Aggregation-Morphology-Dependent Electrochemical Performance of Co3O4 Anode Materials for Lithium-Ion Batteries

The aggregation morphology of anode materials plays a vital role in achieving high performance lithium-ion batteries. Herein, Co3O4 anode materials with different aggregation morphologies were successfully prepared by modulating the morphology of precursors with different cobalt sources by the mild coprecipitation method. The fabricated Co3O4 can be flower-like, spherical, irregular, and urchin-like. Detailed investigation on the electrochemical performance demonstrated that flower-like Co3O4 consisting of nanorods exhibited superior performance. The reversible capacity maintained 910.7 mAh·g−1 at 500 mA·g−1 and 717 mAh·g−1 at 1000 mA·g−1 after 500 cycles. The cyclic stability was greatly enhanced, with a capacity retention rate of 92.7% at 500 mA·g−1 and 78.27% at 1000 mA·g−1 after 500 cycles. Electrochemical performance in long-term storage and high temperature conditions was still excellent. The unique aggregation morphology of flower-like Co3O4 yielded a reduction of charge-transfer resistance and stabilization of electrode structure compared with other aggregation morphologies.


Introduction
Recently, secondary batteries, especially lithium-ion batteries (LIBs) have become significant components in portable devices, electric vehicles, and exploitation systems of clean energy, due to their satisfactory energy density, lifetime, and environmental friendliness [1,2]. However, the booming society is imposing a strong demand on energy density of LIBs [3,4]. The current LIBs utilizing traditional electrode materials, such as lithium cobalt oxides and graphite, can only achieve limited enhancement of energy density in the range of 150-300 Wh·kg −1 [5]. Thus, exploiting high-capacity electrode materials has been the primary solution [6]. Regarding anodes, transition metal oxides, such as NiO [7], Fe 2 O 3 [8], ZnO [9], CuO [10], MnO [11], Co 3 O 4 [12], ZnCo 2 O 4 [13], and NiCo 2 O 4 [14], are promising future anode materials to substitute for the widely used carbon materials. Typically, Co 3 O 4 , with its high theoretical capacity (890 mAh·g −1 ) owing to storage for eight lithium ions per molecule based on the reversible conversion reaction between Co 3 O 4 /Li and Co/Li 2 O (Co 3 O 4 + 8Li ↔ 3Co + 4Li 2 O), has been an important candidate as a next generation anode. Additionally, high volumetric capacity of Co 3 O 4 over graphite is another crucial advantage [15].However, the intrinsic low electrical conductivity, poor Coulombic efficiency [16], high voltage hysteresis [17], high average

Experimental Results and Discussion
The aggregation morphology of Co 3 O 4 materials was flexibly regulated by constructing various precursors, as shown in Figure 1. Specifically, the homogeneous precipitation method was applied in mild conditions, using urea as precipitant and cobalt salts as cobalt sources. SEM images of the synthesized precursors proved the successful regulation of the morphology (Figure 2a). It was proven that different cobalt salts produced precursors in various shapes. Bunches of flower-like precursors consisting of nanorods were inclined to generate when using cobalt chlorate. The configuration resulting from a cobalt sulfate precursor was spherical. Micro-clusters made of flakes tended to form for cobalt acetate. Urchin-like particles were prepared by applying cobalt nitrate. The formation process of the precursors involves nuclei and growth, which can be largely influenced by the microenvironment of the reaction solution. The CO 3 2− and OH − produced from the decomposition of urea contributed to the initial nuclei, and the formed crystal nuclei were inclined to adsorb the existing anions in the solution, leading to a negatively charged surface [43]. The existing Cl − , SO 4 2− , AC − , and NO 3 − may then exert influence on the following growth behavior based on electrostatic interactions [43]. Thus, the morphology of the precursors corresponding to different cobalt sources could be easily adjusted. The crystal structures of the obtained precursors originating from different cobalt salts were confirmed and are shown in Figure 2b. The XRD pattern of the precursor from cobalt chlorate demonstrated  . Thermal behavior analysis of the as-prepared precursors was conducted by TG and is displayed in Figure 2c. During the testing process, the main weight loss occurred in the range of 200-300 • C, and the residual weight remained stable until 550 • C. The total loss in weight was calculated to be 26.2%, 26.3%, 25.3%, and 27.7%, respectively, values which are basically consistent with the theoretical values [40,41]. Thus, the calcination temperature was fixed at 550 • C, and the precursors transformed into oxides. Thermal behavior analysis of the as-prepared precursors was conducted by TG and is displayed in Figure 2c. During the testing process, the main weight loss occurred in the range of 200-300 °C, and the residual weight remained stable until 550 °C. The total loss in weight was calculated to be 26.2%, 26.3%, 25.3%, and 27.7%, respectively, values which are basically consistent with the theoretical values [40,41]. Thus, the calcination temperature was fixed at 550 °C, and the precursors transformed into oxides.    