Enhancing Microwave Dynamic Effects via Surface States of Ultrasmall 2D MOF Triggered by Interface Confinement for Antibiotics‐Free Therapy

Abstract Microwave (MV)‐trigged dynamic therapy based on MV‐responsive materials is promising for treating deep infection diseases that cannot be effectively treated by antibiotics, like life‐threatening osteomyelitis. Surface states of materials affect the generation of free charges under the excitation source with energy less than the band gap, consequently influencing the MV dynamic effects. Herein, an MV responsive system with interface confined 2D metal–organic framework (2D MOF) on oxidized carbon nanotube (CNT) is prepared, in which the ultrasmall Cu‐based 2D MOF possesses sufficient surface/interface defects, endowing the system a large number of surface states. Under MV irradiation, the synthesized CNT‐2D MOF not only efficiently absorbs and converts the microwave into heat for microwaveocaloric therapy (MCT) via enhanced hetero‐interfacial polarization, but also generates excited electrons via surface state for microwave dynamic therapy (MDT). This biocompatible CNT‐2D MOF exhibits highly effective broad‐spectrum antimicrobial activity against seven pathogenic bacteria, including Gram‐negative and Gram‐positive pathogens, under 7 min MV irradiation. And this system is proven to efficiently eradicate Staphylococcus aureus infected rabbit tibia osteomyelitis. Significantly, MV‐excited MCT and MDT of CNT‐CuHHTP developed in this study makes a major step forward in antibiotic‐free MV therapy in deep tissue bacterial infection diseases.


Introduction
Deep tissue infection diseases, for example, osteomyelitis, induced by pathogenic bacteria has threatened human health. [1] Without conducting a proper treatment in time, it may lead to life-threatening sepsis with severe organ failure. [2] Clinical treatment always requires multiple invasive surgeries and long-term antibiotic therapy, which brings not only pains to patients, but also serious side effects and even drug resistance caused by antibiotics. [3] In addition, an antibiotic-free therapy, phototherapy, including photothermal therapy and photodynamic therapy, is ineffective for the treatment of deep tissue infections limited by the poor penetration of light. [4] Hence, it is urgent to develop an non-invasive antibiotic-free strategy for deep tissue infections.
Microwave (MV), as an electromagnetic wave with a longer wavelength than light, has a strong penetration ability and is a promising exogenous energy source to assist treatment of deep tissue infection with CuHHTP, CNT will limit the infinite epitaxial growth of CuHHTP, so that the CuHHTP sheet smaller and has more surface states, which is conducive to efficient microwave dynamic therapy.
Herein, we designed a CNT-CuHHTP heterojunction with rich surface states caused by confinement effect, which showed a synergistic effect of microwaveocaloric and microwave dynamic therapy. CNT-CuHHTP has excellent microwaveocaloric effect due to its excellent electrical conductivity, multiple reflections, heterointerfacial polarization, and dipole polarization. Importantly, under MV excitation, the CNT-CuHHTP can generate free electrons through surface energy level transitions, while the built-in electric field formed in the heterojunction accelerates the electron transfer, further enabling efficient MV dynamic property and generating a large amount of ·O 2 − . Therefore, CNT-CuHHTP can rapidly and efficiently eradicate osteomyelitis by MCT and MDT. The surface states engineering-guided generation of microwave-excited electrons may bring a novel insight for developing microwave energy absorption and converting materials for various kinds of applications such as microwaveocaloric and microwave dynamic therapy of deep tissue infectious diseases.

