1. Introduction
With the development of human society, a large amount of CO
2 is emitted into the atmosphere with the inflammation of fossil fuels, resulting in global climate change, which has become an urgent problem to be solved. An important solution is to increase the application of new energy sources, including wind energy, solar energy, and wave energy [
1]. However, due to the randomness and intermittency of environmental factors, it is difficult for new energy to generate electricity continuously and stably. Therefore, in the large-scale grid connection of new energy generation, in order to prevent the power grid from being impacted, a large amount of energy is wasted [
2], thus the development of energy storage technology has become the key to realizing the large-scale application of new energy [
3]. Redox flow battery (RFB) is efficient, cheap, safe, and stable, and has broad application prospects in the field of new energy storage [
4]. In 1974, Thaler of the NASA Lewis Research Center proposed the concept of ICRFB [
5], and in 1987, M Skyllas Kazacos et al. proposed the vanadium redox battery (VRFB) [
6]. At present, VRFB is the most commercially developed RFB energy storage technology [
7,
8]. However, due to the scarcity of vanadium, it is difficult to further reduce the cost of VRFB [
9,
10]. ICRFB has the characteristics of low cost, wide temperature adaptability, and environmental friendliness [
4,
11]. Therefore, the ICRFB with lower cost has received extensive attention from researchers again in recent years [
11,
12,
13].
GF with low cost and good conductivity is often used as the electrode of ICRFB [
14,
15,
16] and super capacitor [
17]. However, the hydrophilicity and electrochemical activity of GF are poor, resulting in an insufficient number of reactive sites on the electrode. Moreover, the reversibility of GF to Cr
3+/Cr
2+ is poor, and the reaction rate is slower than Fe
2+/Fe
3+, which is prone to hydrogen evolution side reactions and affects the battery efficiency [
18]. The above two points have hindered the application of ICRFB. Therefore, the study of electrode materials with controllable cost, easy preparation, and excellent performance is of great significance to improve the performance of ICRFB and promote its commercial application. In the existing studies, scholars have proposed many methods of electrode modification, which can effectively enhance the specific surface area of the electrode by changing the morphology of the electrode surface through heat treatment [
19], acid, etc., [
16] or biomass modification [
20]. These three methods can also introduce more oxygen-containing functional groups on the surface of the electrode so as to improve the hydrophilicity. The above two changes can increase the number of reactive sites on the electrode surface, thereby effectively improving the electrochemical activity of the electrode [
16,
19,
20]. Another modification method is to deposit metal nanoparticles or metal oxides on the surface of the electrode to improve the catalytic activity of the electrode for redox reactions. Su used In
3+ to modify the PAN-GF electrode. The results showed that the PAN-GF electrode modified by In
3+ of 0.2 M had the best electrochemical performance [
21]. Ren used KMnO
4 to modify the GF electrode, and the results showed that the introduced MnO
x could adsorb the active substances in the solution. Moreover, the discharge capacity and energy efficiency of the optimal modified GF electrode were 41% and 21% higher than that of the GF electrode, respectively [
22]. Zhang deposited metal bismuth (Bi) on the GF negative electrode, effectively improving the reaction reversibility of Cr
3+/Cr
2+ on the negative electrode [
23,
24]. Che used amorphous bismuth nanoparticles (NPS) and N dopants to modify the GF electrode. The optimal energy efficiency of the modified GF electrode reached 85.8% at 60 mA·cm
−2 [
25]. For the negative electrode of the battery, combined with the above two modification methods of increasing electrode specific surface area, hydrophilicity, and introducing the metal catalyst to modify GF, it is promising to enhance the reversibility of the negative electrode for Cr
3+/Cr
2+ while promoting the electrochemical activity. Nevertheless, few related studies have been reported for the ICRFB applications.
In this paper, for the negative electrode of ICRFB, based on the existing research, Bio-GF-O [
26] was used as the matrix, and Bi
3+ was introduced by soaking and drying. Then, the optimal BiCl
3 concentration in the preparation process was determined, and the optimal composite modified graphite felt (Bi-Bio-GF-O) negative electrode was prepared. After analyzing the reversibility of redox reactions and electrochemical activity, the surface morphology, structure, element composition, and hydrophilicity were characterized. Finally, using Bio-GF-O as the positive and Bi-Bio-GF-O as the negative electrode, high-efficiency ICRFB is realized. Under various current densities, the efficiency of ICRFB assembled with Bi-Bio-GF-O as the negative electrode was higher than that of other ICRFBs. When the current density was 100 mA·cm
−2, the CE of ICRFB with Bi-Bio-GF-O as the negative electrode was 97.83%, VE was 85.21%, EE was 83.36% after 100 cycles, and its capacity attenuation was the smallest.
