- Email: [email protected]
tungsten-doped vanadium dioxide nanocomposite coatings at ambient temperature
Thin Solid Films 534 (2013) 231–237
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Preparation of transparent, hard thermochromic polysiloxane/tungsten-doped vanadium dioxide nanocomposite coatings at ambient temperature Yinfeng Lu, Shuxue Zhou ⁎, Guangxin Gu, Limin Wu Department of Materials Science, Advanced Coatings Research Center of Ministry of Education of China, Fudan University, Shanghai 200433, China
a r t i c l e
i n f o
Article history: Received 22 June 2012 Received in revised form 26 February 2013 Accepted 27 February 2013 Available online 13 March 2013 Keywords: Vanadium dioxide Nanocomposite coatings Polysiloxane De-agglomeration Thermochromic Mechanical properties
a b s t r a c t Polysiloxane/tungsten-doped vanadium dioxide [VO2(W)] nanocomposite coatings were prepared by de-agglomerating and modifying the self-made VO2(W) particles with 3-methacryloxypropyltrimethoxysilane in butyl acetate, then mixing MPS-functionalized VO2(W) nanoparticles with polysiloxane oligomers and curing the product at ambient temperature with the aid of 3-aminopropyltriethoxysilane. The VO2(W) particles were obtained by hydrolysis of vanadyl sulfate mingled with tungstate dopant and subsequent calcination. The structure and properties of the VO2(W) particles and nanocomposite coatings were characterized by X-ray diffraction analysis, differential scanning calorimetry, visible-near infrared spectroscopy, pendulum hardness tests, and nanoindentation. The effects of the synthesis conditions and the de-agglomeration process on the properties of the VO2(W) particles were investigated. Crystalline VO2(W) particles were obtained only with an appropriate amount of air and temperature during the calcination step and were easily reduced to nanometer size by bead-milling. The obtained nanocomposite coatings exhibited high transparency, good thermochromic performance, and ultra-high hardness (~1.0 GPa). © 2013 Elsevier B.V. All rights reserved.
1. Introduction Vanadium dioxide (VO2) has intrigued researchers due to its thermochromic properties at a relatively low transition temperature (Tc = ~68 °C) [1]. Moreover, the transition temperature can be adjusted to near room temperature through doping [2,3], demonstrating its promising application in resistive switching elements, thermal relays, opticalswitching devices, variable reflectivity mirrors, smart-window coatings, holographic recording media, and so on [4–7]. Generally, VO2 or doped VO2 is fabricated in the form of thin films by sputtering [8–11], chemical vapor deposition [12–14], pulsed laser deposition [15,16], or sol–gel processes [17–20]. However, the first three methods are gas phase-based routes that have to employ special equipment and the area of the film is limited. As for the sol–gel method, the high-temperature annealing process restricts the in situ application of this method on windows. The low luminous transmittance (usually b 50%) is another significant problem for practical application of VO2 films on windows. Up to now, many approaches have been tried to improve the luminous transmittance. Decreasing the thickness of the VO2 film may be the simplest method of enhancing the transparency, but this method invariably deteriorates the thermochromic properties of the film in the near-infrared region [21]. Doping of the film with fluorine or magnesium is another method of increasing the visible transmittance [22,23]. Also, deposition of a proper anti-reflection layer (i.e., SiO2, CeO2, or TiO2) has been frequently reported to improve the luminous transmittance ⁎ Corresponding author. Tel.: +86 21 65643417; fax: +86 21 55664033. E-mail address: [email protected] (S. Zhou). 0040-6090/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tsf.2013.02.130
[6,24–28]. Recently, Gao et al. employed poly(vinylpyrrolidone) to aid the formation of a film from a VOCl2 solution. The doped VO2 film displayed both excellent luminous transmittance and solar-modulating ability [29,30]. However, an annealing process is still necessary in their solution-based approaches, and the visible transparency at a wavelength of 550 nm is still less than 60% for most of the thermochromic films reported. VO2-containing organic coatings may be an ideal alternative for fabrication of thermochromic coatings at mild conditions. Because the filler, consisting of doped or undoped VO2 particles, is prepared ex situ, the annealing process can be eliminated in the formation of thermochromic films, which is absolutely beneficial for their in situ application on windows. In addition, it has been theoretically proven that the VO2 nanoparticles in dielectric hosts can improve the luminous transmittance relative to VO2 films [31]. So far, very little attention has been paid to the fabrication of organic coatings containing VO2 particles. Valmalette et al. [32] embedded VO2 pigment into a polymeric coating and analyzed the optical thermochromic behavior of the coating. In a previous work, we primarily prepared an acrylic resin/tungsten-doped VO2 [VO2(W)] coating with 2 wt.% of VO2(W) particles in order to demonstrate the thermochromic properties of the VO2(W) particles, but the transmittance (550 nm) of the coatings was as low as 45% and did not meet the requirements desired [33]. In this work, we employed an ambient-curable polysiloxane binder and VO2(W) particles in order to fabricate thermochromic nanocomposite coatings with high visible transmittance (~ 60% at a wavelength of 550 nm), good solar-modulation ability (23.2% at a wavelength of 2500 nm), excellent mechanical properties, and a
232
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
transition temperature near room temperature. The VO2(W) particles were synthesized by hydrolysis of vanadyl sulfate mingled with a tungstate dopant and subsequent calcination and de-agglomeration in butyl acetate (BAc) via bead-milling in the presence of 3-methacryloxypropyltrimethoxysilane (MPS). The effect of the de-agglomeration process on the properties (crystal size, particle size, and surface character) of VO2(W) particles, as well as the thermochromic and mechanical properties of the polysiloxane/VO2(W) coatings, was investigated in detail. 2. Experimental details 2.1. Materials Vanadyl sulfate hydrate (VOSO4·xH2O, 23.5 wt.% V content) was obtained from Shanghai Luyuan Fine Chemical Factory. Ammonium bicarbonate (AR), sodium tungstate dehydrate (AR), aqueous ammonia solution (28 wt.%), absolute ethanol, BAc (CR), n-octane (CR), and 3-aminopropyltriethoxysilane (APS, 98%) were all purchased from Sinopharm Chemical Reagent Co., Ltd. 3-Methacryloxypropylmethyldimethoxysilane (MPDS, 98%), MPS (98%), and methyltriethoxysilane (MTES, 98%) were obtained from Zhangjiagang Guotai-Huarong New Chemical Materials Co., Ltd. of China. 2.2. Preparation of VO2(W) crystals The crystalline VO2(W) powder was prepared as follows. A stoichiometric volume of aqueous sodium tungstate solution (0.50 M) was first dropped into an aqueous solution of VOSO4 (0.1 mol dissolved in 150 mL deionized water), followed by addition of an aqueous solution of ammonium bicarbonate (0.2 mol dissolved in 150 mL deionized water) using an automatic syringe pump over a period of 1.5 h at room temperature. The slurry-like suspension was magnetically stirred for another hour and subsequently filtered with a #4 sand funnel. The as-obtained precipitate was carefully washed with deionized water and absolute ethanol until no sulfate was detected, and then it was dried at 40 °C under vacuum for about 4 h. The dried precursor (~11.0 g) was put in ceramic boats and further moved to a quartz glass tube (87 × 1300 mm). The quartz glass tube was purged with nitrogen and then heated to 800 °C at a heating rate of 10 °C min −1 under a flow of nitrogen. Subsequently, a certain amount of air was injected into the tube within 1 min and thereafter the nitrogen flow was stopped. The temperature was held at 800 °C for another 2 h. Crystalline VO2(W) powder was obtained through thermal decomposition of the precursor. 2.3. Preparation of MPS-functionalized VO2(W) nanoparticles The MPS-functionalized VO2(W) [MPS–VO2(W)] nanoparticles were obtained as follows. VO2(W) powder (7 g) was mixed with 70 g of pre-made BAc/MPS solution. The mixture was sonicated at room temperature for about 30 min and then bead-milled for about 10 h at a rotation speed of 3000 rpm using zirconia beads with diameters of 300 μm. MPS was added into the slurry in three batches, namely, 50, 25, and 25% of the total MPS [MPS/VO2(W) = 0.2:1 mol/mol] at 0, 4, and 7 h of milling time, respectively. The temperature of the slurry was carefully controlled in the range of 20–30 °C using an ice/water bath. The as-obtained VO2(W) slurry was then centrifuged at either 8000 rpm or 14,000 rpm for 10 min to remove any large aggregates. The supernatant fluid was then precipitated with an approximately fivefold volume of n-octane, then centrifuged to get the MPS–VO2(W) nanoparticles. The obtained nanoparticles were further washed with n-octane three times to remove the residual MPS molecules and dried at 40 °C for 2 h under vacuum to evaporate the remained n-octane.
