Temperature-responsive tungsten doped vanadium dioxide thin film starves bacteria to death

1 Department of Orthopaedics and Traumatology, Li Ka Shing Faculty of Me dicine, The University of Hong Kong, Pokfulam, Hong Kong 999077, China Department of Physics and Department of Materials Science and Engineering, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong 999077, China Department of Orthopaedics, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai Jiao Tong University, Shanghai 200233, China 4 Shenzhen Key Laboratory for Innovative Technology in Orthopaedic Trauma, The University of Hong Kong-Shenzhen Hospital, Shenzhen 518053, China


Introduction
Biofilms produced by bacteria, which are a matrix of extracellular polymeric substances, always contaminate infrastructure devices and systems (i.e., biofouling), which produce huge social and economic burdens [1][2][3]. For instance, medical biofouling initiates biofilm accumulation on biomedical devices such as orthopedic and dental implants, catheters, biosensors, and surgical instruments, in which biofilms are the common cause of persistent infections [4][5][6]. Once mature biofilms are developed, they frequently challenge the efficacy of antibacterial treatments, specifically among antibiotic-resistant bacteria [7][8][9]. Although Temperature-responsive tungsten doped vanadium dioxide thin film starves bacteria to death applying antibiotics can effectively prevent or treat bacterial infections, antibiotic resistance has become a serious issue with the development of multidrug-resistant pathogens [10,11]. The bacterial contamination of various surfaces remains a worldwide research challenge [12,13]. Consequently, researchers are constantly attempting to design new antifouling/antimicrobial coatings to prevent bacterial attachment and subsequent biofilm formation.
Bacteria perform respiration to produce energy to conduct vital movements [14]. Bacterial respiration requires extracellular electron transportation, in which electrons are donated by nicotinamide adenine dinucleotide (reduced form of NADH). Then, the electrons are transferred to a terminal acceptor through a range of redox cofactors located in the bacterial membrane [15][16][17][18][19][20]. Inspired by this bioenergetic process, we assume that the interruption of extracellular electron transportation by engineering the specific nanoparticles in contact with bacteria can disrupt membrane respiration and cause bacteria inactivation. The equivalence of bandgap and Fermi energy levels plays a critical role in allowing electron transportation between the semiconductor and bystander. If a conductive circuit between the bacterium and semiconductor surface can be established, electrons generated within the bacterium will continuously drain out via electron transportation and then interrupt bacterial viability.
Vanadium dioxide (VO 2 ) is a thermochromic material that undergoes a reversible semiconductor-metal transition from a high-temperature rutile phase (P4 2 /mnm, R) to a lowtemperature monoclinic phase (P2 1 /c, M1) at a critical temperature (Tc) of $340 K under external temperature stimuli or nearinfrared irradiation [21][22][23][24][25]. Concomitant with the phase transition are discontinuous jumps in electrical conductivity, optical transmittance, magnetic susceptibility, specific heat, and the Seebeck coefficient [26,27]. To consider the significant features and advantages of VO 2 , we propose to tune the thermochromic property of VO 2 by doping tungsten into its matrix to modulate the electrical conductive property, which can potentially interrupt the extracellular electron transportation of bacteria upon contacting with VO 2 thin film, and thus achieve a superior antifouling effect.
The aim of this study is to design a phase-change-tunable VO 2 antifouling coating that is triggered by the change in the temperature of the micro-environment. The VO 2 thin films were prepared using a customized magnetron-sputtering technique, and tungsten doping was applied simultaneously. The changes in the thermochromic and electrical properties of the VO 2 thin films after tungsten doping were systematically investigated. Then, the in vivo antimicrobial ability was analyzed using a rat subcutaneous infection model induced by methicillin-resistant Staphylococcus aureus (MRSA). To understand how the antimicrobial effect of this newly developed thin film occurs in an in vivo condition, additional in vitro experiments were performed to investigate the interactions between the bacteria and VO 2 . Finally, a series of physical parameters were determined using ultraviolet photoemission spectroscopy (UPS), activation energy, and the Hall effect measurements to support the proposed antifouling mechanism.

Materials and methods Sample preparation
Pure VO 2 and tungsten-doped VO 2 thin films were prepared by a magnetron-sputtering apparatus (Plasma Technology Ltd., MS400). Fused quartz glasses with the size of 10 Â 10 Â 0.5 mm were adopted as substrates. To obtain tungsten-doped VO 2 thin films, vanadium dioxide and tungsten trioxide ceramic targets were used. During the deposition process, the DC power of VO 2 target maintained at 80 W, while the RF power of WO 3 varied (5 W, 15 W, and 30 W). The thicknesses of the films were detected by an F20 thin-film analyzer (FILMSTRICS Corp.). All undoped and doped thin films had the same thickness of 50 nm implemented by regulation of the deposition time. Bare quartz glass functioned as the control group. For simplicity, undoped VO 2 thin-film sample is denoted as VO 2 , while the doped samples with RF power = 5 W, 15 W, and 30 W are labeled as VO-W1, VO-W2, and VO-W3, respectively.