Synthetic Procedure and Microscopic Structural of CNT-CuHHTP
The "electric-wire like" CNT-CuHHTP was prepared via hydrothermal method. In this reaction, the C/O functional groups (-COOH, -C-OH) of CNT could absorb the free Cu 2+ via electrostatic attraction, thus forming nucleation centers to guide the crystal and growth of Cu-MOF. After HHTP ligands coordinated to Cu 2+ , the bridge chain has self-assembled into 2D Cu-MOF along with CNT skeleton (Figure 1a). Instead, the free nucleation and growth of CuHHTP is formed by layer-by-layer stacking structures comprised of aggregated 2D nanosheets with a lateral size of 0.2-1 μm, as observed in Figure 1b and Figure S1, Supporting Information. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of CuH-HTP showed that Cu atoms are arranged in order ( Figure 1c) and the lattice spacing is uniform (Figure 1c, inset). When CNT ( Figure S2, Supporting Information) was introduced to reaction system, the ultrasmall CuHHTP is formed and grow along the CNT skeleton (Figure 1d), revealing the CNT can confine the epitaxial growth of CuHHTP. The dark-field TEM image and the corresponding elemental mapping images of CNT-CuHHTP in Figure S3, Supporting Information, displayed a similar distribution area of individual elements of C, O, and Cu, preliminarily confirming that the CuHHTP covered the surface of CNT. The high-resolution transmission electron microscopy (HRTEM) image of the CNT-CuHHTP in Figure 1e exhibited the lattice distance of 0.312 nm and 0.337 nm, which belonged to the (022) and (002) lattice plane of CuHHTP and CNT, respectively. Notably, obvious steps and defects appears in the CNT-CuHHTP. Additionally, the lattice mismatch at the heterojunction leads to the appearance of interface states. These surface/interface effects enhanced by CNT confinement are the main source of CNT-CuHHTP surface states.

Structural Characterization of CNT-CuHHTP
The X-ray diffraction (XRD) pattern of the CNT-CuHHTP (Figure 2a) exhibits almost the same peaks as those of the pure CuHHTP except for the peak at 2 = 26°, which is assigned to the (002) plane of the graphite-like structure of CNT, indicating the coexistence of CuHHTP and CNT in the CNT-CuHHTP. Especially, the position of CuHHTP (022) peak after CNT doping moves from 28.0°to 27.7°, and the corresponding lattice spacing increases from 0.309 to 0.312 nm. This is mainly due to the lattice distortion of ultrasmall CuHHTP caused by the confinement effect. Besides, the X-ray photoelectron spectroscopy (XPS) full spectrum ( Figure S4, Supporting Information) of CNT-CuHHTP demonstrated the existence of C, O, and Cu in CNT-CuHHTP, which is consistent with the element mapping results. The highresolution spectrum of Cu 2p 3/2 spectrum was deconvoluted into Cu 2+ and Cu 1+ , and the two corresponding satellite peaks were shown in Figure 2b. A clear negative shift of Cu 1+ (≈0.49 eV) binding energy was observed in CNT-CuHHTP compared to the one in CuHHTP, further suggesting a strong chemical interaction between CuHHTP and CNT. These results illustrate the successful construction of CNT-CuHHTP heterostructure.
To explore the intrinsic atomic structure of CNT-CuHHTP, synchrotron radiation-based X-ray absorption fine structure The Cu 2p XPS spectra in CuHHTP and CNT-CuHHTP. c) Synchrotron radiation Cu K-edge XANES. d) The k 2 (k)-oscillation curves and e) the Fourier transform curves of Cu K-edge EXAFS spectra for Cu foil, Cu 2 O, CuO, and CNT-CuHHTP. f) FT-EXAFS fitting curve of CNT-CuHHTP in R space. g) Electron holograms of CNT-CuHHTP. h) Reconstructed phase image of (g). i) Relative thickness image of CNT-CuHHTP was converted by amplitude. j) Reconstructed potential map. k) Enlarged view of the black boxed region in (j).