2. Methods and Materials
First, before preparing the composite modified GF electrode, a certain amount of Bio-GF-O was prepared according to the previous research content [
26]. Cut several groups of GF samples into the dimension of 5.0 × 5.0 cm (carbon content ≥ 99.9%, bulk density: 0.12~0.13 g cm
−3, thickness: 3 mm, Gansu Haoshi Carbon Fiber Co., Ltd., Baiyin, China) and rinse with deionized water for three times. Then, the GF samples were immersed in ethanol for 10 min, and they were taken out and rinsed with deionized water for half an hour, and finally dried in a vacuum oven (Yixin Scientific Instrument Factory, Shanghai, China) for 6 h at 80 °C. A total of 15.0 g of the treated pomelo peel powder [
26] was mixed with 200.0 mL of deionized water and 10.0 mL of 0.3 M sucrose solution. The acquired mixture was agitated and ultrasonically for 10 min, respectively, to form slurry. The GF sample was immersed in the slurry for 5 min, and it was dried in the vacuum oven for 6 h at 80 °C. After the samples were taken out, they were pyrolyzed at 1050 °C for 3 h in an N
2 atmosphere in a tubular furnace. The obtained samples were recorded as Bio-GF-O.
After that, Bi3+ was further introduced into Bio-GF-O, and the prepared Bio-GF-O was washed with deionized water. The carbon debris on the surface of the sample was washed off and then dried in a vacuum oven for 6 h at 80 °C, and later it was taken out. Meanwhile, the electronic analytical balance was used to weigh three groups of BiCl3 drugs with different masses, which were fully dissolved in three groups of 200 mL of 3 M HCl solution, so that the concentration of Bi3+ in the solution was 0.1, 0.2, and 0.3 M, respectively. The prepared Bio-GF-O electrodes were divided into three groups. They were soaked in HCl and different concentrations of BiCl3 for 6 h and dried in a vacuum oven for 6 h at 80 °C, and it was finally taken out and labeled as Bi-Bio-GF.
To observe the surface morphology of the modified electrode, scanning electron microscopy (SEM, JEOL JSM-7610F, Tokyo, Japan) was employed. For element mapping analysis, the energy dispersive spectrometer (EDS, JEOL JSM-7610F, Tokyo, Japan) was adopted. Raman spectroscopy (Horiba Jobin Yvon LabRAM HR800, Paris, French) was used to evaluate the order degree of the crystal structure of the sample. X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific K-Alpha, Waltham, MA, USA) and X-ray diffraction (XRD, Bruker D8 advance, Billerica, MA, USA) were used to study the chemical information on the surface of the electrode. The static contact angle (SCA, DataPhysics OCA20, Filderstadt, Germany) was measured to assess the hydrophilicity of the electrode.
The electrochemical performance of the modified electrode was detected by electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) of the electrochemical workstation (CHI760E, CH Instrument, Shanghai, China). The electrolyte was a mixture of 1 M FeCl
2 + 1 M CrCl
3 + 3 M HCl [
27]. The EIS test was conducted in a standard double-electrode system at room temperature. The comparison electrode was a platinum sheet with a specification of 1.0 cm × 1.0 cm, and the working electrode is 1.5 cm × 2.0 cm. During the test, the AC voltage amplitude is set as 5 mV, and the scanning frequency changes from 0.01 Hz to 100 kHz. A CV test is conducted in a standard three-electrode system at room temperature. In addition to the working electrode and the comparison electrode, a reference electrode composed of saturated calomel electrode (SCE) and KCl salt bridge is also required. During the test, the scanning window is limited to 0 V~0.8 V and −0.8 V~0 V. To make the curve change trend more obvious and the comparative analysis more significant, the potential scanning rate is set as 1 mV·s
−1 [
26].