Thermal behavior analysis of the as-prepared precursors was conducted by TG and is displayed in Figure 2c. During the testing process, the main weight loss occurred in the range of 200-300 °C, and the residual weight remained stable until 550 °C. The total loss in weight was calculated to be 26.2%, 26.3%, 25.3%, and 27.7%, respectively, values which are basically consistent with the theoretical values [40,41]. Thus, the calcination temperature was fixed at 550 °C, and the precursors transformed into oxides.   After calcinating at 550 • C for 2 h, the acquired samples had the structure shown in Figure 3a. All the diffraction peaks corresponded to the characteristic peaks of cubic Co 3 (Figure 3b). Moreover, the average particle size of the as-prepared Co 3 O 4 could be calculated based on XRD data and Scherrer equation. Co 3 O 4 -Cl possessed a larger particle size of 27.8 nm than that of Co 3 O 4 -SO 4 (16.9 nm), while the values for Co 3 O 4 -AC and Co 3 O 4 -NO 3 were identical (25.2 nm). FT-IR and Raman spectra further verified the formation of the desired oxides. The absorption peaks at 576 and 662 cm −1 in Figure 3c stand for the characteristic stretching vibration peaks of Co-O band, and the typical peaks seen at 464, 507, 604, and 674 cm −1 (Figure 3d) belong to crystalline Co 3 O 4 [44,45], which provided sufficient evidence for successful synthesis. Å, and 8.060 Å, respectively, which were close to the value (a = 8.084 Å) of the standard Co3O4 ( Figure 3b). Moreover, the average particle size of the as-prepared Co3O4 could be calculated based on XRD data and Scherrer equation. Co3O4-Cl possessed a larger particle size of 27.8 nm than that of Co3O4-SO4 (16.9 nm), while the values for Co3O4-AC and Co3O4-NO3 were identical (25.2 nm). FT-IR and Raman spectra further verified the formation of the desired oxides. The absorption peaks at 576 and 662 cm −1 in Figure 3c stand for the characteristic stretching vibration peaks of Co-O band, and the typical peaks seen at 464, 507, 604, and 674 cm −1 (Figure 3d) belong to crystalline Co3O4 [44,45], which provided sufficient evidence for successful synthesis. Aggregation morphology of the Co3O4 obtained with different cobalt salts was investigated by SEM. The images in Figure 4 present the distinct differences. The morphology of the calcinated products was maintained to some degree from the precursors, but obvious changes occurred. For Co3O4-Cl, flower-like particles were assembled into bamboo-like rods, and the rods consisted of nanograins (Figure 4a,b). Spherical precursors of Co3O4-SO4 were sintered into ellipsoids (Figure 4c), which consisted of numerous small grains (Figure 4d). Flake-like precursors corresponding to Co3O4-AC were transformed to monodispersed nanoparticles, but serious aggregation occurred and led to irregular clusters (Figure 4e,f). The urchin-like shape still existed for Co3O4-NO3 (Figure 4g), but the nanorods changed into bamboo-like rods (Figure 4h). Using the EDS detecting system, the distributions of elemental oxygen and cobalt were acquired for the as-prepared Co3O4 with different micromorphologies, as shown in Figure 5. The results showed that elemental O and Co were distributed homogenously in the various aggregates. The above results demonstrate that Co3O4 materials with different aggregation morphologies can be successfully prepared by the rational use of cobalt sources in the synthesis process. Aggregation morphology of the Co 3 O 4 obtained with different cobalt salts was investigated by SEM. The images in Figure 4 present the distinct differences. The morphology of the calcinated products was maintained to some degree from the precursors, but obvious changes occurred. For Co 3 O 4 -Cl, flower-like particles were assembled into bamboo-like rods, and the rods consisted of nanograins (Figure 4a,b). Spherical precursors of Co 3 O 4 -SO 4 were sintered into ellipsoids (Figure 4c), which consisted of numerous small grains (Figure 4d). Flake-like precursors corresponding to Co 3 O 4 -AC were transformed to monodispersed nanoparticles, but serious aggregation occurred and led to irregular clusters (Figure 4e,f). The urchin-like shape still existed for Co 3 O 4 -NO 3 (Figure 4g), but the nanorods changed into bamboo-like rods (Figure 4h). Using the EDS detecting system, the distributions of elemental oxygen and cobalt were acquired for the as-prepared Co 3 O 4 with different micromorphologies, as shown in Figure 5. The results showed that elemental O and Co were distributed homogenously in the various aggregates. The above results demonstrate that Co 3 O 4 materials with different aggregation morphologies can be successfully prepared by the rational use of cobalt sources in the synthesis process.  