(XAFS) was performed. The Cu K-edge shows the absorption edge positions of copper in CNT-CuHHTP is between Cu foil and CuO references, means that the average valance state of copper in CNT-CuHHTP is between metallic Cu 0 and oxidized Cu 2+ (Figure 2c), which is consistent with the XPS results. CNT-CuHHTP maintains higher energy beyond Cu 2 O, suggesting more positive charges carried by copper atoms of the CNT-CuHHTP. The k 2 (k) oscillation curve of CNT-CuHHTP at Cu K-edge shows a different trend in shape and oscillating frequency to that of control samples over the whole range of 0-12 Å −1 (Figure 2d; Figure S5 Notably, CNT-CuHHTP shows a weaker Cu-O peak relative to that of CuO, possibly indicating the presence of partial coor-dinatively unsaturated Cu nodes. Moreover, the average oxidation state of Cu species in CNT-CuHHTP is confirmed as 1.7 based on the X-ray absorption near-edge spectra (XANES) analyses ( Figure S6, Supporting Information), obviously lower than that of CuO (about 2), which again suggests the existence of unsaturated Cu nodes in CNT-CuHHTP. From the EXAFS fitting results ( Figure 2f; Table S1, Supporting Information), the coordination number of Cu-O bonds is quantitatively determined to be about 3.46 for CNT-CuHHTP, evidently confirming the coexistence of unsaturated Cu-O 2 and saturated Cu-O 4 centers. The wavelet transforms (WT) contour plots of Cu foil displayed the maximum intensity at 7.0 Å −1 ( Figure S7a, Supporting Information) due to their corresponding Cu-Cu coordination. And the WT contour plots of Cu 2 O and CuO displayed the maximum intensity at 4.2 Å −1 ( Figure S7b, Supporting Information) and 5.0 Å −1 ( Figure S7c, Supporting Information), respectively, which could be due to their corresponding Cu-O coordination. Compared to this, the maximum intensity was found to be positioned at 4.8 Å −1 in the WT contour plot of CNT-CuHHTP, which is ascribed to the coordination between C/O (CNT) and Cu ( Figure S7d, Supporting Information). Accordingly, on the basis of the XAFS results, CNT-CuHHTP have abundant unsaturated Cu nodes. This defective structure endow CNT-CuHHTP with abundant surface energy levels and higher oxygen adsorption capability, which might be beneficial to the microwave dynamic process.
In our work, electron holography measurements with high sensitivity (0.36 nm spatial resolution and 0.1 V voltage resolution, Figure S8, Supporting Information) were aimed to detect the electrical potential of CNT-CuHHTP. Specifically, according to the following formula: We can deduce that the projected average electric potential (Vp) of the sample is determined by its reconstructed phase shift (Δ ) and thickness (t), where C E is the interaction constant (0.00728 rad V −1 nm −1 for 200 kV accelerating voltage). [13] Following this method, a representative electron hologram of CNT-CuHHTP (sample) is shown in field of view with a vacuum region ( Figure 2g). The reconstructed phase map ( Figure 2h) with the desired information about the potential distribution shows the phase shift relative to vacuum. Moreover, the phase shift of CNT and CuHHTP can be more clearly observed from their local magnified images ( Figure S9a,b, Supporting Information) and corresponding local linear phase images ( Figure S9c,d, Supporting Information). The relative thickness information of the sample obtained from the amplitude is shown in Figure 2i. Therefore, all parameters for obtaining the potential map are available. In this way, an electrostatic potential difference ranging between −2 and 2 V, which can be demonstrably observed across the interfaces inside CNT-CuHHTP (Figure 2j). And a distinct potential distribution can be noticed near the interface of CNT and CuH-HTP, showing a positive potential on the CNT side and negative potential on the CuHHTP side, which facilitates the separation and transfer of free electrons at the heterointerface to enhance the microwave kinetics effect. In addition, the inhomogeneous potential distribution (Figure 2k) favors the dense interfacial polarization at phase boundary for enhanced MCT.

Microwaveocaloric Performance and Mechanism of CNT-CuHHTP
We tested the microwaveocaloric performance of different materials under medical low-power MV intensity (2.45 GHz, 0.1 W cm −2 ). As shown in Figure 3a, the temperature of CNT-CuHHTP solution rose from 25.9°C to as high as 53.1°C within 7 min, higher than that of the group of physiological saline (Ctrl, 44.3°C), pure CuHHTP (49.8°C), and CNT (50.5°C). The temperature can be controlled by adjusting the MV power. When the MV power is changed from 0.1 to 0.2 W cm −2 , the saline solution of CNT-CuHHTP can even reach up to 68°C after 7 min MV irradiation (Figure 3b, 0 mm). To verify whether the microwaveocaloric effect of CNT-CuHHTP can be used for in situ hyperthermia after penetration different thicknesses of tissue, pork tissues of 5, 10, and 25 mm thicknesses were selected as the model tissues. As the corresponding infrared thermal images shown in Figure 3b, under MV, the temperature of CNT-CuHHTP solution can still rise up to 68°C even though the pork tissue is 25 mm thick by increasing MV power to 0.4 W cm −2 , which is close to the temperature of CNT-CuHHTP solution without pork. Importantly, under the same conditions, MV irradiation does not cause hyperthermia in the surrounding tissues of the CNT-CuHHTP (Figure 3b). That is to say, by adjusting the MV power, the CNT-CuHHTP solution can achieve desired microwaveocaloric performance even at 25 mm penetration depth without causing hyperthermia in the surrounding tissues, which is the key for the successful treatment of deep tissue infection.