The battery testing system (CT3002K, Blue Electron, Wuhan, China) is employed to test the ICFRB cell (Chuxin Technology Co., Ltd., Wuhan, China). There are three groups in total. The gradient current density of the charge discharge test and the constant current density cycle of the charge discharge test are completed for each group. The first group used GF as positive and negative electrodes; the second group used Bio-GF-O as positive and negative electrodes. The third group used Bio-GF-O as the positive electrode and Bi-Bio-GF-O as the negative electrode. The entire battery system except for the peristaltic pump is located in a thermostat at 65 °C. The electrolyte is still a mixed solution [
27], consisting of 35 mL in the catholyte and anolyte storage tanks, respectively. The specification of the ion exchange membrane is 6.0 cm × 6.0 cm (Nafion 212, DuPont, Wilmington, NC, USA). The electrode was cut into a square with a side length of 5.0 cm. The cut-off voltage of battery discharging and charging is 0.8 V and 1.2 V. The current density range for the charge–discharge cycling is between 40 and 140 mA·cm
−2, the interval is 20 mA·cm
−2, and the battery charges and discharges for 5 cycles at each current density. The current density of the long-term charge–discharge cycling is 100 mA·cm
−2. On the premise that the battery performance conditions are met, 100 cycles of charge–discharge operations are performed on the battery to test the cycle stability of ICRFB under different electrode composition schemes.
4. Discussion
The electrochemical features of GF, Bio-GF-O and Bi-Bio-GF were studied according to CV curves, and the optimal Bi
3+ concentration for the preparation of Bi-Bio-GF-O was determined. The CV curve with a scan window of −0.8 V~0 V reflects the reactivity of the anode to Cr
3+/Cr
2+, while the scan window of −0.8 V~0 V reflects the reactivity of the cathode to Fe
2+/Fe
3+. First of all, in
Figure 3a, for the negative electrode, since the redox reaction of Cr
3+/Cr
2+ on the GF electrode is less reversible than Fe
2+/Fe
3+, the reduction peak of the GF electrode is not as obvious as the oxidation peak. Although the I
pa (106.7 mA) of the Bio-GF-O electrode was higher than that of the GF electrode (56.2 mA), the reduction peak of the Bio-GF-O electrode became more difficult to observe after the modification of pomelo peel powder. When Bi
3+ was further introduced into Bio-GF-O, the reduction peak potential of the prepared Bi-Bio-GF electrode increased significantly, the position of the peak moved to the right, and the I
pc also increased. Due to the obvious increase in the reduction peak potential, the ∆E
p of the three groups of Bi-Bio-GF decreased significantly, indicating that the oxidation-reduction reaction of Cr
3+/Cr
2+ was more likely to occur on the electrode surface after the introduction of Bi
3+. According to
Table 2, in the three groups of Bi-Bio-GF, the ∆E
p of Bi-Bio-GF-0.2M is 0.119 V, which is less than the ∆E
p of the GF electrode (0.296 V); I
pa and I
pc were 119.5 mA and −148.9 mA, respectively, which were higher than those of the GF electrode (56.2 mA and −56.6 mA). Moreover, the absolute value of I
pa/I
pc of Bi-Bio-GF-0.2M is the closest to 1, which indicates that Bi-Bio-GF-0.2M has the best reversibility of Cr
3+/Cr
2+ [
28], so the optimal Bi
3+ concentration in the process of preparing the composite modified electrode is determined to be 0.2 M, and Bi-Bio-GF-0.2M is Bi-Bio-GF-O. Comparing the negative CV curves of GF, Bio-GF-O, and Bi-Bio-GF-O (
Figure 3b), it can be seen that Bi-Bio-GF-O has larger I
pa and I
pc and smaller ∆E
p, which is a better negative material for batteries.
For the positive electrode, in
Figure 3c, the peak current of Bio-GF-O electrode (I
pa = 171.3 mA, I
pc = −163.4 mA) is larger than that of GF electrode (I
pa = 121.5 mA, I
pc = −127.4 mA), and the absolute value of I
pa/I
pc also approaches 1, indicating that the electrochemical activity of Bio-GF-O as cathode is higher than that of GF. Moreover, Bio-GF-O has good reversibility for Fe
2+/Fe
3+ redox reactions. The reduction peak potential of the Bi-Bio-GF electrode decreased significantly, the peak position moved to the left, and the I
pc also decreased. At the same time, the oxidation peak on the CV curve disappeared. The reason is that the introduction of bismuth catalyst is not conducive to improving the electrochemical activity of the Fe
2+/Fe
3+ redox reaction [
29]. By comparing the negative CV curves of GF, Bio-GF-O, and Bi-Bio-GF-O (
Figure 3d), it can be seen that Bio-GF-O is a better cathode material for batteries.
