Towards Room Temperature Phase Transition of W-Doped VO2 Thin Films Deposited by Pulsed Laser Deposition: Thermochromic, Surface, and Structural Analysis

Vanadium dioxide (VO2) with an insulator-to-metal (IMT) transition (∼68 °C) is considered a very attractive thermochromic material for smart window applications. Indeed, tailoring and understanding the thermochromic and surface properties at lower temperatures can enable room-temperature applications. The effect of W doping on the thermochromic, surface, and nanostructure properties of VO2 thin film was investigated in the present proof. W-doped VO2 thin films with different W contents were deposited by pulsed laser deposition (PLD) using V/W (+O2) and V2O5/W multilayers. Rapid thermal annealing at 400–450 °C under oxygen flow was performed to crystallize the as-deposited films. The thermochromic, surface chemistry, structural, and morphological properties of the thin films obtained were investigated. The results showed that the V5+ was more surface sensitive and W distribution was homogeneous in all samples. Moreover, the V2O5 acted as a W diffusion barrier during the annealing stage, whereas the V+O2 environment favored W surface diffusion. The phase transition temperature gradually decreased with increasing W content with a high efficiency of −26 °C per at. % W. For the highest doping concentration of 1.7 at. %, VO2 showed room-temperature transition (26 °C) with high luminous transmittance (62%), indicating great potential for optical applications.


Introduction
VO 2 has the closest phase transition to room temperature of any the thermochromic oxide materials and has consequently been extensively studied for a variety of applications including electronic switches, smart windows, memory devices, RF microwave switches, and terahertz metamaterial devices [1][2][3][4][5]. Indeed, VO 2 enables insulator-to-metal transition (IMT) as well as transition from a monoclinic (M1) to a rutile structure (R) at around 68 • C [6]. However, the phase transition temperature of pristine VO 2 thin films is too high for room temperature applications. To meet the demand for a broad range of room temperature applications, considerable efforts have been made to reduce the phase transition temperature. One way to do so is replacing V 4+ ions by metal ions with higher valences such as W 6+ , Nb 5+ , and Mo 6+ [7][8][9][10], among which W has been considered one of the most effective dopants. Interestingly, the doping of W into the VO 2 matrix affects the transition characteristics. The substitutional doping of the W 6+ to replace V 4+ has been shown to lead to a remarkable reduction in the phase transition temperature, i.e., 20-28 • C per W at. % [7,[10][11][12][13][14]. VO 2 -based thin films can be synthesized using several techniques including sol-gel [15], magnetron sputtering [16,17], chemical vapor deposition [18], and pulsed laser deposition (PLD). However, there have been few reports on synthesizing W-doped VO 2 films using PLD [19][20][21][22], which is known to be a versatile method to control doping in thin films. Komal et al. [23] used mixed V 2 O 3 and WO 3 pellets as targets for the PLD process and observed that the VO 2 doped with 1.5 at. % of W reduced the phase transition temperature towards room temperature (27 • C). Émond et al. [24] used PLD to prepare W-doped VO 2 films with a phase transition temperature of 36 • C. Soltani et al. [25] manufactured 1.6 at. % W-doped VO 2 thin films using a W-doped vanadium oxide target. The phase transition temperature was about 36 • C and all-optical switching was demonstrated at a telecommunication wavelength of 1.55 µm. S. S. Majid et al. [26] used a W-doped target and obtained W-doped VO 2 films with a significantly lower phase transition temperature (∼19 • C for 1.3% W) by stabilizing the metallic rutile phase R. All these studies demonstrate that PLD is a versatile method capable of tuning the properties of VO 2 . However, they used either a target mixture of tungsten and vanadium oxide or a co-ablation process and mainly focused on reducing the phase transition temperature. In the present work, an innovative method to synthesize W-doped VO 2 thin films using pulsed laser deposition is proposed. This method consists in fabricating multilayers composed of V/W (+O 2 ) and V 2 O 5 /W bilayers, followed by rapid thermal annealing to allow both crystallization of VO 2 and the diffusion of W through the VO 2 layers to synthetize homogeneous W-doped VO 2 thin films. The main advantage of this process over co-ablation deposition is its repeatability. Indeed, co-ablation can cause problems of thickness and of non-homogeneous composition as well as contamination between the targets. In addition to the synthesis process, this work not only aims to reduce the phase transition temperature but also to improve our understanding of the surface chemistry of W-doped films. For the latter, angle-resolved X-ray photoelectron spectroscopy (ARXPS), which is rarely used in VO 2 -reported studies, was used to measure the concentration of vanadium, the depth of tungsten oxidation, as well as to assess their depth distribution.