To reveal the mechanism of the efficient microwaveocaloric performance of CNT-CuHHTP, we tested its MV absorption and energy conversion properties using the reflection loss (RL) value, where a lower RL value means stronger MV absorption and more efficient energy conversion capacity. [14] CNT-CuHHTP has the lower RL value than CNT and CuHHTP under different test conditions ( Figure 3c). Specifically, the RL value of CNT-CuHHTP reached −18.42 dB (≈98.56% absorption) at 2.45 GHz, while the RL value of CNT and CuHHTP is only −0.98 dB (≈20.20% absorption) and −2.29 dB (≈40.98% absorption), respectively (Figure 3d). The above results indicated the excellent MV response properties of the synthesized CNT-CuHHTP, suggesting the great potential of this material to achieve efficient MCT. Generally, |Z in /Z 0 | and are used to evaluate the absorption and consume MV ability of the MV absorber, respectively. [7a,b] When |Z in /Z 0 | value closing to 1.0, represents an excellent impedance matching for efficient MV absorption. [15] Especially, the |Z in /Z 0 | value of CNT-CuHHTP is 0.95, which is closer to 1 than that of CuHHTP (0.59) and CNT (0.07), indicating that CNT-CuHHTP absorbed more incident MV than CuHHTP or CNT (Figure 3e). Besides, CNT-CuHHTP with moderate value is able to convert the absorbed MV into heat (Figure 3f). Although CNT has excessive value, the incident MV is reflected on the surface of CNT due to the skin effect, resulting in poor MV absorption performance. Meanwhile, CuHHTP exhibits poor MV absorption performance due to the inappropriate |Z in /Z 0 | value and small value. To sum up, owning to good impedance matching and moderate attenuation constant, the CNT-CuHHTP has excellent microwaveocaloric performance.
In order to explore the MV attenuation mechanism of CNT-CuHHTP, we further studied imaginary permittivity ( ″) and imaginary magnetic loss (μ″), which represent the ability to dissipate electromagnetic energy. [16] Specifically, CNT-CuHHTP with suitable dielectric loss ( Figure 3g) and negligible magnetic loss (Figure 3h), indicate that the attenuation mechanism of CNT-CuHHTP is dominated by dielectric loss. Generally, the dielectric loss capability of CNT-CuHHTP is related to conduction loss and polarization relaxation. On the one hand, the high electrical conductivity of CNT-CuHHTP endows them with efficient conduction loss. On the other hand, heterointerface polarization and dipole polarization are the main types of CNT-CuHHTP polarization relaxation. And the polarization relaxation behaviors of CNT-CuHHTP can be illustrated by the relationship between ′ and ″ curve in Figure 3i, the ′-″ curve of CNT-CuHHTP exhibits obvious Cole-Cole semicircles compared to the CNT (Figure 3i, inset) and CuHHTP ( Figure S10, Supporting Information), demonstrating that several Debye relaxations processes occur in CNT-CuHHTP. Further, through the electron hologram we observe a distinct charge distribution at the heterogeneous interface of CNT-CuHHTP, demonstrating the intensively interfacial polarization in CNT-CuHHTP (Figure 2j,k). These charge distributions caused by interface polarization are forced to re-arrange orderly with the propagation direction of the alternating electric field to dissipate the MV energy. Besides, some dangling bonds on CNT and CuHHTP can quickly adjust their orientation under MV to adapt to the changing electric field direction, resulting in dipole polarization to dissipate MV energy. In conclusion, the efficient microwaveocaloric effect of CNT-CuHHTP is attributed to the reasonable wave-resistance matching and attenuation con-stant caused by multiple reflections, interface polarization, dielectric loss, and dipole polarization (Figure 3j).