Sample characterization
A field-emission scanning electron microscope (SEM; HITACHI S-4800, Japan) was used to observe the surface morphology of samples VO 2 , VO-W1, VO-W2, and VO-W3. X-ray diffractometer (XRD; Rigaku Ultima IV) was used to study the crystalline structure of films, fitted with Cu Ka radiation (k = 1.541 Å) at a voltage of 40 kV and current of 40 mA. The XRD patterns were recorded in the range of 20°-80°(2h) with a step size of 0.02°and scanning rate of 2°/min. For the X-ray diffraction test, the glancing incidence angle was fixed at 1°. Phase identification was performed with the help of the standard JCPDS database. The chemical composition and chemical state were studied by X-ray photoelectron spectroscopy (XPS; Thermo Scientific Escalab 250Xi, US) with Al K a (1486.6 eV) illumination. Argon ion sputtering with an acceleration voltage of 2 kV for 30 s was performed prior to the XPS measurement to obtain a clean surface of each sample. Ultraviolet photoelectron spectroscopy (UPS) measurements were performed on all the samples at room temperature with the use of XPS testing system. Emission of He I with photon energy of 21.2 eV was used as a light source. The AFM image was collected using a Nanocute SII scanning probe microscopy operated in tapping mode under ambient conditions. Etched N-type Si nanoprobe tips with a spring constant of 40 N/m and resonance frequency of 200-400 kHz were adopted. Resistance versus temperature curves were conducted with a four-probe testing system (JANIS Research co., Model VPF-100) equipped with a temperature controller (Cryogenic Control System, Inc., Model 32/32B, USA) and a Keithley 2400 digital source meter. The schematic of this customized testing system and experimental details can be found in this Ref. [28]. The optical transmittance characteristics were monitored by the UV-visible-near-IR spectrophotometer (Hitachi U-3010) equipped with an attachment to control the temperature of the films. The precise temperature was measured using a PT100 temperature sensor in contact with the surface of films, and the temperature was controlled via the temperature controlling unit. The transmittance spectra were collected from 350 nm to 2600 nm, and the transmittance curve of the quartz glass was used as a baseline calibration for the transmittance measurements. Hysteresis loops were measured by collecting the transmittance spectra of samples at the fixed wavelength of 2000 nm with a heating-cooling rate of 2.0°C/min. The electron transport properties of the films were characterized by the Hall effect measurement system (Lakeshore, 7704A) at room temperature using a four-terminal van der Pauw configuration. To obtain the work function of the prepared thin films, a series of Kelvin Probe Force Microscopy (KPFM) measurements were performed using an atomic force microscope (Dimension ICON, Bruker co., USA), and the temperature of each sample was maintained at 37°C (incubator temperature) using a heating stage.

In vivo antimicrobial test
Sample preparation: The samples Control, VO 2 , VO-W1, VO-W2, and VO-W3 in size of 10 Â 10Â0.5 mm were used in the in vivo test. Rat implant-related subcutaneous infection model. The animal experiment was approved by the Animal Care and Experiment Committee of Sixth People's Hospital Affiliated to Shanghai Jiao Tong University. Fifty specific pathogen-free male Sprague-Dawley rats that were six months old and had an average weight of 450 g were randomly divided into five groups (Control, VO 2 , VO-W1, VO-W2, and VO-W3). Methicillin-resistant Staphylococcus aureus (MRSA, ATCC43300, biofilm-positive) was used to develop the infection model. Prior to surgery, these rats were first anesthetized with 3% pentobarbital sodium via intraperitoneal injection; the hair on the anterior upper backs was shaved, and the skin area was sterilized with povidone iodine. Under sterile conditions, the skin was incised layer by layer and the samples were placed into the subcutaneous soft tissue immediately. Afterward, the suture of skin incisions was carefully completed. Then, 70 ll of prepared MRSA suspension ($10 7 CFU/ml) in phosphate-buffered saline (PBS) was added into the subcutaneous surgical sites upon the implants.
Histological observation: The soft tissues around the implants of each group were collected and fixed in 10% buffered formalin at 4°C for 3 d. Then, they were washed with fresh PBS, dehydrated in gradient alcohol solutions, and embedded in the paraffin. The samples were cut into sections using a sledge microtome (Leica, Hamburg, Germany). Finally, they were deparaffinized in the xylene solution, and Hematoxylin and Eosin (H&E) and May-Grünewald Giemsa staining was used to detect the histological changes on optical microscopy.
Microbiological evaluation: The subcutaneous implants in each group were aseptically taken out to quantitatively determine the numbers of bacteria in vivo. The implants were washed gently twice with fresh PBS to remove the planktonic bacteria. Afterward, they were immersed in 1 ml of PBS, sonicated for 10 min at 150 kHz (150 W, B3500S-MT, Branson Ultrasonics Co., Shanghai, China), and vortexed for 2 min (Vortex Genie 2, Scientific Industries, Bohemia, NY) to dislodge the adherent bacteria. Meanwhile, the soft tissues around the implants were acquired, weighed, and homogenized in sterile tubes containing 1 ml of PBS using a high-speed homogenizer (Jingxin Industrial Limited Company, Shanghai, China). The soft tissue homogenates were serially diluted in PBS, and 100 ll of diluent was evenly distributed on sheep blood agar (SBA). Finally, these plates were cultivated in an incubator statically at 37°C overnight, and the numbers of bacterial colonies were counted using the spread plate method in line with the National Standard of China GB/T 4789.2 protocol.

In vivo toxicity evaluation
The major organs (heart, liver, kidney, and spleen) were harvested, fixed in 10% buffered formalin, processed routinely into paraffin, cut into sections, stained with H&E, and examined by optical microscopy.