Synthesis of W-Doped VO 2 Thin Films Using Pulsed Laser Deposition
W-doped VO 2 thin films were synthesized by rapid thermal annealing of multilayers composed of VO X /W bilayers as illustrated in the figure inserted in Table 1. The bilayers were deposited on fused silica substrates using pulsed laser deposition (PLD) at room temperature. The PLD chamber was evacuated to a background pressure of 2 × 10 −7 mbar. Three types of targets were used in the investigation: Vanadium (V), Vanadium pentoxide (V 2 O 5 ), and Tungsten (W), all at 99.9% purity. A KrF excimer laser (wavelength 248 nm, repetition rate 10 Hz, fluence 13 J/cm 2 ) was used to ablate the targets. The target/substrate distance was 5 cm. The laser beam was oriented with an angle of 45 • to the target. For the W-doped VO X using metallic vanadium as a precursor, the oxygen pressure inside the deposition chamber was 1 × 10 −2 mbar when ultrapure O 2 was used. For the W-doped VO X using the V 2 O 5 target, ablation was performed without oxygen gas. During the deposition process, the targets were rotated to enable uniform ablation. For all depositions, the deposition rate obtained by profilometry was around 6.3 nm/min for V+O 2 , 6.4 nm/min for V 2 O 5 , and 1.37 nm/min for W. The alternative ablation sequences of V2O5 (or V in O 2 ) and W were adjusted to obtain three multilayers (A, B, C) with different W contents to obtain three W-doped VO X thin films with a similar total thickness of 20 nm (Table 1). Table 1. Condition of deposited multilayers using pulsed laser deposition. The ablated thickness per layer and the number of individual layers were adjusted to obtain three different W contents (next investigated using XPS). Total thickness (nm) 20 20 20 The rapid thermal annealing process was performed using an RTP furnace (AS-One RTP system from Annealsys, Montpellier, France). For samples A and B, the annealing temperature was 400 • C for 120 s, while for sample C, the annealing temperature was 450 • C for 60 s. The time of rapid thermal annealing was adjusted from preliminary experiments, depending on the two different precursors, to ensure the observation of the thermochromic transition in both cases. All the annealing processes were carried out at an oxygen partial pressure of 1 mbar and a 50 sccm flow. The heating rate was 5 • C/s and the cooling was natural. The annealing chamber was evacuated to a 1 × 10 −2 mbar before the oxygen was introduced. Undoped VO 2 thin films were obtained using a similar process for the purpose of comparison.

