Fabrication of high-surface-area, SiO2 supported polyimide carbon aerogel microspheres: electrochemical application

A series of polyimide (PI)/SiO2 aerogel microspheres were prepared by using polyamide acid salt and hydrolyzed tetraethyl orthosilicate based on the reverse-phase emulsion method. Then, PI/SiO2 aerogel microspheres were carbonized and etched to obtain carbon aerogel microspheres (CAMs). Scanning electron microscope, transmission electron microscope and nitrogen isothermal adsorption were used to characterize the micro-morphology and pore structure of the microspheres; and electrochemical workstation was used to test the electrochemical performance of the CAMs. The results showed that CAMs with different pore structures and specific surface area were obtained by adjusting the content of SiO2. Highest specific surface area of 1166.9 m2 g−1 and a total pore volume of 1.2369 cm3 g−1 were achieved at a SiO2 content of 50%. When used as the electrode materials for supercapacitors, these CAMs demonstrated a maximum specific capacitance of 125.1 F g−1 in a three-electrode system and a maximum capacitance of 53.3% at 30 A g−1. This article provides a new strategy for the preparation of CAMs with high specific surface area by using linear PI precursor and SiO2 support skeleton.


Introduction
Porous carbon materials have been extensively studied for gas storage and separation, supercapacitors, catalysis, catalyst support, and other applications. It has been shown that the morphology and size of carbon can influence its function. For example, the combined advantages of porosity and colloidal particles can provide faster molecular diffusion [1][2][3][4]. General, chemical or physical strategies are used to increase the surface area and porosity of carbonized organic precursors, this process is referred to as 'activation' . Physical activation is usually carried out at high temperatures under oxidizing gases, such as steam, carbon dioxide, air, or a mixture of gases. In contrast, the chemical activation process can be carried out at lower temperatures in the presence of certain dehydration reagents, such as phosphoric acid and basic hydroxides [5][6][7].
Recently, carbon aerogel microsphere (CAM) as a porous carbon material with high porosity and high specific surface area has attracted researchers' attention because of its regular spherical shape and adjustable size. CAM has better liquidity and higher taping density than carbon powder with irregular shape which is usually prepared by the crushing of carbon aerogel bulk [8,9]. Because of its high surface area (or porosity), rich regulatable multi-stage pore structure, and good electrical conductivity, CAM has enormous potential in the field of hydrogen storage [10], adsorption [11], batteries [12], and capacitors [13,14]. Especially in the field of energy storage, CAM is a promising electrode material because of its regular spherical structure, good dispersibility and adjustable microsphere size [14]. CAM can be derived from the carbonized organic precursors; these precursors can be prepared from polymer aerogel such as resorcinol formaldehyde (RF) aerogel. Li et al [15] used phenolic resin to prepare CAMs with high specific surface area and high porosity, and used different carbonization temperatures to regulate the pore structure of CAMs. Although many literatures have conducted related research, the lack of nitrogen heteroatoms for conduction and ion adsorption limits the application of RF aerogel microspheres as the precursors of CAM for electrode material. Since nitrogen doping can adjust the energy band structure of the metal-semiconductor transition, and the nitrogen-doped carbon structure can provide better electrochemical performance [4,16,17]. As the electronegativity of N (3.04) is higher than that of carbon (2.55), N can better adsorb electrolyte ions. In addition, nitrogen doping can inhibit the formation of hydrocarbons, which can form dielectric dead layers that negatively affect the performance of supercapacitor [18]. Well-bound nitrogen atoms can increase electronic conductivity, provide additional active sites and enhance the interaction between carbon structure and ions [19,20]. Polyimide (PI) aerogels are of interest as high-performance polymers due to their low density, heat resistance, low dielectric constant and low thermal conductivity. The N atoms on the PI molecular skeleton form in-situ N doping in carbonized PI aerogel microspheres, which makes PI aerogel microspheres an ideal precursor of CAMs for electrochemical performance [21].