Microwave Dynamics Performance and Mechanism of CNT-CuHHTP
Further, we tested the microwave dynamics performance of CNT-CuHHTP. The transient MV-current response was observed over three repeated cycles under 20 s MV pulse irradiation (a 20-s pulse cycle consists of 20 s on, 20 s off). As shown in Figure 4a, the CNT-CuHHTP exhibits an obvious microwave response current, indicating that MV can excite the space charge separation in the CNT-CuHHTP. In addition, the MV-current density (ΔJ) of CNT-CuHHTP was obviously stronger than that of CNT, demonstrating the more effective charge transfer in the CNT-CuHHTP heterointerface under MV irradiation. Notably, the MV-current of CNT-CuHHTP has shrilly front-peak and back-peak, which is consistent with the characteristics of surface states in the photocurrent, [17] indicating CNT-CuHHTP contains many surface states and defects, which are favorable for multi-level transition of electron under the excitation of MV. In addition, the MVcurrent density of CNT-CuHHTP under different MV pulse conditions were almost similar, while the MV-current also exhibited the obvious front-peak and back-peak (Figure 4b), indicating that the intensity of MV response current is independent of MV pulse conditions. Notably, the MV-current intensity of CNT-CuHHTP exhibits positively correlated with MV power, that is, the MVcurrent becomes stronger as the MV power increases (Figure 4c). It indicates that we can significantly increase the amount of the excited electrons for ROS production by increasing the MV power. Besides, the photoluminescence (PL) spectra (Figure 4d) of CuHHTP and CNT-CuHHTP revealed one absorption peak at ≈455 nm, and the PL signal of CNT-CuHHTP is lower than that of CuHHTP indicated the heterostructure of CNT-CuHHTP can efficiently reduce the recombination of electron−hole pairs, which are used to generate ROS. Therefore, CNT-CuHHTP has higher ROS yield than CNT and CuHHTP under MV excitation (Figure 4e), which fully illustrates the enhanced contribution of CNT in microwave dynamics. And it is determined that the type of ROS is only ·O 2 − (Figure 4f), not hydroxyl radicals (·OH) and singlet oxygen ( 1 O 2 ) ( Figure S11, Supporting Information). Importantly, CNT-CuHHTP hardly produces ·O 2 − without MV, but a large amount of ·O 2 − is produced under MV irradiation, which is beneficial for the specific responsiveness of MDT treatment.
According to the UV−vis diffuse reflectance spectra ( Figure S12a In order to better illustrate the mechanism of CNT-CuHHTP to generate ·O 2 − under MV excitation, we have summarized the oxygen adsorption energies for different sites of CNT, CuHHTP, and CNT-CuHHTP. According to the principle of minimum energy, it can be seen that the C site of CNT is more likely to adsorb oxygen, and Cu (Cu-O 2 ) site of CuHHTP and CNT-CuHHTP is more likely to adsorb oxygen ( Figure S13, Supporting Information). In addition, the adsorption energy of CNT-CuHHTP (−2.96 eV) is lower than that of CNT (−1.76 eV) and CuHHTP (−1.95 eV) (Figure 4h), indicating that CNT-CuHHTP is easier to adsorb oxygen than CNT and CuHHTP. With the local electric polarization, it is thermodynamically favorable for the Cu-O 2 centers to strongly couple with O 2 molecules, and thus beneficial for react with free electrons toward rapid formation of ·O 2 − for antibacterial.
The MV-dynamic mechanism of CNT-CuHHTP is summarized in Figure 4i. Before CNT and CuHHTP contact, the surface energy band of CuHHTP bends downward due to the existence of surface states. [18] After contact, the interaction of the energy band and interfacial structure promote MV-dynamic effects. According to the pinning theory, the Fermi level of CNT will be pinned at the surface level of CuHHTP regardless of the work function of CNT. [19] Thereby, a built-in electric field (E) is formed, and the direction of the electric field is from the CNT side to the CuHHTP, that is, the CNT side has a positive charge and the CuHHTP side has a negative charge, which is consistent with the result of electron hologram charge distribution. Moreover, the built-in electric field can drive electrons rapidly transferred from CuHHTP to CNT. As a result, under MV irradiation, MV can excite surface energy levels matching their energy to generate more MV-excited electrons, which are transferred to CNT driven by a built-in electric field to reduce recombination. These free electrons react with oxygen to generate ·O 2 − for eradication of bacteria.