Characterization of the Undoped VO 2 and W-Doped VO 2 Films
The composition and dopant profile were investigated using XPS analysis using a Thermo VG Theta probe spectrometer (Thermo Fisher Scientific, Invitrogen, Waltham, MA, USA) with a focused monochromatic AlKα source (hν = 1486.68 eV, 400 µm spot size) and photoelectrons were collected using a concentric hemispherical analyzer. A constant ∆E mode and a 2D channel plate detector (Thermo Fisher Scientific, Invitrogen, Waltham, MA, USA) were used. The energy scale was calibrated with sputter-cleaned pure reference samples of Au, Ag, and Cu such that Au4f 7/2 , Ag3d 5/2 , and Cu3p 3/2 were positioned at binding energies of 83.98, 386.26, and 932.67 eV, respectively. Angle-resolved XPS analyses were performed thanks to the ability of the spectrometer to simultaneously collect several photoelectron emission angles in the acceptance range of 60 • without tilting the sample. Charge neutralization was applied during analysis. High-resolution spectra (i.e., O1s-V2p, V3p-W4f) were fitted using AVANTAGE software version 5.9 by Thermo Fisher Scientific. The structure of the films was analyzed by Raman analysis (Jobin-Yvon ARAMIS) using a Helium-Neon laser source (Horiba Jobin Yvon, Gières, France) at an excitation wavelength of 633 nm, a laser power of 0.1 mW, focused with a 100× objective, consistent with a spot diameter of less than 1 mm. Atomic force microscopy (AFM) (Icon BRUKER, Berlin, Germany) and scanning electron microscopy (SEM) (JEOL IT 800 SHL, Tokyo, Japan) were used to analyze the topography, roughness, and morphology of the films. The thermochromic properties of the films were measured by collecting the transmittance in a temperature range between 15 and 100 • C using a fiber optic spectrometer (Ocean Insight, Duiven, The Netherlands) equipped with handmade heating units in the visible (400-800 nm) and IR (900-2500 nm) wavelength ranges. For the undoped and W-doped VO 2 thin films, the integrated luminous transmittance (T lum , 380-780 nm) and solar modulation efficiency (T sol , 300-2500 nm) were deduced from the following equation: T lum, sol = ( ϕ lum, sol (λ)T(λ)dλ)/( ϕ lum, sol (λ)dλ) (1) where T(λ) is film transmittance at wavelength (λ), ϕ lum (λ) is the standard luminous efficiency depending on the photopic vision of human eye, ϕ sol (λ) is the solar irradiance spectrum (air mass 1.5) corresponding to the sun at an angle of 37 • to the horizon [23]. ∆T sol is obtained from ∆T sol = T sol (15 • C) − T sol (100 • C) and T lum = (T lum (15 • C) + T lum (100 • C)/2).

Thermochromic Properties
To investigate the influence of W-doping on thermochromic properties, the transmittance of the samples during a heating and cooling phase were recorded and plotted the hysteresis loops of undoped and W-doped thin films. Figure 1a shows the hysteresis loop produced by plotting the temperature dependence of transmittance of undoped and W-doped VO 2 thin films at a fixed wavelength of 1500 nm. The switching temperatures during heating (T t,h ) and cooling (T t,c ) were determined from the half-value width of each curve and the average temperature of commutation (T t ) representative of thermochromic behavior was defined as T t = (T t,h + T t,c )/2. Both undoped VO 2 films had a phase transition temperature Tt of 70-71 • C (quite similar whatever the precursors).

Thermochromic Properties
To investigate the influence of W-doping on thermochromic properties, the transmittance of the samples during a heating and cooling phase were recorded and plotted the hysteresis loops of undoped and W-doped thin films. Figure 1a shows the hysteresis loop produced by plotting the temperature dependence of transmittance of undoped and W-doped VO2 thin films at a fixed wavelength of 1500 nm. The switching temperatures during heating (Tt, h) and cooling (Tt, c) were determined from the half-value width of each curve and the average temperature of commutation (Tt) representative of thermochromic behavior was defined as Tt = (Tt, h + Tt, c)/2. Both undoped VO2 films had a phase transition temperature Tt of 70-71 °C (quite similar whatever the precursors).
The dependence of the phase transition temperature on the tungsten content was linear ( Figure 1b). W doping significantly reduced Tt to ~−26 °C per at. % W, as shown in Figure 1b, which is in line with reported values of ~20-28 °C per at. % W [7,[12][13][14]. The Tt reduction mechanism of the W-doped VO2 can be ascribed to free electron carriers generation [27] as well as to the extra strain resulting from the atom replacement of V 4+ by W 6+ . This induced high symmetry around the W atom, implying transformation into a rutile structure [28,29], which was subsequently corroborated by Raman analysis.
