In Situ Reduced Multi-Core Yolk–Shell Co@C Nanospheres for Broadband Microwave Absorption

The preparation of yolk–shell microwave absorption materials with low density and excellent microwave absorption property requires reasonable design and economical manufacture. In this study, an efficient strategy without any templates or reducing gases has been designed to fabricate multi-core yolk–shell Co@C nanospheres by high temperature carbonization. The results showed that Co3O4 was completely reduced by the carbon shell to metal cobalt at temperatures above 750 °C. This unique multi-core yolk–shell structure with shell of 600 nm and multiple cores of tens of nanometers can provide sufficient interface and space to reflect and scatter electromagnetic waves. At the same time, the metal cobalt layer and carbon layer provide magnetic loss ability and dielectric loss ability, respectively, making the composite show good wave absorption performance. The minimal RL value of samples carbonized at 750 °C reaches −40 dB and the efficient absorption band reaches 9 GHz with the thickness ranges from 2–9 mm. Therefore, this is a facile, effective and economical strategy to prepare yolk–shell structure, which provides a new idea for the preparation of microwave absorption materials.


Introduction
With the rapid development of electronic devices, the microwave with a frequency in the 2~18 GHz ranges has become a main part of electromagnetic pollution, so it is urgent to investigate the microwave absorbing materials (MAM) [1][2][3][4]. Among these materials, transition magnetic metals (Fe, Co, Ni) with high specific magnetism, outstanding electric conductivity and low cost has attracted a lot of attention as MAM because of its excellent magnetic loss capability endowed by natural resonance and eddy current effects [5][6][7][8]. However, in order to achieve good impedance matching, so that the electromagnetic wave can incident into the material as much as possible rather than be reflected, it is required that the impedance of the material surface should be as close as possible to the characteristic impedance of free space, that is, the values of relative complex permeability and relative complex dielectric constant should be as close as possible [9][10][11][12]. Hence, the poor impedance matching which causes an undesirable microwave absorption capacity is still a great challenge to transition metals.
In general, it is a common strategy that combines the magnetic loss materials with dielectric loss materials to form core-shell or yolk-shell composites with excellent impedance matching [13][14][15][16]. Yolk-shell structure is a special kind of core-shell structure and the structure can also be called a core@void@shell structure, which means that there is not direct contact between the core and the shell. The existence of void provides unique advantages such as low density and large surface area [17][18][19]. Furthermore, the outside dielectric shell (SnO 2 [20], MnO 2 [21], carbon [22]) cannot only effectively protect naked magnetic

Characterization
The Crystalline structures and phase composition of obtained products were characterized by X-ray diffraction (XRD, smartlab9) with a Cu Kα source (40 KV,200 mA). The morphology and size were obtained by scanning electron microscopy (SEM, JSM-7001F) with an accelerating voltage of 20 kV. The transmission electron microscopy (TEM) images were characterized on JEM-2100F with an accelerating voltage of 200 kV. The magnetic hysteresis loops were measured by vibrating sample magnetometer (VSM, Lake Shore Cryotronics, Westerville, OH, USA) at room temperature. The X-ray photoelectron spectrometer (XPS, Axis Supra) was measured on an Axis Supra with an Al Kα X-ray source. The relative complex permeability and relative permittivity of the sample in the frequency range of 2-18 GHz were characterized by a vector network analyzer (VNA, Agilent N5234A). After mixing the products with the paraffin in the ratio of 3:7 uniformly, the annular sample with outer diameter of 7 mm and inner diameter of 3 mm was pressed, and the thickness was about 2.00 mm.

Results and Discussion
The synthesis process of yolk-shell Co@C nanospheres is schematically illustrated in Figure 1. Co-glycerate nanospheres are formed by solvothermal action of Co 3+ and glycerol, and are then calcined in air to form Co 3 O 4 . Then, resorcinol and formaldehyde formed a phenolic resin layer on the surface of Co 3 O 4 by in situ polymerization. Finally, the microspheres were carbonized and reduced in protective gas at high temperature to form a multi-core yolk-shell Co@C structure.
