Size and Crystal Structure Dependence of Photochromism of Nanocrystalline WO₃ and MoO₃ Prepared by Acid-Precipitation Method

Abstract

Nanocrystallne WO₃ and MoO₃ with several different sizes and crystal structures were prepared by simple acid precipitation and subsequent heat treatment. The photochromic (PC) properties of these samples were comparatively investigated in powder state by monitoring diffuse reflectance spectral changes after bandgap irradiation. The PC effect of hexagonal WO₃ and monoclinic WO₃ strongly depended upon crystallite size rather than crystal structure. The smaller the crystallite size, the better the PC effect. However, orthorhombic WO·H₂O and MoO₃ having hexagonal and orthorhombic structures did not follow this trend. One consistent result for all WO₃ and MoO₃ samples is that the heat treatment in air, which changes crystallinity, whether it changes the crystal structure or only the crystallite size, reduces the PC effect. Since the thermal treatment reduces the surface oxygen defect sites, we believe that the PC effect of WO₃ and MoO₃ depends critically on the surface oxygen defect sites that serve as deep trap sites for photogenerated electrons and oxygen radical holes. We also found that the proton insertion claimed by double charge injection model is not critical for the PC effect.

INTRODUCTION

Photochromism is a very useful property for modulating solar radiation.¹ Two types of photochromic materials have been widely investigated: organic materials and inorganic transition metal oxides. Although organic photochromic materials exhibit relatively large modulation and fast response, their lack of durability makes them less useful in practical applications. The transition metal oxides such as WO₃,² MoO₃,³ Nb₂O₅,⁴ V₂O₅⁵ and TiO₂⁶ exhibit photochromism by band gap irradiation. Their photochromic characteristics are so widely variable depending on crystal size, structure, morphology and impurities. Among the photochromic transition metal oxides, WO₃ and MoO₃ have been studied most extensively. The photochromic mechanisms of these oxides have been described in several different ways, such as color center model,⁷ double charge injection model⁸ and small polaron model.⁹ Despite considerable research, the most important factor determining the overall photochromic (PC) properties of the oxides is still unclear. This is partly due to the lack of uniform preparation method for comparison. Here, we prepared various sizes and crystal structures of nanocrystalline WO₃ and MoO₃ through simple acid precipitation and only subsequent heat treatment with the same type of precursor. And the crystallite-size and crystal structural dependence of their photochromic properties are comparatively investigated.

EXPERIMENTAL

Materials and Preparation

Ammonium paratungstate pentahydrate (APT, (NH₄)₁₀W₁₂O·5H₂O, 99%), was purchased from Nippon Chemical Co. Ammonium heptamolybdate tetrahydrate (AHM, (NH₄)₆Mo₇O₂₄·H₂O, >98%) and ~66% HNO₃ were obtained from Dae-Jung Chemical Co. Both hexagonal nanocrystalline WO₃ and MoO₃ were prepared by an acid precipitation method using APT or AHM and HNO₃ as precursor and acid, respectively.¹⁰ For hexagonal WO (h-WO₃) 35.0 g of APT was suspended in 250 mL H₂O at 90 ℃ and 100 mL of ca. 0.03 M HNO₃ solution was added slowly and the reaction mixture was vigorously stirred for two hours. The yellowish precipitate was filtered and washed three times with a copious amount of water. The resulting precipitate was dried at 80 ℃ overnight. This sample has hexagonal crystal structure and denoted as W- 1. In order to increase the crystallinity of hexagonal WO₃ without structural transformation, W-1 was annealed at 350 ℃ in air for 3 h and the annealed sample was denoted as W-2. Since increasing crystallinity of W-1 by annealing is limited to the maximum temperature of about 370 ℃, hexagonal WO₃ with further improved crystallinity was prepared by this way: the product mixture in the W-1 sample preparation was evaporated to dryness without filtering and washing and the remaining powder was calcined at 350 ℃ in air for 3 h. The resulting powder was washed with water and dried at 80 ℃ overnight. This sample was denoted as W-3. The monoclinic WO₃ (γ-WO₃) was obtained by thermal treatment of W-1 sample at 400 ℃ in air for 3 h and denoted as W-4. For orthorhombic WO₃·H₂O (denoted as W-5), the procedure was the same as for W-1, except that the concentration of HNO₃ was about 0.1 M. Hexagonal MoO₃ (h-MoO₃) was prepared by the same way as in W-1 except that the precursor of Mo was AHM and the half amount of nitric acid was used. Orthorhombic MoO₃ (α-MoO₃) was obtained by annealing h-MoO₃ at 500 ℃ in air for 3 h. The resulting samples were labeled Mo-1 and Mo-2 for the former and the latter, respectively.