2.4. Synthesis of polysiloxane oligomers Polysiloxane oligomers were synthesized as per our previous method [34]. That is, a 50 mL round-bottom flask equipped with a condenser was put in an oil bath. Then 2.32 g (0.01 mol) of MPDS and 7.12 g (0.04 mol) of MTES were added into the flask and magnetically stirred, followed by dropwise addition of a diluted aqueous ammonia (0.1 mol/kg, 0.9 g)/ethanol (1 g) solution over 15 min at room temperature. Afterwards, the reaction mixture was heated to about 90 °C and refluxed for 4 h to obtain the polysiloxane oligomers. 2.5. Preparation of polysiloxane/VO2(W) nanocomposite coatings The MPS–VO2(W) nanoparticles were re-dispersed in BAc based on a solid content of 10% and then mixed with the as-obtained polysiloxane oligomers (theoretical solid content: 83.25%). The mass ratio of the MPS–VO2(W)/BAc dispersion to the as-obtained polysiloxane oligomer was set as 5.7/50, 14.5/50, 23.6/50, 29.8/50, 45.9/50, 62.9/50, and 141.6/50, corresponding to the VO2(W) content of 1.0, 2.5, 4.0, 5.0, 7.5, 10.0, and 20.0 wt.% in the dried films. The mixture was then sonicated for about 20 min and evaporated under vacuum at room temperature (or further charged with a certain amount of BAc) to obtain coatings with a solid content of about 50%. After 30 wt.% of APS was added (based on the weight of the as-obtained polysiloxane oligomer solution), the coating was cast on glass slides using a #8 wire-wound applicator (about 80 μm in wet film thickness) and dried at room temperature for one week. The samples were then used for further optical and mechanical measurements. 2.6. Characterization Fourier transform infrared (FTIR) spectra were obtained using a Nicolet Nexus 470 spectrometer (ThermoFisher, USA) in the wavenumber range of 400–4000 cm −1 with a resolution of 4 cm −1 and accumulation of 32 scans. Liquid MPS solution was dropped onto the surface of NaCl plate and quickly dried by infrared irradiation for direct characterization. The dried nanoparticles were blended with KBr to form sample flakes. X-ray powder diffraction (XRD) measurements were carried out at room temperature using a Bruker Nanostar U System, with incident X-ray wavelength (λ = 0.1542 nm). The collimation system consisted of two cross-coupled Gobel mirrors and four pinholes. A Hi-Star two-dimensional (2-D) area detector (Siemens) filled with pressurized xenon gas was used to record the one-dimensional (1-D) XRD patterns at a voltage of 40 kV and a current of 40 mA. Transmission electron microscopy (TEM) was performed on an H600 transmission electron microscope (Hitachi Corp., Japan) at an accelerated voltage of 100 kV. The as-prepared VO2 dispersion was diluted with BAc and directly dried on copper grids. The particle size and distribution of the MPS–VO2(W)/BAc dispersions were detected by a Nano ZS90 particle-size analyzer (Malvern Instruments Co., Ltd., U.K.). Thermogravimetric analysis (TGA) was carried out using a PerkinElmer TGA-7 instrument (USA) from ambient temperature to 750 °C at a heating rate of 20 °C/min under ultrapure nitrogen atmosphere. Differential scanning calorimetry (DSC) was performed on a PerkinElmer Pyris 1 system with a heating rate of 10 °C/min and ultrapure nitrogen protection. Thermochromic switching characteristics were monitored on a Hitachi U-4100 UV–visible-near IR (NIR) spectrophotometer equipped with a film-heating unit. The transmittance spectra at normal incidence were recorded in the wavelength range of 400–2600 nm at a scanning rate of 600 nm/min. Hysteresis loops were measured by collecting the transmittance of the films at a fixed wavelength (1550 nm) at approximate intervals of 2.0 °C. The diffuse reflectance at normal incidence
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
and the specular reflectance at a 10° incidence angle were determined using a sheet of BaSO4 powder as the standard. The temperature was measured using a thermocouple in contact with the films and controlled through a temperature-controlling unit. The pendulum hardness (Koenig hardness) of the nanocomposite coatings was measured with a 707 KP pendulum hardness rocker (Sheen Instrument Ltd., UK). The time of the Koenig pendulum swinging from 6° to 3° was automatically recorded. The average value from three measurements at different sites on the same sample was adopted. The micro-mechanical performance was determined using a nanoindentation tester (CSM Instruments, Switzerland) with a Berkovich diamond indenter. After contact with the surface of sample, the indenter was pressed into the coatings at a constant strain rate of 0.05 s−1 until 2000 nm of depth was reached, held at the maximum load for 50 s, and then withdrawn from the surface at the same rate as loading. About 5 indents were performed on each sample and the averaged values were finally adopted. The hardness and elastic modulus were calculated by the Oliver–Pharr method [35]. 3. Results and discussion 3.1. Preparation of crystalline VO2(W) powder Preparation of the crystalline VO2 and VO2(W) powder was demonstrated through thermal decomposition of the hydrolytic product of VOSO4 [33]. In this work, it was found that introduction of a small amount of oxygen into the system during the calcination stage was necessary to get VO2 (W). Fig. 1 shows the XRD patterns of the vanadium oxide prepared at various calcination atmospheres. V2O3 (JCPDS 65-9474) rather than VO2 was obtained when air was not charged into the nitrogen flow. A typical polycrystalline diffraction pattern of monoclinic P21/c VO2 (JCPDS 43-1051) was exhibited when 700 mL air was added into the quartz glass tube, indicating the formation of crystalline VO2(W). Nevertheless, the reflection angle of (110) of VO2(W) was slightly smaller than that of pure VO2 and the adjacent interplanar distance was just the opposite, owing to the larger size of the W atom relative to the V atom. However, when either an excess (1000 mL) or an insufficient amount (100 mL) of air was added, a vanadium oxide mixture with various vanadium valence was obtained, as seen from XRD patterns b and d. Therefore, it is crucial to control the amount of the air introduced during the calcination step. Fig. 2 displays the typical DSC curves of the VO2(W) with various W fractions and calcination temperatures. It can be seen that the critical phase transition temperature (Tc) of the undoped VO2 was 67 °C
233
(the peak temperature), very close to the 68 °C reported by Morin [1]. When the W atom was doped, the Tc decreased to 36 and −13 °C at 1.5 and 3.5 atm.% of W, respectively. A low and wide exothermic peak is revealed at a calcination temperature of 600 °C, and the peak temperature shifts to high temperature, suggesting a weakening of the phase transition of VO2(W). This may be attributed to the incomplete and nonhomogeneous doping of W and weak crystallization of VO2(W) particles. 3.2. De-agglomeration and modification of VO2(W) particles Because the as-synthesized VO2(W) particles consist of aggregates owing to the calcination step at 800 °C, de-agglomeration of VO2(W) is necessary for its application. It has been demonstrated that MPS is an efficient surface modifier for the de-agglomeration of TiO2 nanoparticles in BAc [36]. Thus, MPS was herein adopted for the de-agglomeration and modification of VO2(W) particles. Fig. 3 shows the mean particle size and distribution of VO2(W) after the VO2(W) powder was bead-milled in BAc for different lengths of time in the presence of MPS. The mean particle size of VO2(W) was about 792 nm just after sonication for 30 min. After 1 h of bead-milling, the particle-size distribution shifted to a smaller size and a new small peak with an average diameter of 105 nm appeared, indicating the breakup of the agglomerates. VO2(W) dispersion with a mean particle size of 200 nm was finally obtained after 10 h of bead-milling. TEM images of the corresponding VO2(W) dispersions are also presented in Fig. 3. The as-obtained VO2(W) particles were significantly agglomerated, with particle sizes at the micrometer level, and were effectively broken up after bead-milling, being well in agreement with the size distribution. To remove the large aggregates, the as-obtained MPS–VO2(W)/BAc dispersion was further centrifuged at 8000 rpm for 10 min. The supernatant, with a solid content of about 4 wt.%, was collected. Its particle size distribution and TEM images are also given in Fig. 3. It can be seen that the MPS–VO2(W) particles in the supernatant actually belong to nanoparticles with a mean particle size of 47 nm. These MPS–VO2(W) nanoparticles were used in the preparation of transparent polysiloxane/ VO2(W) nanocomposite coatings. Fig. 4 compares the XRD patterns of VO2(W) particles before and after de-agglomeration and centrifugation. The intensity of the XRD patterns obviously declined after bead-milling, indicating the reduced crystallinity and crystal size of the VO2(W) particles. Theoretical calculations based on the Scherrer equation applied to the (110) reflection indicates that the crystal size of the VO2(W) particles (given in Fig. 4) changed from 78.9 to 14.2 nm after bead-milling. Moreover,
0.0 ∗ V3O5
V6O13
∗
-0.2
a ∗
∗ ∗ ∗
b
V2O3
Heat Flow Endo Up (W/g)
Intensity (counts/s)
♣
V2O3+V3O5
∗
011 200
c
210
220
VO2(W)
♣♣ ♣♣
10
20
♣
♣ ♣
30
40
V2O5+V6O13
♣d
50
60
70
80
2θ (degree) Fig. 1. XRD patterns of thermal decomposition products calcined at 800 °C for 2 h with different amounts of air added into the quartz glass tube: (a) 0, (b) 100, (c) 700, and (d) 1000 mL. (Molar ratio of W/V: 0.015:1.)
d
-0.6
o
b
-0.4
44 C
c
o
-13 C
o
36 C
-0.8
a
-1.0 -1.2 -1.4 -20
o
67 C 0
20
40
60
80
100
Temperature (oC) Fig. 2. DSC curves of pure VO2 particles (calcination temperature: 800 °C) (a) and VO2(W) particles synthesized with different tungsten-doping fractions and calcination temperatures: (b) 3.5%, 800 °C, (c) 1.5%, 800 °C, and (d) 1.5%, 600 °C.