The charge transfer resistance (R
ct) in the redox reaction process is evaluated according to the EIS diagram and its fitting results. Wherein, R
s represents the intrinsic resistance of the system, R
ct represents charge transfer resistance, C
pe represents double-layer capacitance of electrode/electrolyte liquid interface, and W
0 represents Warburg impedance. The Nyquist diagram is divided into two parts, including the semicircular part (high-frequency section) and the linear part (low-frequency section), which, respectively, represent the charge transfer at the electrode/electrolyte interface and the diffusion of redox active substances [
30,
31,
32]. R
ct can be estimated from the semicircular diameter [
33]. As shown in
Figure 4, the impedance spectrum is fitted according to the equivalent circuit. The data with -cal are the fitting calculation results, and the data with -msd are the EIS experimental test results. The results show that the radius of the high-frequency semicircular part of the Bio-GF-O electrode is smaller than that of the GF electrode, while the radius of the high-frequency semicircular part of the Bi-Bio-GF-O electrode is the smallest. As shown in
Table 3, the R
ct (2.422 Ω) of the Bio-GF-O electrode is less than that of GF electrode (2.961 Ω), and the R
ct (0.57 Ω) of the Bi-Bio-GF-O electrode is the smallest, far less than the first two. This shows that compared with the GF electrode, Bio-GF-O and Bi-Bio-GF-O electrodes have smaller polarization resistance in the mixed electrolyte, and the performance of the Bi-Bio-GF-O electrode is better.
The surface morphologies of GF, Bio-GF-O and Bi-Bio-GF-O were studied by scanning electron microscopy. According to
Figure 5a,d, the GF electrode is composed of disordered graphite fibers, which is beneficial to increase the porosity of the electrode surface and then increase the reactive active sites compared with the ordered structure; However, the surface of the fiber is relatively smooth, which is not conducive to increasing the specific surface area of the electrode. According to
Figure 5b,e, after the Bio-GF-O electrode was modified by pomelo peel powder and high temperature in the N
2 atmosphere, the gullies on the graphite fiber surface were more obvious and uneven, and the surface was uniformly encapsulated with a stratum of biomass nanoparticles exhibiting a heterogeneous distribution. This modification is anticipated to augment the specific surface area of the Bio-GF-O electrode, thereby enhancing its electrochemical performance. From
Figure 5c,f, crystalline bulges were observed on the Bi-Bio-GF-O electrode surface. These crystals were BiCl
3 crystals naturally precipitated during the drying process when preparing Bi-Bio-GF-O. It is worth noting that some precipitated BiCl
3 crystals can also be seen on the relatively internal graphite fiber, implying that the preparation method of Bi-Bio-GF-O, which is fully soaked and then dried, can uniformly and deeply introduce Bi
3+.
The elemental compositions of GF (
Figure 6a,b), Bio-GF-O (
Figure 6c,d), and Bi-Bio-GF-O (
Figure 6e–h) were preliminarily analyzed by energy dispersive spectroscopy. The results showed that more O elements were enriched on the biomass particles that adhered to the Bio-GF-O surface, and these O elements came from the oxygen-containing functional groups in the biomass particles. For Bi-Bio-GF-O, Bi is enriched on the crystalline bulge adhered to the electrode surface, which is consistent with the analysis results of SEM images. It can be proved that the modification method adopted in this study effectively introduces Bi on the electrode surface.
The chemical elements of GF, Bio-GF-O, and Bi-Bio-GF-O were further determined by XPS analysis, and the oxygen-containing functional groups and Bi elements incorporated into the modified electrode were determined. From
Figure 7a, the XPS spectrum of Bio-GF-O shows a stronger O1s signal than that of GF, which indicates that there are larger quantities of oxygen-containing functional groups in Bio-GF-O. The O1s signal of Bi-Bio-GF-O is the strongest and shows a strong Bi4f signal, which indicates that a huge amount of Bi elements are incorporated into the surface of Bi-Bio-GF-O and confirms the conclusion that Bi
2O
3 is produced during electrode drying.