The relationship between the aggregation states of Co3O4 and its performance were illustrated in detail by electrochemical analysis. Cyclic voltammograms (CV) were first collected in the coin cells from 0.01 V to 3.0 V, and shown in Figure 6. In the initial negative scanning process, the obvious cathodic peaks located around 0.80 V were ascribed to the transformation from Co3O4 to Co and some side reactions, including the generation of the SEI layer [46,47] The reduction potentials for Co3O4-Cl, Co3O4-SO4, Co3O4-AC, and Co3O4-NO3 were 0.854 V, 0.765 V, 0.850 V, and 0.817 V (Table 1), respectively, and the difference in potential may indicate that small grains benefited the electrochemical reaction. In the subsequent positive scanning, broad anodic peaks near 2.1 V occurred, and were related to the transition of Co and Li2O [48]. The corresponding oxidation potentials were 2.055 V, 2.118 V, 2.061 V, and 2.070 V (Table 1), respectively. The slight increase of anodic potential for Co3O4-SO4 may have been caused by excessive formation of SEI on the smaller primary particles. As the CV test proceeded, the potential for reduction reaction inclined to shift to 1.1 V, and the values of Co3O4-SO4 displayed a larger increase, while the oxidation potential stayed at almost 2.1 V. The tested curves were basically overlapped, revealing preferable reversibility during the electrochemical conversion. The CV results demonstrate that Co3O4 materials exhibit similar electrochemical behavior in spite of particle morphology.  The relationship between the aggregation states of Co3O4 and its performance were illustrated in detail by electrochemical analysis. Cyclic voltammograms (CV) were first collected in the coin cells from 0.01 V to 3.0 V, and shown in Figure 6. In the initial negative scanning process, the obvious cathodic peaks located around 0.80 V were ascribed to the transformation from Co3O4 to Co and some side reactions, including the generation of the SEI layer [46,47] The reduction potentials for Co3O4-Cl, Co3O4-SO4, Co3O4-AC, and Co3O4-NO3 were 0.854 V, 0.765 V, 0.850 V, and 0.817 V (Table 1), respectively, and the difference in potential may indicate that small grains benefited the electrochemical reaction. In the subsequent positive scanning, broad anodic peaks near 2.1 V occurred, and were related to the transition of Co and Li2O [48]. The corresponding oxidation potentials were 2.055 V, 2.118 V, 2.061 V, and 2.070 V (Table 1), respectively. The slight increase of anodic potential for Co3O4-SO4 may have been caused by excessive formation of SEI on the smaller primary particles. As the CV test proceeded, the potential for reduction reaction inclined to shift to 1.1 V, and the values of Co3O4-SO4 displayed a larger increase, while the oxidation potential stayed at almost 2.1 V. The tested curves were basically overlapped, revealing preferable reversibility during the electrochemical conversion. The CV results demonstrate that Co3O4 materials exhibit similar electrochemical behavior in spite of particle morphology. The relationship between the aggregation states of Co 3 O 4 and its performance were illustrated in detail by electrochemical analysis. Cyclic voltammograms (CV) were first collected in the coin cells from 0.01 V to 3.0 V, and shown in Figure 6. In the initial negative scanning process, the obvious cathodic peaks located around 0.80 V were ascribed to the transformation from Co 3 O 4 to Co and some side reactions, including the generation of the SEI layer [46,47] The reduction potentials for Co 3 O 4 -Cl, Co 3 O 4 -SO 4 , Co 3 O 4 -AC, and Co 3 O 4 -NO 3 were 0.854 V, 0.765 V, 0.850 V, and 0.817 V (Table 1), respectively, and the difference in potential may indicate that small grains benefited the electrochemical reaction. In the subsequent positive scanning, broad anodic peaks near 2.1 V occurred, and were related to the transition of Co and Li 2 O [48]. The corresponding oxidation potentials were 2.055 V, 2.118 V, 2.061 V, and 2.070 V (Table 1), respectively. The slight increase of anodic potential for Co 3 O 4 -SO 4 may have been caused by excessive formation of SEI on the smaller primary particles. As the CV test proceeded, the potential for reduction reaction inclined to shift to 1.1 V, and the values of Co 3 O 4 -SO 4 displayed a larger increase, while the oxidation potential stayed at almost 2.1 V. The tested curves were basically overlapped, revealing preferable reversibility during the electrochemical conversion. The CV results demonstrate that Co 3 O 4 materials exhibit similar electrochemical behavior in spite of particle morphology.