234
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
Fig. 3. The mean particle size and distribution of VO2(W) particles at various lengths of milling time: (a) 0 h with sonication for about 30 min, (b) 1 h, (c) 10 h, and (d) after centrifugation at 8000 rpm for 10 min. (The values marked on the peak represent the average particle size and intensity fraction of the corresponding peak. The insert pictures are the corresponding TEM images of original VO2(W) powder and VO2(W) dispersion.)
a further decrease in crystal size was revealed after centrifugation, which should be attributed to the removed of the VO2(W) particles with large crystal sizes. Therefore, in addition to the de-agglomeration of the as-synthesized VO2(W) particles, bead-milling can crack VO2(W) crystals. The surface properties of MPS-modified VO2(W) nanoparticles were characterized by FTIR, as shown in Fig. 5. The absorption peaks at 2840–3000, 1720, and 1637 cm −1, corresponding to the stretching vibrations of C\H, C_O, and C_C, respectively, appear in the FTIR spectrum of MPS–VO2(W), clearly demonstrating the attachment of MPS to VO2(W) nanoparticles. Furthermore, an absorption peak at 1122 cm −1 due to the stretching vibration of Si\O\V appears. Additionally, we found that the modified VO2(W) nanoparticles became hydrophobic after modification, while the original VO2(W) particles were hydrophilic. Both these facts demonstrate that MPS has been grafted onto the surface of VO2(W) particles.
The MPS-modified VO2(W) (nano)particles were quantitatively characterized by TGA, as displayed in Fig. 6. Clearly, MPS-functionalized VO2(W) particles have an increased weight loss relative to the assynthesized VO2(W) particles, which is attributed to the organic components of the attached MPS. The modified VO2(W) nanoparticles from the supernatant have a much higher weight loss than those from the dispersion without centrifugation. It is rational to assume that those VO2(W) nanoparticles in the supernatant have a higher specific surface area and thus higher amount of MPS attached, in accordance with the surface modification of the TiO2 nanopowder with MPS [36]. 3.3. Preparation of polysiloxane/VO2(W) nanocomposite coatings An ambient curable polysiloxane oligomer was adopted as the polymer binder for the preparation of the nanocomposite coatings, because the oligomer can be cured in the presence of APS to form
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
100
10000 7500
78.9 nm
a Remaining weight (%)
5000 2500 3500
Intensity (counts/s)
3000
235
b
14.2 nm
2500 2000 1750
11.5 nm
As-synthesized VO2 a
90
b
85
b,c: MPS modified VO2
80
c
75
c
1500
95
100
200
300
1250 1000 1250
10.7 nm
500
600
700
Fig. 6. TGA curves of (a) as-synthesized VO2(W), (b) modified VO2(W) from dispersion after 10 h bead-milling, and (c) modified VO2(W) from the supernatant fluid obtained by 10 h bead-milling and centrifugation at 8000 rpm for 10 min.
d
1000 750 500
10
400
Temperature (oC)
20
30
40
50
60
70
80
2θ (degree) Fig. 4. XRD patterns of W-doped VO2 from (a) the original sample, (b) dispersion after 10 h of bead-milling, and dispersions centrifuged at (c) 8000 rpm and (d) 14,000 rpm for 10 min each. (The data correspond to the grain size calculated based on the Scherrer equation.)
result is also proven by the decline of the absorption at 2750–3140 cm−1 due to C\H. Moreover, a new peak at 1570 cm − 1, assigned to the bending vibration of \NH2, occurs in spectrum b. Surprisingly, it weakens when the MPS–VO2 nanoparticles are embedded. The band at 3250–3400 cm−1 due to stretching vibration is also quite different from that of pure polysiloxane coatings. These peak changes may mean the occurrence of an aza-Michael reaction between the \NH2 group and the C_C group. 3.4. Thermochromic property of nanocomposite coatings
c Transmittance (%)
Fig.8 displays the transmittance spectra and corresponding photographs of the ambient-cured polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents. The transmittance of the nanocomposite film at high temperature increases as the wavelength increases at wavelengths above 1500 nm. This trend is the converse to that of the conventional VO2 or doped VO2 film. In any case, the results are very similar to those theoretically predicted by Li et al. [31]. Solar modulation of 4.3, 8.8, and 23.2% at the wavelength of 2500 nm, and luminous transmittance of 89.3%, 77.2%, and 59.4% at the wavelength of 550 nm, are achieved for the films containing 1, 4, and 10 wt.% of VO2(W), respectively. The difference of transparency
a Transmittance (%)
thick polysiloxane coatings with good mechanical performance [34]. The MPS–VO2(W) nanoparticles obtained from the supernatant were redispersed in BAc to get a MPS–VO2(W)/BAc dispersion and then blended with the polysiloxane oligomer. After being cured at room temperature for about one week, the nanocomposite coatings were employed for structure and performance characterization. Fig. 7 shows the FTIR spectra of the polysiloxane oligomers, pure polysiloxane coatings, and polysiloxane/VO2(W) nanocomposite coatings. The peaks at 1720 and 1640 cm−1 are attributed to the stretching vibrations of C_O and C_C, respectively. A broad absorption band at 3140–3720 cm−1 in the spectrum due to the stretching vibration of \OH indicates the hydrolysis of MPDS/MTES in the pre-hydrolysis step. However, the intensity of the absorption band significantly decreases and even disappears in spectra b and c, demonstrating the further condensation of the silanol groups of polysiloxane oligomers. This
b
c
a
4000
4000
b
3500
3000
2500
2000
1500
1000
500
-1
Wavelength (cm ) Fig. 5. FTIR spectra of (a) MPS, (b) as-synthesized VO2(W), and (c) MPS-modified VO2(W) nanoparticles from the supernatant.
3500
3000
2500
2000
1500
1000
Wavelength (cm-1) Fig. 7. FTIR spectra of (a) polysiloxane oligomer, (b) polysiloxane coating, and (c) polysiloxane/VO2(W) nanocomposite coating. (The ratios of polysiloxane oligomers to APS in b and c are the same and the VO2(W) content is 20 wt.% based on the dried polysiloxane coating.)
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
can also clearly be distinguished from their photographs, as shown in Fig. 8b, namely, higher VO2(W) content results in better solar modulation and low visible transmittance. Therefore, a balance has to be considered between the luminous transmittance and the solar modulation ability. In comparison with literature data reported, our nanocomposite film has higher transparency than previous VO2 films [18,37], doped VO2 films [14,23,38], or even multilayer thermochromic films [25,39]. However, the solar modulation of the nanocomposite film is lower relative to the literature data. This is thought to be caused by the doping of the W atom [40] and the reduced crystallinity of VO2(W) nanoparticles by bead-milling. If VO2(W) particles with better dispersibility and lower primary particle size are adopted, we are convinced that the solar modulation of the nanocomposite film would be greatly improved. Fig. 9 displays the hysteresis loop for the nanocomposite film with 10 wt.% VO2(W) content. As seen from the figure, the film exhibits a temperature centered at about 40 °C and a width of 13 °C. The centered temperature is close to the metallic/semiconductor transition temperature of VO2(W) determined by DSC. Fig. 10 presents the reflectance spectra of the nanocomposite film. Stronger specular reflectance at 60 °C can be clearly seen relative to that at room temperature in the near-infrared range. However, the specular reflectance does not change significantly with temperatures in the visible range. This result is the same as the reflectance spectra of W-doped VO2 film [37]. The diffuse reflectance, being due to the scattering of light by large VO2(W) particles in the films, is also exhibited in Fig. 10b. However, it is less impacted by the temperature and thus does not contribute to the solar modulation of the nanocomposite film.
(a) 1% Transmittance (%)
4% 80
10% 60
0.70
0.65
Transmittance (%)
236
heating 0.60
cooling 0.55
0.50
0.45 10
20
30
40
50
60
70
Temperature (oC) Fig. 9. Hysteresis loop of the nanocomposite film with 10 wt.% VO2(W) sample at a wavelength of 1550 nm.