Characterization

Powder X-ray diffraction (XRD) patterns were collected by Phillips X’Pert MPD diffractometer using CuKα radiation source at a scan rate of 0.02° s^-1. Field emission electron microscopy (FE-SEM, Jeol JEM-6700F) was used to get the morphologies of the prepared WO₃ and MoO₃ powders. Diffuse reflectance spectra of the powder samples were measured by Shimadzu UV-2600 with integrating sphere attachment that has two detectors of PMT and PbSe. The photochromic color changes were performed by illuminating the powder samples with 300 W Xe-arc lamp (ILC Tech). The incident light power was measured about 300 mW/cm² by power meter (LI Model 7100).

RESULTS AND DISCUSSION

Five different samples were prepared for WO₃ and two samples were prepared for MoO₃. WO₃ is known to have various polymorphic phases such as tetragonal, orthorhombic, monoclinic, triclinic, cubic and hexagonal phases.¹¹ Among them, monoclinic WO₃, known as γ-WO₃ is the most stable phase at ambient condition. Metastable hexagonal WO₃ (h-WO₃) has received particular attention for electrochromism¹² and photochromism¹³ as well as γ-WO₃. Here we prepared 3 different phases of WO₃: hexagonal, monoclinic and orthorhombic hydrated ones for comparison. And three different crystallite sizes for h-WO₃ were prepared.

Fig. 1 shows the XRD patterns of all samples of WO₃ and MoO₃. The WO₃ samples labeled W-1 to W-3 show hexagonal patterns in good agreement with JCPDS 33-1387 data. The crystallite size was estimated by Scherrer equation using the highest intensity peak for all samples. W-1 obtained by acid precipitation alone had a crystallite size of about 5 nm. Annealing of W-1 at 350 ℃ in air increased the crystallite size to about 8 nm without structural transformation (W-2 sample). Since the crystal structure changes from hexagonal to monoclinic when annealing temperature increased higher, h -WO₃ with crystallite larger than W-2 could not be prepared only by increasing annealing temperature further. Thus, as mentioned in experimental section, the largest crystalline h-WO₃ (W-3) was obtained by thermal treatment of W-1 at 350 ℃ with the decomposable byproduct salt NH₄NO₃ in the precipitation reaction. This thermal treatment increased the crystallite size from 5 nm to 25 nm, whereas only simple annealing increased it from 5 nm to 8 nm at the same temperature. h-WO₃ with good crystallinity has been reported by hydrothermal reaction at elevated temperature,^12(b),14 however, the thermal treatment method used here for h-WO₃ has never been reported previously to our knowledge. The reason for increasing the crystallinity is not clear at this point. Monoclinic WO₃ (γ-WO₃, W-4) was obtained by thermal transformation of h-WO₃ by simple annealing at 400 ℃ in air. The crystallite size of W-4 was estimated to be ca. 29 nm. Orthorhombic hydrated WO₃ (W-5) was prepared by the same precipitation reaction as W-1, but with a much larger amount of acid. Interestingly, only the amount of acid gave a dramatic effect on the crystal structure of the product WO₃. The crystallinity of W-5 as shown in the lower part of Fig. 1(b) is much better than those of W-1 to W-4 shown in Fig. 1(a). The crystallite size of W-5 was estimated to be about 55 nm from (111) peak.

Figure 1. XRD patterns of the synthesized WO₃ and MoO₃ samples (Some major peaks are marked).

MoO₃, which has the same octahedral motif as WO₃ is known to have three main polymorphic structures: orthorhombic, monoclinic and hexagonal. Orthorhombic MoO₃ (α-MoO₃) is thermodynamically preferred and the other two phases are metastable. Hexagonal MoO₃ (h-MoO₃) labeled Mo-1 was prepared by the same acid precipitation method as h-WO₃. XRD pattern of Mo-1 was shown in the upper part of Fig. 1(b) and coincident with JCPDS 47-0871. The crystallinity of Mo-1 is very high even without annealing, as noticeable from the intensity of XRD peaks. Although the preparation method of MoO₃ is identical to that for WO₃, there are significant differences in the results. The results of this precipitation reaction are quite interesting considering that good crystalline h-MoO₃ was often obtained through hydrothermal reaction,¹⁵ because the precipitation reaction even at room temperature gave fairly good crystalline h-MoO₃, although the results were not shown here. The estimated crystallite size of Mo-1 was about 49 nm. Orthorhombic MoO₃ (Mo-2) was prepared by thermal conversion of Mo-1 at 500 ℃ in air. Partial conversion occurred at annealing temperature below 500 ℃, while complete conversion was achieved above 500 ℃. The XRD pattern of Mo-2 shown in the upper part of Fig. 1(b) shows pure orthorhombic phase coincident with JCPDS 75-0912.