In vitro antimicrobial assay
The biofilm-positive MRSA (ATCC43300) were used to evaluate the antimicrobial property of samples Control, VO 2 , VO-W1, VO-W2, and VO-W3. The strain was first cultured in a fresh trypticase soy broth (TSB) medium at 37°C overnight, and then according to the McFarland method, the inoculation concentration was adjusted to 10 6 CFU/ml in TSB. Each sample was positioned in a sterile 24-well plate, and 500 ll of prepared bacterial suspension was added into the wells containing samples. The 24-well plate was then kept in an incubator statically at 37°C overnight.
Quantitative evaluation by SPM: After overnight culturing, the samples were taken out from the suspension with sterile tweezers and then gently dislodged with fresh PBS twice. Subsequently, each sample was placed in a sterile test tube containing 1 ml of fresh PBS and oscillated by an ultrasonic vibration instrument for 15 min and a vortex mixer for 1 min. The suspension with bacteria was acquired, and 100 ll of diluent bacterial suspension was evenly spread on SBA and cultured at 37°C overnight. Then, the CFU number of viable bacteria on SBA was counted by SPM.
Microbial biofilm evaluation by confocal laser scanning microscopy (CLSM 510 meta; Zeiss, Germany). The surfaces of incubated specimens were gently washed with sterile PBS three times. Then, these samples were stained with 500 ll of mixed dye (LIVE/DEAD BacLight bacteria viability kits, Invitrogen) in a new 24-well plate, which remained in the dark for 20 min. Finally, the samples were lightly rinsed with fresh PBS and subsequently observed using CLSM. The viable microorganisms were stained in green, and the dead ones were stained in red.
Microbial biofilm evaluation by SEM: The incubated specimens were gently washed with fresh PBS three times, fixed with 2.5% glutaraldehyde overnight in a 4°C freezer, and subsequently dehydrated in the following ethanol series (30,50,70,80,90,95 v/v%) for 10 min and in the absolute ethanol solution (twice for 10 min). Then the samples were desiccated , gold sprayed and examined using scanning electron microscopy (JEOL JSM-6310LV, Japan).

Physical measurements
The room-temperature electrical resistance and carrier density were acquired by the Hall effect measurements with a Lakeshore 7704A Hall System. The samples were tested with a size of 1 Â 1 cm 2 , and metal contacts were made with indium (In) dots using a four-point probe in the Van der Pauw configuration. UPS measurements were performed for all samples at room temperature in order to analyze electron states near Fermi level. Emission of He II with photon energy of 40.8 eV was used as a light source.

Statistical analysis
Statistically significant differences (P) between various groups were measured using a one-way analysis of variance and Tukey's multiple comparison tests on a GraphPad Prism 5 statistical software package. All the data were expressed as mean ± standard deviation (SD). A value of P < 0.05 was considered to be statistically significant and denoted as " ⁄ "; P < 0.01 was " ⁄⁄ "; and P < 0.001 was " ⁄⁄⁄ ".