XRD patterns of Co 3 O 4 , S1, S2 and S3 are shown in Figure 2. Co 3 O 4 nanospheres' diffraction peaks can well match the spinel phase of Co 3 O 4 (JCPDS 73-1701), indicating that the precursor transformed into the Co 3 O 4 phase after being calcined. The same diffraction peaks at 44 • , 51 • , 76 • in all of the samples can be in good agreement with (111), (200), and (220) planes of fcc metallic cobalt (JCPDS 89-7093), and the remaining diffraction peaks of S1 are matched with that of CoO (JCPDS 43-1004), which indicates that the calcination temperature of 650 • C is not enough to help the carbon layer completely reduce Co 3 O 4 to zero-valent metallic cobalt. At the same time, although the formed metal cobalt can catalyze the transition from amorphous carbon layer to graphite, the diffraction peak of graphite carbon does not appear in the XRD patterns [47,48]. XRD patterns of Co3O4, S1, S2 and S3 are shown in Figure 2. Co3O4 nanospheres' diffraction peaks can well match the spinel phase of Co3O4 (JCPDS 73-1701), indicating that the precursor transformed into the Co3O4 phase after being calcined. The same diffraction peaks at 44°, 51°, 76° in all of the samples can be in good agreement with (111), (200), and (220) planes of fcc metallic cobalt (JCPDS 89-7093), and the remaining diffraction peaks of S1 are matched with that of CoO (JCPDS 43-1004), which indicates that the calcination temperature of 650 °C is not enough to help the carbon layer completely reduce Co3O4 to zero-valent metallic cobalt. At the same time, although the formed metal cobalt can catalyze the transition from amorphous carbon layer to graphite, the diffraction peak of graphite carbon does not appear in the XRD patterns [47,48].  Figure 3 illustrates the morphology evolution of Co3O4 to yolk-shell Co@C nanospheres at different calcination temperatures. Co3O4 nanospheres are initially synthesized with an average size of 700 nm (Figure 3a). After in situ polymerization and calcination at 650 °C for 1 h, the shape of the nanoparticles is distorted and the internal particles shrink to form several nanospheres of different sizes. Combined with the XRD patterns, it is found that the inner spheres of the carbon layer are composed of metal cobalt and CoO.  XRD patterns of Co3O4, S1, S2 and S3 are shown in Figure 2. Co3O4 nanospheres' diffraction peaks can well match the spinel phase of Co3O4 (JCPDS 73-1701), indicating that the precursor transformed into the Co3O4 phase after being calcined. The same diffraction peaks at 44°, 51°, 76° in all of the samples can be in good agreement with (111), (200), and (220) planes of fcc metallic cobalt (JCPDS 89-7093), and the remaining diffraction peaks of S1 are matched with that of CoO (JCPDS 43-1004), which indicates that the calcination temperature of 650 °C is not enough to help the carbon layer completely reduce Co3O4 to zero-valent metallic cobalt. At the same time, although the formed metal cobalt can catalyze the transition from amorphous carbon layer to graphite, the diffraction peak of graphite carbon does not appear in the XRD patterns [47,48].     (Figure 3a). After in situ polymerization and calcination at 650 • C for 1 h, the shape of the nanoparticles is distorted and the internal particles shrink to form several nanospheres of different sizes. Combined with the XRD patterns, it is found that the inner spheres of the carbon layer are composed of metal cobalt and CoO. Compared with the sample calcinated at 650 • C, the outer carbon layer is preserved better in spherical shape at 750 and 850 • C, and the inner cobalt oxide is completely reduced to several irregular metal cobalt particles with similar size (Figure 3c,d). The morphology and structure of the S2 is further characterized by TEM. As can be clearly seen from the image, there are many small particles of tens of nanometers evenly distributed in the inner space wrapped by the carbon layer ( Figure 3e). HRTEM images show that the lattice fringes of the internal small particle can match well with the (111) crystal plane of metal cobalt, further proving that cobalt particles are formed after in situ reduction of Co 3 O 4 in the carbon shell (Figure 3f). several irregular metal cobalt particles with similar size (Figure 3c,d). The morphology and structure