Fig. 2 shows FE-SEM images of the synthesized WO₃ and MoO₃ samples, showing the particle morphology. Hexagonal WO₃ samples with different crystallite sizes do not differ greatly from each other, but appear to be composed of agglomerates composed of fine nanoparticles, and appear to be slightly larger as the particles are fused after heat treatment. Monoclinic WO₃ (W-4) prepared by annealing W-1 at 400 ℃ shows slightly larger aggregates without appreciable morphological change, too. The orthorhombic hydrated WO₃ (W-5) exhibits a platy morphology and is significantly different and larger than the other WO₃ samples, as expected from the peak intensities of the XRD pattern. Hexagonal MoO₃ sample (Mo-1) is rod-shaped with a length of 5-10 μm, much larger than all WO₃ samples. (See the difference between the scale bars in the figure) The orthorhombic MoO₃ (Mo-2) prepared by sintering the Mo-1 sample is more plate-shaped than rod-shaped and has a smaller size than Mo-1.

Figure 2. FE-SEM images of the synthesized WO₃ and MoO₃ samples (×80,000).

Diffuse reflectance spectra were collected for all samples to obtain bandgap energy information for these samples. Fig. 3 shows the UV-Vis diffuse reflectance spectra of all samples as Kubelka-Munk function (F(R)). As shown in Fig. 3(a), the absorption edges of all the WO₃ samples are different. The edge wavelength increased with increasing sample number. Since the crystallite size of WO₃ samples increased in order of sample number, the band gap energy of all samples decreased as the crystallite size increased regardless of the crystal structure. The bandgap energy (E_g) of WO₃, which is an indirect bandgap semiconductor,¹⁶ can be estimated by Tauc plot.¹⁷ From the extrapolation of the absorption edge in the plot of (αhν)^1/2 or (F(R)hν)^1/2 vs. hv (Fig. 4), the bandgap energies of all the samples were estimated and listed in Table 1 along with the unit cell structure and crystallite size. The bandgap energy values of all the WO₃ samples systematically decreased from 3.4 to 2.5 eV with increasing crystallite size, regardless of crystal structure. W-5 sample, which is orthorhombic monohydrated WO₃, had the smallest E_g as expected from its distinct yellow color. Although E_g value of bulk crystalline WO₃ is not well reported for each crystal structure, some reported E_g values were 3.3,¹⁸ 2.76¹⁹ and 2.75 eV²⁰ for hexagonal, monoclinic and orthorhombic hydrated WO₃, respectively. The values we obtained are in some agreement with these values. Bandgap energies of the MoO₃ samples were estimated as 3.2 and 3.1 eV for Mo-1 and Mo-2, respectively. After the thermal conversion of h-MoO₃ to α-MoO₃, E_g decreased as the crystallite size increased regardless of crystal structure in the same way as WO₃. The reported E_g values for h-MoO₃ and α-MoO₃ that were prepared in a method similar to ours were 3.05 and 3.18 eV, respectively.²¹ The result was in the opposite direction. The calculated Eg values for h-MoO₃ and α-MoO₃ were in line with our result.²²

Figure 3. UV-Vis diffuse reflectance spectra of the synthesized WO₃ and MoO₃ powder samples.

Figure 4. Tauc plots for (a) WO₃ and (b) MoO₃ samples.

