The CNT papers were prepared by a vacuum filtration method. In a typical process, the pristine CNTs were sheared into CNT cotton by high-speed shearing, and then immersed into the solution of hydrochloric acid (5 mol/L) for 48 h to remove catalyst particles. The purified CNTs were sheared into CNT cotton again and dispersed into deionized water by ultrasonic treatment with the help of Tween-80 (as the dispersant). Then, the dispersed CNT solutions were filtered through a microporous cellulose filter membrane using vacuum filtration and washed by deionized water repeatedly to remove remnant dispersants. After dissolving the cellulose filter membrane by acetone, a freestanding CNT paper with a diameter of 40 mm was obtained. All of the chemical reagents were purchased from Sinopharm Chemical reagent Co., Ltd with analytical reagent grade (AR).

Ni_xCo_2-x(OH)₂CO₃ was grown on CNT paper by a modified chemical bath deposition method. Typically, NiCl₂ and CoCl₂ were dissolved in 20 ml water with Co/Ni ratio 1/0, 2/1, 1/1 and 1/2, are denoted as Co, Co2Ni, CoNi and CoNi2, respectively. The total concentration of NiCl₂ and CoCl₂ were kept at 1 M. Then, a piece of CNT paper (20 mg) was immersed into the solution, respectively. At last, 0.62 g urea was solved into the solution. The mixed solution was put into a glass bottle and its top was screwed. The bottle was put into an oven with the temperature of 80°C for 24 h. After that, the CNT paper was taken out and washed by water. The washed paper was dried at 60°C in air.

The morphology and microstructure of the samples were systematically investigated by field emission scanning electron microscopy (FE-SEM, Quanta 400 FEG, FEI), high resolution transmission electron microscopy (HRTEM, Tecnai G2 F20S-Twin, FEI), and X-ray diffraction (XRD, D8 Advance Powder X-ray diffractometer, Bruker AXS), X-ray photoelectron spectroscopy (XPS, EscaLab 250Xi). The prepared CNT paper was weighed of 10 mg for each and pressed onto the Ni-foam of 1.0 ⅹ 1.0 cm² as the electrode. Electrochemical experiments were carried out in CHI-660C electrochemical workstation and LANHE CT 2001A electrochemical cell test equipment. A three-electrode system was chosen to test the electrochemical performance of the materials. A platinum wire was used as the counter electrode, and a calomel electrode was used as the reference electrode. A 6 M KOH aqueous solution was chosen as the electrolyte. CV performances were tested in a potential range of 0 V up to 0.8 V under scan rates of 5 mV/s. The cycling stability was tested in LANHE 2001A (5V 50 mA) battery station.

As shown in Figure 1A, for the carbonate nanowires with Co/Ni ratio 1/0, 2/1, 1/1 and 1/2 denoted as Co, Co2Ni, CoNi and CoNi2, the loading mass of Co/Ni basic carbonate nanowires on CNT network increases along with the decreasing Co/Ni mol ratio. After the deposition of Co/Ni basic carbonate nanowires, the masses of CNT network become 244%, 254%, 331% and 375% of the pristine mass for Co, Co2Ni, CoNi and CoNi2, respectively. It means that the loading ratios of basic carbonate in the composite network are 59, 61, 70 and 73% for Co, Co2Ni, CoNi and CoNi2, respectively. To enhance the performance of the composite network, high loading of active materials is beneficial. However, the formation mechanism of the loading mass under different Co/Ni ratios should be further investigated. The thermal decomposition of Co/Ni carbonate compounds was monitored by TGA, which indicated that Co/Ni carbonate with different Co/Ni ratio exhibit quite similar thermal evolution as shown in Supplementary Figure S1A. As it can be seen in Supplementary Figure S1B, the CoNi2 carbonate sample experience three weight losses : CNT paper caused the first 20% mass loss in the interval [250–550]°C ; a second 12% weigh loss at [550–700]°C is attributed to the decomposition of OH in the composition; a third 6% progressive weight loss at [700–900]°C caused by the degradation of CO₃.

