Characterization of Hexagonal Tungsten Bronze Cs_xWO₃ Nanoparticles and Their Thin Films Prepared by Chemical Coprecipitation and Wet-Coating Methods

Tungsten trioxide is the typical well-known chromogenic material that can change its color from colorless transparent to deep blue electrochromically,¹ photochromically² and gasochromically³ by the reduction of W⁶⁺. The reduced tungsten oxide, WO_3-x can also absorb or reflect near-infrared (NIR) light strongly as well as visible light absorption by a polaronic absorption mechanism⁴ or the localized surface plasmon resonance.⁵ Although this NIR shielding property of WO_3-x is promising for the application to heat-insulating windows, it is limited by the low visible light transmittance. Takeda et al. reported that the hexagonal tungsten bronzes with alkaline metal ions such as Cs, Rb, Tl have much higher visible transparency and excellent NIR shielding property than WO_3-x.⁶ Among these tungsten bronzes Cs_xWO₃ has been most practically used for solar shielding films of automotive and building windows.⁷

The most cost-effective method of preparing thin films on any kind of substrate for practical applications is wet-coating method without high-temperature thermal treatment using coating solution including well-dispersed functional materials. It is essential to prepare functional nanoparticles as smaller as possible for this coating method. There have been reported only two types of methods of preparing nanoparticles of cesium tungsten bronze Cs_xWO₃: one is a solid-state synthesis of direct heating the mixtures of Cs and W precursor compounds.^6,8 The other one is hydrothermal or solvothermal synthesis of the precursors solution.⁹ Although the traditional solid state synthesis is very facile and simple, it usually gives large fused polycrystalline particles and long-time milling is necessary to break down into smaller particles and this energetic milling often changes their original properties. The hydrothermal and solvothermal methods are difficult for a large-scale synthesis because of the limited volume of container and it usually takes long synthetic time. In this study we report, for the first time as far as we know, a facile chemical coprecipitation method of preparing cesium tungsten bronze nanoparticles. The subsequent annealing effect of the prepared tungsten bronze nanoparticle was investigated. The thin films of Cs_xWO₃ were also prepared by wet-coating method using the well-dispersed Cs_xWO₃ solution and their NIR shielding properties are investigated.

The addition of nitric acid into the mixture solution of polytungstate and cesium ion gave an immediate precipitation. The synthesis of WO₃ nanoparticles by the hydrolysis of tungstate in acidic medium has been known¹⁰ but the synthesis of alkaline tungsten bronze by the acidic hydrolysis of tungstate has not been reported as far as we know. Since the theoretical maximum ratio of Cs/W is 1/3 in cesium tungsten bronze,¹¹ we used excess cesium ions by keeping the Cs/W ratio as about 1. Fig. 1 shows the XRD patterns of the prepared Cs_xWO₃ powder samples that was as precipitated and annealed in N₂ atmosphere at various temperatures. All the samples show the hexagonal crystal structure of the well-known tungsten bronze.¹² The sample as precipitated without annealing also show the weak crystallinity even though it was very poor. The crystallinity systematically increased as the annealing temperature increased. The crystallite sizes that were estimated by Scherrer equation using (200) peak were listed in Table 1.¹³ The sizes were in the range of 4−70 nm depending on the annealing temperature and systematically increased as increasing the temperature. The crystallinity and size were changed largely at the temperatures in the range of 500-800 °C. The change was not very appreciable at the temperature higher than 800 °C, which was not shown here.

Figure1.

XRD patterns of the Cs_xWO₃ samples that were annealed at the various temperatures in N₂ atmosphere.

Fig. 2 shows the morphologies of Cs_xWO₃ powders that was as precipitated without annealing and annealed at various temperatures in N₂ atmosphere. The aggregated particles have the sizes of several hundred nm in all the cases and the morphologies were not much different irrespective of annealing temperatures up to 800 °C. This means that the annealing in N₂ did not seriously affect the fusion of crystalline particles in micrometer scale but it clearly influenced the crystallinity and crystalline size in nanometer scale as shown in Fig. 1 and Table 1. The cesium contents in the samples were analyzed with EDS. The data from three different sites for each sample were averaged and listed in Table 1. The atomic ratio of Cs/W was about 0.3 in all the samples, which is very close to the maximum value in hexagonal tungsten bronze structure of Cs_xWO₃. This ratio was approximately maintained if the ratio in the precursor solution was kept higher than 0.5. This EDS data confirmed the incorporation of cesium ion into WO₃ during the acid-precipitation of the tungstate by hydrolysis.

Figure2.

FE-SEM images of Cs_xWO₃ powders that were annealed at the various temperatures.

In order to study the oxidation states of W in Cs_xWO₃, we measured XPS data for the annealed samples. Fig. 3 shows the XPS survey spectrum of Cs_xWO₃ nanoparticles annealed at 400 °C in N₂. The peaks at binding energies corresponding to Cs, W and O were clearly seen. No extra element except carbon was detected. The atomic composition from the high resolution XPS spectra of Cs_3d, W_4f and O_1s was listed in Table 2. The atomic ratio of Cs/W was constant as about 0.22 irrespective of the annealing temperatures. According to Liu et al’s study of Cs_xWO₃ that were prepared by a solvothermal method,^9d the Cs/W ratio increased in the samples annealed at the higher temperature. Their explanation was that Cs⁺ ion diffused out to the surface by the annealing. This phenomena didn’t happen for our samples that were differently prepared by the acid coprecipitation. The value of CS/W ratio was lower than that from EDS data. That is obvious as expected because XPS data comes from only surface atoms whereas EDS, which is X-ray fluorescence, can measure a bulk average property. The high resolution XPS spectra of W_4f in the Cs_xWO₃ samples that were annealed at various temperatures were shown in Fig. 4(a). As the annealing temperature increased, the signal intensity and the splitting into 4f_5/2 and 4f_7/2 were decreased. The decreased intensity was due to the decreased powder density of lager particles at the higher annealing temperature and the decreased slitting was due to the increased component of lower oxidation states. Fig. 4(b) shows the W_4f peak of the sample annealed at 400 °C that was deconvoluted into three oxidation states of W⁶⁺, W⁵⁺ and W⁴⁺. The deconvolution was not fitted well without W⁴⁺ component. The composition of W⁶⁺ and the reduced components (W⁵⁺+W⁴⁺) was listed in Table 2. As the temperature increased, the reduced tungsten ion components systematically increased as expected with the colors of the powders.