of the S2 is further characterized by TEM. As can be clearly seen from the image, there are many small particles of tens of nanometers evenly distributed in the inner space wrapped by the carbon layer ( Figure 3e). HRTEM images show that the lattice fringes of the internal small particle can match well with the (111) crystal plane of metal cobalt, further proving that cobalt particles are formed after in situ reduction of Co3O4 in the carbon shell ( Figure 3f). The magnetization properties of all the samples are characterized by VSM with a magnetic field of −10,000 Oe-10,000 Oe at room temperature. Figure 4a illustrates the magnetic hysteresis loops of S1, S2 and S3. The saturation magnetization (Ms) of S1, S2, S3 are 67.7, 81.0, 107.2 emu/g, respectively. It can be found that the Ms of the sample increases with the increase in the reduction temperature. The increase in Ms from S1 to S2 can be attributed to the enhancement of magnetism caused by the reduction of more antiferromagnet Co3O4 to ferromagnetic cobalt metal, while the increase in Ms from S2 to S3 may be due to the increase in the crystallinity of cobalt metal with the increase in reduction temperature. The coercivity (Hc) values of S1, S2, S3 are 98.1, 210.3, 90.9 Oe, respectively. For ferromagnetic MAM, initial permeability (μi) can be expressed by the following formula [24,49]: where a and b are two constants determined by the material composition, k is the magnetostriction constant, λ is the magnetostriction constant, and ξ is an elastic strain parameter of the crystal. An increase in μi usually means an increase in magnetic loss capacity, and both high MS and low Hc are beneficial to the increase in μi. The magnetization properties of all the samples are characterized by VSM with a magnetic field of −10,000 Oe-10,000 Oe at room temperature. Figure 4a illustrates the magnetic hysteresis loops of S1, S2 and S3. The saturation magnetization (M s ) of S1, S2, S3 are 67.7, 81.0, 107.2 emu/g, respectively. It can be found that the M s of the sample increases with the increase in the reduction temperature. The increase in M s from S1 to S2 can be attributed to the enhancement of magnetism caused by the reduction of more antiferromagnet Co 3 O 4 to ferromagnetic cobalt metal, while the increase in M s from S2 to S3 may be due to the increase in the crystallinity of cobalt metal with the increase in reduction temperature. The coercivity (H c ) values of S1, S2, S3 are 98.1, 210.3, 90.9 Oe, respectively. For ferromagnetic MAM, initial permeability (µ i ) can be expressed by the following formula [24,49]: where a and b are two constants determined by the material composition, k is the magnetostriction constant, λ is the magnetostriction constant, and ξ is an elastic strain parameter of the crystal. An increase in µ i usually means an increase in magnetic loss capacity, and both high M S and low H c are beneficial to the increase in µ i . In order to further elucidate the surface element composition and atomic configuration of S2, X-ray photoelectron spectroscopy (XPS) was characterized in Figure 4b-d. As can be seen from Figure 4b, the survey spectrum of S2 contains C 1s, Co 2p and O 1s. Figure 2 exhibits high resolution Co 2p spectrum, where the peaks at 781.9 eV and 797.6 eV are attributed to Co 2+ 2p 3/2 and Co 2+ 2p 1/2 spectrums, respectively [50,51]. Two peaks at 785.5 eV and 803.6 eV are the satellite peaks that accompany the two peaks of Co 2+ . The peak located at 795.0 eV can be assigned to Co 0 . Combined with the XRD spectrum of S2, it can be inferred that the peaks of Co 2+ are due to the oxidation of surface of sample. The C 1s peaks at 284.8, 285.7, 287.4, 288.4, 289.5 eV are assigned to aromatic ring structure, phenolic hydroxyl group, C-O linkage, carbonyl groups, carbonyl groups, respectively (Figure 4d). It is noteworthy that there is a peak that can be assigned to O 1s in the survey spectrum. According to the XRD result of S2, Co 3 O 4 was completely reduced to metal cobalt, so it was speculated that the occurrence of the O 1s peak might be due to the incomplete carbonization of phenolic resin and the oxidation of the surface of metal cobalt.  