XRD patterns of the four Co/Ni basic carbonate/CNT composites are shown in Figure 1B. Different from Ni basic carbonate nanowires with typical Ni₂(OH)₂CO₃ structure in our previous discussion,[31] the four samples of Co basic carbonate with Ni doping exhibits typical Co₂(OH)₂CO₃ crystal structure (PDF: 48-0083). It means that even the Ni doping degree is high to 67%, the existence of Co can still keep the crystal structure of the basic carbonate as pure Co basic carbonate. Furthermore, the shapes of Co, Co2Ni, CoNi and CoNi2 are similar as shown in Figure 2, which also indicates the similar crystal structure and consistent with TGA results.

According to our previous research, the Ni₂(OH)₂CO₃ formed a nanowire array with thin diameter on individual CNTs (Chen et al., 2015a) Figure 2 shows the micro morphologies of Co/Ni basic carbonates with different Co/Ni ratios on CNT network. The pure Co₂(OH)₂CO₃ nanowire has larger diameter with large particles as the center as shown in Figures 2A,B. Only large particles (1–5 μm) with large-diameter nanowires (100–200 nm) were grown on CNT paper without close contact to individual CNTs on it. When doping the basic carbonate with Ni to achieve Co2Ni, as shown in Figures 2C,D Co2Ni basic carbonate forms dense nanowires on CNT paper, largely different from pure Co basic carbonate with low growth density on CNT paper (see Figure 2A). It means that the main loading form of pure Co basic carbonate on CNT paper may be the particles. However, the diameters of pure Co and Co2Ni basic carbonate nanowires exhibit no obvious difference. The increasing content of Ni will induce thinner basic carbonate nanowire formation as shown in Figures 2E,F (Co/Ni = 1/1). The CoNi nanowire with small diameter can form sea-urchinlike structures on individual CNTs. With higher Ni doping degree, Figures 2G,H reveal that the CoNi2 basic carbonate nanowire forms a mace-like structure on individual CNTs in the network. The nanowires can still be nucleated on CNTs, which should be attributed to their thin diameter. However, the diameter of CoNi2 basic carbonate nanowire is obviously larger than Ni₂(OH)₂CO₃ (Chen et al., 2015b) Generally speaking, the diameter of the Co/Ni basic carbonate nanowires decreases along with the increasing Ni doping, and only high Ni content could ensure the nanowires growth on individual CNTs to form hierarchical core/shell nanowires as shown in Figure 2. TEM images in Figures 3A,B prove that the CoNi2 basic carbonate nanowires are grown on individual CNTs. However, the nanowire is instable under electron beam, and is separated into particles as shown in Figures 3C,D.

Co/Ni basic carbonates can be electrochemically activated from charge/discharge cycles, especially in conductive network such as graphene foam and CNT film. (Chen et al., 2015a; Chen et al., 2016b) The four samples of Co-Ni basic carbonates could be also activated by electrochemical cycles as shown in Figure 4. However, the cyclic voltammetry tests of the four samples exhibit different active extents after 50 cycles, though the loading ratio of basic carbonates in the four composites has no large difference (59% for Co, 61% for Co2Ni, 70% for CoNi, 73% for CoNi2). The Co and Co2Ni with low Ni doping degree have no obvious enlargement after 50 cycles of CV scanning (see Figures 4A,B). However, CoNi2 and CoNi hybrids exhibit effective activation in 50 cycles of CV scanning with enlarged cyclic areas (see Figures 4C,D). This difference cannot be simply explained from the increase of basic carbonate loading on CNT films. It means that the chemical composition or the morphology of the nanowires have large influence on the electrochemical activation of the basic carbonates.