CNT-CuHHTP Has Broad-Spectrum Bactericidal Effect under MV Irradiation
We tested the synergistic antibacterial effect of CNT-CuHHTP with MV in vitro, including against typical Gram-negative (G − ) and Gram-positive (G + ) bacteria. It can be easily observed that the number of E. coli after CNT-CuHHTP synergistic MV (CNT-CuHHTP+MV) treatment for only 7 min was significantly reduced compared with other groups (Figure 5a). Specifically, the number of E. coli (10 7 CFU mL −1 , Figure 5b) decreased from the original 0.4971 (Ctrl) to 0.0085 (CNT-CuHHTP+MV), far lower than the CNT-CuHHTP (0.3148) and MV (0.448) groups alone demonstrating that 98.29% E. coli were killed by CNT-CuHHTP+MV (Figure 5c). Similarly, compared with Ctrl (not treated), CNT-CuHHTP and MV groups, the treatment group of CNT-CuHHTP+MV had the best antibacterial effect against S. aureus (Figure 5d), reducing the counts of S. aureus (10 7 CFU mL −1 , Figure 5e) from 0.4467 (Ctrl) to 0.0785 with 82.43% anti-S. aureus efficiency (Figure 5f). Besides, we also tested four G − bacteria including enteropathogenic E. coli (E. coli EPEC), Salmonella typhimurium (S. typhimurium), Aeromonas veronii (A. veronii), and Proteus vulgaris (P. vulgaris), and G + bacteria methicillin-resistant S. aureus (MRSA). In addition, the antibacterial effects of different treatment groups against different bacteria are shown in Figures S14-S18, Supporting Information, and the antibacterial results are summarized in Figure 5g.  Figure S19, Supporting Information). The CNT-CuHHTP+MV group had the highest antibacterial rate, whether against G − or G + bacteria, it is worth noting that the antibacterial effect of CNT-CuHHTP+MV against G − bacteria (more than 95.6%) is higher than that against G + bacteria (82.43% for S. aureus and 48.5% for MRSA) under the same antibacterial conditions. We observed surface details of E. coli and S. aureus before (Ctrl) and after CNT-CuHHTP+MV treatment by atomic force microscopy (AFM). Specifically, E. coli in the Ctrl group were rod-like shape and plumply cell (Figure 5h). The complete profile presented by the corresponding height profile also proves its completeness (Figure 5i). Conversely, the morphology of E. coli in CNT-CuHHTP+MV group showed severe collapses (Figure 5j) with a depth of 81 nm (Figure 5k). Similarly, the morphology of S. aureus was spherical and chubby (Figure 5l,m), but shrinkage and deformation in the CNT-CuHHTP+MV group (Figure 5n,o). In addition, the SEM images of the four typical bacteria ( Figure S20, Supporting Information) also showed a similar phenomenon: the Ctrl group showed a plump morphology, the CNT-CuHHTP and MV groups both showed slight damage, and the CNT-CuHHTP+MV group showed different degrees of collapse. Notably, the damage of CNT-CuHHTP+MV to G − bacteria is more obvious than that of G + bacteria, which is consistent with the results of antibacterial rate. We speculated that this different antibacterial effect is related to the cell wall structure of these two types of bacteria. The cell membranes thickness of G + bacteria is up to 15-30 nm due to the presence of a thick peptidoglycan layer, [20] while the thickness of G − bacteria is only 10-15 nm. [21] The thick cell membrane of G + bacteria weaken the destructive effect of high temperature on it, so the antibacterial efficiency of G + is lower than that of G − bacteria, even though exogenous

·O 2
− is easier to penetrate the G + membrane and enter the cell to play a bactericidal role. [22] This speculation suggests that CNT-CuHHTP+MV eliminates bacteria mainly by destroying the bacterial cell membranes.
Furthermore, taking E. coli as an example, we visualize the permeability of E. coli membranes using the membrane dye FM4-64 and the nucleic acid stain SYTOX Green. Generally, SYTOX Green will not pass through the intact cell membrane, and only when the bacterial membrane is damaged, SYTOX Green will enter the cell and cause cells to display green fluorescence. E. coli in the Ctrl group (without treatment) showed the outline of the bacterial membrane without the green fluorescent signal (Figure 5p). The CNT-CuHHTP ( Figure 5q) and MV (Figure 5r) groups showed strong red signals and weak green signals indicating that CNT-CuHHTP and MV had a slight damaging effect on E. coli cell membranes. Conversely, in the CNT-CuHHTP+MV group (Figure 5s), a clear intracellular green signal was displayed, the linear scan results showed good colocalization, and the green signal intensity was higher than other groups indicating that the cell membrane of E. coli was seriously damaged and leads to its death.