In order to further elucidate the surface element composition and atomic configuration of S2, X-ray photoelectron spectroscopy (XPS) was characterized in Figure 4b-d. As can be seen from Figure 4b, the survey spectrum of S2 contains C 1s, Co 2p and O 1s. Figure 2 exhibits high resolution Co 2p spectrum, where the peaks at 781.9 eV and 797.6 eV are attributed to Co 2+ 2p3/2 and Co 2+ 2p1/2 spectrums, respectively [50,51]. Two peaks at 785.5 eV and 803.6 eV are the satellite peaks that accompany the two peaks of Co 2+ . The peak located at 795.0 eV can be assigned to Co 0 . Combined with the XRD spectrum of S2, it can be inferred that the peaks of Co 2+ are due to the oxidation of surface of sample. The C 1s peaks at 284.8, 285.7, 287.4, 288.4, 289.5 eV are assigned to aromatic ring structure, phenolic hydroxyl group, C-O linkage, carbonyl groups, carbonyl groups, respectively (Figure 4d). It is noteworthy that there is a peak that can be assigned to O 1s in the survey spectrum. According to the XRD result of S2, Co3O4 was completely reduced to metal cobalt, so it was speculated that the occurrence of the O 1s peak might be due to the incomplete carbonization of phenolic resin and the oxidation of the surface of metal cobalt.
Complex permittivity (ε r =ε' -jεʹʹ) and complex permeability (μ r =μ'-jμʹʹ) of the MAM are measured by the vector network analyzer, which determines the absorbing performance of the MAM at the GHz frequency band (Figure 5a-f). The real parts of complex permittivity (ε′) and complex permeability (μ′) decide the storage capacity of electromagnetic energy, while the imaginary parts of complex permittivity (ε″) and complex permeability (μ″) represent the loss capacity of electromagnetic energy. Figure 5a shows the ε′ of all samples in the range of 2-18 GHz; it can be found that the ε′ of Co3O4 and S1 are Complex permittivity (ε r = ε − jε ) and complex permeability (µ r = µ − jµ ) of the MAM are measured by the vector network analyzer, which determines the absorbing performance of the MAM at the GHz frequency band (Figure 5a-f). The real parts of complex permittivity (ε ) and complex permeability (µ ) decide the storage capacity of electromagnetic energy, while the imaginary parts of complex permittivity (ε") and complex permeability (µ") represent the loss capacity of electromagnetic energy. Figure 5a shows the ε of all samples in the range of 2-18 GHz; it can be found that the ε of Co 3 O 4 and S1 are both around 2.5, and there is no change with the increase in frequency. The ε of S2 and S3 are relatively high and fluctuate around 6 as the frequency changes. Similarly, the ε" values of Co 3 O 4 and S1 fluctuate around 0 with frequency changes, while the ε" values of S2 and S3 fluctuate around 1 with frequency increasing, with a peak value of around 3.5 at 17.5 GHz (Figure 5b). According to the free electron theory, ε" ≈ 1/2πρfε o [52,53], where ρ stands for resistivity. It can be found that high conductivity can lead to the increase in ε". Therefore, the increase in ε" value of S2 and S3 may be due to the increase in conductivity caused by the increase in metal cobalt content and the deepening of graphitization degree of carbon layer. Dielectric tangent loss factors (tanδ ε = ε"/ε ) represents the dielectric loss capacity of a MAM. The tanδ ε of S2 and S3 varies around 0.2 and reaches the peak value of 0.7 at 17.5 GHz, while the values of Co 3 O 4 and S1 stay at about 0 with the change in frequency, which proves that the dielectric loss capacity of S2 and S3 is stronger than that of Co 3 O 4 and S1 in the frequency range of 2-18 GHz (Figure 5c). In terms of complex permeability, the values of both the real part and the imaginary part of the four samples are very close. The real part stays at about 1 with the increase in frequency, while the imaginary part stays at about 0 (Figure 5d,e). Similarly, the magnetic loss tangent factors (tanδµ = µ"/µ ) of the four samples are also very close, and stay at about 0 with the change in frequency, which proves that the four samples only have very weak magnetic loss ability close to each other, and the increase in saturation magnetization does not bring substantial enhancement of magnetic loss ability (Figure 5f). tivity caused by the increase in metal cobalt content and the deepening of graphitization degree of carbon layer. Dielectric tangent loss factors (tanδε = ε″/ε′) represents the dielectric loss capacity of a MAM. The tanδε of S2 and S3 varies around 0.2 and reaches the peak value of 0.7 at 17.5 GHz, while the values of Co3O4 and S1 stay at about 0 with the change in frequency, which proves that the dielectric loss capacity of S2 and S3 is stronger than that of Co3O4 and S1 in the frequency range of 2-18 GHz (Figure 5c). In terms of complex permeability, the values of both the real part and the imaginary part of the four samples are very close. The real part stays at about 1 with the increase in frequency, while the imaginary part stays at about 0 (Figure 5d,e). Similarly, the magnetic loss tangent factors (tanδμ = μ″/μ′) of the four samples are also very close, and stay at about 0 with the change in frequency, which proves that the four samples only have very weak magnetic loss ability close to each other, and the increase in saturation magnetization does not bring substantial enhancement of magnetic loss ability (Figure 5f). For magnetic microwave absorbing materials, the mechanisms of magnetic loss mainly include natural resonance, eddy current loss, hysteresis loss and domain wall resonance [54,55]. The hysteresis loss can be ignored in the case of a weak electromagnetic field, while the domain wall resonance has a good response behavior, mainly in the MHz frequency band. Therefore, the main mechanisms in the GHz frequency band are natural For magnetic microwave absorbing materials, the mechanisms of magnetic loss mainly include natural resonance, eddy current loss, hysteresis loss and domain wall resonance [54,55]. The hysteresis loss can be ignored in the case of a weak electromagnetic field, while the domain wall resonance has a good response behavior, mainly in the MHz frequency band. Therefore, the main mechanisms in the GHz frequency band are natural resonance and eddy current loss. The eddy current loss can be expressed by the following formula [56]: where σ is the electrical conductivity and µ 0 is the permeability in vacuum. The formula can be converted to C 0 = µ(µ ) −2 f −1 , which means that C 0 will remain constant if the magnetic loss of the material is caused solely by eddy current loss. Figure 6a shows the curve of C 0 of the samples changing with frequency in the range of 2-18 GHz. When the frequency is in the range of 6-18 GHz, the C 0 values of the four samples remain around 0 as the frequency changes, which proves that the magnetic loss of the four samples in this frequency band is completely caused by the eddy current loss. Meanwhile, according to the previous report, natural resonance generally occurs in the frequency band below 6 GHz. As shown in Figure 6a, there are certain resonance peaks in the C 0 values of the four samples in the frequency band between 2-6 GHz. Therefore, it can be speculated that these resonance peaks are caused by natural resonance.
quency is in the range of 6-18 GHz, the C0 values of the four samples remain around 0 as the frequency changes, which proves that the magnetic loss of the four samples in this frequency band is completely caused by the eddy current loss. Meanwhile, according to the previous report, natural resonance generally occurs in the frequency band below 6 GHz. As shown in Figure 6a, there are certain resonance peaks in the C0 values of the four samples in the frequency band between 2-6 GHz. Therefore, it can be speculated that these resonance peaks are caused by natural resonance. According to the transmission line theory, the attenuation constant α determines the ability of the MAM to attenuate the incident electromagnetic waves, which can be defined as [57]: where f is the frequency of microwave and c is the velocity of electromagnetic wave in vacuum. As demonstrated in Figure 6b, the α values of the four samples have different increasing trends with the increase in frequency. Among them, the α values of S2 and S3 have a more obvious increasing trend, which can reach 240 at 18 GHz, while the α values of Co3O4 and S1 can only reach 25 at 18 GHz. Obviously, in the frequency band of 2-18 GHz, the ability of samples S2 and S3 to attenuate electromagnetic waves is much higher than that of Co3O4 and S1. Such a result is consistent with the results of dielectric loss factor and magnetic loss factor in Figure 5.