Hysteresis behavior is another important thermochromic parameter of VO2-based thin films. The hysteresis loop width gradually narrowed from 11 °C for the VO2 film to 4 °C for the film doped with 1.7 at. % W, as can be seen in Figure 2a. This illustrates that the W dopant not only reduces the transition temperature but also reduces the width of the hysteresis loop. Previous studies suggested that ΔT is closely linked to the grain size, lattice stress, impurity phase, and crystallinity of VO2 film [30,31]. Structural defects induced by W doping act as nucleation sites of the phase transition [32,33]. Therefore, the activation energy of the phase transition would be reduced, with a decrease in hysteresis width. The result is promising, as narrow hysteresis loops are required to create devices with rapid commutations.  The dependence of the phase transition temperature on the tungsten content was linear ( Figure 1b). W doping significantly reduced T t to~−26 • C per at. % W, as shown in Figure 1b, which is in line with reported values of~20-28 • C per at. % W [7,[12][13][14]. The T t reduction mechanism of the W-doped VO 2 can be ascribed to free electron carriers generation [27] as well as to the extra strain resulting from the atom replacement of V 4+ by W 6+ . This induced high symmetry around the W atom, implying transformation into a rutile structure [28,29], which was subsequently corroborated by Raman analysis.
Hysteresis behavior is another important thermochromic parameter of VO 2 -based thin films. The hysteresis loop width gradually narrowed from 11 • C for the VO 2 film to 4 • C for the film doped with 1.7 at. % W, as can be seen in Figure 2a. This illustrates that the W dopant not only reduces the transition temperature but also reduces the width of the hysteresis loop. Previous studies suggested that ∆T is closely linked to the grain size, lattice stress, impurity phase, and crystallinity of VO 2 film [30,31]. Structural defects induced by W doping act as nucleation sites of the phase transition [32,33]. Therefore, the activation energy of the phase transition would be reduced, with a decrease in hysteresis width. The result is promising, as narrow hysteresis loops are required to create devices with rapid commutations. Moreover, the luminous transmittance (Tlum) and the solar modulation ability (ΔTsol) are plotted in Figure 2b. Compared to undoped VO2 thin films, the luminous transmittance of the W-doped VO2 thin films was higher. Indeed, Tlum increased with increasing W content. The luminous transmittance of the W-rich sample was 62.2%, a rather very good performance for W-doped VO2 thin films. In the literature, the luminous transmittance of VO2 after W-doping usually either remained unchanged or decreased, as reported in Table 2. In the present study, our W-doped VO2 thin films had higher Tlum. Further investigations would be necessary to identify the exact origin of such a significant increase of Tlum at the two highest W contents. The solar modulation ability ΔTsol decreased gradually with increasing W content. This is consistent with the results of previous works [34,35], in which ΔTsol decreased with increasing W content. The variation of ΔTsol depending on W content is in good agreement with the variation of Tt. Two explanations are possible: the incorporation of W induces a destabilization of VO2(M) lattice, and reduces the phase transition temperature. On the other hand, too much W doping leads to an excess of free electrons, thereby deteriorating the phase transition property [36]. Overall, as shown in Table 2, our W-doped VO2 thin films outperformed other reported works on W-doped VO2 using different synthesis methods in terms of reducing Tt and improving luminous transmittance.  Moreover, the luminous transmittance (T lum ) and the solar modulation ability (∆T sol ) are plotted in Figure 2b. Compared to undoped VO 2 thin films, the luminous transmittance of the W-doped VO 2 thin films was higher. Indeed, T lum increased with increasing W content. The luminous transmittance of the W-rich sample was 62.2%, a rather very good performance for W-doped VO 2 thin films. In the literature, the luminous transmittance of VO 2 after W-doping usually either remained unchanged or decreased, as reported in Table 2.