Sample
1st Cycle 2nd Cycle 3rd Cycle  The practical capacity for lithium storage of the as-prepared Co3O4 was evaluated through galvanostatic test. The relevant results are listed in Figure 7. The initial discharge-charge curves at 200 mA·g −1 , shown in Figure 7a, showed a similar rule to the cathodic/anodic potentials of the CV data. The plateau potential in discharging of to Co3O4-SO4 was slightly lower, combined with a larger capacity (1281.8 mAh·g −1 ). The initial coulombic efficiencies (CE) of Co3O4-Cl, Co3O4-SO4, Co3O4-AC, and Co3O4-NO3 were close, and irreversible capacity can occur due to undesired side reactions. It seemed that the spherical Co3O4-SO4 with smaller grains displayed superior electrochemical performance. Long-term cycling was conducted at 200 mA·g −1 , 500 mA·g −1 , and 1000 mA·g −1 . Figure 7b reveals that Co3O4-SO4 delivered a higher capacity in the first 30 cycles, which demonstrates the advantage of decreasing grain size compared with morphology. The capacity then showed abrupt decay, which may have been due to the collapse of secondary particles. By contrast, Co3O4-Cl, Co3O4-AC, and Co3O4-NO3 possessed stable cyclic traits. The capacity of Co3O4-Cl was relatively smaller than those of Co3O4-AC and Co3O4-NO3 during the 100 cycles, which indicates that grain size may be the key at low current density, regardless of aggregation morphology. Reaction sites can be maximized along with the reduction on particle size. Obviously, the discharged capacities for Co3O4-Cl, Co3O4-AC, and Co3O4-NO3 samples were still higher than the theoretical value even after 100 cycles, and the extra capacity be in part due to surface pseudocapacitive behavior, which is commonly discovered in transition metal oxides with high surface area and nano size [47]. Moreover, the reversible capacities at 200 mA·g −1 for Co3O4-Cl, Co3O4-AC, and Co3O4-NO3 samples increased along with the cycling, originating from the progressive activation and gradual formation of a gel-like surface film [47][48][49][50]. Cycle properties at higher currents (500 mA·g −1 and 1000 mA·g −1 ) were then analyzed. At 500 mA·g −1 (Figure 7c), spherical Co3O4-SO4 materials still had superior capacity before 30 cycles, then the capacity sharply decreased and fluctuated. By contrast, the flower-like Co3O4-Cl sample showed preferable cycle stability. The capacity remained at 910.7 mAh·g −1 after 500 cycles, and the capacity retention rate reached to 92.7% with only tiny decay (0.015%  Table 1. The data originating from the CV curves in Figure 6.

Sample
1st Cycle 2nd Cycle 3rd Cycle The practical capacity for lithium storage of the as-prepared Co 3 O 4 was evaluated through galvanostatic test. The relevant results are listed in Figure 7. The initial discharge-charge curves at 200 mA·g −1 , shown in Figure 7a, showed a similar rule to the cathodic/anodic potentials of the CV data. The plateau potential in discharging of to Co 3 O 4 -SO 4 was slightly lower, combined with a larger capacity (1281.8 mAh·g −1 ). The initial coulombic efficiencies (CE) of Co 3 O 4 -Cl, Co 3 O 4 -SO 4 , Co 3 O 4 -AC, and Co 3 O 4 -NO 3 were close, and irreversible capacity can occur due to undesired side reactions. It seemed that the spherical Co 3 O 4 -SO 4 with smaller grains displayed superior electrochemical performance. Long-term cycling was conducted at 200 mA·g −1 , 500 mA·g −1 , and 1000 mA·g −1 . Figure 7b reveals that Co 3 O 4 -SO 4 delivered a higher capacity in the first 30 cycles, which demonstrates the advantage of decreasing grain size compared with morphology. The capacity then showed abrupt decay, which may have been due to the collapse of secondary particles. By contrast, Co 3 O 4 -Cl, Co 3 O 4 -AC, and Co 3 O 4 -NO 3 possessed stable cyclic traits. The capacity of Co 3 O 4 -Cl was relatively smaller than those of Co 3 O 4 -AC and Co 3 O 4 -NO 3 during the 100 cycles, which indicates that grain size may be the key at low current density, regardless of aggregation morphology. Reaction sites can be maximized along with the reduction on particle size. Obviously, the discharged capacities for Co 3 O 4 -Cl, Co 3 O 4 -AC, and Co 3 O 4 -NO 