Sample characterization
The crystal structures of the as-prepared samples were characterized by X-ray diffraction (XRD). Fig. 1a shows XRD patterns of the as-prepared thin films with different W-doping fractions. For VO 2 thin film, only XRD peaks corresponding to the signal of the monoclinic VO 2 phase were observed, and all the positions of peaks were concordant with the standard reference data (JCPDS No. 72-0514), which indicated the exclusive formation of single-phase VO 2 polycrystalline films. For W-doped VO 2 thin films (sample VO-W1, VO-W2, and VO-W3), the peaks belonging to WO 3 could not be observed because W was doped into vanadium oxide as the solute donor, which indicated that W entered into the crystal lattice of the VO 2 matrix. In addition, most peaks of the doped VO 2 samples shifted slightly toward lower 2h values. To clearly demonstrate the shift of (0 1 1) the main peak, a scan rate of 0.3°min À1 was applied to record the pattern of samples VO 2 , VO-W1, and VO-W2 in the 2h range from 25°to 30°, as shown in the inset of Fig. 1a. Because the atom radius of W (r = 1.41 Å) is larger than that of V (r = 1.35 Å), when W atoms substitute V atoms in the VO 2 lattice, the adjacent interplanar distance d(h k l) becomes larger and results in the shift of the peak toward a low 2 Theta angle [29]. The surface morphologies of prepared samples are presented in Fig. 1b-e. Although doping with W makes the grains smaller to some extent, all the samples have similar morphology features. All crystalline grains have irregular shapes and sharp outlines. The RMS values obtained from the three-dimensional atomic force microscopy (AFM) images for VO 2 , VO-W1, VO-W2, and VO-W3 are 0.9, 4.9, 3.6, 3.1, and 4.2 nm, respectively (Fig. S1). The small RMS values indicate that the surface of the VO 2 thin film is extraordinarily smooth.
X-ray photoelectron spectroscopy (XPS) was also adopted to elucidate the valence state and content information. For pure VO 2 thin films (Fig. 2a), the binding energies of 515.6 eV and 523.0 eV correspond to the V 2p 3/2 and V 2p 1/2 core levels, respectively, of tetravalent vanadium (V 4+ ) [30]. The XPS result indicates that the prepared VO 2 thin film is pure without other valances, which is in accordance with the XRD results. The binding energy of 530.0 eV is the O 1s core level. Moreover, three W-doped VO 2 thin films also have the same binding energy of V and O elements (Fig. S2), which indicates that doping does not change the valance of V. For W-doped VO 2 thin films ( Fig. 2b-d), the binding energies of 35.7 ± 0.1 eV and 37.8 ± 0.1 eV correspond to the W 4f 7/2 and W 4f 5/2 core levels [31,32], respectively. This implies that the valence of tungsten ion in the W-doped VO 2 thin film is +6. Moreover, the relative intensity of W 4f 7/2 and W 4f 5/2 core levels increases from sample VO-W1 to VO-W3, which is ascribed to the increase in the W-doping concentration. The binding energy of 40.0 eV is the V 3p core level of tetravalent vanadium (V 4+ ) ( Fig. 2b-d) [33]. The XPS composi-tional results also indicate the phase purity of the W-doped VO 2 sample. The atom percentage of elements is shown in Table S1. The doping atom percentages of samples VO-W1, VO-W2, and VO-W3 are 0.26, 0.83, and 1.89 at.%, respectively.
The electrical resistance versus temperature curves were plotted to explore the electrical transport characteristics of undoped and doped VO 2 thin films under different temperatures. As shown in Fig. 3a, the resistance of all the samples presents a huge contrast between the low-temperature semiconductor phase and the high-temperature metallic phase. The difference is especially remarkable across the transition region. The resistance ratio, as shown in Table S2, is defined as the ratio of the resistance at 40°C below the respective transition temperature Tc (R (Tc-40) ) and the resistance at 40°C above the Tc (R (Tc+40) ), which changes from 385 to 84 from sample VO 2 to VO-W2. The transition temperature (T c ) is defined as the center temperature of the thermochromic hysteresis. The doping content of the W element plays a great role in the transition temperature of VO 2 . For pure VO 2 thin film, the transition temperature is 61°C (334 K). The transition temperature of VO 2 thin film is lower than that of bulk crystal (T c = 340 K); this is attributed to the internal stress produced by the ion bombardment effect during the film growth process. After doping 0.26, 0.83, and 1.89 at.% of W element, the T c was 53°C, 36°C, and 8°C, respectively. Obviously, the T c values for pure VO 2 thin film and sample VO-W1 are above the incubator temperature (37°C = 310 K) of bacteria and cells. The T c value for VO-W2 approaches to the incubator temperature, while the T c value for VO-W3 is below that, as shown in Fig. 3b. The results are apparent that both pure VO 2 thin film and VO-W1 are in semiconducting state at the incubating temperature, while VO-W3 sample is in metallic state. It is believed that the intermediate VO-W2 appears in a mixed state between semiconducting and metallic phase. The depression in phase transition upon W doping arises from destabilization of the V-V dimers in the M1 phase probed by local X-ray absorption fine structure spectroscopy [34]. Tungsten is found to be displacively doped within VO 2 as W 6+ and forms two distinctive pairs with valences of V 3+ -W 6+ and V 3+ -V 4+ to maintain the charge neutrality of the lattice. The destabilization of the M1 phase thus reduces the potential energy barrier for transformation to the rutile structure. Additionally, the ionization of the tungsten dopant to W 6+ adds two electrons to the band structure of VO 2 , which further increases the carrier density and coaxes the system toward transformation to the metallic state [26].
Transmittance is another indicator of the variation of thermochromic performance after doping treatment. The transmittance of thin films measured at relatively low temperatures (30°C , solid curves) and high temperatures (90°C, dashed curves) are exhibited in Fig. 3c. Pure VO 2 and three W-doped VO 2 thin films (especially VO-W1 and VO-W2) exhibit an obvious transmittance contrast between the low-temperature semiconductor phase and high-temperature metallic phase, especially in the near-infrared region, which is ascribed to the thermochromic nature of VO 2 . The plasma frequency (x p ) of the metallic state VO 2 is reported to be 1.0 eV ($1240 nm) or 1.6 eV ($775 nm); thus, the rutile phase has a lower transmittance in the IR region because the incident light at a frequency below the plasma frequency leads to motion in the charge carriers, which screens out the incident field, in other words, reflects the incident waves. The increase in W-doping content results in the shrink of the transmittance contrast between the two phases. This difference is mainly caused by a relatively low testing temperature (30°C) because all the samples have completely transformed into the metallic phase under a relatively high testing temperature (90°C). For sample VO-W2, the transmittance in both phases is low and the contrast is minimum because the transition temperature of VO-W2 is 8°C, which is much lower than the test temperature (30°C); thus, the metallic phase fails to completely transform into the semiconductor phase. On the contrary, for sample VO-W1, almost all the grains are in the semiconductor phase because its transition temperature is much higher than the test temperature (30°C), while for sample VO-W2, whose transition temperature is near the test temperature, part of the grains is in the semiconductor. The optical transmittance of the sputtered films at the fixed wavelength of 2000 nm was studied as a function of temperature from 35°C to 95°C. The thermal hysteresis loops are clearly shown in Fig. 3d. The loops in the transmittance spectra are similar to those in the resistance curves and confirm the depression of transition temperature after W-doping treatment.
In vivo antifouling performance H&E and Giemsa staining were performed for histological evaluation to observe morphology change and microbe amounts in    peri-implant soft tissue. For the control (Fig. 4a1-d1), the typical representation of soft tissue infection was observed at the interface between quartz glass and circumambient soft tissue, and there existed apparent acute inflammation, tissue necrosis, and extensive neutrophil infiltration into the soft tissue. In the meantime, the corresponding Giemsa staining slice confirmed the presence of large numbers of microbes at the implant-tissue interface (Fig. 4e1). To some extent, those situations were improved for sample VO 2 (Fig. 4a2-e2). With a sharp contrast, according to the H&E and Giemsa staining slices, sample VO-W1 (Fig. 4a3-e3) and VO-W2 (Fig. 4a4-e4) showed more moderate inflammatory response, lower neutrophil infiltration, and less microbe residue. Sample VO-W3 displayed the mildest inflammatory reaction and lowest amount of microbe residue at the implant-soft tissue interface (Fig. 4a5-e5). As a result, the in vivo anti-infection effect showed a strong dependence on the conductivity of the underlying substrate (i.e., undoped and doped VO 2 films). We then used the spread plate method to quantify the microbial burden from the quartz glass and peri-implant skin tissue. Representative photographs of microbial colonies are given in Fig. 4f. According to the corresponding counting results of the colony forming units (CFU) per implant (Fig. 4g) and CFU per g skin tissue around the implant (Fig. 4h), the average number of CFU per implant demonstrated the following trend (Â10 5 ): Control (19.69 ± 1.27) > VO 2 (5.90 ± 0.52) > VO-W1 (4.05 ± 0.63) > VO-W2 (1.21 ± 0.15) > VO-W3 (0.57 ± 0.15), and there was an apparent statistical difference among the groups (P < 0.001). The mean number of CFU per g peri-implant skin tissue showed a similar trend (Â10 5 ): Control (49.63 ± 0.84) > VO 2 (14.41 ± 0.44) > VO-W1 (11.59 ± 0.40) > VO-W2 (6.16 ± 0.26) > VO-W3 (3.85 ± 0.24), and there was a significant difference among the groups (P < 0.01). Interestingly, when we put them together for comparison, as shown in Fig. S3, the number of CFU in skin tissue was obviously larger than the corresponding number of CFU on material with significant statistical difference (P < 0.001), thus indicating that the antimicrobial incident mainly occurred on the surface of VO 2 and W-doped VO 2 films. This has suggested that the conductivity of the film materials plays an important role in the antimicrobial behavior; in other words, high conductivity contributes to diverting electrons from bacteria.

