The carbon nanotube arrays are confirmed to form in the first-step pyrolysis in the H
2/Ar atmosphere. As shown in Fig. S3c (Supporting information), uniformly grown carbon nanotubes can already be observed on the surface of carbonized nanosheets (derived from Co
0.2/Zn-MOF nanosheets) before the nitridation treatment. During the thermal treatment in the H
2/Ar atmosphere, cobalt ions are firstly reduced into metallic Co that nucleate and grow to tiny nanocrystals. Then, organic residuals acting as the carbon source grow into carbon nanotubes, which are catalyzed by the
in-
situ formed Co nanocatalysts. When Zn-MOF nanosheets without Co incorporation were used as the precursor, only carbonized nanosheets with porous surface were obtained (Fig. S3e in Supporting information). With the use of Co
0.1/Zn-MOF nanosheets precursor containing a low amount of Co, none of the carbon nanotubes could be observed (Fig. S3d in Supporting information) due to the lack of Co nanocatalysts. As the Co/Zn molar ratios increase to 0.3 and 0.4 in the Co/Zn-MOF precursors, the TEM images in Figs. S3a and b (Supporting information) clearly show the formation of carbon nanotubes after thermal treatment in H
2/Ar atmosphere. However, in comparison with the products (Fig. S3c) prepared from Co
0.2/Zn-MOF nanosheets, the sizes of embedded Co nanoparticles obviously become larger due to the sintering of Co species under the relatively high content. These results indicate the critical role of
in-
situ formed Co nanocrystals for catalyzing the growth of carbon nanotubes, which has been widely confirmed in previous reports [
35-
37]. Moreover, an optimized amount of incorporated Co is benefit for the construction of hierarchical structures with tiny embedded Co nanoparticles. The effect of first-step calcination temperature on the composites derived from Co
0.2/Zn-MOF nanosheets was also investigated and shown in Fig. S4 (Supporting information). At a low temperature of 700 ℃, tiny Co nanoparticles were observed throughout the nanosheets, while only a few short carbon nanotubes appeared on the surface (Fig. S4a). The incomplete pyrolysis of Co
0.2/Zn-MOFs under such a low temperature mainly leads to the insufficient growth of carbon nanotubes. As the temperatures increased to 900 ℃ and 1000 ℃, enhanced pyrolysis promoted the catalytic growth of longer carbon nanotubes, while the Co nanoparticles also grew into larger sizes (Figs. S4c and d). As a result, a Co/Zn molar ratio of 0.2 and a pyrolysis temperature of 800 ℃ were set as the optimized conditions in the first-step thermal treatment. For comparison, Co
0.2/Zn-MOF rhombic dodecahedrons with the size of ~250 nm were also synthesized in methanol media for the use as a precursor (Fig. S5a in Supporting information). After the same first-step treatment, only rhombic dodecahedrons with embedded Co nanoparticles were obtained (Figs. S5b and c in Supporting information), further indicating the advantages of the Co
0.2/Zn-MOF nanosheets for the construction of hierarchical structure. The compositions and structures of the series of Co/N doped carbon nanosheets (denoted as Co
x/NC/NSs-y; x=0.1, 0.2, 0.3 and 0.4, notes the Co/Zn molar ratio in the precursors; y=700, 800, 900 and 1000, notes the first-step pyrolysis temperature) were firstly identified and compared
via X-ray diffraction (XRD) measurements. As exhibited in Fig. S6 (Supporting information), no diffraction peaks belonging to metal Zn are observed due to the evaporation of low boiling point Zn [
38,
39]. Co
0.1/NC/NSs-800 solely give the signals of the carbonaceous component, which is consistent with the observed lack of Co nanocatalysts. All the other samples present the components of metallic Co and carbon. When the pyrolysis temperature is fixed, the intensity of the peaks of metallic Co obviously increases with the increase of Co amount. In the case with constant Co/Zn molar ratio in the precursors, the degree of graphitization shows gradual enhancement with the increase in calcination temperature. The contents of N atoms in the carbon matrixes are confirmed by the results of element analysis in Table S2 (Supporting information). The X-ray photoelectron spectroscopy (XPS) spectra for N 1s of the Co
x/NC/NSs-y reveal the types of N species in each sample. All the N 1s spectra can be fitted into four peaks corresponding to pyridinic N, Co-N
x, pyrrolic N and graphitic N (Fig. S7a and b in Supporting information). Fig. S8 and Table S3 (Supporting information) summaries the relative content of each type of N species in the samples. Notably, the Co
0.2/NC/NSs-800 possess a high relative content of Co-N
x (20.57%) which is confirmed to be the critical active center for ORR [
40]. The Co 2p spectrum of Co
0.2/NC/NSs-800 (Fig. S7c in Supporting information) can be deconvoluted into three characteristic peaks of Co° (778.6 eV), Co
2+ (781.8 eV) and Co
3+ (780.3 eV) [
41-
43]. Moreover, the Brunauer–Emmett–Teller (BET) calculations from N
2 sorption isotherms curves assign the largest specific surface area of 685.2 m
2/g to the Co
0.2/NC/NSs-800 (Fig. S9 and Table S4 in Supporting information). The Co
0.2/NC/NSs-800 with well-defined hierarchical structure, uniformly dispersed Co nanoparticles, high content of Co-N
x and the largest specific surface area is thus chosen as the optimized precursor for nitridation treatment to investigate the composition and electrocatalytic performance of the resultant Co/NCNTs/NSs.