3 samples were still higher than the theoretical value even after 100 cycles, and the extra capacity be in part due to surface pseudocapacitive behavior, which is commonly discovered in transition metal oxides with high surface area and nano size [47]. Moreover, the reversible capacities at 200 mA·g −1 for Co 3 O 4 -Cl, Co 3 O 4 -AC, and Co 3 O 4 -NO 3 samples increased along with the cycling, originating from the progressive activation and gradual formation of a gel-like surface film [47][48][49][50]. Cycle properties at higher currents (500 mA·g −1 and 1000 mA·g −1 ) were then analyzed. At 500 mA·g −1 (Figure 7c), spherical Co 3 O 4 -SO 4 materials still had superior capacity before 30 cycles, then the capacity sharply decreased and fluctuated. By contrast, the flower-like Co 3 O 4 -Cl sample showed preferable cycle stability. The capacity remained at 910.7 mAh·g −1 after 500 cycles, and the capacity retention rate reached to 92.7% with only tiny decay (0.015% per cycle). Co 3 O 4 -AC clusters and urchin-like Co 3 O 4 -NO 3 performed unstably in the long-term cycle. The same situation appeared at the current density of 1000 mA·g −1 (Figure 7d). The Co 3 O 4 -Cl electrode achieved a capacity of 717 mAh·g −1 in the 500th cycle, with an acceptable retention rate (78.27%). Admittedly, nanoparticles benefitted from a reducing path length of Li + , but side reactions were also enhanced [43]. In addition, the volume changes may have resulted in the irreversible destruction of secondary particles. Therefore, the flower-like Co 3 O 4 -Cl with relatively larger grains displayed more satisfactory stability, especially at high current density. per cycle). Co3O4-AC clusters and urchin-like Co3O4-NO3 performed unstably in the long-term cycle. The same situation appeared at the current density of 1000 mA·g −1 (Figure 7d). The Co3O4-Cl electrode achieved a capacity of 717 mAh·g −1 in the 500th cycle, with an acceptable retention rate (78.27%). Admittedly, nanoparticles benefitted from a reducing path length of Li + , but side reactions were also enhanced [43]. In addition, the volume changes may have resulted in the irreversible destruction of secondary particles. Therefore, the flower-like Co3O4-Cl with relatively larger grains displayed more satisfactory stability, especially at high current density. The rate performance in Figure 8a and low-temperature test in Figure 8c also demonstrated that Co3O4 with small grains (Co3O4-SO4) had advantages on capacity, but the capacity stability seemed slight deficient. As to the cells with a standing time of 60 days (Figure 8b), the reversible capacity at 500 mA·g −1 decreased compared to the cells without standing (Figure 7c  Electrochemical kinetics and reaction mechanism were then investigated by implementing CV tests at different scan rates from 0.1 mV·s −1 to 1.0 mV·s −1 (Figure 9a-d). The relationship between peak current (Ip) and scan rate (ν) can be described with Equation (1) to determine the contributions of diffusion-limited and capacitive effects during the cathodic and anodic process.
where a and b refer to variable parameters. B = 1 implies capacitive behavior, while b = 0.5 corresponds to a diffusion-controlled process. The specific value of b can be calculated by fitting the line of logIp vs. logν and obtaining the slope. The detailed data for cathodic and anodic peaks are shown in Figure 9e,f, including the fitted slopes. The b values of the Co3O4-SO4 sample were 0.764 and 0.827, respectively, related to the cathodic and anodic process, which were higher than those of other Co3O4 samples. A higher b value reflects the enhanced capacitive effect and improved rate property [51]. In the same way, the lithium ion diffusion property can also be calculated according to Equation  Electrochemical kinetics and reaction mechanism were then investigated by implementing CV tests at different scan rates from 0.1 mV·s −1 to 1.0 mV·s −1 (Figure 9a-d). The relationship between peak current (I p ) and scan rate (ν) can be described with Equation (1) to determine the contributions of diffusion-limited and capacitive effects during the cathodic and anodic process.