Based on the complex permittivity and complex permeability of MAM, the reflection loss (RL) can be calculated by the transmission line theory. The calculation method is shown as follows [58,59]: where Zin represents the input impedance of the material, and the calculation formula is as follows: According to the transmission line theory, the attenuation constant α determines the ability of the MAM to attenuate the incident electromagnetic waves, which can be defined as [57]: where f is the frequency of microwave and c is the velocity of electromagnetic wave in vacuum. As demonstrated in Figure 6b, the α values of the four samples have different increasing trends with the increase in frequency. Among them, the α values of S2 and S3 have a more obvious increasing trend, which can reach 240 at 18 GHz, while the α values of Co 3 O 4 and S1 can only reach 25 at 18 GHz. Obviously, in the frequency band of 2-18 GHz, the ability of samples S2 and S3 to attenuate electromagnetic waves is much higher than that of Co 3 O 4 and S1. Such a result is consistent with the results of dielectric loss factor and magnetic loss factor in Figure 5.
Based on the complex permittivity and complex permeability of MAM, the reflection loss (RL) can be calculated by the transmission line theory. The calculation method is shown as follows [58,59]: where Z in represents the input impedance of the material, and the calculation formula is as follows: where c represents the speed of light in free space, f represents the frequency of electromagnetic waves, and d represents the thickness of the absorber. Figure 6 shows the three-dimensional diagrams and contour diagrams of the RL values of Co 3 O 4 , S1, S2 and S3 in the frequency range of 2-18 GHz and d in the range of 1-9 mm. It can be seen from the figure that Co 3 O 4 and S1 do not show distinct microwave absorption performance (Figure 7a-d). When the range of the thickness of the absorber is 1-9 mm and the range of frequency of microwave is 2-18 GHz, no RL values reach effective value of −10 dB (absorbing 90% electromagnetic waves). The minimum RL value of Co 3 O 4 is −4.4 dB when d is 7.6 mm and f is 18 GHz and, the minimum RL value of S1 is −7.4 dB when d is 8.9 mm and f is 15.8 GHz. The performance of such weak absorbing properties is consistent with their poor values of dielectric and magnetic loss factors. In contrast, S2 and S3 both have good wave absorption performance. When the d of S2 is 7.4 mm, the minimum RL value reaches −40 dB (at 13 GHz) and the EAB also reaches 2 GHz (from 11.9 GHZ to 13.9 GHz), while when the d of S3 is 7.9 mm, the minimum RL value is −36 dB and the EAB is 2.2 GHz (from 11.5 GHz to 13.7 GHz). In general, materials with high microwave absorption performance are requested to possess low absorber thickness, low density, wide efficient absorption bandwidth and strong absorption ability. According to Figure 8, compared with other reported yolk-shell structure composites, yolk-shell Co@C nanospheres have good absorption ability and efficient absorption bandwidth, but they need a high absorber thickness, which limits their practical application ability. However, considering the development potential of low-density yolk-shell materials in the field of microwave absorption materials, this simple and controllable method of in situ reduction can be extended to other metal materials to prepare the yolk-shell materials with better microwave absorption properties.   The microwave absorption performance can be due to the unique yolk-shell structure and the composite loss mechanism of metal cobalt and carbon shell. The microwave absorption mechanism of Co@C nanospheres is shown in Figure 9. First of all, the void space of the multi-core yolk-shell Co@C nanospheres could cause multiple reflection and scattering of electromagnetic wave, resulting in the absorption and exhaustion of incident microwave. Second, as shown in Figure 9b, there are abundant interfaces between multiple metal cobalt cores and carbon layers. Due to the difference in electrical conductivity, a large number of positive and negative charges will be accumulated at these interfaces. In the high-frequency electromagnetic environment, the accumulated charges at these interfaces will generate strong polarization and promote the attenuation of electromagnetic wave. Third, a