Different from chemical conversion (Zhu et al., 2013) , the electrochemical conversion mechanism can be demonstrated as the equations below, which has also been discussed in our previous reports (Chen et al., 2015a; Chen et al, 2016a). The Ni_xCo_2-x(OH)₂CO₃ crystal has its pseudocapacitive core Ni²⁺ and Co²⁺, which could be converted to Ni³⁺ and Co³⁺ in electrochemical charge process, and the latter ion exited as Ni_x/2Co_1-x/2OOH, which was converted by Ni_xCo_2-x(OH)₂CO₃. CO³⁻ was resolved into the electrolyte and Ni_x/2Co_1-x/2(OH)₂ kept as a solid on the CNT paper. After that, in the discharge process, Ni_x/2Co_1-x/2OOH converted to Ni_x/2Co_1-x/2(OH)₂ and CO₃ ²⁻ was not existed in the solid phase on the CNT papers Eqs (1), (2) N i x C o 2 − x ( O H ) 2 C O 3 + 4 O H − → 2 N i x / 2 C o 1 − x / 2 O O H + C O 3 2 − + 2 H 2 O + 2 e - (1) N i x / 2 C o 1 - x / 2 O O H + H 2 O + e − → N i x / 2 C o 1 − x / 2 ( O H ) 2 + O H − (2)

X-ray photoelectron spectroscopy (XPS) was utilized to evaluate the surface chemical state of the converted Co/Ni hydroxide materials in the range of 0–1000 eV. As presented in Figure 5A, the diffraction peaks located at 284.3, 529.9, 779.1 and 856.5 eV correspond to C, O, Co and Ni elements in the converted CoNi2 hydroxide materials. The results further indicate that Co/Ni hydroxide were successfully prepared. In the high-resolution O 1 s spectrum in Figure 5B, the peaks at 529.6,531.5 and 532.8 indicate the presence of metal-oxygen bond, O-C-O, and the O-H groups, respectively. The Co 2p and Ni 2p XPS peak spectra were computer fitted using a Gaussian fitting method considering two spin-orbit doublets and two shakeup satellites (marked as “Sat.”). The high-resolution XPS spectrum of Ni 2p (Figure 5C.) reveals that two obvious shakeup satellites (indicated as “Sat”) close to two spin-orbit doublets at 855.7 and 873.1 eV, that can be assigned to Ni 2p_3/2 and Ni 2p_1/2 signals, respectively. It suggests the existence of both Ni²⁺ and Ni³⁺. The intense satellite peaks indicate that Ni³⁺ is the majority. In the case of Co 2p XPS spectrum (Figure 5D), the spin-orbit splitting value of Co 2p_1/2(796.1 eV) and Co 2p_3/2(781.2 eV) indicated both Co³⁺ and Co²⁺ in the Co/Ni carbonate sample. The weak satellite peaks indicate that the majority of cobalt is Co³⁺. The XPS further demonstrated Ni²⁺ and Co²⁺ could be converted to Ni³⁺ and Co³⁺ in electrochemical charge process, that consistent with our proposed mechanism.

The morphologies of the activated composite films are shown in Figure 6. It reveals that the converted hydroxides have different shapes. The pure Co hydroxide has hexagonal shape with large thickness as shown in Figures 6A,B. Furthermore, such nanosheet has no contact with CNT network. For Co2Ni, the converted hydroxide has a particle-like shape and poor contact with CNT network (see Figures 6C,D). Furthermore, CoNi with the increasing Ni doping degree, the converted hydroxide nanosheets form small porous balls as shown in Figures 6E,F, but the thickness of the nanosheets seems still large. At last, CoNi2 hydroxide has thin nanosheet structures with flower-like structures as shown in Figures 6G,H which could largely utilize the active areas of the hydroxides. Figure 7A indicates the electrochemical conversion from CoNi2 basic carbonate to relative hydroxide, that thin nanowires on the CNTs results in tied contact on individual CNTs. Combining the morphologies in Figure 6 and the activation effects in Figure 4, it reveals that the tied contact between Co-Ni and individual CNTs can ensure basic carbonate being converted to thin nanosheets with high electrochemical performance. Although the basic carbonate with low Ni doping degree can also be electrochemically converted to hydroxide, the relative increase of electrochemical performance is not definite.

