), respectively.

Results and discussion

Preparation of aluminium nitride nanofibers obtained from polyvinyl alcohol/boehmite precursors

Firstly, we examined the effect of the PVA/alumina ratio of precursors on AlN productivity. Figure 2 shows XRD curves of precursors (PVA/alumina = 50/50 wt% and 70/30 wt%) after heating at 1500°C. AlN peaks can be seen in both curves, but the crystallinity of AlN obtained from the precursor (PVA/alumina = 70/30 wt%) is much higher than AlN obtained from the precursor (PVA/alumina = 50/50 wt%). The nitridation will typically proceed in the presence of carbon as follows [33,34,35]:

$${\text{Al}}_{2} {\text{O}}_{3} + \, 3{\text{C }} + {\text{ N}}_{2} \to \, 2{\text{AIN }} + \, 3{\text{CO}}$$
(1)
Figure 2
figure 2

XRD curves of resultant nanofibers obtained by 1500°C (a) PVA:alumina = 50:50wt.% and (b) PVA:alumina = 70:30wt.%.

Full size image

According to this scheme, the amount of carbon atom in the precursor (PVA/alumina = 50/50 wt%) is insufficient to proceed with the reaction formula (1), and the peaks of α-alumina can be seen in the curve. Thus, we selected the precursor (PVA/alumina = 70/30 wt%) to form AlN nanofibers in this experiment.

The reaction's Gibbs free energy change, ΔG°, was given using the available thermodynamic data [36,37,38]. If the ΔG° < 0, the reaction will proceed. The initial nitridation temperature calculated from thermodynamics data is 1664°C for AlN. This temperature is higher than the observed temperature, 1500°C (as shown in Fig. 2), indicating that the precursor nanofiber is a nanoscale mixture of metal oxide and carbon sources, and thus the reaction formula (1) would proceed effectively, or another reaction route might be created.

Figure 3 displays the SEM images and histogram of PVA/boehmite precursor, γ-alumina heated at 500°C, α-alumina heated at 1200°C and AlN nanofibers heated at 1500°C. The electrospinning technique created the precursor nanofibers at a mass ratio of 70/30 wt% with an average fiber diameter of 277 nm, shown in Fig. 3a. After calcinating at 500°C, PVA/boehmite precursor was converted into γ-alumina, and the average diameter creased to 284 nm. During calcination, applying heat can induce physical changes in the nanofibers. At 500°C, the thermal energy causes the fibers to expand or undergo sintering, where neighboring fibers fuse. These processes can lead to an increase in fiber diameter as the individual fibers merge or coalesce. The heating process can also cause some fibers to grow or elongate, contributing to the larger average diameter (Fig. 3b). PVA/boehmite precursor nanofibers were calcined at 1200°C, producing α-alumina with an average diameter of 229 nm. Calcinating at 1200°C may have resulted in more extensive sintering and grain growth, leading to smaller fiber diameters (Fig. 3c). In the presence of nitrogen gas and high calcination temperatures of 1500°C, PVA/boehmite precursor was converted to AlN nanofibers. The fiber diameters are mainly distributed from 100 to 200 nm. The average diameter of AlN nanofibers became 156 nm under the influence of sintering (Fig. 3d).

Figure 3
figure 3

SEM images and histogram of a PVA/boehmite precursor nanofibers, and after calcinating at different temperatures of precursor nanofibers, b γ-alumina nanofibers at 500°C, c α-alumina nanofibers at 1200°C, and d AlN nanofibers at 1500°C.

Full size image

In this comprehensive study, we have undertaken XRD analyses at varying heating durations, precisely at 2 h, 4 h, 10 h, and 20 h, to delve into the dynamic changes in the crystalline properties of AlN nanofibers, as visually depicted in Fig. 4. To measure the average crystallite size (D) Scherrer's equation was used. Our XRD data has unveiled an intriguing and highly significant trend in the evolution of the material's crystallinity in response to varying heating times. The XRD spectra vividly illustrate distinct crystallite sizes, each corresponding to the crystalline phases of the AlN nanofibers, with measured values of 9.0 nm, 10.3 nm, 10.7 nm, and 13.8 nm, respectively. At the outset, with just 2 h of heating, the recorded crystallite size registered at 9.0 nm, indicating an initial level of crystalline order within the AlN nanofibers. As we extended the heating duration to 4 h, we observed a subtle yet discernible increase in the crystallite size to 10.3 nm, signifying a modest improvement in crystallinity. However, the most remarkable transformation was honored at the 10 h mark, where the crystallite size substantially expanded to 10.7 nm. This observation unequivocally reflects a significant enhancement in the crystal structure. Upon reaching the 20 h mark, the crystallite size had surged to 13.8 nm, marking a remarkable crystallinity improvement. These striking changes in crystallite sizes, critical indicators of crystallite dimensions, emphasize the progressive evolution of AlN nanofibers' crystallinity with increasing heating duration.

Figure 4
figure 4

XRD curves of AlN nanofibers obtained by each sintering time (a) 2 h, (b) 4 h, (c) 10 h, and (d) 20 h at 1500°C temperature.