In vitro antimicrobial performance
To further reveal the in vivo antimicrobial behaviors of the VO 2 and W-doped VO 2 samples, we performed the in vitro antimicrobial test. As seen in Fig. 5a, the viability of MRSA was visualized by live/dead staining after 24 h of culture on various samples, and the red fluorescent intensity increased clearly in the order of Control, VO 2 , VO-W1, VO-W2, and VO-W3, thus confirming the significant loss of MRSA viability upon direct physical contact. SEM observation was also used to examine the morphology of MRSA on various samples. Large numbers of microbes were observed on the control sample with continuous biofilms (Fig. S4a). After VO 2 deposition, this situation was ameliorated to some degree (Fig. 5b). By W doping, the number of microbes sharply decreased in VO-W1 (Fig. S4c), and this antimicrobial effect became increasingly powerful for VO-W2 (Fig. S4d) and VO-W3 (Fig. S4e). Meanwhile, significant microbial lysis was also observed on the W-doped VO 2 films. Therefore, the SEM observations were in accordance with the live-dead staining results, thus confirming the conductivity-dependent antimicrobial activity of the VO 2 and W-doped VO 2 films.
The spread plate method was used to quantify the number of MRSA on materials. Representative photographs of microbial colonies are shown in Fig. 5b. The mean CFU/ml values of Control, VO 2 , VO-W1, VO-W2, and VO-W3 were (Â10 5 ) as Histological sections of important organs including the heart, liver, kidney, and spleen after the implantation of samples Control, VO 2 , VO-W1, VO-W2, and VO-W3, stained with H&E.
Overall, it was found that tuning the thermochromism of VO 2 films by W doping can effectively modulate the film conductivity, and thus determine the antimicrobial behavior. In addition, the corresponding histological changes of main organs, including the heart, liver, kidney, and spleen, were examined by H&E staining. As seen in Fig. 6, no apparent signals of organ damage can be observed from all the groups. Meanwhile, the proliferation results of rat bone mesenchymal stem cells (rBMSCs) in Fig. S5 also show the samples had no significant cytotoxicity on rBMSCs. In accordance with the endosymbiotic theory (symbiogenesis), eukaryotic mitochondrion is an analog of prokaryote bacteria. However, the electron transport chain of mammalian cell for respiration is located in the intracellular mitochondria. Therefore, its electron transport chain is not interrupted, and the cell can survive on the surfaces.

Physical parameters
To further unravel the mechanism of the enhanced antibiofouling capability subject to phase-change temperature of VO 2 films, we then investigated the influence of W doping on the band structure and electrical properties of VO 2 films. For semiconductor-phase VO 2 , the bandgap between the top of the filled a 1g band (E v ) and the bottom of the empty e g p band (E c ) is produced by the splitting of V 4+ 3d states. Intrinsic VO 2 thin film is n type [35]. In thermal equilibrium, the electron density in the conduction band P(N) = 2[2pm p(n) kT/h 2 ] 1/2 exp(À4E a /kT) is derived from the Fermi-Dirac distribution function, where m p(n) is the electron effective, k is the Boltzmann constant, and h is the Planck constant. The activation energy is defined as 4E a = E C À E F . The temperature scanning in the resistance measurements allows us to acquire the activation energy in the semiconductor phase. Based on the Arrhenius' law R = R 0 exp (ÀE a /k B T), the activation energy values are obtained by fitting the resistance (Fig. 7a). For the undoped VO 2 thin film, the calculated activation energy in the semiconductor phase is 0.29 eV (Fig. 7b). After W doping, the activation energy ranged from 0.18 eV (VO-W1) to 0.14 eV (VO-W3). The VO-W2 (0.17 eV) and VO-W3 had relatively lower activation energy and thus overall higher conductivity, which may be ascribed to donate electrons from d shells upon doping with tungsten.
To develop a further understanding of the electrical properties of these films, carrier concentrations were measured at room temperature using the Hall effect device, while the work function values were estimated by Kelvin Probe Force Microscopy at 37°C. For pure VO 2 thin film and W-doped films, the electrons are the major contributors, as indicated by the sign of the Hall voltage. As shown in Fig. 7c, the carrier densities of samples VO 2 , VO-W1, VO-W2, and VO-W3 at room temperature were 2.5 Â 10 18 , 3.6 Â 10 19 , 7.0 Â 10 20 , and 2.4 Â 10 21 , respectively. These results are consistent with that of resistance value. Moreover, the increase in W-doping concentration may lead to the increment of work function slightly. For instance, the value of pure VO 2 thin film is 5.08 eV only. When 1.89 at.% W is doped on the film (sample VO-W3), the value has increased by 0.25 eV. The influence of W doping on the work function is similar to that of temperature rise. For example, Ko et al. [36] estimated that the work function of VO 2 is $5.15 eV in the semiconductor phase. When it enters to the semiconductor-metal transition phase, the value increases by $0.15 eV measured by KPFM. Martens et al. [37] also found that the work function of VO 2 evolves from 5.2 eV (room temperature) to 5.3 eV (100°C), assessed by electrical characterization of VO 2 -metal contact resistivity. However, a couple of previous studies presented the dichotomized results on the exact value and temperature dependence of the work function of VO 2 . For instance, Sohn et al. [38] demonstrated that the work function drops from 4.88 eV to 4.70 eV when the semiconductor-metal transition phase is obtained. Similarly, Zhou et al. [39] suggested that work function of VO 2 decreases by $0.2 eV when it goes through semiconductor-metal transition by the measurement of the electrical properties of GaN/VO 2 p-n heterojunctions. These discrepancies in the literatures may attribute to the microstructure, strain state [40], and surface stoichiometry of VO 2 when exposed to air/humidity [36] or in contact with other materials and the limitation of measuring machine (e.g., resolution limit). To confirm the aforementioned hypothesis, UPS measurement was carried out. The UPS spectra recorded from various films (Fig. 7d) represent successively increasing intensity near the Fermi level. For the samples of VO 2 and VO-W1 in the semiconductor phase at room temperature and incubating temperature (37°C), the UPS intensity, resembling approximately the density of occupied initial states, does not disappear near the Fermi level, presumably due to a strong V 3d band tailing into the bandgap, which is induced by the defects in the polycrystalline films. When the sample of VO-W3 is fully in the metallic phase at either room temperature or 37°C, the intensity in the range of the V 3d band does not disappear and remains higher than the intensity at the Fermi level. As a result, in the semiconducting state, fewer electrons occupy the Fermi level, and the occupancy increases as a function of W concentration. The occupancy in the metallic state is higher than that in the semiconducting state near the Fermi level [41]. Therefore, the change in the band structure of samples VO 2 , VO-W1, VO-W2, and VO-W3 agrees well with the development of the aforementioned electrical properties.