As shown in Figure 7B, the electrochemical capacitive performance of the converted hydroxides/CNT composites exhibit large difference from the pristine basic carbonates with similar loading ratio. It reveals that the cyclic area of CoNi2 hybrid is larger than other three samples. The area of CoNi hybrid is a little smaller than CoNi2. Two other samples, Co and Co2Ni with low Ni doping degree exhibit much smaller area than the samples of CoNi and CoNi2. This result is consistent with the activation CV curves in Figure 4. From the charge/discharge curves under different current densities in three-electrode system, the rate performance of the four samples can be evaluated and shown in Figure 7C. The introducing of CNT network could effectively ensure the good rate performance of the four samples. All of the four samples have a capacitance decrease of about 20% from current density of 0.5–10 A g⁻¹. For the composite film with converted CoNi2 hydroxide, the electrochemical capacitance decreases from 1497 F g⁻¹ at 0.5A·g⁻¹–1192 F g⁻¹ at 10 A g⁻¹. For the converted CoNi hybrid, the relative electrochemical capacitances are 921 F g⁻¹ at 0.5 A g⁻¹ and 762 F g⁻¹ at 10 A g⁻¹. For the other two samples with low Ni doping degrees in the hydroxide, the practical capacitance of their composites is lower than 300 F g⁻¹, which is much lower than the samples with high Ni ratio (Co2Ni and CoNi). It means that high Ni doping degree in Co/Ni basic carbonates can ensure the high practical capacitance of the converted hydroxide. However, comparing with the rate performance of pure Ni converted hydroxide/CNT composite film reported by our previous research, the introduce of Co enhances the rate performance of the hydroxide. For both directly grown Co-Ni hydroxides and converted hydroxides from basic carbonates (Chen et al., 2015b) a core-shell structure with suitable-density arrays of hydroxide/basic carbonate nanosheets or nanowires on individual CNTs in the CNT network usually exhibited higher specific capacitance comparing with other samples. We have summarized electrochemical performance data of related Co/Ni based electrode materials for supercapacitors in Table 1. Even at an ultrahigh current density of 10 A g⁻¹, the electrode exhibited a high capacitance of 1192 F g⁻¹, which is comparatively higher than those reported earlier in literature using Co/Ni/carbon system (Liu et al., 2020; Hosseini and Shahrokhian 2019; Rahimi et al., 2018; Chen et al., 2015a) It indicates the importance of high-conductive CNT network in the composites. Only close contact between nanostructures with suitable porous structures and individual CNTs can ensure the high electrochemical performance of the composites.

Figures 7D,E compare the discharge curve of the four samples under low (0.5 A g⁻¹) and high (10 A g⁻¹) current densities. The CoNi2, Co2Ni and Co converted hydroxides display the sole discharge plateau potential, but Co/Ni 1/1 converted hydroxide has two plateau potentials, which agrees with two pairs of oxidation/reduction peaks in CV curves (Figure 4 and Figure 7B). It might be attributed to the special oxidation/reduction pair of Co ion in such Co/Ni ration hydroxide. Although Co has more oxidation/reduction pairs than Ni, it cannot be exhibited in many composites. In the CoNi 1/1 hybrid, the combination of Ni and Co with such ratio 1/1 can largely utilize the oxidation/reduction pairs. When increasing or decreasing the Ni ratio in the hydroxide, only one oxidation/reduction pair appears. The full charge/discharge curves of pure Co, Co2Ni, CoNi, CoNi2 hybrids and pure Ni are shown in Supplementary Figure S2. For the CoNi hybrid, which seems a triangle, just similar as electrochemical double-layer capacitor electrodes, shows typical platforms for Ni²⁺/Ni³⁺ conversion. Thus, although the practical capacitance of CoNi hybird is not as high as that of CoNi2 hybrid, it is still valuable to be combined with carbonous materials to assemble asymmetric supercapacitors.

Figure 8 shows the cyclic performance of pristine CoNi2 carbonate/CNT hybird at the current density of 2 A g⁻¹. It reveals a slow activation process comparing with pure Ni carbonate/CNT hybrid. After 700 cycles of charge/discharge, the specific capacitance of the sample increases to a maximum value and still kept around 1600 F g⁻¹ after 5,000 cycles with a stable Coulombic efficiency.