In the Raman spectrum, the three electrodes have two manifest characteristic peaks at 1597 and 1353 cm
−1, corresponding to the G and D band, respectively [
34]. The D-band arises from the vibrational modes associated with sp
3 hybridized carbon atoms or other structural irregularities at disordered grain boundaries, whereas the G-band is attributed to the in-plane tensile vibrations of sp
2 hybridized carbon atoms that are interconnected within the carbon lattice (
Figure 7b) [
35]. The I
D/I
G values of Bi-Bio-GF-O (1.46) and Bio-GF-O (1.43) were higher than GF (1.28), which indicated that the quantity of defective carbon/edge carbon increased, heteroatoms increased, and the crystal structure order decreased after biomass modification and composite modification.
The crystal structures of GF, Bio-GF-O, and Bi-Bio-GF-O were studied by XRD and Raman spectroscopy. According to XRD analysis, first of all, after biomass modification, the main structure of the Bio-GF-O electrode is still graphite. From
Figure 7c, GF and Bio-GF-O both show wide peaks at 43.0° and 25.5°, which are attributed to the (100) and (002) planes of graphite [
36,
37,
38]. For Bi-Bio-GF-O, besides the characteristic peaks of (110) and (002) crystal planes, there are more characteristic peaks. Among them, the characteristic peaks at 26.21°, 36.85°, 48.54°, 55.37°, 60.79°, and 63.21° represent the (012), (104), (202), (024), (116), and (122) crystal planes of hexagonal system Bi (JCPDS No. 44–1246). Other characteristic peaks correspond to some crystal planes of Bi
2O
3 (JCPDS No. 45–1344), which indicates that BiCl
3 crystals are precipitated on the Bi-Bio-GF-O electrode surface during the drying process, and a small amount of Bi
2O
3 is also produced during this process. Moreover, the characteristic peaks of (110) and (002) crystal planes representing graphite shifted to the left, which was caused by the introduction of BiCl
3 crystal on the electrode surface.
The static contact angle (SCA) was used to characterize the change in hydrophilicity of electrode materials. SCA refers to the angle that comes into being when the liquid phase is sandwiched between the solid–liquid and gas–liquid interface at the gas–liquid-solid three-phase junction on the solid surface. The smaller the SCA, the better the hydrophilicity of the electrode sample. From
Figure 8a–c, the SCA of Bio-GF-O (135.827°) is smaller than that of GF (153.977°), which indicates that the hydrophilicity of Bio-GF-O is better because higher amounts of oxygen-containing functional groups are located in the biomass particles on the Bio-GF-O surface. The SCA (127.515°) of Bi-Bio-GF-O is the smallest because more O elements are introduced into the Bi
2O
3 on its surface.
The performance of ICRFB with different electrodes was compared. In the battery test diagram, the GF curve refers to GF as positive and negative electrodes; the Bio-GF-O curve refers to Bio-GF-O as positive and negative electrodes; and the Bi-Bio-GF-O curve indicates that Bio-GF-O is used as the positive electrode and Bi-Bio-GF-O as the negative electrode. First, the charge and discharge capacities of the batteries in each group were tested. According to
Figure 9a, when the current density was 100 mA·cm
−2, the performance of Bio-GF-O (charge capacity was 631 mAh, discharge capacity was 601 mAh) and Bi-Bio-GF-O (charge capacity was 584 mAh, discharge capacity was 534 mAh) were better than those of the GF group (charge capacity was 253 mAh, discharge capacity was 208 mAh). The charge and discharge capacity of Bi-Bio-GF-O was lower than that of Bio-GF-O because the voltage changed faster when the charge and discharge voltage were close to the cut-off voltage. The gradient tests of three groups of ICRFBs with different current densities were carried out. After calculation, the electrolyte utilization rate of the GF group charging was 18.88%, Bio-GF-O was 39.85%, and Bi-Bio-GF-O was 47.09%.