In addition, the normal saline solution of CNT-CuHHTP (1 mg mL −1 ) can release 11.56 ppm copper ion in 48 h. After MV irradiation for 7 min, the concentration of copper ion can reach 10.18 ppm ( Figure S21, Supporting Information). This shows that MV irradiation of CNT-CuHHTP will release a small amount of copper ions, which is conducive to the elimination of bacteria.
To sum up, these results demonstrate that CNT-CuHHTP generated ·O 2 − and microwaveocaloric under MV irradiation,destroy the bacterial cell membrane structure, and conducive to the entry of copper ions into the cell, thus causing the efflux of intracellular substances and the collapse of the cytoskeleton, eventually leading to bacterial death.

Safety Evaluation of CNT-CuHHTP
We investigated the cytocompatibility of CNT-CuHHTP by observing the morphological changes and cell viability of cells after co-culture with CNT-CuHHTP. Specifically, after MC3T3-E1 was co-cultured with CNT-CuHHTP for 3 days ( Figure S22, Supporting Information), the cells were spindle-shaped, conical, or cubic, with obvious nuclei, and short protrusions on the cell surface were connected to adjacent cells, which had no significant difference from the cell morphology of the control group (Ctrl). Likewise, CNT-CuHHTP did not have a significant effect on the morphology of A549 cells. In addition, the cell viabilities of MC3T3-E1 were concentration-dependent, and the cell viability increased with the decrease of CNT-CuHHTP concentration. The cell viabilities of MC3T3-E1 remained above 94% even with CNT-CuHHTP concentration of 1 mg mL −1 ( Figure S23, Supporting Information). These results demonstrate the excellent cytocompatibility of CNT-CuHHTP in vitro.
Moreover, we have evaluated the effect of CNT-CuHHTP+MV (consistent with the irradiation conditions in animal experiments) on the viability of A549. As show in Figure S24, Supporting Information, after CNT-CuHHTP+MV treatment for one day, the morphology of A549 cells changed from a polygonal shape with filamentous pseudopodia to a fusiform antenna that shortened and shrank, and the cells shrank into a spherical structure. Continue to culture to the third day, the cell morphology returned to normal, and the number of cells increased. These results indicate that cell proliferation will be inhibited after CNT-CuHHTP+MV treatment, but after a short period of recovery, the cells can gradually recover and begin to proliferate.
Besides, CNT-CuHHTP has good blood compatibility and in vivo safety. Specifically, there was no obvious hemolysis after coculture of CNT-CuHHTP with rabbit erythrocytes (Figure 6a, inset). The hemolysis rate of CNT-CuHHTP (1 mg mL −1 ) was only 6.3% (Figure 6a), which was much lower than that of the positive control group (1% TritonX-100). Meanwhile, after injection of CNT-CuHHTP, there was no abnormality in the corresponding indicators of blood routine (WBC, Lymph, Mon, Gran, RBC, HCT, MCV, MCH, RDW, and MPV) in rabbits (Figure 6b). There was no significant difference in the hepatic function ( Figure S25a,b, Supporting Information) and renal function ( Figure S25c,d, Supporting Information) between the Ctrl and CNT-CuHHTP group at the given dose. Hematoxylin and eosin (H&E) sections ( Figure S26, Supporting Information) of major organs (heart, liver, spleen, lung, kidney) show that CNT-CuHHTP has no organ toxicity, further demonstrating the excellent in vivo safety of CNT-CuHHTP.

Eradication of Osteomyelitis
The surgical procedure for modeling rabbit osteomyelitis is shown in Figure S27, Supporting Information, which mainly includes disinfection of the skin surface, drilling of the tibia, injection of bacteria, bone wax sealing, and suture of tissue. The treatment process of osteomyelitis, taking the CNT-CuHHTP+MV group as an example, was to add CNT-CuHHTP after adding the bacterial solution, suture the tissue, and then irradiate MV for 7 min. The details are described in Section 4. The temperature distribution of rabbit tissue before and after CNT-CuHHTP+MV treatment is shown in Figure S28, Supporting Information. The temperature rises from 28 to 42°C, which will not cause serious tissue damage. The MRI image (Figure 6c) exhibited the inflammation of the bacterial-infected bone marrow site is hyperintense (indicated by blue arrows) and accompanied by bone defects (indicated by red arrows). After CNT-CuHHTP+MV treatment, the inflammatory signals were attenuated, and bone defect symptoms were reduced. According to the Wright's staining of bone marrow (Figure 6d), it can be seen that the Ctrl, CNT-CuHHTP, and MV group have different degrees of red blood cell damage, accompanied by a large number of lymphocytes (indicated by green arrows). In contrast, the morphology of erythrocytes was complete and the number of inflammatory cells was less in the CNT-CuHHTP+MV group. In addition, the number of S. aureus (10 6 CFU g −1 ) in the bone marrow in the CNT-CuHHTP+MV group was 0.619, which was also the least compared to the Ctrl (7.181), CNT-CuHHTP (4.635), and MV (2.277) groups (Figure 6e). These results demonstrate that CNT-CuHHTP+MV has an excellent antibacterial effect in vivo.