3.5. Mechanical properties of the nanocomposite coatings Fig. 11 shows the effect of MPS–VO2(W) content on the pendulum hardness (Koenig hardness) of the polysiloxane/VO2(W) nanocomposite films. Clearly, the pendulum hardness significantly increases with increasing VO2(W) content at the VO2(W) content less than 10 wt.%, which should be due to the reinforced role of inorganic nanoparticles and the cross-linking role of MPS–VO2(W) nanoparticles via an aza-Michael reaction. However, when the MPS–VO2(W) content is greater than 10%, a slight reduction of the pendulum hardness can be observed. This may result from the high quantity of free MPS segments present, because these segments attached to nanoparticles act as a flexible component for the nanocomposites [41,42]. The loading–hold–unloading curves of polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents are shown in Fig. 12. Upon loading, the force increases with increasing depth and creep behavior is observed for all coatings when the loading remains at the maximum during a period of holding time. However, the deformation almost completely recovers after unloading all the coatings, indicating the high elasticity of the polysiloxane/VO2(W) nanocomposite coatings. The microhardness and elastic modulus increase with increasing VO2(W) content at VO2(W) content less than 10%. However, a slight decrease of the microhardness and elastic
40 1000
1500
2000
2500
Wavelength (nm)
Reflectance, %
(b)
Reflectance, %
500
12 11 10 9 8 7 6 5 4 7 6 5 4 3 2 1
a
o
20
b
o
20
o
60
500 Fig. 8. (a) Vis-NIR transmittance spectra and (b) photographs of polysiloxane/VO2(W) nanocomposite coatings with 1, 4, and 10 wt.% VO2(W) content. (Molar ratio of W: V = 0.15:100; film thickness: ~20 μm, determined by a digital microscope; The solid and dashed lines in (a) are for the films at 20 and 60 °C, respectively.).
o
60
1000
1500
2000
2500
Wavelength, nm Fig. 10. (a) Specular and (b) diffuse reflectance spectra of polysiloxane/VO2(W) nanocomposite coatings with 10 wt.% VO2(W) content.
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
nanocomposite coatings. Moreover, the higher the VO2(W) content in the coating, the more obvious the thermochromic phenomenon is. Luminous transmittance (550 nm) of about 60% and solar modulation (2500 nm) of 23% was achieved at 10% VO2(W) load. In addition, the polysiloxane/VO2(W) nanocomposite coatings possess ultrahigh hardness. Thus, transparence, thermochromic properties at near room temperature, and ultrahigh hardness certainly favor their practical application on windows.
Pendulum hardness (s)
175 170 165 160 155 150
Acknowledgements
145 140 135 0
5
10
15
20
VO2(W) content (wt%) Fig. 11. Pendulum hardness of polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents.
modulus were observed when the VO2(W) content increased to 20%. This trend is completely consistent with the trend of pendulum hardness as discussed above. Among these nanocomposite coatings, the microhardness (990 MPa) of the coatings with 10 wt.% of VO2(W) content is close to that of the sol–gel-derived silica coatings (1.0 GPa) [43], suggesting that the thermochromic coating has an ultrahigh hardness. It is thus thought that these hard polysiloxane/VO2(W) nanocomposite coatings are promising for in situ application on windows. 4. Conclusions Transparent polysiloxane/VO2(W) nanocomposite coatings with excellent thermochromic properties were prepared by mixing a VO2(W) dispersion with a polysiloxane oligomer and ambient-curing with APS as a curing agent. The VO2(W) particles were synthesized by alkaline hydrolysis of VOSO4 mingled with a certain amount of Na2WO4 as dopant and subsequent calcination in an oxidative atmosphere. A certain amount of air is crucial to guarantee the valence of V. The as-obtained VO2(W) particles can be efficiently de-agglomerated in BAc by a bead-milling process in the presence of MPS. Bead-milling can crack the VO2(W) crystals. Nevertheless, the cracked VO2(W) crystals still exhibit good thermochromic properties. Removing partial large VO2(W) aggregates by centrifugation can produce a MPS–VO2(W) dispersion dispersed at the nano-size level, resulting in high transparency of the H(MPa)
20
Load (mN)
237
0% 2.5% 10% 20%
15
473 647 990 780
13 14 53 93
E(GPa)
1.26 1.77 2.74 2.35
0.06 0.05 0.02 0.10
10
5
0 0
500
1000
1500
2000
Depth (nm) Fig. 12. The loading–hold–unloading curves of the polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents.
This work was supported by the Foundation of Science and Technology of Shanghai (09DJ1400205) and the innovative team of Ministry of Education of China (IRT0911). Prof. Q. Z. Zhu from Shanghai University of Electric Power is also appreciated for the supply of a temperature controller for thermochromic tests. References [1] F.J. Morin, Phys. Rev. Lett. 3 (1959) 34. [2] J.B. Goodenough, J. Solid State Chem. 3 (1971) 490. [3] C. Batista, R. Ribeiro, J. Carneiro, V. Teixeira, J. Nanosci, Nanotechnol. 9 (2009) 4220. [4] M. Gurvitch, S. Luryi, A. Polyakov, M. Dudley, G. Wang, S. Ge, et al., J. Appl. Phys. 102 (2007) 033504. [5] S.H. Chen, H. Ma, X.J. Yi, H.C. Wang, X. Tao, M.X. Chen, et al., Infrared Phys. Technol. 45 (2004) 239. [6] M.H. Lee, Sol. Energy Mater. Sol. Cells 71 (2002) 537. [7] I. Balberg, S. Trokman, J. Appl. Phys. 46 (1975) 2111. [8] F. Guinneton, L. Sauques, J.C. Valmalette, F. Cros, J.R. Gavarri, J. Phys. Chem. Solids 62 (2001) 1229. [9] P. Jin, S. Tanemura, Jpn. J. Appl. Phys. 33 (1994) 1478, Part 1. [10] G.V. Jorgenson, J.C. Lee, Sol. Energy Mater. 14 (1986) 205. [11] N.Y. Shishkin, A.A. Komarov, D.V. Kosov, V.A. Cherkasov, L.A. Bashkirov, U. Bardi, et al., Sensors Actuators B Chem. 108 (2005) 113. [12] T.D. Manning, I.P. Parkin, R.J.H. Clark, D. Sheel, M.E. Pemble, D. Vernadou, J. Mater. Chem. 12 (2002) 2936. [13] T.D. Manning, I.P. Parkin, M.E. Pemble, D. Sheel, D. Vernardou, Chem. Mater. 16 (2004) 744. [14] R. Binions, G. Hyett, C. Piccirillo, I.P. Parkin, J. Mater. Chem. 17 (2007) 4652. [15] T.H. Yang, S. Nori, H.H. Zhou, J. Narayan, Appl. Phys. Lett. 95 (2009) 102506. [16] D.H. Kim, H.S. Kwok, Appl. Phys. Lett. 65 (1994) 3188. [17] K.R. Speck, H.S.-W. Hu, M.E. Sherwin, R.S. Potember, Thin Solid Films 165 (1988) 317. [18] S.W. Lu, L.S. Hou, F.X. Gan, Adv. Mater. 9 (1997) 244. [19] T.J. Hanlon, J.A. Coath, M.A. Richardson, Thin Solid Films 436 (2003) 269. [20] J.H. Cho, Y.J. Byun, J.H. Kim, Y.J. Lee, Y.H. Jeong, M.P. Chun, et al., Ceram. Int. 38 (2012) S589. [21] G. Xu, P. Jin, M. Tazawa, K. Yoshimura, Jpn. J. Appl. Phys. 43 (2004) 186. [22] C.G. Granqvist, Thin Solid Films 193–194 (1990) 730. [23] N.R. Mlyuka, G.A. Niklasson, C.G. Granqvist, Appl. Phys. Lett. 95 (2009) 171909. [24] P. Jin, G. Xu, M. Tazawa, K. Yoshimura, Jpn. J. Appl. Phys. 41 (2002) 278. [25] P. Jin, G. Xu, M. Tazawa, K. Yoshimura, Appl. Phys. A 77 (2003) 455. [26] S. Saitzek, F. Guinneton, L. Sauques, K. Aguir, J.R. Gavarri, Opt. Mater. 30 (2007) 407. [27] N.R. Mlyuka, G.A. Niklasson, C.G. Granqvist, Sol. Energy Mater. Sol. Cells 93 (2009) 1685. [28] Z. Chen, Y.F. Gao, L.T. Kang, J. Du, Z.T. Zhang, H.J. Luo, et al., Sol. Energy Mater. Sol. Cells 95 (2011) 2677. [29] L.T. Kang, Y.F. Gao, H.J. Luo, ACS Appl. Mater. Interfaces 10 (2009) 2211. [30] L.T. Kang, Y.F. Gao, H.J. Luo, Z. Chen, J. Du, Z.T. Zhang, ACS Appl. Mater. Interfaces 3 (2011) 135. [31] S.Y. Li, G.A. Niklasson, C.G. Granqvist, J. Appl. Phys. 108 (2010) 063525. [32] J.C. Valmalette, J.R. Gavarri, Sol. Energy Mater. Sol. Cells 33 (1994) 135. [33] J.Q. Shi, S.X. Zhou, B. You, L.M. Wu, Sol. Energy Mater. Sol. Cells 91 (2007) 1856. [34] X.B. Chen, S.X. Zhou, B. You, L.M. Wu, J. Sol–Gel Sci. Technol. 58 (2011) 490. [35] G.M. Pharr, W.C. Oliver, J. Mater. Res. 7 (1992) 1564. [36] Y.F. Lu, S.X. Zhou, L.M. Wu, J. Disper. Sci. Technol. 33 (2012) 497. [37] C. Piccirillo, R. Binions, I.P. Parkin, Thin Solid Films 516 (2008) 1992. [38] C.S. Blackman, C. Piccirillo, R. Binions, I.P. Parkin, Thin Solid Films 517 (2009) 4565. [39] Z.T. Zhang, Y.F. Gao, Z. Chen, J. Du, C.X. Cao, L.T. Kang, et al., Langmuir 26 (2010) 10738. [40] M.A. Sobhan, R.T. Kivaisi, B. Stjerna, C.G. Granqvist, Sol. Energy Mater. Sol. Cells 44 (1996) 451. [41] S.X. Zhou, L.M. Wu, Macromol. Chem. Phys. 209 (2008) 1170. [42] G.D. Chen, S.X. Zhou, G.X. Gu, H.H. Yang, L.M. Wu, J. Colloid. Interf. Sci. 281 (2005) 339. [43] A. Sellinger, P.M. Weiss, A. Nguyen, Y.F. Lu, R.A. Assink, W.L. Gong, et al., Nature 394 (1998) 256.