large number of dipoles caused by defects in the amorphous carbon layer will produce strong relaxation polarization under the action of high frequency external electromagnetic field, which enhances the dielectric loss ability of the composite absorption material. Fourth, with the increase in reduction temperature, the degree of graphitization of the carbon layer increases, and a large number of electrons in the carbon layer can generate micro-current under the action of the external electromagnetic field, which is conducive to enhancing the loss of conductance and converting electromagnetic energy into heat energy. Lastly, natural resonance and eddy current loss will be caused by magnetic metal cobalt under the action of electromagnetic wave in the GHz band, so the electromagnetic wave can be reduced by magnetic loss. These mechanisms above work together to provide excellent microwave absorption performance of Co@C nanospheres. The microwave absorption performance can be due to the unique yolk-shell structure and the composite loss mechanism of metal cobalt and carbon shell. The microwave absorption mechanism of Co@C nanospheres is shown in Figure 9. First of all, the void space of the multi-core yolk-shell Co@C nanospheres could cause multiple reflection and scattering of electromagnetic wave, resulting in the absorption and exhaustion of incident microwave. Second, as shown in Figure 9b, there are abundant interfaces between multiple metal cobalt cores and carbon layers. Due to the difference in electrical conductivity, a large number of positive and negative charges will be accumulated at these interfaces. In the high-frequency electromagnetic environment, the accumulated charges at these interfaces will generate strong polarization and promote the attenuation of electromagnetic wave. Third, a large number of dipoles caused by defects in the amorphous carbon layer will produce strong relaxation polarization under the action of high frequency external electromagnetic field, which enhances the dielectric loss ability of the composite absorption material. Fourth, with the increase in reduction temperature, the degree of graphitization of the carbon layer increases, and a large number of electrons in the carbon layer can generate micro-current under the action of the external electromagnetic field, which is conducive to enhancing the loss of conductance and converting electromagnetic energy into heat energy. Lastly, natural resonance and eddy current loss will be caused by magnetic metal cobalt under the action of electromagnetic wave in the GHz band, so the electromagnetic wave can be reduced by magnetic loss. These mechanisms above work together to provide excellent microwave absorption performance of Co@C nanospheres.

Conclusions
In summary, in this study, hollow Co@C yolk-shell nanospheres with multiple metal cores were prepared by in situ reduction of Co3O4 in the carbon shell. This special structural design can not only reflect and scatter electromagnetic waves multiple times to achieve the purpose of the electromagnetic wave energy dissipation, but also can combine the dielectric loss and magnetic loss, making the composite Co@C absorption materials exhibt excellent absorption performance: the minimal RL value of samples carbonized at 750 °C reaches −40 dB and the EABD reaches 9 GHz with the thickness ranges from 2-9 mm. This unique microstructural design scheme provides a new idea for the preparation of new MAM.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to funder data retention policies.

Conflicts of Interest:
The authors declare no conflict of interest.

Conclusions
In summary, in this study, hollow Co@C yolk-shell nanospheres with multiple metal cores were prepared by in situ reduction of Co 3 O 4 in the carbon shell. This special structural design can not only reflect and scatter electromagnetic waves multiple times to achieve the purpose of the electromagnetic wave energy dissipation, but also can combine the dielectric loss and magnetic loss, making the composite Co@C absorption materials exhibt excellent absorption performance: the minimal RL value of samples carbonized at 750 • C reaches −40 dB and the EABD reaches 9 GHz with the thickness ranges from 2-9 mm. This unique microstructural design scheme provides a new idea for the preparation of new MAM.