In the present study, our W-doped VO 2 thin films had higher T lum . Further investigations would be necessary to identify the exact origin of such a significant increase of T lum at the two highest W contents. The solar modulation ability ∆T sol decreased gradually with increasing W content. This is consistent with the results of previous works [34,35], in which ∆T sol decreased with increasing W content. The variation of ∆T sol depending on W content is in good agreement with the variation of T t . Two explanations are possible: the incorporation of W induces a destabilization of VO 2 (M) lattice, and reduces the phase transition temperature. On the other hand, too much W doping leads to an excess of free electrons, thereby deteriorating the phase transition property [36]. Overall, as shown in Table 2, our W-doped VO 2 thin films outperformed other reported works on W-doped VO 2 using different synthesis methods in terms of reducing T t and improving luminous transmittance.

Surface Chemistry Analysis: Composition and Depth Distribution
To analyze the surface chemistry, the high-resolution XPS spectra of V 2p and W 4d core-levels of W-doped VO X (as-deposited) and W-doped VO 2 (after annealing) thin films are shown in Figure 3a,b, respectively.

Surface Chemistry Analysis: Composition and Depth Distribution
To analyze the surface chemistry, the high-resolution XPS spectra of V 2p and W 4d core-levels of W-doped VOX (as-deposited) and W-doped VO2 (after annealing) thin films are shown in Figure 3a,b, respectively. The W peaks of W 4d5/2 and W 4d3/2 were located at 246.8 and 259.8 eV, respectively, reflecting the + 6 oxidation state of W ions in both thin films [11,12,41,44]. The V 2p corelevel was split into two regions, V 2p3/2 located at 516.9 eV and V 2p1/2 located around 524 eV [45,46], with the O1s located at 530 eV. The intensity of vanadium and oxygen peaks is not the same for all films, especially when the two deposition precursors V/W (O2) and V2O5/W are compared. Since the intensity of the tungsten peak differed from one sample to another, the W content differed. The presence of W can also be confirmed by its W 4f valence state [11]. The angle-resolved XPS spectra of W 4f core-levels at 23 and 68° of sample C are shown in Figure 2a after thermal annealing. Two strong peaks of W 4f core- The W peaks of W 4d 5/2 and W 4d 3/2 were located at 246.8 and 259.8 eV, respectively, reflecting the + 6 oxidation state of W ions in both thin films [11,12,41,44]. The V 2p corelevel was split into two regions, V 2p 3/2 located at 516.9 eV and V 2p 1/2 located around 524 eV [45,46], with the O1s located at 530 eV. The intensity of vanadium and oxygen peaks is not the same for all films, especially when the two deposition precursors V/W (O 2 ) and V 2 O 5 /W are compared. Since the intensity of the tungsten peak differed from one sample to another, the W content differed. The presence of W can also be confirmed by its W 4f valence state [11]. The angle-resolved XPS spectra of W 4f core-levels at 23 and 68 • of sample C are shown in Figure 2a after thermal annealing. Two strong peaks of W 4f core-level located at 34.7 eV and 36.9 eV were assigned to W 4f 7/2 and W 4f 5/2 , respectively. This revealed that W 6+ cations were present in W-doped VO 2 thin films in line with previous reported work [47], and confirmed the successful W-doping of VO 2 thin films. Furthermore, the atomic concentrations of W were calculated based on the XPS spectra of V 3p and W 4f and the results of the quantification are shown in Figure 4b.   Figure 4b reports W content in each sample before and after annealing. The a process did not significantly affect W content when the V2O5 target was used for W VO2 synthesis (samples A and B) because almost identical W contents were found utive to the annealing treatment. This suggests that V2O5 may act as a barrier to th diffusion of W. On the contrary, using the vanadium target, an increase in the W after thermal annealing (sample C) was observed. This means that W diffused wards the VO2 surface during annealing and suggests that the V+O2 environmen the surface diffusion of W. Such an increase in W content during annealing was cently reported by Ström et al. [48]. In their study, the W content in steel increased thermal annealing and the modified composition was attributed to a near-surfa This might also be the case in our study, with surface diffusion of W towards surface during the annealing process.