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Preparation of transparent, hard thermochromic polysiloxane/tungsten-doped vanadium dioxide nanocomposite coatings at ambient temperature Yinfeng Lu, Shuxue Zhou ⁎, Guangxin Gu, Limin Wu Department of Materials Science, Advanced Coatings Research Center of Ministry of Education of China, Fudan University, Shanghai 200433, China
a r t i c l e
i n f o
Article history: Received 22 June 2012 Received in revised form 26 February 2013 Accepted 27 February 2013 Available online 13 March 2013 Keywords: Vanadium dioxide Nanocomposite coatings Polysiloxane De-agglomeration Thermochromic Mechanical properties
a b s t r a c t Polysiloxane/tungsten-doped vanadium dioxide [VO2(W)] nanocomposite coatings were prepared by de-agglomerating and modifying the self-made VO2(W) particles with 3-methacryloxypropyltrimethoxysilane in butyl acetate, then mixing MPS-functionalized VO2(W) nanoparticles with polysiloxane oligomers and curing the product at ambient temperature with the aid of 3-aminopropyltriethoxysilane. The VO2(W) particles were obtained by hydrolysis of vanadyl sulfate mingled with tungstate dopant and subsequent calcination. The structure and properties of the VO2(W) particles and nanocomposite coatings were characterized by X-ray diffraction analysis, differential scanning calorimetry, visible-near infrared spectroscopy, pendulum hardness tests, and nanoindentation. The effects of the synthesis conditions and the de-agglomeration process on the properties of the VO2(W) particles were investigated. Crystalline VO2(W) particles were obtained only with an appropriate amount of air and temperature during the calcination step and were easily reduced to nanometer size by bead-milling. The obtained nanocomposite coatings exhibited high transparency, good thermochromic performance, and ultra-high hardness (~1.0 GPa). © 2013 Elsevier B.V. All rights reserved.
1. Introduction Vanadium dioxide (VO2) has intrigued researchers due to its thermochromic properties at a relatively low transition temperature (Tc = ~68 °C) [1]. Moreover, the transition temperature can be adjusted to near room temperature through doping [2,3], demonstrating its promising application in resistive switching elements, thermal relays, opticalswitching devices, variable reflectivity mirrors, smart-window coatings, holographic recording media, and so on [4–7]. Generally, VO2 or doped VO2 is fabricated in the form of thin films by sputtering [8–11], chemical vapor deposition [12–14], pulsed laser deposition [15,16], or sol–gel processes [17–20]. However, the first three methods are gas phase-based routes that have to employ special equipment and the area of the film is limited. As for the sol–gel method, the high-temperature annealing process restricts the in situ application of this method on windows. The low luminous transmittance (usually b 50%) is another significant problem for practical application of VO2 films on windows. Up to now, many approaches have been tried to improve the luminous transmittance. Decreasing the thickness of the VO2 film may be the simplest method of enhancing the transparency, but this method invariably deteriorates the thermochromic properties of the film in the near-infrared region [21]. Doping of the film with fluorine or magnesium is another method of increasing the visible transmittance [22,23]. Also, deposition of a proper anti-reflection layer (i.e., SiO2, CeO2, or TiO2) has been frequently reported to improve the luminous transmittance ⁎ Corresponding author. Tel.: +86 21 65643417; fax: +86 21 55664033. E-mail address: [email protected] (S. Zhou). 0040-6090/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tsf.2013.02.130
[6,24–28]. Recently, Gao et al. employed poly(vinylpyrrolidone) to aid the formation of a film from a VOCl2 solution. The doped VO2 film displayed both excellent luminous transmittance and solar-modulating ability [29,30]. However, an annealing process is still necessary in their solution-based approaches, and the visible transparency at a wavelength of 550 nm is still less than 60% for most of the thermochromic films reported. VO2-containing organic coatings may be an ideal alternative for fabrication of thermochromic coatings at mild conditions. Because the filler, consisting of doped or undoped VO2 particles, is prepared ex situ, the annealing process can be eliminated in the formation of thermochromic films, which is absolutely beneficial for their in situ application on windows. In addition, it has been theoretically proven that the VO2 nanoparticles in dielectric hosts can improve the luminous transmittance relative to VO2 films [31]. So far, very little attention has been paid to the fabrication of organic coatings containing VO2 particles. Valmalette et al. [32] embedded VO2 pigment into a polymeric coating and analyzed the optical thermochromic behavior of the coating. In a previous work, we primarily prepared an acrylic resin/tungsten-doped VO2 [VO2(W)] coating with 2 wt.% of VO2(W) particles in order to demonstrate the thermochromic properties of the VO2(W) particles, but the transmittance (550 nm) of the coatings was as low as 45% and did not meet the requirements desired [33]. In this work, we employed an ambient-curable polysiloxane binder and VO2(W) particles in order to fabricate thermochromic nanocomposite coatings with high visible transmittance (~ 60% at a wavelength of 550 nm), good solar-modulation ability (23.2% at a wavelength of 2500 nm), excellent mechanical properties, and a
232
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
transition temperature near room temperature. The VO2(W) particles were synthesized by hydrolysis of vanadyl sulfate mingled with a tungstate dopant and subsequent calcination and de-agglomeration in butyl acetate (BAc) via bead-milling in the presence of 3-methacryloxypropyltrimethoxysilane (MPS). The effect of the de-agglomeration process on the properties (crystal size, particle size, and surface character) of VO2(W) particles, as well as the thermochromic and mechanical properties of the polysiloxane/VO2(W) coatings, was investigated in detail. 2. Experimental details 2.1. Materials Vanadyl sulfate hydrate (VOSO4·xH2O, 23.5 wt.% V content) was obtained from Shanghai Luyuan Fine Chemical Factory. Ammonium bicarbonate (AR), sodium tungstate dehydrate (AR), aqueous ammonia solution (28 wt.%), absolute ethanol, BAc (CR), n-octane (CR), and 3-aminopropyltriethoxysilane (APS, 98%) were all purchased from Sinopharm Chemical Reagent Co., Ltd. 3-Methacryloxypropylmethyldimethoxysilane (MPDS, 98%), MPS (98%), and methyltriethoxysilane (MTES, 98%) were obtained from Zhangjiagang Guotai-Huarong New Chemical Materials Co., Ltd. of China. 2.2. Preparation of VO2(W) crystals The crystalline VO2(W) powder was prepared as follows. A stoichiometric volume of aqueous sodium tungstate solution (0.50 M) was first dropped into an aqueous solution of VOSO4 (0.1 mol dissolved in 150 mL deionized water), followed by addition of an aqueous solution of ammonium bicarbonate (0.2 mol dissolved in 150 mL deionized water) using an automatic syringe pump over a period of 1.5 h at room temperature. The slurry-like suspension was magnetically stirred for another hour and subsequently filtered with a #4 sand funnel. The as-obtained precipitate was carefully washed with deionized water and absolute ethanol until no sulfate was detected, and then it was dried at 40 °C under vacuum for about 4 h. The dried precursor (~11.0 g) was put in ceramic boats and further moved to a quartz glass tube (87 × 1300 mm). The quartz glass tube was purged with nitrogen and then heated to 800 °C at a heating rate of 10 °C min −1 under a flow of nitrogen. Subsequently, a certain amount of air was injected into the tube within 1 min and thereafter the nitrogen flow was stopped. The temperature was held at 800 °C for another 2 h. Crystalline VO2(W) powder was obtained through thermal decomposition of the precursor. 2.3. Preparation of MPS-functionalized VO2(W) nanoparticles The MPS-functionalized VO2(W) [MPS–VO2(W)] nanoparticles were obtained as follows. VO2(W) powder (7 g) was mixed with 70 g of pre-made BAc/MPS solution. The mixture was sonicated at room temperature for about 30 min and then bead-milled for about 10 h at a rotation speed of 3000 rpm using zirconia beads with diameters of 300 μm. MPS was added into the slurry in three batches, namely, 50, 25, and 25% of the total MPS [MPS/VO2(W) = 0.2:1 mol/mol] at 0, 4, and 7 h of milling time, respectively. The temperature of the slurry was carefully controlled in the range of 20–30 °C using an ice/water bath. The as-obtained VO2(W) slurry was then centrifuged at either 8000 rpm or 14,000 rpm for 10 min to remove any large aggregates. The supernatant fluid was then precipitated with an approximately fivefold volume of n-octane, then centrifuged to get the MPS–VO2(W) nanoparticles. The obtained nanoparticles were further washed with n-octane three times to remove the residual MPS molecules and dried at 40 °C for 2 h under vacuum to evaporate the remained n-octane.