Besides, we analyzed the blood samples and bone marrow tissue histopathological section of the rabbit after treatment. Specifically, after 2 days of treatment, the number of WBC (Figure 6f) in the treatment group showed a decreasing trend, and the CNT-CuHHTP+MV group decreased to 8.0 (10 9 L −1 ) reached the normal level. A similar trend emerged at after 5 days of treatment. To evaluate the inflammatory response of different treatment groups the levels of TNF-and IL-6 were measured. The CNT-CuHHTP+MV group had the lowest concentrations of TNF- (Figure 6g) and IL-6 ( Figure 6h) whether on the second day (Day 2) or the fifth day (Day 5) after treatment, indicating that the treat-ment of CNT-CuHHTP+MV did not cause obvious inflammatory response, on the contrary, the other groups had caused a significant inflammatory storm.
Further, HE sections of bone marrow tissue (Figure 6i) showed that the CNT-CuHHTP+MV group had fewer inflammatory cells at the early stage of treatment (Day 2), while the Ctrl group had more multinucleated giant cells (indicated by green arrows), neutrophils (indicated by back arrows), and lymphocytes (indicated by yellow arrows) in the bone marrow. The presence of large numbers of inflammatory cells indicates a serious infection in the bone marrow. At Day 5, the bone marrow of the Ctrl group showed a large number of adipocyte rupture (indicated by orange arrows), and the degree of rupture was less in the treatment group, especially in the group of CNT-CuHHTP+MV. Systemic infection caused by osteomyelitis can lead to lesions of the organs. The pathological sections of the main organs in different groups are shown in Figure S29, Supporting Information. S. aureus infection causes punctate necrosis of cardiomyocytes (indicated by red arrows), diffuses distribution of inflammatory cells in the liver and spleen (indicated by yellow), and alveolar epithelial hyperplasia with red mucus (indicated by black arrows). These symptoms almost disappeared in group of CNT-CuHHTP+MV, demonstrating the excellent antibacterial properties of MCT and MDT in vivo. Notably, the improvement effect of CNT-CuHHTP+MV on osteomyelitis is comparable to that of traditional antibiotics gentamicin (Gent) ( Figure S30, Supporting Information).

Conclusion
CNT confined the epitaxial growth of CuHHTP, resulting in the formation of ultra-small CuHHTP growing along the CNT skeleton, accompanied by abundant surface states which were mainly derived from the surface/interface defects (surface steps, unsaturated bonds and lattice mismatch) of CNT-CuHHTP. CNT-CuHHTP can efficiently kill five Gram-negative bacteria (E. coli, E. coli EPEC, S. typhimurium, A. veronii, and P. vulgaris) and two Gram-positive bacteria (S. aureus and MRSA) with only 7 min of MV irradiation. Its robust broad-spectrum antibacterial effectiveness is mainly attributed to the destruction of bacterial membranes caused by microwaveocaloric and microwave dynamics effects (·O 2 − ). Specifically, the efficient microwaveocaloric effect of CNT-CuHHTP is own to the reasonable wave-resistance matching and attenuation constant caused by multiple reflections, interface polarization, dielectric loss and dipole polarization. Besides, under microwave irradiation, CNT-CuHHTP generates excited electrons through the surface energy level, and the built-in electric field accelerates the transfer of excited electrons to CNT and promotes the generation of ·O 2 − . The in vivo results confirm that this antibiotic-free microwave therapy can cure osteomyelitis by reducing bacterial counts with topical treatments. This platform may bring further insight and understanding of deep tissue infection diseases treatment through antibiotic-free microwave therapy.