PI aerogel microspheres can be prepared by template method [22], precipitation method, emulsion method, etc. Microspheres prepared by the template method are capable of maintaining a good spherical shape and size distribution. For example, Groenewolt et al [22] prepared PI nanoparticles with a size of 13 nm using mesoporous SiO 2 as a template. However, this method is still limited by the size and shape of the template. The emulsion method is a preparation strategy using a polymer solution as a dispersed phase and a solvent as a continuous phase. This strategy follows a general sol-gel strategy in which a sol is added to an emulsion system and dispersed into emulsion microspheres, which are then gelled to prepare organic aerogel microspheres. Lu et al [23] prepared PI microspheres with high solid content by the inverse emulsion method. The results showed that the average diameter decreased from 33.9 µm to 25.6 µm and the size distribution became narrower when the amount of emulsifier was increased. Gu et al [24] reported non-aqueous emulsion micron-sized aerogel microspheres using an oil-in-oil emulsion method. They used dimethylformamide as the dispersed phase and cyclohexane as the continuous phase to prepare PI aerogel microspheres. However, the PI microspheres cannot maintain its spherical shape because of its low skeleton strength. As a result, there are few reports on the preparation of CAM with PI, because it is difficult to prepare PI aerogel microspheres by general emulsion method, which requires PI bearing rigid molecular chains to dissolve in water.
In a previous study, we reported that PI microspheres can be prepared by reversed-phase emulsification using water-soluble polyamide acid salt (PAAs) precursors. Furthermore, PI/SiO 2 microspheres with a SiO 2 content of 30% were prepared in situ by the sol-gel process, and the obtained SiO 2 would not aggregate and form nanoparticles [25]. Herein, we adopt the above-mentioned in-situ sol-gel idea and introduce a larger SiO 2 content to support PI microspheres during the carbonization process. CAMs with high specific surface area and high porosity are obtained after carbonization and etching. In this work, hydrolyzed tetraethyl orthosilicate (TEOS) was used to form an in-situ growth SiO 2 network to assist in the gelation of PI and to form PI/SiO 2 composite aerogel microspheres in inverse emulsion systems. The pore structures of aerogel microspheres with different SiO 2 content at each preparation stage were studied.

Preparation of PAAs
Under nitrogen, 2.0024 g ODA and 47 ml NMP were added to a three-neck flask, and mixed, and then 2.230 g PMDA was added in three portions. After the reaction was carried out for 6 h, 2.81 ml TEA was added to form PAAs. Then, PAAs powder was obtained after precipitation, solvent replacement, and drying in a vacuum oven at 45 • C for 10 h.

Preparation of PI/SiO 2 aerogel microspheres
First, 1.05 g PAAs, 28.95 g deionized water, and 0.21 ml TEA were added into the beaker, the mixed solution was magnetically stirred at room temperature until the solution is completely transparent. Hydrolyzed TEOS, prepared through the reaction of 3.64 g TEOS and 1.26 ml hydrochloric acid solution (pH = 2.00) for 4 h, was then added into the PAAs solution. 30 ml of the above solution was added to a three-necked flask containing 300 ml liquid paraffin and 30 ml span-80, stirred at 400 rpm for 2 h, and then 11.6 ml acetic anhydride and 0.58 ml pyridine were added to induce chemical imidization. The above product was washed with acetone and cyclohexane and then freeze-dried to obtain the desired PAAs/SiO 2 -50 aerogel microspheres. 50 means the 50% weight ratio of SiO 2 . By changing the amount of TEOS, the amount of SiO 2 can be adjusted. For PAAs/SiO 2 -37.5, PAAs/SiO 2 -62.5, and PAAs/SiO 2 -75, the amounts of TEOS and hydrochloric acid solutions added were 2.18 g and 0.75 ml, 6.07 g, and 2.10 ml, and 10.92 g and 3.77 ml, respectively. The PI/SiO 2 aerogel microspheres were vacuum heat treated at 80, 150, 250, and 350 • C for 1 h, 2 h, 1 h, and 0.5 h to complete imidization.