Discussion
The temperature of microbiological incubators is usually set at 37°C, which mimics human body temperature. During surgery, medical devices such as implants, catheters, and surgical tools can undergo an ambient temperature change from room temperature to body temperature. The intrinsic thermochromic property can make VO 2 material undergo the reversible phase transition from a semiconducting state to a metallic state at the critical temperature (Tc) as an active response to ambient temperature stimuli or near-infrared irradiation, which is accompanied by discontinuous jumps in electrical conductivity. The extracellular electron transportation is a normal living movement for bacterial respiration. The adjustment of the phase-change/ electrical property of VO 2 is important because it can potentially interfere with the extracellular electron transportation process of bacteria when contacting with them, thereby producing temperature-responsive VO 2 antifouling coating that can block bacteria colonization and prevent subsequent biofilm formation on medical devices.
First, we tuned the transition temperature of VO 2 thin films by altering the W-doping content to the desired range, namely, from 61°C of VO 2 , 53°C of VO-W1, 36°C of VO-W2 to 8°C of VO-W3, which led to the manipulation of the electrical properties of VO 2 thin films (Fig. 3, Table S2). This design reassures that (i) VO 2 and VO-W1 incubated at either room temperature or human body temperature are in a semiconducting state; (ii) VO-W2 is certainly in a semiconducting state at room temperature, but a mixed state of semiconducting and metallic appears at human body temperature; and (iii) VO-W3 is entirely in a metallic state, whether it is subject to room temperature or human body temperature. The resistance ratio between the two phases is slightly degraded with the increase in W-doping concentration, which is in accordance with previous reports [31,42]. The W-doping-induced resistivity decrease can be primarily attributed to several reasons. On the one hand, the substitution of V atoms with W in the VO 2 matrix leads to the tungsten oxidation state of 4+ and causes the valence band to be populated with tungsten d electrons. On the other hand, addition of tungsten may break V 4+ -V 4+ bonds and result in the formation of two new bonds: V 3+ -W 6+ and V 3+ -V 4+ . In this case, the resistivity decrease is due to the addition of released vanadium electrons from the d electronic shell [43]. Electrical conductivity (r) describes how the ease of material (e.g., thin films) allows charges (e.g., electrons) to move or get through the material and it totally depends on the resistivity (q), as: r = 1/q. Hence, the orders of electrical conductivity of the samples are: VO-W3 > VO-W2 > VO-W1 > VO 2 . The carrier concentration of the samples presents the same tendency as plotted in Fig. 7c. After that, we used a rat subcutaneous infection model to examine whether such a material design can influence the interactions between VO 2 coatings and microbial cells, and thus combat biofouling. As evidenced in Fig. 4, through tuning the semiconductor-tometal transition of VO 2 thin films to alter electron concentration by W doping, the antifouling outcome can be very different among groups. With enhancing the film conductivity, the colonization of microorganisms on the coated implant or in the skin tissue adjacent to the implant can be sharply reduced, thus displaying a mild inflammatory reaction and a low amount of microbial residue at the implant-skin tissue interface.
To determine the mechanism, we first cultured MRSA in vitro to investigate their response to pure and W-doped VO 2 thin films. As evidenced in Fig. 5, along with improving the film conductivity, the amounts of MRSA colonization on the coated samples were acutely decreased and exhibited favorable antibiofouling activity. These in vitro results agreed with and partially explained the in vivo outcomes. To further understand the underlying mechanism, we conducted a series of physical measurements on the band structures and electrical properties of the VO 2 and W-doped VO 2 thin films. As evidenced in Fig. 3, by virtue of W doping, the phase-change VO 2 thin films can be responsive to a range of environment temperatures, including body temperature, room temperature, and seawater temperature (30-8°C [44]). Accompanied by this, the electrical properties of VO 2 thin films were modulated in a large range. The activation energies were significantly lowered from the VO 2 group to VO-W3 group ( Fig. 7a and b). Accordingly, the carrier concentrations were apparently elevated from the VO 2 group to VO-W3 group, while the work function values were increased by 0.25 eV (Fig. 7c). Obviously, the change in the electrical conductivity also indicated the variation in the band structure of VO 2 thin films, which was further demonstrated by the results of UPS measurements (Fig. 7d). With reference to the values of activation energy and work function previously described, the band diagrams of doped and undoped VO 2 thin films can be interpreted.
In addition, the VO-W3 group has exhibited superior antifouling ability than that of the VO-W2 group. This observation can attribute to the higher electrical conductivity of VO-W3 as compared with the sample of VO-W2. In fact, the VO-W3 sample is already in a metallic state at either room temperature or human body temperature, whereas the VO-W2 only appears in a mixed state of metallic and semiconducting at human body temperature. Hence, the antifouling abilities of the samples should be in the order of VO-W3 > VO-W2 > VO-W1 > VO 2 .
These results confirm a close relationship between the electrical conductivity (band structure) and antifouling activity of VO 2 thin films. Regarding the in vivo toxicity, the histological images of main organs (heart, liver, kidney, and spleen) shown in Fig. 6 demonstrate that all the implanted samples did not cause significant abnormalities or lesions, thus showing no apparent in vivo toxic side effects; moreover, as shown in Fig. S5, there was no significant cytotoxicity on rat bone mesenchymal stem cells. In fact, vanadium has been documented on the list of 40 essential micronutrients and are required in trace amounts for normal metabolism, proper growth, and development of mammals [45]. Vanadium deficiency causes growth inhibition, tooth, bone and cartilage growth impairment, generative function disorder, and lipids and carbohydrate metabolic pathways disturbance [45,46]. Some studies show that vanadium can enhance the mineralization of bones and teeth in laboratory animals [47] and can improve the integration of implant and soft tissue [48].
Our results demonstrated for the first time that the antimicrobial property of VO 2 correlates with its electrical conductivity or band structure. Extracellular electron transfer along the respiratory chain on the plasma membrane of microorganisms natu-rally proceeds through a sequence of the transmembrane protein complex, including complex I (consisting of the motifs for binding NADH, FMN and Fe-S cluster), complex II (succinate ubiquinone reductase), complex III (ubiquinone cytochrome c reductase), and complex IV (cytochrome c oxidase), as illustrated in Scheme 1a. Therefore, it is reasonable to consider that the leakage of electrons out of the electron transfer chain will jeopardize energy metabolism and microbial viability. The stronger antimicrobial activity of W-doped VO 2 films indicated that, after microbe-VO 2 physical contact and membrane perturbation, VO 2 exerted conductivity-dependent antifouling function. Due to the electrically conductive property, the VO 2 films of metallic state resulting from W doping could be able to "short-circuit" the microorganisms by acting as a conductive electron pump or electron trap that can destructively extract electrons from the electron transfer chain of microorganisms and release microbial energy into the extracellular environment. That is, not only can VO 2 films serve as conduits for extracellular electron transfer out of microbial plasma membrane but also the occurrence of the electron leakage shows strong dependence on the electrical conductivity or band structure of the thin films. Apparently, the aforementioned transmembrane protein complex is highly redox-active, as seen in Table S3. Thus, by tuning the electrical conductivity or band structure of VO 2 films, it is feasible to make extracellular electron transfer occur between the VO 2 films and redox-active bystanders [49,50].