Figure3.

XPS survey spectrum of Cs_xWO₃ samples that were annealed at 400 °C in N₂.

Figure4.

(a) The high resolution XPS spectra of W_4f in Cs_xWO₃ samples that were annealed at various temperatures and (b) the deconvoluted spectra of the sample that were annealed at 400 °C in N₂.

The thin films of Cs_xWO₃ that were annealed at the various temperatures were prepared on PET substrate by a wet-coating method. The coating solution was prepared by the ball-milling of Cs_xWO₃ powders in ethanol. The solution ball-milling is one of the common methods of dispersing the aggregated particles and making homogeneous suspension or clear solution without giving a strong stress. Fig. 5 shows TEM images of Cs_xWO₃ nanoparticles, which were annealed at various temperatures, obtained from the ball-milled coating solution. After the ball-milling the aggregated particles that were shown in FE-SEM images of Fig. 2 were well dispersed and the individual crystalline particles were shown. The particle sizes systematically increased as the annealing temperature increased. The estimated sizes were in the range of 10-60 nm and pretty much matched with those estimated from XRD peaks mentioned above. The cast films prepared by a bar-coater from the coating solution without any polymer binder were very homogeneous and transparent with a thickness of about 150 nm. Fig. 6 shows FE-SEM images of Cs_xWO₃ films on PET substrate. The submicroscopic morphology of the films were nanoporous with the pores of less than 50 nm size for all the cases regardless of the annealing temperature. However, there was a slight difference for the surface roughness between the samples prepared at different annealing temperature. For the film prepared with the nanoparticles annealed at the higher temperature it had the more pores and the rougher the surface was as shown in the SEM images.

Figure5.

FE-TEM images of Cs_xWO₃ nanoparticles that were annealed at various temperatures. The samples were obtained from the ball-milled coating solutions in ethanol.

Figure6.

FE-SEM images of Cs_xWO₃ films on PET substrate, of which Cs_xWO₃ samples were annealed at various temperatures.

Fig. 7 shows the optical transmission spectra of Cs_xWO₃ thin films on PET substrate. The near-infrared transmittance of the films prepared with the annealed Cs_xWO₃ decreased as the annealing temperature increased. The decrease was almost linear until 600 °C. After 600 °C the decrease was not prominent and saturated above 800 °C. The NIR shielding effect of Cs_xWO₃ and WO_3-x is ascribed to the reduced tungsten ions whether it may be explained by a polaronic absorption mechanism or the localized surface plasmon resonance as mentioned in Introduction. Therefore our results of this NIR-shielding effect is consistent with XPS data that showed the increase of W⁵⁺ and W⁴⁺ in the samples with the increased temperatures. However, interestingly, this tendency is very different from the thin films of Cs_xWO₃ that were prepared by the solvothermal^9d, and hydrothermal^9g methods. In both their cases, the best NIR shielding property was obtained from the sample that was annealed at the optimized temperature, which was 500 °C, much lower than 800 °C. Indeed, the samples annealed at 800 °C showed the very poor NIR absorption in their results. Their explanation for this optimized temperature is that the O/W atomic ratio was minimized at 500 °C from XPS analysis and NIR-shielding was maximized due to the highest free carrier concentration for surface plasmon resonance. We doubt seriously their explanation because their XPS analysis showed that the ratio difference between 500 °C-annealed and 800 °C-annealed samples was only 0.08 and 700 °C-annealed sample, which had much better shielding property than 800 °C-sample, have the higher value than 800 °C-sample by 0.09. Considering the reliability of elemental analysis by XPS and their data fluctuation, their explanation cannot be acceptable at all.

Figure7.

Visible and NIR transmission spectra of the Cs_xWO₃ thin films on PET substrate, of which Cs_xWO₃ samples were annealed at various temperatures.

We report for the first time the synthesis of Cs_xWO₃ nanoparticles that have a hexagonal tungsten bronze structure by a chemical coprecipitation method. The content of Cs in Cs_xWO₃ was about 0.3 vs W in atomic ratio, which is very close to the theoretical maximum value, 1/3 in hexagonal tungsten bronze structure. Although the Cs_xWO₃ nanoparticles as prepared have poor crystallinity with a size of ca. 4 nm in diameter, the annealing in N₂ atmosphere increased systematically the crystallinity and crystalline size as the temperature increased up to 800 °C. The annealing increased the ratio of reduced tungsten ions in both forms of W⁵⁺ and W⁴⁺ relative to W⁶⁺. We have also prepared the Cs_xWO₃ thin films on PET substrate by a wet-coating method using the ball-milled solution of the Cs_xWO₃ nanoparticles annealed in N₂ at various temperatures. The near-infrared shielding property of these thin film systematically increased as the annealing temperature increased and it was almost saturated at 700 °C. This result was quite different from those of Cs_xWO₃ thin film that was prepared from the hydrothermal and solvothermal methods by Shi et al.^9g and Liu et al.^9d, respectively.