Preparation of CAMs
The following carbonization of PI/SiO 2 aerogel microspheres was achieved by using a muffle furnace. The PI/SiO 2 aerogel microspheres were heat treated at 1000 • C for 1 h at a heating rate of 5 • C min −1 under nitrogen condition. After heat treatment at 80 • C for 8 h, the CAMs were prepared by etching with HF acid and washed by deionized water.

Characterization
Scanning electron microscope (SEM) and transmission electron microscope (TEM) were used to observe the microstructure of aerogel microspheres. Energy dispersive spectrometer (EDS) was used to analyze the distribution of element before and after etching. X-ray photoelectron spectroscopy (XPS) was used to analyze the element content and configuration. Nitrogen isothermal adsorption was used to analyze the pore structure of microspheres. Brunauer-Emmett-Teller method was carried out to calculate the specific surface area, and the non-local density functional theory (NLDFT) model was used to calculate the pore size distribution. The electrochemical measurements were tested using an electrochemical work-station. The test was under a three-electrode system with 6 mol l −1 KOH electrolyte, the thickness of each electrode is about 70 µm with a about 7 mg cm −2 mass loading. Cyclic voltammetry (CV) curves were obtained under scan rate from 20 to 500 mV s −1 . Galvanostatic charge discharge (GCD) curves were tested ranging from 0.5 to 30 A g −1 . Electrochemical impedance spectroscopy (EIS) was tested with frequencies from 10 −2 to 10 5 Hz combined with an amplitude of 5 mV.

Result and discussion
The synthesis of PI follows a two-step strategy, as shown in figure 1(a). First, ODA and PMDA undergo polycondensation to form PAA. Then, PAA can react with TEA to form PAAs, which can be further converted to PI through thermal imidization. This two-step strategy facilitates the formation of water-soluble PAAs. The water solubility of PAAs makes it possible to prepare PI aerogel microspheres by the inverse emulsion method. The whole fabrication process is shown in figure 1(c). The water-soluble PAAs and the hydrolyzed TEOS are mixed to form a dispersed aqueous phase of the inverse emulsion system; while liquid paraffin forms the continuous oil phase. TEOS is hydrolyzed and condensed to form SiO 2 to gel the microspheres. After the gelation, the PAAs are imidized, washed with acetone and cyclohexane, and the solvent is removed by freeze-drying to form aerogel microspheres. Then CAMs are produced by carbonization and etching. The overall preparation process is still a sol-gel strategy. In the hole fabrication process, in-situ growth of SiO 2 makes itself form nanoparticles, which helps the gelation of PI aerogel microspheres at high SiO 2 content. Besides, SiO 2 at a high content acts as structural support in gelation and carbonization. The removal of SiO 2 after etching leads to high specific surface area and high pore volume.