2.4. Synthesis of polysiloxane oligomers Polysiloxane oligomers were synthesized as per our previous method [34]. That is, a 50 mL round-bottom flask equipped with a condenser was put in an oil bath. Then 2.32 g (0.01 mol) of MPDS and 7.12 g (0.04 mol) of MTES were added into the flask and magnetically stirred, followed by dropwise addition of a diluted aqueous ammonia (0.1 mol/kg, 0.9 g)/ethanol (1 g) solution over 15 min at room temperature. Afterwards, the reaction mixture was heated to about 90 °C and refluxed for 4 h to obtain the polysiloxane oligomers. 2.5. Preparation of polysiloxane/VO2(W) nanocomposite coatings The MPS–VO2(W) nanoparticles were re-dispersed in BAc based on a solid content of 10% and then mixed with the as-obtained polysiloxane oligomers (theoretical solid content: 83.25%). The mass ratio of the MPS–VO2(W)/BAc dispersion to the as-obtained polysiloxane oligomer was set as 5.7/50, 14.5/50, 23.6/50, 29.8/50, 45.9/50, 62.9/50, and 141.6/50, corresponding to the VO2(W) content of 1.0, 2.5, 4.0, 5.0, 7.5, 10.0, and 20.0 wt.% in the dried films. The mixture was then sonicated for about 20 min and evaporated under vacuum at room temperature (or further charged with a certain amount of BAc) to obtain coatings with a solid content of about 50%. After 30 wt.% of APS was added (based on the weight of the as-obtained polysiloxane oligomer solution), the coating was cast on glass slides using a #8 wire-wound applicator (about 80 μm in wet film thickness) and dried at room temperature for one week. The samples were then used for further optical and mechanical measurements. 2.6. Characterization Fourier transform infrared (FTIR) spectra were obtained using a Nicolet Nexus 470 spectrometer (ThermoFisher, USA) in the wavenumber range of 400–4000 cm −1 with a resolution of 4 cm −1 and accumulation of 32 scans. Liquid MPS solution was dropped onto the surface of NaCl plate and quickly dried by infrared irradiation for direct characterization. The dried nanoparticles were blended with KBr to form sample flakes. X-ray powder diffraction (XRD) measurements were carried out at room temperature using a Bruker Nanostar U System, with incident X-ray wavelength (λ = 0.1542 nm). The collimation system consisted of two cross-coupled Gobel mirrors and four pinholes. A Hi-Star two-dimensional (2-D) area detector (Siemens) filled with pressurized xenon gas was used to record the one-dimensional (1-D) XRD patterns at a voltage of 40 kV and a current of 40 mA. Transmission electron microscopy (TEM) was performed on an H600 transmission electron microscope (Hitachi Corp., Japan) at an accelerated voltage of 100 kV. The as-prepared VO2 dispersion was diluted with BAc and directly dried on copper grids. The particle size and distribution of the MPS–VO2(W)/BAc dispersions were detected by a Nano ZS90 particle-size analyzer (Malvern Instruments Co., Ltd., U.K.). Thermogravimetric analysis (TGA) was carried out using a PerkinElmer TGA-7 instrument (USA) from ambient temperature to 750 °C at a heating rate of 20 °C/min under ultrapure nitrogen atmosphere. Differential scanning calorimetry (DSC) was performed on a PerkinElmer Pyris 1 system with a heating rate of 10 °C/min and ultrapure nitrogen protection. Thermochromic switching characteristics were monitored on a Hitachi U-4100 UV–visible-near IR (NIR) spectrophotometer equipped with a film-heating unit. The transmittance spectra at normal incidence were recorded in the wavelength range of 400–2600 nm at a scanning rate of 600 nm/min. Hysteresis loops were measured by collecting the transmittance of the films at a fixed wavelength (1550 nm) at approximate intervals of 2.0 °C. The diffuse reflectance at normal incidence
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
and the specular reflectance at a 10° incidence angle were determined using a sheet of BaSO4 powder as the standard. The temperature was measured using a thermocouple in contact with the films and controlled through a temperature-controlling unit. The pendulum hardness (Koenig hardness) of the nanocomposite coatings was measured with a 707 KP pendulum hardness rocker (Sheen Instrument Ltd., UK). The time of the Koenig pendulum swinging from 6° to 3° was automatically recorded. The average value from three measurements at different sites on the same sample was adopted. The micro-mechanical performance was determined using a nanoindentation tester (CSM Instruments, Switzerland) with a Berkovich diamond indenter. After contact with the surface of sample, the indenter was pressed into the coatings at a constant strain rate of 0.05 s−1 until 2000 nm of depth was reached, held at the maximum load for 50 s, and then withdrawn from the surface at the same rate as loading. About 5 indents were performed on each sample and the averaged values were finally adopted. The hardness and elastic modulus were calculated by the Oliver–Pharr method [35]. 3. Results and discussion 3.1. Preparation of crystalline VO2(W) powder Preparation of the crystalline VO2 and VO2(W) powder was demonstrated through thermal decomposition of the hydrolytic product of VOSO4 [33]. In this work, it was found that introduction of a small amount of oxygen into the system during the calcination stage was necessary to get VO2 (W). Fig. 1 shows the XRD patterns of the vanadium oxide prepared at various calcination atmospheres. V2O3 (JCPDS 65-9474) rather than VO2 was obtained when air was not charged into the nitrogen flow. A typical polycrystalline diffraction pattern of monoclinic P21/c VO2 (JCPDS 43-1051) was exhibited when 700 mL air was added into the quartz glass tube, indicating the formation of crystalline VO2(W). Nevertheless, the reflection angle of (110) of VO2(W) was slightly smaller than that of pure VO2 and the adjacent interplanar distance was just the opposite, owing to the larger size of the W atom relative to the V atom. However, when either an excess (1000 mL) or an insufficient amount (100 mL) of air was added, a vanadium oxide mixture with various vanadium valence was obtained, as seen from XRD patterns b and d. Therefore, it is crucial to control the amount of the air introduced during the calcination step. Fig. 2 displays the typical DSC curves of the VO2(W) with various W fractions and calcination temperatures. It can be seen that the critical phase transition temperature (Tc) of the undoped VO2 was 67 °C
233
(the peak temperature), very close to the 68 °C reported by Morin [1]. When the W atom was doped, the Tc decreased to 36 and −13 °C at 1.5 and 3.5 atm.% of W, respectively. A low and wide exothermic peak is revealed at a calcination temperature of 600 °C, and the peak temperature shifts to high temperature, suggesting a weakening of the phase transition of VO2(W). This may be attributed to the incomplete and nonhomogeneous doping of W and weak crystallization of VO2(W) particles. 3.2. De-agglomeration and modification of VO2(W) particles Because the as-synthesized VO2(W) particles consist of aggregates owing to the calcination step at 800 °C, de-agglomeration of VO2(W) is necessary for its application. It has been demonstrated that MPS is an efficient surface modifier for the de-agglomeration of TiO2 nanoparticles in BAc [36]. Thus, MPS was herein adopted for the de-agglomeration and modification of VO2(W) particles. Fig. 3 shows the mean particle size and distribution of VO2(W) after the VO2(W) powder was bead-milled in BAc for different lengths of time in the presence of MPS. The mean particle size of VO2(W) was about 792 nm just after sonication for 30 min. After 1 h of bead-milling, the particle-size distribution shifted to a smaller size and a new small peak with an average diameter of 105 nm appeared, indicating the breakup of the agglomerates. VO2(W) dispersion with a mean particle size of 200 nm was finally obtained after 10 h of bead-milling. TEM images of the corresponding VO2(W) dispersions are also presented in Fig. 3. The as-obtained VO2(W) particles were significantly agglomerated, with particle sizes at the micrometer level, and were effectively broken up after bead-milling, being well in agreement with the size distribution. To remove the large aggregates, the as-obtained MPS–VO2(W)/BAc dispersion was further centrifuged at 8000 rpm for 10 min. The supernatant, with a solid content of about 4 wt.%, was collected. Its particle size distribution and TEM images are also given in Fig. 3. It can be seen that the MPS–VO2(W) particles in the supernatant actually belong to nanoparticles with a mean particle size of 47 nm. These MPS–VO2(W) nanoparticles were used in the preparation of transparent polysiloxane/ VO2(W) nanocomposite coatings. Fig. 4 compares the XRD patterns of VO2(W) particles before and after de-agglomeration and centrifugation. The intensity of the XRD patterns obviously declined after bead-milling, indicating the reduced crystallinity and crystal size of the VO2(W) particles. Theoretical calculations based on the Scherrer equation applied to the (110) reflection indicates that the crystal size of the VO2(W) particles (given in Fig. 4) changed from 78.9 to 14.2 nm after bead-milling. Moreover,
0.0 ∗ V3O5
V6O13
∗
-0.2
a ∗
∗ ∗ ∗
b
V2O3
Heat Flow Endo Up (W/g)
Intensity (counts/s)
♣
V2O3+V3O5
∗
011 200
c
210
220
VO2(W)
♣♣ ♣♣
10
20
♣
♣ ♣
30
40
V2O5+V6O13
♣d
50
60
70
80
2θ (degree) Fig. 1. XRD patterns of thermal decomposition products calcined at 800 °C for 2 h with different amounts of air added into the quartz glass tube: (a) 0, (b) 100, (c) 700, and (d) 1000 mL. (Molar ratio of W/V: 0.015:1.)