where a and b refer to variable parameters. B = 1 implies capacitive behavior, while b = 0.5 corresponds to a diffusion-controlled process. The specific value of b can be calculated by fitting the line of logI p vs. logν and obtaining the slope. The detailed data for cathodic and anodic peaks are shown in Figure 9e,f, including the fitted slopes. The b values of the Co 3 O 4 -SO 4 sample were 0.764 and 0.827, respectively, related to the cathodic and anodic process, which were higher than those of other Co 3 O 4 samples. A higher b value reflects the enhanced capacitive effect and improved rate property [51]. In the same way, the lithium ion diffusion property can also be calculated according to Equation (2).  Since the spherical Co3O4 with small grains (Co3O4-SO4) had stronger Li + diffusion ability than that of flower-like Co3O4 with nanorods (Co3O4-Cl), the relevant electrochemical performance should be superior. However, the above electrochemical results reveal that flower-like Co3O4 exhibited more excellent performance. Therefore, it can be concluded that grain size plays an important role in delivering capacity, and aggregation morphology decides the cycle performance under various working conditions. Irregular clusters with nanoparticles (Co3O4-AC) may promote excessive side reactions. Spherical Co3O4 with small grains (Co3O4-SO4) may easily suffer from the fracture of secondary particles, as might urchin-like Co3O4 with nanorods (Co3O4-NO3). Flower-like Co3O4 with nanorods (Co3O4-Cl) may preferably balance stability of structure and capacity.
Further measurement on the cycled electrodes by EIS and SEM provided more proof of the performance difference. EIS spectra ( Figure 10) were collected on fresh electrodes and cycled electrodes after 500 cycles at 500 mA·g −1 and 1000 mA·g −1 . All the spectra comprised a semicircle (medium-high frequency) and an inclined line (low frequency region) before and after cycling. The charge-transfer resistance corresponding to the semicircle of the cycled Co3O4-Cl electrodes was obviously smaller than those of other electrodes at 500 mA·g −1 and 1000 mA·g −1 . The reduced resistance to charge transfer facilitates rapid electrochemical reaction. Meanwhile, the active particles were clearly seen on the fresh electrodes (Figure 11a-d). After 500 cycles at 500 mA·g −1 , the surface on the electrode was covered with SEI films, which has also been detected in other studies [47]. The surface for the Co3O4-Cl electrode remained intact (Figure 11e), while pores and fractures emerged on the electrodes of Co3O4-SO4, Co3O4-AC, and Co3O4-NO3. In turn, this indicates the enhanced structural stability of the electrode, which may be radically ascribed to the active materials. Flower-like Co3O4 with nanorods (Co3O4-Cl) exhibited advantages in constructing a stable electrode with superior performance. Therefore, aggregation morphology of the Co3O4 materials played a vital role in achieving high performance and stability. Further measurement on the cycled electrodes by EIS and SEM provided more proof of the performance difference. EIS spectra ( Figure 10) were collected on fresh electrodes and cycled electrodes after 500 cycles at 500 mA·g −1 and 1000 mA·g −1 . All the spectra comprised a semicircle (medium-high frequency) and an inclined line (low frequency region) before and after cycling. The charge-transfer resistance corresponding to the semicircle of the cycled Co 3 O 4 -Cl electrodes was obviously smaller than those of other electrodes at 500 mA·g −1 and 1000 mA·g −1 . The reduced resistance to charge transfer facilitates rapid electrochemical reaction. Meanwhile, the active particles were clearly seen on the fresh electrodes (Figure 11a-d). After 500 cycles at 500 mA·g −1 , the surface on the electrode was covered with SEI films, which has also been detected in other studies [47]. The surface for the Co 3 O 4 -Cl electrode remained intact (Figure 11e), while pores and fractures emerged on the electrodes of Co 3 O 4 -SO 4 , Co 3 O 4 -AC, and Co 3 O 4 -NO 3 . In turn, this indicates the enhanced structural stability of the electrode, which may be radically ascribed to the active materials. Flower-like Co 3 O 4 with nanorods (Co 3 O 4 -Cl) exhibited advantages in constructing a stable electrode with superior performance. Therefore, aggregation morphology of the Co 3 O 4 materials played a vital role in achieving high performance and stability.  