For more insight into the depth distribution of W, Figure 5a,b show the dep of the W 6+ proportion as a function of the detection angle (23 to 76°) for all the before and after thermal annealing.  Figure 4b reports W content in each sample before and after annealing. The annealing process did not significantly affect W content when the V 2 O 5 target was used for Wdoped VO 2 synthesis (samples A and B) because almost identical W contents were found consecutive to the annealing treatment. This suggests that V 2 O 5 may act as a barrier to the surface diffusion of W. On the contrary, using the vanadium target, an increase in the W content after thermal annealing (sample C) was observed. This means that W diffused more towards the VO 2 surface during annealing and suggests that the V+O 2 environment favors the surface diffusion of W. Such an increase in W content during annealing was also recently reported by Ström et al. [48]. In their study, the W content in steel increased during thermal annealing and the modified composition was attributed to a near-surface effect. This might also be the case in our study, with surface diffusion of W towards the VO 2 surface during the annealing process.
For more insight into the depth distribution of W, Figure 5a,b show the dependence of the W 6+ proportion as a function of the detection angle (23 to 76 • ) for all the samples before and after thermal annealing. cently reported by Ström et al. [48]. In their study, the W content in steel increased during thermal annealing and the modified composition was attributed to a near-surface effect. This might also be the case in our study, with surface diffusion of W towards the VO2 surface during the annealing process.
For more insight into the depth distribution of W, Figure 5a,b show the dependence of the W 6+ proportion as a function of the detection angle (23 to 76°) for all the samples before and after thermal annealing.

Similar W distribution between as-deposited and annealed films in samples A and B
showed that the inter-diffusion between VO X and W layers occurred at room temperature during deposition. In other words, the inter-diffusion process and the crystallization process occurred separately. Diffusion occurred during PLD deposition while crystallization occurred during the thermal annealing stage. Conversely, with sample C, the different W distribution between as-deposited and annealed films showed that the inter-diffusion between VO X and W layers was not completed during PLD deposition, but was completed during the thermal annealing stage. This surface diffusion leads to an increase in W content toward the surface after thermal annealing. Therefore, along the first nanometers probed by ARXPS, sample C was richer in W after thermal annealing, while samples A and B had lower W contents. In addition, W distribution was almost constant with few discrepancies irrespective of the sample (before and after annealing), meaning that the W distribution was homogeneous in all the samples.
It is generally accepted that in addition to the doping element, the valence states of V can significantly affect the thermochromic characteristics of VO 2 -based thin films. Figure 6a,b show the high-resolution spectra of V 2p-O 1s recorded in XPS angle-resolved mode at two photoelectron take-off angles (23 • and 68 • ) before and after thermal annealing for the W-rich sample (C). Deconvolution analysis of the V 2p 3/2 and V 2p 1/2 regions revealed two vanadium components in the samples, corresponding to V 4+ and V 5+ valence states: V 4+ located at 516.1 eV for 2p 3/2 , and 522.8 eV for 2p 1/2 with V 5+ centered at 517.1 eV for 2p 3/2 , and 524.5 eV for 2p 1/2 [45,46]. The binding energies of the peaks at 515.8 eV and 517.2 eV are close to those reported for VO 2 (515.7-516.2 eV) and V 2 O 5 (516.9-517.2 eV), respectively [45,49,50]. There was no significant shift in the position of the peak in terms of the binding energy of the as-deposited and annealed samples. However, no difference in the intensity of the two oxidation states V 4+ and V 5+ was observed: the intensity of the V 4+ oxidation state increased while that of V 5+ decreased after thermal annealing. The O 1s peak is deconvoluted into two peaks, one corresponding to the O-C/O-H bond, which appeared at 531.2 eV due to surface contamination of the films. The other peak was due to the O-V bond located at 529.9 eV [51].