d
-0.6
o
b
-0.4
44 C
c
o
-13 C
o
36 C
-0.8
a
-1.0 -1.2 -1.4 -20
o
67 C 0
20
40
60
80
100
Temperature (oC) Fig. 2. DSC curves of pure VO2 particles (calcination temperature: 800 °C) (a) and VO2(W) particles synthesized with different tungsten-doping fractions and calcination temperatures: (b) 3.5%, 800 °C, (c) 1.5%, 800 °C, and (d) 1.5%, 600 °C.
234
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
Fig. 3. The mean particle size and distribution of VO2(W) particles at various lengths of milling time: (a) 0 h with sonication for about 30 min, (b) 1 h, (c) 10 h, and (d) after centrifugation at 8000 rpm for 10 min. (The values marked on the peak represent the average particle size and intensity fraction of the corresponding peak. The insert pictures are the corresponding TEM images of original VO2(W) powder and VO2(W) dispersion.)
a further decrease in crystal size was revealed after centrifugation, which should be attributed to the removed of the VO2(W) particles with large crystal sizes. Therefore, in addition to the de-agglomeration of the as-synthesized VO2(W) particles, bead-milling can crack VO2(W) crystals. The surface properties of MPS-modified VO2(W) nanoparticles were characterized by FTIR, as shown in Fig. 5. The absorption peaks at 2840–3000, 1720, and 1637 cm −1, corresponding to the stretching vibrations of C\H, C_O, and C_C, respectively, appear in the FTIR spectrum of MPS–VO2(W), clearly demonstrating the attachment of MPS to VO2(W) nanoparticles. Furthermore, an absorption peak at 1122 cm −1 due to the stretching vibration of Si\O\V appears. Additionally, we found that the modified VO2(W) nanoparticles became hydrophobic after modification, while the original VO2(W) particles were hydrophilic. Both these facts demonstrate that MPS has been grafted onto the surface of VO2(W) particles.
The MPS-modified VO2(W) (nano)particles were quantitatively characterized by TGA, as displayed in Fig. 6. Clearly, MPS-functionalized VO2(W) particles have an increased weight loss relative to the assynthesized VO2(W) particles, which is attributed to the organic components of the attached MPS. The modified VO2(W) nanoparticles from the supernatant have a much higher weight loss than those from the dispersion without centrifugation. It is rational to assume that those VO2(W) nanoparticles in the supernatant have a higher specific surface area and thus higher amount of MPS attached, in accordance with the surface modification of the TiO2 nanopowder with MPS [36]. 3.3. Preparation of polysiloxane/VO2(W) nanocomposite coatings An ambient curable polysiloxane oligomer was adopted as the polymer binder for the preparation of the nanocomposite coatings, because the oligomer can be cured in the presence of APS to form
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
100
10000 7500
78.9 nm
a Remaining weight (%)
5000 2500 3500
Intensity (counts/s)
3000
235
b
14.2 nm
2500 2000 1750
11.5 nm
As-synthesized VO2 a
90
b
85
b,c: MPS modified VO2
80
c
75
c
1500
95
100
200
300
1250 1000 1250
10.7 nm
500
600
700
Fig. 6. TGA curves of (a) as-synthesized VO2(W), (b) modified VO2(W) from dispersion after 10 h bead-milling, and (c) modified VO2(W) from the supernatant fluid obtained by 10 h bead-milling and centrifugation at 8000 rpm for 10 min.
d
1000 750 500
10
400
Temperature (oC)
20
30
40
50
60
70
80
2θ (degree) Fig. 4. XRD patterns of W-doped VO2 from (a) the original sample, (b) dispersion after 10 h of bead-milling, and dispersions centrifuged at (c) 8000 rpm and (d) 14,000 rpm for 10 min each. (The data correspond to the grain size calculated based on the Scherrer equation.)
result is also proven by the decline of the absorption at 2750–3140 cm−1 due to C\H. Moreover, a new peak at 1570 cm − 1, assigned to the bending vibration of \NH2, occurs in spectrum b. Surprisingly, it weakens when the MPS–VO2 nanoparticles are embedded. The band at 3250–3400 cm−1 due to stretching vibration is also quite different from that of pure polysiloxane coatings. These peak changes may mean the occurrence of an aza-Michael reaction between the \NH2 group and the C_C group. 3.4. Thermochromic property of nanocomposite coatings
c Transmittance (%)
Fig.8 displays the transmittance spectra and corresponding photographs of the ambient-cured polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents. The transmittance of the nanocomposite film at high temperature increases as the wavelength increases at wavelengths above 1500 nm. This trend is the converse to that of the conventional VO2 or doped VO2 film. In any case, the results are very similar to those theoretically predicted by Li et al. [31]. Solar modulation of 4.3, 8.8, and 23.2% at the wavelength of 2500 nm, and luminous transmittance of 89.3%, 77.2%, and 59.4% at the wavelength of 550 nm, are achieved for the films containing 1, 4, and 10 wt.% of VO2(W), respectively. The difference of transparency
a Transmittance (%)
thick polysiloxane coatings with good mechanical performance [34]. The MPS–VO2(W) nanoparticles obtained from the supernatant were redispersed in BAc to get a MPS–VO2(W)/BAc dispersion and then blended with the polysiloxane oligomer. After being cured at room temperature for about one week, the nanocomposite coatings were employed for structure and performance characterization. Fig. 7 shows the FTIR spectra of the polysiloxane oligomers, pure polysiloxane coatings, and polysiloxane/VO2(W) nanocomposite coatings. The peaks at 1720 and 1640 cm−1 are attributed to the stretching vibrations of C_O and C_C, respectively. A broad absorption band at 3140–3720 cm−1 in the spectrum due to the stretching vibration of \OH indicates the hydrolysis of MPDS/MTES in the pre-hydrolysis step. However, the intensity of the absorption band significantly decreases and even disappears in spectra b and c, demonstrating the further condensation of the silanol groups of polysiloxane oligomers. This
b
c
a
4000
4000
b
3500
3000
2500
2000
1500
1000
500
-1
Wavelength (cm ) Fig. 5. FTIR spectra of (a) MPS, (b) as-synthesized VO2(W), and (c) MPS-modified VO2(W) nanoparticles from the supernatant.
3500
3000
2500
2000
1500
1000
Wavelength (cm-1) Fig. 7. FTIR spectra of (a) polysiloxane oligomer, (b) polysiloxane coating, and (c) polysiloxane/VO2(W) nanocomposite coating. (The ratios of polysiloxane oligomers to APS in b and c are the same and the VO2(W) content is 20 wt.% based on the dried polysiloxane coating.)