The observations of live/dead staining (Fig. 5a) and SEM (Fig. S5) collectively demonstrate that the inactivation and death of MRSA should be largely attributed to the destruction of plasma membrane integrity. The sequence of transmembrane protein complex can shuttle electrons along the respiratory chain, normally to the terminal electron acceptor, dioxygen, and produce H 2 O through complex IV. However, under external stimuli, such as exposure to metallic VO 2 , the overproduction of reactive oxygen species (ROS) can be caused by electron leakage from the impeded electronic transport. For instance, electron leakage from the ubiquinone (CoQ) can induce the overproduction of Å O 2 [51] and other species in Table S4. The mechanism of electrically conductive VO 2 thin films acting as antifouling coatings on a substrate to prevent biofilm formation is illustrated by a schematic diagram (Scheme 1). The elevated ROS production can cause oxidative damage to the plasma membrane, which may account for the observed destruction of MRSA membrane integrity [52]. Meanwhile, interruption of electron transport will also disturb the synthesis of adenosine triphosphate (ATP) [53]. At this stage, the observed antifouling process can be summarized in three steps. The first step is that microorganisms initially adhere or deposit onto the metallic VO 2 films to form direct microbe-VO 2 physical contact, followed by the destructive extraction of electrons from the transmembrane protein complex, a discharge process of respiratory chain, by the electrically conductive VO 2 films, which causes ROS overproduction and ATP de-synthesis, thus leading to oxidative damage and energy starvation. Finally, the intact microorganisms collapse and fragment, which causes membrane integrity destruction and intracellular matter leakage, eventually resulting in the loss of microbial viability, as illustrated in Scheme 1b. To further elucidate it, the energy band diagrams in Scheme 2 have been drawn in order to highlight the interactions between pure/ W-doped VO 2 and the respiration chain in bacterial membrane. The biological redox potential (BRP) values of transmembrane protein complex in respiration chain are listed in Table S3. Based on the conversion equation of standard hydrogen electrode (E 0 , V) and absolute potential level (E, eV) [54], E 0 = À4.44 À E, the BRP therefore ranges from À4.12 eV to À4.84 eV. Table S5 suggests the estimated energy levels of un-doped and doped VO 2 thin films at human body temperature (37°C). Referring to our findings, Scheme 2a demonstrates the energy band diagram of VO 2 thin films and respiration chain, indicating that the VO 2 film in semiconducting phase can rarely extract electrons from the respiration chain, and then interrupt electron transfer process due to the conduction band in the range of BRP. In contrast, the W-doped VO 2 thin films can spontaneously extract electrons from the respiration chain and disrupt the BRP because of the Fermi level lower than BRP (Scheme 2b), thereby leading to the interruption of electron transfer process.
As a result, by W doping, the electrical conductivity of VO 2 films can be further enhanced to possess powerful electron trapping capability, which contributes to the observed antifouling activity to prevent biofilm formation. In fact, altering the electrical conductivity of the underlying substrate can also significantly influence the microbial behavior on the surface [55]. In fact, VO 2 is well known for its ultrafast reversible first-order semiconductor-to-metal transition at the critical temperature of $68°C, which serves as an active response to environmental temperature stimuli or near-infrared light irradiation (NIR) from infrared-transparent semiconductor phase (M1) to infraredreflective metallic phase (R) [24,56]. This distinctive temperature-dependent phase change constitutes VO 2 to be a promising candidate of temperature-responsive or NIR-RESEARCH: Original Research SCHEME 1 Schematic illustration elucidating the mechanism of electrically conductive VO 2 thin film acting as antifouling coating on a substrate to thwart biofilm formation: (a) When microorganisms initially adhere onto a bare substrate e.g., titanium or stainless steel, they can normally carry out extracellular electron transfer to terminal acceptor oxygen and ATP synthesis through respiration. (b) When microorganisms initially adhere onto VO 2 :W film that can respond to environmental temperature e.g., body temperature 37°C and transform from semiconductor phase to metallic phase, the direct physical contact induces the destructive electron extraction from transmembrane protein complex, a discharge process of respiratory chain, which thus causes ROS overproduction and ATP dyssynthesis, thereby leading to the oxidative damage and energy starvation and eventual electrocution of microorganisms. responsive intelligent material. As evidenced by this study, W doping can effectively modulate the VO 2 phase change initiated by temperature alteration and the thermochromic optical property just comes concomitantly. While exposing W-doped VO 2 to human body temperature (i.e., 37°C) from room temperature, the semiconductor-to-metal phase change will be induced and lead to the concomitant dramatic change in electrical conductivity or band structure. Indeed, the change of electrical conductivity or band structure serves as the key factor of antifouling function. As a result, W doping can manipulate the phase change behaviors of VO 2 thin films. When the phase change of VO 2 film occurs at desired temperature, the electrical conductivity will be largely upregulated. For instance, when phase change of the film occurs at human body temperature and in contact with microor-ganisms, metallic VO 2 can perturb the normal physiological function of transmembrane proteins and consequently interrupt the electron transport along the respiratory chain. This will impede energy metabolism, cause oxidative damage and membrane leakage, electrocute microorganisms, and prevent biofilm formation. Most importantly, due to the advantage of using phase-change VO 2 , the stimuli-responsive (temperatureresponsive or NIR-responsive) antifouling function at a desired temperature can be achieved using W doping. As a result, the mechanism of increasing loss of MRSA viability on W-doped VO 2 films of metallic state is caused by the metallic, electrically conductive, VO 2 -mediated electron leakage out of the microbial plasma membrane. In addition, the specific nano-patterns appearing on the thin-film surfaces may also contribute to the SCHEME 2 Schematic illustration of the energy band diagrams of VO 2 (a) and VO-W3 (b) at human body temperature (37°C) while interacting with the respiratory chain in bacterial membrane with BRP. (a) the semiconductor phase of VO 2 thin film can rarely disrupt BRP and electron transfer, whereas (b) the metal phase of VO-W3 film can spontaneously disrupt BRP and electron transfer, thereby leading to electron leakage. antimicrobial effect. For instance, nickel nanostructures are able to alter the adhesion behaviors of Staphylococcus aureus on material surface [57]. Puckett et al. [58] also investigated the effects of nanorough, nanotubular and nanotextured titanium surfaces against the adhesion of Staphylococcus aureus. It was found that the bacteria unlikely adhered onto the nanorough surfaces, whereas the nanotubular surfaces provided a good environment for bacterial adhesion.