By utilizing the prepared Co3O4 as anode active materials, full batteries were assembled to evaluate a realistic application scenario. The cathodes were composed of commercial NCM811 with an areal loading of 4.0-5.0 mg·cm −1 . The working voltage window was set at 0.01-4.3 V, and the specific capacity was calculated according to the weight of NCM. Moreover, the capacity of the Co3O4 electrode was almost equal to that of the positive electrode, and moderate pre-lithiation on the Co3O4 negative electrode was adopted before assembling full battery. Figure 12a reveals the initial charge-discharge curves of the full batteries at 40 mA·g −1 . The inclined charging and discharging platforms in the full batteries were similar to those of the NCM/Li half-cell. Figure 12b presents the cycle performance of the assembled full batteries at 100 mA·g −1 . The cycle stability was relatively good, and the capacities of the batteries using Co3O4-Cl, Co3O4-SO4, Co3O4-AC, and Co3O4-NO3 remained at 137.2, 124.3, 126.4, and 107.1 mAh·g −1 , respectively, after 20 cycles. The full battery adopting the flower-like Co3O4 delivered a comparatively higher capacity, indicating the structural advantage of flower-like Co3O4 in practical applications.   By utilizing the prepared Co3O4 as anode active materials, full batteries were assembled to evaluate a realistic application scenario. The cathodes were composed of commercial NCM811 with an areal loading of 4.0-5.0 mg·cm −1 . The working voltage window was set at 0.01-4.3 V, and the specific capacity was calculated according to the weight of NCM. Moreover, the capacity of the Co3O4 electrode was almost equal to that of the positive electrode, and moderate pre-lithiation on the Co3O4 negative electrode was adopted before assembling full battery. Figure 12a reveals the initial charge-discharge curves of the full batteries at 40 mA·g −1 . The inclined charging and discharging platforms in the full batteries were similar to those of the NCM/Li half-cell. Figure 12b presents the cycle performance of the assembled full batteries at 100 mA·g −1 . The cycle stability was relatively good, and the capacities of the batteries using Co3O4-Cl, Co3O4-SO4, Co3O4-AC, and Co3O4-NO3 remained at 137.2, 124.3, 126.4, and 107.1 mAh·g −1 , respectively, after 20 cycles. The full battery adopting the flower-like Co3O4 delivered a comparatively higher capacity, indicating the structural advantage of flower-like Co3O4 in practical applications. By utilizing the prepared Co 3 O 4 as anode active materials, full batteries were assembled to evaluate a realistic application scenario. The cathodes were composed of commercial NCM811 with an areal loading of 4.0-5.0 mg·cm −1 . The working voltage window was set at 0.01-4.3 V, and the specific capacity was calculated according to the weight of NCM. Moreover, the capacity of the Co 3 O 4 electrode was almost equal to that of the positive electrode, and moderate pre-lithiation on the Co 3 O 4 negative electrode was adopted before assembling full battery. Figure 12a reveals the initial charge-discharge curves of the full batteries at 40 mA·g −1 . The inclined charging and discharging platforms in the full batteries were similar to those of the NCM/Li half-cell. Figure 12b presents the cycle performance of the assembled full batteries at 100 mA·g −1 . The cycle stability was relatively good, and the capacities of the batteries using

Synthesis of the Precursors
The precursors were prepared using the homogeneous precipitation method [52]. In the typical preparation process, 0.004 mol cobalt salts and 0.02 mol urea were weighed and transferred into a 100 mL Teflon liner with 50 mL distilled water, followed by 15 min ultrasonic treatment. The liner was then placed in a stainless steel autoclave and heated at 95 °C for 8 h. The precipitates in the cooling autoclave were separated, washed repeatedly with distilled water and ethanol, and dehydrated at 60 °C for 12 h. Four kinds of precursors were used, corresponding to four cobalt salts, respectively, and the preparation procedure was the same.

Synthesis of Cobalt Oxides
The dried precursors were calcined in the muffle furnace under air atmosphere. The sintering temperature and holding time were set at 550 °C and 2 h, respectively. The heating rate was fixed at 5 °C·min −1 . The black cobalt oxide powders were obtained after the furnace cooling to ambient temperature. The four cobalt oxides were marked as Co3O4-Cl, Co3O4-SO4, Co3O4-AC, and Co3O4-NO3, respectively.

Material Characterization
The crystal structure, morphologies, and elemental distribution of the synthesized materials and electrodes were characterized by X-ray diffraction (XRD, Bruker AXS D8, Karlsruhe, Germany) at 4°·min −1 , and scanning electron microscopy (SEM, Hitachi S-4800, Tokyo, Japan) combined with EDS detector. The chemical states of the obtained materials were tested by applying Fourier transformation infrared spectroscopy (FT-IR, Bruker V80, Karlsruhe, Germany) at 4000-400 cm −1 and Raman spectroscopy (Thermo Scientific DXR, Waltham, MA, USA, λ = 532 nm). Thermal behavior analysis was carried out in air using a thermal gravimetric analyzer (TGA, Thermo Scientific, Waltham, MA, USA) at 10 °C·min −1 from 30 °C to 800 °C.