As evident from figures 2 (a) and S1, the microspheres prepared by the inverse emulsion method possess a spherical structure with relatively narrow particle size distribution and average diameter of about 20 µm, and the size of the microspheres does not change significantly with the increase of SiO 2 content. The factors affecting the size of the microspheres in the inverse emulsion method mainly include the emulsifier content and the stirring speed. It can be seen from figure S1 that the particle size of the microspheres is less uniform at low SiO 2 content, while the adhesion of the microspheres occurs at high SiO 2 content, and the PI/SiO 2 aerogel microspheres exhibit better sphericity at an intermediate SiO 2 content. As can be seen from the SEM images (figures S1(e)-(h)), the surface of the microspheres has a porous structure, which is mainly formed via the volatilization of the solvent and the detachment of the surface SiO 2 particles. Due to the lack of strong interaction between SiO 2 and PI, the PI phase shrinks during freeze-drying and imidization, and the surface SiO 2 falls off. As the SiO 2 content increases, the SiO 2 particles on the surface of the microspheres increase, which form richer pores after detachment.  The C/SiO 2 composite microspheres are obtained after carbonization at 1000 • C. C/SiO 2 -50 remains a good spherical shape after carbonization (figures 2(b) and S2(b)), while microspheres with low or high SiO 2 content show some breakage (figures S2(a), (c) and (d)). At low SiO 2 content, due to the lack of the support of the SiO 2 gel network, the C/SiO 2 aerogel microspheres are deformed and destroyed during carbonization ( figure S2(a)). At high SiO 2 content, the PI phase becomes discontinuous and shrinks during carbonization, while SiO 2 network remains unchanged, which leads to the fracture of microspheres (figures S2(c) and (d)). At 50% SiO 2 content, PI can form a continuous phase and be supported by SiO 2 network, which allows C/SiO 2 -50 to remain a spherical structure during carbonization. From the SEM images (figures S2(e)-(h)), the pores on the surface of C/SiO 2 microspheres gradually become smaller with the decrease of PI content. The high PI content increases the shrinkage of the microspheres, thereby forming larger pores.
The SEM images of C aerogel microspheres are shown in figures 2(c)-(f) and S3. C-50 still maintains a good spherical shape, indicating that the carbon skeleton structure has sufficient strength to support and maintain the morphology after etching. EDS mapping of silicon (figures 3(a)-(d) and table S1) and TEM images (figure 2(g)) indicate most SiO 2 is removed during etching. It is difficult to avoid a small amount of  SiO 2 residue after direct etching with HF acid because aerogel is a highly porous material, which makes it difficult for the solvent to completely penetrate into the aerogel during etching. TEM image (figure 2(h)) shows a complete carbon aerogel skeleton of C-50, which proves that CAMs were successfully prepared.
Nitrogen adsorption-desorption isotherms were investigated to further characterize the effect of SiO 2 addition, carbonization and etching on the pore structure of the microspheres. Figures 4 and S4 show the nitrogen adsorption-desorption isotherms and NLDFT curves of aerogel microspheres. First, nitrogen adsorption-desorption isotherms at each stage of aerogel microspheres were investigated. As shown in figure 4, the total amount of nitrogen adsorption of PI aerogel does not have a significant change after imidization and carbonization. But as the SiO 2 is etched, the CAMs form more micropores. Therefore, CAMs have a higher total nitrogen adsorption at both low p/p 0 and high p/p 0 . Figure 4(b) also shows that the specific surface area of carbon aerogel increases sharply. These results indicate that we have successfully constructed porous polymer aerogel microspheres with cross-linked network. Subsequently, the effect of SiO 2 content on the pore structure of microspheres was investigated. The results ( figure S4) show that the total adsorption amount displays a decreasing trend with the increase of SiO 2 addition, which is because the addition of SiO 2 network promotes the formation of pores. In the low p/p 0 region, a sharp increase of nitrogen adsorption occurred in 75%, 62.5%, and 50% SiO 2 , indicating the appearance of micropores. This shows that as the SiO 2 content increases, more micropores appear. The earlier hysteresis loop indicates that the mesopores tend to decrease gradually with the increase of SiO 2 content, which is confirmed by the NLDFT curves ( figure S4(b)). The specific surface area tends to increase gradually with the increase of SiO 2 content, since the main contribution to the specific surface area value comes from the micropores. This indicates that we have successfully constructed porous polymer aerogel microspheres with SiO 2 as cross-linked network. The carbonization of the PI/SiO 2 aerogel microspheres are carried out at 1000 • C. Figure S4(c) shows the pore structure of the carbonized microspheres. C-SiO 2 microspheres with low SiO 2 content