Structural and Morphological Analysis
The structural change that occurred with W doping at room temperature was tigated using Raman analysis. Figure 7 shows the Raman spectra of the undoped and W-doped thin films a temperature. The two undoped VO2 films (black and red curves in Figure 7 [52][53][54]. No vib peak of V2O5 was observed, perhaps due to the low concentration of V2O5 resul weaker molecular vibration. Concerning the W-doped VO2 films, a decrease in the sities of the Raman active modes (low-frequency and high-frequency mode) and ening of the peaks with an increase on the percentage of W doping were observe can be explained by the fact that W doping starts to favor a more symmetric rutile ture. A similar effect has already been reported for W-doped VO2 films [23,55] an doped VO2 films [56]. Moreover, the 617(Ag) cm −1 phonon mode was slightly up and broadened as well as being reduced in intensity with W doping up to 625 cm −1 curve), indicating a distorted M1 phase and the nucleation of rutile domains at thi [57]. In the W-rich sample (1.7 at. % W), some modes started to disappear (purple indicating the beginning of the metallic rutile phase as observed in the above-men analysis of the phase transition temperature. To investigate the depth distribution of the V 5+ oxidation state of vanadium, the spectra of V 2p were measured at various detection angles ranging from 23 • to 76 • using angle-resolved X-ray photoelectron spectroscopy. Figure 6c,d shows the dependence of the concentration of the V 5+ oxidation state as a function of the detection angle for all the samples before and after thermal annealing. It can be seen from Figure 6c that the angle dependence of V 5+ oxidation state concentration is quite similar to that before thermal annealing, indicating homogeneous distribution of V 5+ on the surface of all samples. After thermal annealing (Figure 5d), the proportion of the V 5+ oxidation state increased with the increase in the detection angle in all samples. The results of angle-resolved measurements confirmed that the outer part of the W-doped VO 2 thin films was enriched in V 5+ , especially when the V 2 O 5 target was used for the synthesis of W-doped VO 2 films and hence, that the other V 4+ oxidation state is located in the inner region of the films. To summarize, sample C was poorer in V 5+ and richer in W while samples A and B were richer in V 5+ and poorer in W after thermal annealing.

Structural and Morphological Analysis
The structural change that occurred with W doping at room temperature was investigated using Raman analysis. Figure 7 shows the Raman spectra of the undoped and W-doped thin films at room temperature. The two undoped VO 2 films (black and red curves in Figure 7) had typical peaks at 143 (Ag), 192 (Ag), 222 (Ag), 262 (Bg), 306 (Ag), 338 (Ag), 389 (Ag), 441 (Bg), 498 (Ag), and 617 (Ag) cm −1 , corresponding to the monoclinic VO 2 (M1) phase [52][53][54]. No vibration peak of V 2 O 5 was observed, perhaps due to the low concentration of V 2 O 5 resulting in weaker molecular vibration. Concerning the W-doped VO 2 films, a decrease in the intensities of the Raman active modes (low-frequency and high-frequency mode) and broadening of the peaks with an increase on the percentage of W doping were observed. This can be explained by the fact that W doping starts to favor a more symmetric rutile structure. A similar effect has already been reported for W-doped VO 2 films [23,55] and Nbdoped VO 2 films [56]. Moreover, the 617 (Ag) cm −1 phonon mode was slightly upshifted and broadened as well as being reduced in intensity with W doping up to 625 cm −1 (green curve), indicating a distorted M1 phase and the nucleation of rutile domains at this stage [57]. In the W-rich sample (1.7 at. % W), some modes started to disappear (purple curve), indicating the beginning of the metallic rutile phase as observed in the above-mentioned analysis of the phase transition temperature. Changes in the surface morphology and variations in the surface roughness of the undoped and W-doped VO2 thin films were investigated by SEM and AFM, respectively. Regarding the thin films obtained using