Table 1. Structural and photochromic properties of WO₃ and MoO₃ samples

When all these WO₃ and MoO₃ samples were irradiated with UV light, their color change was visually perceptible within a minute, although it varied from sample to sample. Fig. 5 shows the pictures of some samples showing distinct color change before and after UV irradiation. As shown in pictures, W-1 sample shows the most intense visual color change from pale yellow to deep blue under UV irradiation. Fig. 6 shows the diffuse reflectance spectral changes of all powder samples of WO₃ under UV irradiation in an ambient condition. For h-WO₃ samples with different crystallite sizes (W-1~W-3), The crystallite size dependence of the photochromic properties of these samples is clearly revealed. The smaller the size, the better the color change. W-3 has the smallest E_g among the h-WO₃ samples and can absorb more light, but it appears to have the worst photochromic effect. Monoclinic WO₃ (W-4), which is larger in size and has smaller E_g than all h-WO₃ samples, showed a worse result than W-3 consistently in size dependence regardless of crystal structure. This size dependence does not appear to hold for the W-5 sample, which is the largest in crystallite size and has the smallest E_g among all WO₃, as shown in Fig. 5 and 6. W-5 differs from other WO₃ samples not only in its crystal structure but also in being hydrated. This color change of W-5 is very interesting because it has been reported that the orthorhombic WO₃·H₂O did not show photochromism.²³ He and Zhao reported the color change of orthorhombic WO₃·H₂O after irradiation of NIR laser light.²⁴ They attributed this color change to a photoinduced thermochromic effect accompanied by structural change from orthorhombic WO₃·H₂O to monoclinic WO₃. Our result may be related to those of He and Zhao. Further investigation is required. Fig. 7 shows the time dependence of relative reflectance changes at 660 nm due to the photochromic change of all the WO₃ samples. The color change of all samples except W-5 were saturated in a few min. W-5 sample showed slightly slower color change than other WO₃ samples. This fact implies that the photochromic mechanism of the W-5 sample may be different from that of other samples as mentioned earlier. The initial colorization efficiencies of all samples at 660 nm were estimated with the relative diffuse reflectance and light intensity using eq. (1).^23(a)

Figure 5. Photochromic color changes of the hexagonal (W-1), monoclinic WO₃ (W-4), orthorhombic WO₃·H₂O (W-5) and hexagonal MoO₃ (Mo1) samples. Upper and lower pictures are before and after UV irradiation for 10 min, respectively.

Figure 6. Diffuse reflectance spectral changes of all WO₃ samples (W-1 ~ W-5) after UV irradiation.

Figure 7. The time dependence of relative reflectance changes at 660 nm for all WO₃ samples after UV irradiation.

CE = ΔOD/It = log(R_o/R)/It       (1)

where CE is the photochromic colorization efficiency, ΔOD is optical density difference, I is the light intensity, t is irradiation time, R_o and R are the reflectance before and after irradiation. The initial photochromic colorization efficiencies were obtained as 48, 7.2, 5.5, 2.0 and 5.7 cm²/kJ for W-1, W-2, W- 3, W-4 and W-5, respectively. The PC response times of all WO₃ samples were roughly estimated by fitting the decay of relative reflectance to single exponential decay function and the estimated time constants (τ) and CE were together listed in Table 1 with crystallite size and bandgap energy.

The large difference in photochromic effect between W-1 and other samples cannot be explained by the difference in crystallite size alone. One difference between the W-1 sample and the other samples except W-5 is whether it was thermally treated in air or not. Thermal treatment in air is thought to not only reduce the surface area by increasing the crystallite size, but also play a role in reducing defect sites caused by the lack of oxygen atoms on the crystal surface, so the difference in photochromic effect seems to be related to this matter. Among the proposed mechanisms of photochromism of WO₃, briefly mentioned in the introduction, the double charge injection model has been frequently cited. In this model, surfaceadsorbed H₂O is oxidized by photogenerated holes in WO₃ and electron and proton simultaneously are injected into WO₃ to form hydrogen tungsten bronze (H_xWO₃). The blue color of the bronze is explained by the absorption of visible and near infrared light to cause intervalence charge transfer between W⁵⁺ and W⁶⁺ in the bronze. This model necessarily requires an oxidizable proton source like adsorbed H₂O. In order to check the role of the water adsorbed at ambient condition, W-1 sample was dried at 130 ℃, overnight and the photochromic effect was immediately rechecked. The drying at 130 ℃ overnight didn’t change any crystallinity of the sample as confirmed from XRD patterns (not shown here). Fig. 8 shows the photochromic spectral changes before and after drying. The photochromic effect after drying slightly decreased. But this is not enough to say the adsorbed H₂O is critical to the photochromism. Therefore, it is still unclear whether H⁺ intercalates into the WO₃ lattice to form the hydrogen bronze claimed by the double charge injection mechanism. However, our results clearly indicate that heat treatment, which changes crystallinity, reduces the photochromic effect, regardless of size or structure, as shown by the large difference between W-1 and W-2 to W-4.

Figure 8. Photochromic reflectance spectra changes of W-1 before and after drying.