show a sharp adsorption increase in the high p/p 0 region, indicating the presence of macropores, while C-SiO 2 microspheres with high SiO 2 content show a plateau in the high p/p 0 region, indicating the absence of macropores. As seen from the low p/p 0 region, C-SiO 2 microspheres with low SiO 2 content absorb less nitrogen than those with higher SiO 2 content, which indicates that C-SiO 2 microspheres with low SiO 2 content have more micropores and thus larger specific surface area (table S2). The NLDFT curves also confirm this result ( figure S4(d)). As shown in figure 4(a), before carbonization, the curve shape of the microspheres with 50% SiO 2 does not change significantly, but the amount of adsorption decreases after carbonization. This result indicates that the pores at each level decrease after carbonization. This reduction in multi-stage pores is due to the collapse of the cross-linked PI phase during carbonization. The NLDFT curves of figure 4(b) also confirm this result. After etching, the specific surface area of the aerogel microspheres increases sharply, and the maximum specific surface area reaches 1166.9 m 2 g −1 . Among them, the pore volume of C-50 reaches 1.0872 cm 2 g −1 . However, when the SiO 2 content increases, the micropores do not increase much, but the pore size obviously decreases. This is caused by the collapse of the microspheres after SiO 2 etching, which provides mesopores and macropores. As seen from figure S4(e), the adsorption curves of the C-62.5 and C-75 change from type IV to type I, and the mesopores and macropores are severely deficient, leaving only micropores. The NLDFT curves in figure S4(f) further support this phenomenon. C-62.5 and C-75 are rich in micropores, lacking mesopores and macropores, and their average pore sizes are reduced from 9.425 nm to 8.477 nm before etching to 2.428 nm and 2.362 nm after etching, respectively. This indicates that the SiO 2 etching produces more micropores in the microspheres with high SiO 2 content. In the etched microspheres, C-37.5 and C-50 still retain a clear hysteresis loop, indicating that they have a mesoporous structure, while C-50 has the largest adsorption increase in the low p/p 0 region, which indicates that C-50 has a rich microporous structure. But the absence of significant absorption growth in the high p/p 0 region indicates that C-50 lacks macropores and the NLDFT curves also indicate this result.
CAMs are ideal materials for supercapacitor electrodes. We performed electrochemical performance tests on the etched samples. The test used a three-electrode system with an electrolyte solution of 6 mol l −1 KOH solution. The CV and GCD curves are shown in figure 5. At low scanning rate, all CV curves present a nearly rectangular shape, indicating the efficient formation of an electric double layer and fast charge diffusion. When the scanning rate gradually increases to 200 mV s −1 and 500 mV s −1 , the CV curve deviates from the ideal shape and there is an obvious redox peak. A couple of redox peaks can be observed in CV curves, implying that the faradaic reaction occurs in the energy storage process with obvious pseudocapacitive characteristics. The redox peak becomes more obvious with the increase of scanning rate. This may be caused by the oxidation-reduction reaction of the residual SiO 2 and the KOH electrolyte. At the same time, XPS shows that there are a lot of pyridine-N (N-6) and quaternary-N (N-Q) and other forms of N in C-50. N-6 and N-Q have high contributions to pseudocapacitance. Then, CO-type oxygen group contributed positively to the pseudocapacitance of porous carbon [18]. The GCD curves ( figure 5(b)) also exhibit an approximate isosceles triangle shape. These results indicate that the prepared CAMs have good electrical double layer properties, due to the presence of abundant micropores and mesopores. The CV curves and GCD curves of CAMs with different SiO 2 content are shown in figures 6(a) and (b). All CAMs can effectively form an electric double layer and rapidly diffuse charges. Figure 6(c) and table S3 show the specific capacitance results measured by GCD curves at different current densities. Although C-37.5 has the largest specific capacitance, its capacitance retention at high current density is small. The C-50 has a slightly smaller specific capacitance than C-37.5 at low current density, but the C-50 has a better rate performance, and retain 53.3% capacitance at a current density of 30 A g −1 due to its rich micropore adsorption site. For C-62.5 and C-75, they are rich in micropores, lack mesopores and macropores. Mesopores and macropores act as ion transfer channels and ion-buffers reservoirs to reduce ion diffusion distance, respectively. The absence of mesopores and macropores makes C-62.5 and C-75 lack active absorption sites, thus their specific capacitances decrease significantly at high current densities [26]. Due to the lack of mesopores and macropores, the adsorption of  ions occurs mainly on the surface of the C-62.5 and C-75, their specific capacitance is low, and the rate performance is also low at high current densities.