From
Figure 9b−d, when the current density rises from 40 mA·cm
−2 to 140 mA·cm
−2 at an interval of 20 mA·cm
−2, the VE and EE of the battery decrease, and CE increases. Nevertheless, the EE and CE of the GF group exhibited pronounced fluctuations, which were attributed to the concurrent side reactions involving hydrogen evolution occurring on the GF [
26]. It can be seen that the ICRFB assembled with Bio-GF-O significantly reduced the fluctuation caused by the side reactions involving hydrogen evolution and effectively improved the VE, EE and CE of the battery. Although the stability of the Bi-Bio-GF-O group decreased slightly, this scheme further improved the efficiency of ICRFB. After gradient cycling, under the low current density of 40 mA·cm
−2, the VE, EE, and CE of the battery were 93.72%, 88.97%, and 94.93%, respectively.
On the basis of the above three groups of ICRFBs, an additional control group was added. The control group used Bio-GF-O as the cathode and anode of the battery but added 5 mmol/L of BiCl
3 to the negative electrolyte, which was recorded as Bio-GF-O (Bi). Relevant studies show that Bi
3+ with this concentration is directly added to the anode electrolyte, which can better catalyze the oxidation-reduction reaction of the anode [
39]. The four ICRFBs were cyclic charging and discharging at 100 mA·cm
−2. From
Figure 10a,b, the attenuation speed of the GF group began to slow down after nearly 20 cycles, and it was unable to stabilize the cycle for 100 times. After nearly 60 cycles, the battery capacity was 70 mAh, which had been attenuated to the point that it could not continue to support the cyclic charge and discharge. In the Bio-GF-O group, the decay rate of capacity was slower during 100 cycles, and the charging capacity was still 124 mAh after 100 cycles. The discharge capacity is 120 mAh, which is very close to the charging capacity. The main reasons for the capacity attenuation include the side reaction of hydrogen evolution, the transmembrane migration of ions, and the increase in internal resistance caused by the increase in charge and discharge times. Bio-GF-O group ICRFB can be cycled for 100 times, indicating that the side reaction of hydrogen evolution on the Bio-GF-O electrode is reduced. The capacity of the Bi-Bio-GF-O group decreased rapidly at the initial stage of the cycle. But after about 20 times, the attenuation rate began to slow down. After about 60 times, the capacity of the Bio-GF-O group was always higher than that of the Bio-GF-O group, and it was more stable. After 100 cycles, the charging capacity was still 160 mAh, and the discharge capacity was 157 mAh; they are both higher than the Bio-GF-O group. Moreover, in comparison to the control group of Bio-GF-O (Bi), the decay rate of the capacity of the Bi-Bio-GF-O group was slower, indicating that the preparation method of Bi-Bio-GF-O, which was fully soaked and then dried, was better than the modification scheme of adding Bi
3+ directly into the negative electrolyte and could retain the catalytic effect of Bi on Cr
3+/Cr
2+ redox reactions to a greater extent after multiple cycles.
The battery efficiency of the above four groups of ICRFBs during the cyclic test at 100 mA·cm
−2 is recorded. From
Figure 11a–c, After 57 cycles, the CE, VE, and EE of the GF group were 91.04%, 71.80%, and 65.37%, respectively. The CE, VE, and EE of Bio-GF-O group were higher than those of the GF group during the cycle. After 100 cycles, the CE, VE, and EE of the battery were 96.42%, 78.50%, and 75.69%, respectively. After 100 cycles, the Bi-Bio-GF-O battery maintained the highest efficiency among the four ICRFBs, the CE and EE were more stable than those in the control group. This result once again proves that the preparation method of Bi-Bio-GF-O can make the catalytic effect of Bi on Cr
3+/Cr
2+ redox reactions more stable. This is because Bi can only be sedimented on the electrode surface during the charging process by adding Bi
3+ only to the negative electrolyte, which is not deep enough. Moreover, due to the single internal channel structure of the stack and the flow direction of the electrolyte, Bi deposited at the inlet will be more than that at the outlet, which is difficult to deposit evenly. The preparation method of Bi-Bio-GF-O, which is fully soaked and then dried, makes it easier to deposit more Bi
3+ on the Bi-Bio-GF-O. After 100 cycles, the CE, VE, and EE of the Bi-Bio-GF-O battery were 97.83%, 85.21%, and 83.36%, respectively.