MoO₃, like WO₃, is well known for its photochromic properties and has been described with the same photochromic mechanisms.³ Fig. 9 shows photochromic spectral changes of two MoO₃ samples with different crystal structures shown in XRD patterns. While all WO₃ samples show the similar photochromic spectral change, the spectral changes of two MoO₃ samples are quite different. Mo-1 (h-MoO₃) showed a maximum absorption at 660 nm and rapidly decreased absorption at wavelengths above 700 nm, whereas Mo-2 (α-MoO₃) showed a similar spectral change to those of the WO₃ samples, showing stronger absorption in the longer wavelength region. Fig. 10 shows the time dependence of relative reflectance change at 660 nm of two MoO₃ samples. The photochromic changes of both MoO₃ are slower than W-1 to W-4 samples but closer W-5. Since Mo-1, Mo-2 and W-5 have a similar crystallite size that are much larger than W-1 to W-4, the photochromic response time must be definitely related to the crystallite size. Mo-2 showed a lower response time than Mo-1. Since Mo-2 was thermally converted from Mo-1 and had a larger crystallite size than Mo-1 as shown in Table 1, this can be understood as the relationship between response time and crystallite size as mentioned before. The initial colorization efficiencies of Mo-1 and Mo-2 at 660 nm were estimated as 8.3 and 3.9 cm²/kJ, respectively. Despite that bandgap energy of Mo-1 is larger than that of Mo-2 and Mo-2 can absorb more light, Mo-1 showed more than twice better photochromic effect than Mo-2. Xie et al. reported previously the difference of photochromic effect, which is similar to ours, between aqueous α-MoO₃ and h-MoO₃ nanobelt suspensions prepared by hydrothermal method.^22(a) They also explained the photochromic effect of MoO₃ as a double charge injection mechanism like WO₃. The holes formed by photoexcitation oxidized water and the produced hydrogen ions entered the crystal to form hydrogen molybdenum bronze. The difference in photochromic effect between α-MoO₃ and h-MoO₃ was explained by the difference in crystalline structure: the hexagonal channel structure of h-MoO₃ provides a better pathway for hydrogen ion than zigzag channel of α-MoO₃. This explanation may be plausible for their result where the experiment was conducted in an aqueous suspension, However, since all our experiments were performed with powder samples in ambient conditions and showed that adsorbed water is not critical for the photochromic effect, there are considerable questions about the double charge injection mechanism. We therefore believe that the photochromic difference between α-MoO₃ that was thermally converted from h-MoO₃ and h-MoO₃ originates from heat treatment like the WO₃ samples. All our results consistently support that heat treatment causing crystallinity change reduces the photochromic effect. As mentioned earlier, the thermal treatment at more than 300 ℃ in air usually decreases the surface oxygen defect sites as well as changing crystal structure. Thus, the photochromic effect strongly related to surface oxygen defect sites to play a role of deep trap sites for photoexcited electrons and holes that are oxygen radicals.

Figure 9. Diffuse reflectance spectral changes of Mo-1 and Mo-2 after UV irradiation.

Figure 10. The time dependence of relative reflectance changes at 550 nm for Mo-1 and Mo-2 samples.

Reversibility in photochromism is very important not only in terms of applications but also in understanding the mechanism of photochromism. In this study we only focused on the coloring process of WO₃ and MoO₃. Their bleaching process in ambient atmosphere is so slow that comparative studies are not easy. In order to understand the photochromic mechanism of these materials more clearly, the bleaching process under various conditions will be conducted in future studies.

CONCLUSION

In summary, we have investigated comparatively the photochromic (PC) effect of nanocrystalline WO₃ and MoO₃ with various crystallite size and crystal structures prepared by simple acid precipitation and subsequent thermal treatment. The photochromic effect of h-WO₃ and γ-WO₃ is highly dependent on crystallite size rather than crystalline structure. The smaller the crystallite size, the better the PC effect.

However, the orthorhombic WO₃·H₂O and MoO with hexagonal an orthorhombic crystal structures did not follow this trend. One consistent result for all WO₃ and MoO₃ samples is that the thermal treatment in air, which changes crystallinity, whether it changes the crystal structure or only the crystallite size, reduces the PC effect. Since the thermal treatment reduces the surface oxygen defect sites, we believe that PC effect of WO₃ and MoO₃ depends critically on the surface oxygen defect sites and these sites serve as deep trap sites for photogenerated electrons and oxygen radical holes. In this case the proton insertion for charge compensation claimed by double charge injection mechanism may not be necessary.