Conclusions
In summary, we have demonstrated the capability of W-doped VO 2 thin films to act as functional antifouling coatings. Microbial biofilms are ubiquitous on the surfaces of domestic, industrial, medical, and marine infrastructures/devices, and conventional biocides are increasingly problematic due to poor efficacy, nontarget environmental impact, and the evolution of resistance by microorganisms. Here, we show that W-doped VO 2 thin films with tunable phase change can exert temperature-responsive antifouling activity. When in contact with microorganisms, the electrically conductive VO 2 can disrupt membrane function and prevent biofilm formation. Targeting microbial membrane function is a promising strategy for impeding biofilm formation [59]. The deposition method is suitable for batch production for metal, glass, ceramic, or polymer substrates of various shapes, and the VO 2 films can even be deposited in arbitrary microscale patterns by masking specific areas to enable spatially patterned antifouling coatings. It is believed that these VO 2 systems can serve as an alternative strategy to overcome the current limitations of biofouling control.   (Table S2) based on the resistance-temperature curves. Although the △Ea is calculated at specific temperature region (about 390-320 K for VO2 sample, 280-310 K for VO-W1 and 260-290 K for VO-W2), the values of calculated △Ea can be approximatively equal to that at 310 K (37 o C) since the temperature of 310 K is near to the center temperate of fitted temperature region, especially for sample VO2 and VO-W1. Sample VO-W3, which is fully in its metallic phase, is exceptional and therefore its △Ea calculated at 230-260K is useless in terms of band diagram at 310 K (37 o C).

In vitro cytocompatibility evaluation
The rat bone mesenchymal stem cells (rBMSCs) were used to evaluate the cytocompatibility of the various VO2 samples, and the cell viability was determined using Cell Counting Kit-8 (CCK-8, Beyotime Bio-Tech) assay [7]. Five samples were used for each group, positioned in a new 24-well plate with cell seeding density of 10 5 cells/ml. After 1 day and 3 days of incubation at 37 °C in a humidified environment with 5% CO2, the culture media were removed and fresh media containing 10% CCK-8 were added into the wells for additional 3 h of incubation. Finally, 100 μl of the incubated medium from each group was sucked up in a new 96-well plate and the absorbance was measured at 450 nm on a microplate reader.