Synthesis of the Precursors
The precursors were prepared using the homogeneous precipitation method [52]. In the typical preparation process, 0.004 mol cobalt salts and 0.02 mol urea were weighed and transferred into a 100 mL Teflon liner with 50 mL distilled water, followed by 15 min ultrasonic treatment. The liner was then placed in a stainless steel autoclave and heated at 95 • C for 8 h. The precipitates in the cooling autoclave were separated, washed repeatedly with distilled water and ethanol, and dehydrated at 60 • C for 12 h. Four kinds of precursors were used, corresponding to four cobalt salts, respectively, and the preparation procedure was the same.

Synthesis of Cobalt Oxides
The dried precursors were calcined in the muffle furnace under air atmosphere. The sintering temperature and holding time were set at 550 • C and 2 h, respectively. The heating rate was fixed at 5 • C·min −1 . The black cobalt oxide powders were obtained after the furnace cooling to ambient temperature. The four cobalt oxides were marked as Co 3 O 4 -Cl, Co 3 O 4 -SO 4 , Co 3 O 4 -AC, and Co 3 O 4 -NO 3 , respectively.

Material Characterization
The crystal structure, morphologies, and elemental distribution of the synthesized materials and electrodes were characterized by X-ray diffraction (XRD, Bruker AXS D8, Karlsruhe, Germany) at 4 • ·min −1 , and scanning electron microscopy (SEM, Hitachi S-4800, Tokyo, Japan) combined with EDS detector. The chemical states of the obtained materials were tested by applying Fourier transformation infrared spectroscopy (FT-IR, Bruker V80, Karlsruhe, Germany) at 4000-400 cm −1 and Raman spectroscopy (Thermo Scientific DXR, Waltham, MA, USA, λ = 532 nm). Thermal behavior analysis was carried out in air using a thermal gravimetric analyzer (TGA, Thermo Scientific, Waltham, MA, USA) at 10 • C·min −1 from 30 • C to 800 • C.

Electrochemical Measurements
Electrochemical measurements of the prepared anode materials were carried out in 2032-type coin cells, including working electrodes, separators (Celgard 2400, Φ = 17 mm), counter electrodes (Li foils, Φ = 16 mm), and electrolyte (1 M LiPF6 in EC/DEC, EC: DEC = 3:7, mass ratio). The working electrodes (Φ = 12 mm) were fabricated by blending active materials, Super P and poly(vinylidene fluoride) (70:20:10, weight ratio) in the dispersant (N-methyl-2-pyrrolidone, NMP) for 4 h, coating the slurry on the copper foil, evaporating the solvent in the oven, and subsequently punching into disks. The mass loading of active materials was controlled at 0.7-1.0 mg·cm −1 . The cells were assembled in the glove-box and aged for 4 h before testing. Galvanostatic discharge/charge results were acquired using Land instruments (CT-2001A) within the given potential region (0.01-3.0 V, vs. Li/Li + ). Cyclic voltammograms (CV) and electrochemical impedance spectra (EIS) were conducted on an electrochemical workstation (Bio-logic VSP) with a set potential window (0.01-3.0 V, vs. Li/Li + ) and frequency range (10 mHz-100 kHz).

Conclusions
In summary, the aggregation morphology of the Co 3 O 4 materials can be easily regulated by synthesizing precursors in various shapes with different cobalt sources in the coprecipitation process, including spherical, flower-like, irregular, and urchin-like Co 3 O 4 . Applied in a battery system, flower-like Co 3 O 4 with nanorods displayed a superior reversible capacity of 910.7 mAh·g −1 at 500 mA·g −1 and 717 mAh·g −1 at 1000 mA·g −1 after 500 cycles, enhanced cyclic stability with a capacity retention rate of 92.7% at 500 mA·g −1 and 78.27% at 1000 mA·g −1 after 500 cycles, and high performance in harsh environments (long-term storage and high temperature). Although the capacitive character and diffusion property of flower-like Co 3 O 4 were slightly poor, the lower charge-transfer resistance and stable electrode structure owing to the advantage of its unique aggregation morphology contributed a satisfying performance. When constructing Co 3 O 4 -based anode materials, the property of single grains and aggregates should be simultaneously considered, especially the morphology of the aggregates. Thus, the relationship between aggregation morphology and performance of Co 3 O 4 can be applied to synthesize other high-performance anode materials for LIBs.