The EIS test further clarifies the relationship between ion transfer and multi-level pores of CAMs. As seen from figure 7(a), the slope of C-50 is closer to 90 degrees at low frequencies, meaning lower ion transfer resistance. The slope of the C-37.5 is small, indicating that its ion transport ability is inferior to that of C-50, due to the long ion transfer distance caused by the lack of micropores in C-37.5. The smaller slopes of C-62.5 and C-75 are caused by the longer ion transfer distance due to the lack of mesopores and macropores. In the high-frequency region, the semicircle of the C-50 curve (interfacial charge transfer impedance, the smaller the semicircle, the lower the impedance) is smaller than that of C-37.5, presumably because C-37.5 has fewer micropores as adsorption sites and thus longer ion transfer distance than C-50. C-62.5 and C-75 are closer to having no semicircle because of the high electron activity. The C-50 has a lower starting point of the impedance frequency response ( figure 7(b)), which is consistent with the Nyquist plot. In terms of specific capacitance dependence on frequency (figure 7(c)), the total capacitance of C-50 drops slowly from low frequency to high frequency, so it has a better capacitance retention compared with C-37.5, C-62.5 and C-70. In the Bode phase diagram, the C-50 has a larger phase angle in the low-frequency region and exhibits better electrical double-layer performance compared to C-37.5, C-62.5 and C-75. The relaxation time (τ = 1/f ), the minimum time required to release all of the energy from the device with an efficiency of greater than 50%, is another parameter used to evaluate the performance of the supercapacitor. A smaller relaxation time means a faster charge and discharge rate. At a phase angle of 45 • , it is the reciprocal of the frequency. As shown in figure 7(d) and table S4, the CAMs have a short relaxation time. Compared with C-50, C-62.5 and C-75, C-37.5 has a longer relaxation time (0.15 s) because C-37.5 has a smaller specific surface area and pore volume. While C-50 has the largest specific surface area and pore volume, the relaxation time (0.078 s) is longer than that of C-62.5 and C-75. This may be caused by the different pore structures of C-50, C-62.5, and C-75. As shown in figure S4 and table S2, compared with C-62.5 and C-75, C-50 has fewer macropores which act as ion transfer channels that facilitate the efficient ionic diffusion. Meanwhile, the porosity of C-50 is lower than that of C-62.5 and C-75, which may also result in longer relaxation time of C-50 than that of C-62.5 and C-75 [26]. In conclusion, C-50 has a shorter relaxation time and exhibits better capacitor performance than C-37.5. C-62.5 and C-75 have shorter relaxation times, but the lower specific capacitance limits their application in capacitors. Therefore, C-50 has better electrical double-layer capacitor performance than other samples.

Conclusion
In conclusion, PI/SiO 2 aerogel microspheres were prepared by SiO 2 assisted sol-gel method, and CAMs were prepared by subsequent carbonization and etching. By adjusting SiO 2 content, we obtained PI/SiO 2 aerogel microspheres and CAMs with different pore structures. At a SiO 2 content of 50%, the PI/SiO 2 -50 aerogel microspheres can still maintain a good spherical shape after carbonization and etching. After etching, C-50 has the largest specific surface area, reaching 1166.9 m 2 g −1 . Due to its rich pore structure, C-50 also has a excellent electrical double-layer performance. The specific capacitance measured in the three-electrode system is 118 A g −1 at a current density of 0.5 A g −1 , and 53.3% of the capacitance is reserved at 30 A g −1 .
This work provides a new idea for the linear PI gel to prepare CAMs with high specific surface, which can be applied in the field of supercapacitors.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).