Full size image

In this study, we employed silica as a sintering additive in conjunction with AlN nanofibers and conducted XRD analyses at 1500°C for two distinct sintering durations, 10 and 20 h. The XRD curve presented in Fig. 5 reveals significant findings. Specifically, pure AlN nanofibers displayed a progressive improvement in crystallinity with extended sintering, evident from crystallite sizes at 10.7 nm for the 10 h sintering and 13.8 nm for the 20 h sintering. In contrast, AlN/silica nanofibers exhibited distinct crystallite sizes at 21.7 nm and 22.6 nm for the exact sintering durations, showcasing substantially higher AlN crystallinity. This remarkable enhancement is attributed to liquid phase sintering facilitated by forming a glassy AlN/silica phase during heating. These findings underscore the critical role of sintering additives in enhancing material properties and present exciting avenues for controlled synthesis of advanced nanofiber structures with superior crystalline quality.

Figure 5
figure 5

XRD curves of AlN and AlN/silica nanofibers at 1500°C temperature (a) AlN nanofiber, 10 h sintering time, (b) AlN nanofiber, 20 h sintering time, (c) AlN/silica nanofiber, 10 h sintering time, and (d) AlN/silica nanofibers, 20 h sintering time.

Full size image

Figure 6 shows the effect of heating time at 1500°C on the specific surface area of AlN and AlN/silica nanofibers obtained. As for AlN nanofibers, the values of the surface area of AlN nanofibers heated for 2 h was 35.5 m2/g, showing that porous AlN nanofibers were obtained due to the thermal decomposition of PVA. It decreases with increasing heating time due to sintering, but the specific surface area at 20 h still shows a high specific surface area, 22.1 m2/g. No drastic change in the AlN/silica nanofiber's surface areas indicates that silica addition does not affect it.

Figure 6
figure 6

Effect of heat treatment time on AlN/silica nanofibers of specific surface area (temperature, 1500°C)

Full size image

Formation of polyvinyl butyral sheets containing aluminum nitride nanofibers

Figure 7 shows an overview of AlN nanofibers and PVB/AlN nanofibers sheet. Here it is seen that there is no significant difference between them. However, the SEM image of the PVB/AlN nanofibers sheet is more compact than AlN nanofibers.

Figure 7
figure 7

Overviews of a AlN nanofibers mat, b PVB/AlN nanofibers heat dissipation sheet, and SEM images of c AlN nanofibers mat, d PVB/AlN nanofibers heat dissipation sheet.

Full size image

The sheets containing AlN 47, 52, and 54 vol% showed good thermal conductivity, 22, 21, and 19 W/(m·K), for aligned nanofiber direction in the sheet. These values are quietly higher than sheets containing α-alumina nanofibers (10.1–14.6 W/(m·K)) [32]. However, the thermal conductivity in the thickness direction is very low (0.1–2.1 W/(m·K)) which is shown in Fig. 8.

Figure 8
figure 8

Effect of sintering time on thermal conductivities of PVB sheets containing aligned AlN nanofibers at 1500°C temperature.

Full size image

Figure 9 shows the thermal conductivity for the aligned nanofiber direction of each sheet prepared with AlN/silica nanofibers obtained by heating at 1500°C for 2 to 20 h in nitrogen gas flow. The thermal conductivity becomes higher by increasing the heating time. Significantly, the sheet containing 35 vol% of AlN/silica nanofibers obtained by heating for 10 h shows high thermal conductivity, 25.3 W/(m·K). It indicates we succeeded in developing a high thermal conductive sheet with low filler content by using silica (ethyl silicate) as a sintering additive of AlN. The sheet shows 39.2 W/(m·K) at 57 vol% of AlN/silica nanofibers. We measured the sheet's thermal conductivity containing aligned AlN/silica nanofibers in thickness directions. Meanwhile, thermal conductivity in the thickness direction was very low (0.1–1.2 W/(m·K)) because there are few AlN nanofibers heat-transfer pathways in the thickness direction. It is almost the same compared to our previous research, which used the sheet containing α-alumina nanofibers (0.3–2.1 W/(m·K)). However, excellent thermal conductivity in the plane direction compared to the sheet containing α-alumina nanofibers (10.1–14.6 W/(m·K)) [32]. This is because AlN has higher thermal conductivity than alumina. The surface resistance of the sheets containing AlN nanofibers (54 vol%) and AlN/silica nanofibers (57 vol%) showed excellent insulating properties, less than 1.0 × 1012 Ω/cm2.

Figure 9
figure 9

Effect of sintering time on thermal conductivities of PVB sheets containing aligned AlN/silica nanofibers at 1500°C temperature.

Full size image

Conclusion

This study prepared AlN nanofibers by electrospinning technique using PVA solution dispersed boehmite nanoparticles, followed by high-temperature calcination in nitrogen gas flow. The AlN nanofibers mats were then impregnated with PVB solution. The obtained composite sheet containing 47 vol.% of the aligned AlN nanofibers had high thermal conductivity (22 W/(m·K)) in the plane direction. The thermal conductivity of the sheet increased proportionately to the AlN nanofiber contents. Our findings indicate that the elongated shape of AlN nanofibers serves as an effective filler, creating linear thermal pathways in the resin. Adding silica in the precursor is helpful to increase the crystallinity of AlN; thereby, the sheet using AlN/silica nanofibers showed excellent thermal conductivity above 30 W/(m·K). The thermal conductivity would be increased by increasing the AlN crystallinity and purity and decreasing the AlN nanofiber's porosity by adjusting the precursor's heating condition. In addition, the composite sheet showed anisotropic thermal conductivity derived from the fiber direction and electrical insulating property (below 1.0 × 1012 Ω/cm2). This electrically insulating thermally conductive sheet would help design an effective heat-removal path to safeguard heat-sensitive parts in electrics device.