V+O2 deposition, the surface morphology of undoped film (Figure 8a) was flat with trace-like cracks whereas with W doping, the surface of the film contained holes resembling pores. The surface morphology of the films deposited from the V2O5 target was uniform, dense, and compact, the undoped film being slightly smoother than the W-doped film. The difference in morphology between undoped and W-doped VO2 films can be attributed to the general regularity observed in solid solution formation: an increase in the number of constituents can play a role in the crystallization of the films [58,59]. AFM scanning maps of the films obtained from V+O2 exhibited a slight decrease in the root mean square (RMS) from the undoped to W-doped VO2 films. The RMS of the W-doped VO2 film obtained from V2O5, was slightly higher than that of the undoped film. However, the low RMS values indicate that our VO2-based thin films are extraordinarily smooth with improved quality compared to those obtained using other synthesis methods [60,61]. Changes in the surface morphology and variations in the surface roughness of the undoped and W-doped VO 2 thin films were investigated by SEM and AFM, respectively. Regarding the thin films obtained using V+O 2 deposition, the surface morphology of undoped film (Figure 8a) was flat with trace-like cracks whereas with W doping, the surface of the film contained holes resembling pores. The surface morphology of the films deposited from the V 2 O 5 target was uniform, dense, and compact, the undoped film being slightly smoother than the W-doped film. The difference in morphology between undoped and W-doped VO 2 films can be attributed to the general regularity observed in solid solution formation: an increase in the number of constituents can play a role in the crystallization of the films [58,59]. AFM scanning maps of the films obtained from V+O 2 exhibited a slight decrease in the root mean square (RMS) from the undoped to W-doped VO 2 films. The RMS of the W-doped VO 2 film obtained from V 2 O 5 , was slightly higher than that of the undoped film. However, the low RMS values indicate that our VO 2 -based thin films are extraordinarily smooth with improved quality compared to those obtained using other synthesis methods [60,61]. solid solution formation: an increase in the number of constituents can play a role in the crystallization of the films [58,59]. AFM scanning maps of the films obtained from V+O2 exhibited a slight decrease in the root mean square (RMS) from the undoped to W-doped VO2 films. The RMS of the W-doped VO2 film obtained from V2O5, was slightly higher than that of the undoped film. However, the low RMS values indicate that our VO2-based thin films are extraordinarily smooth with improved quality compared to those obtained using other synthesis methods [60,61].

Conclusions
An innovative PLD method was developed to synthesize smooth W-doped thin films using V/W (+O 2 ) and V 2 O 5 /W multilayers with subsequent post-rapid thermal annealing. Our results revealed that the inter-diffusion between the W and VO X layers was completed during the PLD deposition process at room temperature when the V 2 O 5 target was used, whereas when the V+O 2 source was used, inter-diffusion between the W and VO X layers was completed during the thermal annealing stage. Therefore, the V 2 O 5 acted as a W diffusion barrier, while the V+O 2 environment favored surface diffusion of W during the annealing stage. These findings also show that the V 5+ is more surface sensitive and that the distribution of W on the surface of the samples is homogeneous. Furthermore, as expected, W doping leads to a sharp decrease in the VO 2 IMT by a highly efficient reduction of the phase transition temperature −26 • C per at. % W. With W-doping of 1.7 at. %, the concentration of V 5+ decreased, and the transition temperature of VO 2 dropped to room temperature (26 • C), accompanied by a high luminous transmittance (62%) while retaining a narrow hysteresis width. These very good thermochromic properties have high potential in the application of smart windows and optical devices based on thermochromic switching behavior.