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
can also clearly be distinguished from their photographs, as shown in Fig. 8b, namely, higher VO2(W) content results in better solar modulation and low visible transmittance. Therefore, a balance has to be considered between the luminous transmittance and the solar modulation ability. In comparison with literature data reported, our nanocomposite film has higher transparency than previous VO2 films [18,37], doped VO2 films [14,23,38], or even multilayer thermochromic films [25,39]. However, the solar modulation of the nanocomposite film is lower relative to the literature data. This is thought to be caused by the doping of the W atom [40] and the reduced crystallinity of VO2(W) nanoparticles by bead-milling. If VO2(W) particles with better dispersibility and lower primary particle size are adopted, we are convinced that the solar modulation of the nanocomposite film would be greatly improved. Fig. 9 displays the hysteresis loop for the nanocomposite film with 10 wt.% VO2(W) content. As seen from the figure, the film exhibits a temperature centered at about 40 °C and a width of 13 °C. The centered temperature is close to the metallic/semiconductor transition temperature of VO2(W) determined by DSC. Fig. 10 presents the reflectance spectra of the nanocomposite film. Stronger specular reflectance at 60 °C can be clearly seen relative to that at room temperature in the near-infrared range. However, the specular reflectance does not change significantly with temperatures in the visible range. This result is the same as the reflectance spectra of W-doped VO2 film [37]. The diffuse reflectance, being due to the scattering of light by large VO2(W) particles in the films, is also exhibited in Fig. 10b. However, it is less impacted by the temperature and thus does not contribute to the solar modulation of the nanocomposite film.
(a) 1% Transmittance (%)
4% 80
10% 60
0.70
0.65
Transmittance (%)
236
heating 0.60
cooling 0.55
0.50
0.45 10
20
30
40
50
60
70
Temperature (oC) Fig. 9. Hysteresis loop of the nanocomposite film with 10 wt.% VO2(W) sample at a wavelength of 1550 nm.
3.5. Mechanical properties of the nanocomposite coatings Fig. 11 shows the effect of MPS–VO2(W) content on the pendulum hardness (Koenig hardness) of the polysiloxane/VO2(W) nanocomposite films. Clearly, the pendulum hardness significantly increases with increasing VO2(W) content at the VO2(W) content less than 10 wt.%, which should be due to the reinforced role of inorganic nanoparticles and the cross-linking role of MPS–VO2(W) nanoparticles via an aza-Michael reaction. However, when the MPS–VO2(W) content is greater than 10%, a slight reduction of the pendulum hardness can be observed. This may result from the high quantity of free MPS segments present, because these segments attached to nanoparticles act as a flexible component for the nanocomposites [41,42]. The loading–hold–unloading curves of polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents are shown in Fig. 12. Upon loading, the force increases with increasing depth and creep behavior is observed for all coatings when the loading remains at the maximum during a period of holding time. However, the deformation almost completely recovers after unloading all the coatings, indicating the high elasticity of the polysiloxane/VO2(W) nanocomposite coatings. The microhardness and elastic modulus increase with increasing VO2(W) content at VO2(W) content less than 10%. However, a slight decrease of the microhardness and elastic
40 1000
1500
2000
2500
Wavelength (nm)
Reflectance, %
(b)
Reflectance, %
500
12 11 10 9 8 7 6 5 4 7 6 5 4 3 2 1
a
o
20
b
o
20
o
60
500 Fig. 8. (a) Vis-NIR transmittance spectra and (b) photographs of polysiloxane/VO2(W) nanocomposite coatings with 1, 4, and 10 wt.% VO2(W) content. (Molar ratio of W: V = 0.15:100; film thickness: ~20 μm, determined by a digital microscope; The solid and dashed lines in (a) are for the films at 20 and 60 °C, respectively.).
o
60
1000
1500
2000
2500
Wavelength, nm Fig. 10. (a) Specular and (b) diffuse reflectance spectra of polysiloxane/VO2(W) nanocomposite coatings with 10 wt.% VO2(W) content.
Y. Lu et al. / Thin Solid Films 534 (2013) 231–237
nanocomposite coatings. Moreover, the higher the VO2(W) content in the coating, the more obvious the thermochromic phenomenon is. Luminous transmittance (550 nm) of about 60% and solar modulation (2500 nm) of 23% was achieved at 10% VO2(W) load. In addition, the polysiloxane/VO2(W) nanocomposite coatings possess ultrahigh hardness. Thus, transparence, thermochromic properties at near room temperature, and ultrahigh hardness certainly favor their practical application on windows.
Pendulum hardness (s)
175 170 165 160 155 150
Acknowledgements
145 140 135 0
5
10
15
20
VO2(W) content (wt%) Fig. 11. Pendulum hardness of polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents.
modulus were observed when the VO2(W) content increased to 20%. This trend is completely consistent with the trend of pendulum hardness as discussed above. Among these nanocomposite coatings, the microhardness (990 MPa) of the coatings with 10 wt.% of VO2(W) content is close to that of the sol–gel-derived silica coatings (1.0 GPa) [43], suggesting that the thermochromic coating has an ultrahigh hardness. It is thus thought that these hard polysiloxane/VO2(W) nanocomposite coatings are promising for in situ application on windows. 4. Conclusions Transparent polysiloxane/VO2(W) nanocomposite coatings with excellent thermochromic properties were prepared by mixing a VO2(W) dispersion with a polysiloxane oligomer and ambient-curing with APS as a curing agent. The VO2(W) particles were synthesized by alkaline hydrolysis of VOSO4 mingled with a certain amount of Na2WO4 as dopant and subsequent calcination in an oxidative atmosphere. A certain amount of air is crucial to guarantee the valence of V. The as-obtained VO2(W) particles can be efficiently de-agglomerated in BAc by a bead-milling process in the presence of MPS. Bead-milling can crack the VO2(W) crystals. Nevertheless, the cracked VO2(W) crystals still exhibit good thermochromic properties. Removing partial large VO2(W) aggregates by centrifugation can produce a MPS–VO2(W) dispersion dispersed at the nano-size level, resulting in high transparency of the H(MPa)
20
Load (mN)
237
0% 2.5% 10% 20%
15
473 647 990 780
13 14 53 93
E(GPa)
1.26 1.77 2.74 2.35
0.06 0.05 0.02 0.10
10
5
0 0
500
1000
1500
2000
Depth (nm) Fig. 12. The loading–hold–unloading curves of the polysiloxane/VO2(W) nanocomposite coatings with different VO2(W) contents.
This work was supported by the Foundation of Science and Technology of Shanghai (09DJ1400205) and the innovative team of Ministry of Education of China (IRT0911). Prof. Q. Z. Zhu from Shanghai University of Electric Power is also appreciated for the supply of a temperature controller for thermochromic tests. References [1] F.J. Morin, Phys. Rev. Lett. 3 (1959) 34. [2] J.B. Goodenough, J. Solid State Chem. 3 (1971) 490. [3] C. Batista, R. Ribeiro, J. Carneiro, V. Teixeira, J. Nanosci, Nanotechnol. 9 (2009) 4220. [4] M. Gurvitch, S. Luryi, A. Polyakov, M. Dudley, G. Wang, S. Ge, et al., J. Appl. Phys. 102 (2007) 033504. [5] S.H. Chen, H. Ma, X.J. Yi, H.C. Wang, X. Tao, M.X. Chen, et al., Infrared Phys. Technol. 45 (2004) 239. [6] M.H. Lee, Sol. Energy Mater. Sol. Cells 71 (2002) 537. [7] I. Balberg, S. Trokman, J. Appl. Phys. 46 (1975) 2111. [8] F. Guinneton, L. Sauques, J.C. Valmalette, F. Cros, J.R. Gavarri, J. Phys. Chem. Solids 62 (2001) 1229. [9] P. Jin, S. Tanemura, Jpn. J. Appl. Phys. 33 (1994) 1478, Part 1. [10] G.V. Jorgenson, J.C. Lee, Sol. Energy Mater. 14 (1986) 205. [11] N.Y. Shishkin, A.A. Komarov, D.V. Kosov, V.A. Cherkasov, L.A. Bashkirov, U. Bardi, et al., Sensors Actuators B Chem. 108 (2005) 113. [12] T.D. Manning, I.P. Parkin, R.J.H. Clark, D. Sheel, M.E. Pemble, D. Vernadou, J. Mater. Chem. 12 (2002) 2936. [13] T.D. Manning, I.P. Parkin, M.E. Pemble, D. Sheel, D. Vernardou, Chem. Mater. 16 (2004) 744. [14] R. Binions, G. Hyett, C. Piccirillo, I.P. Parkin, J. Mater. Chem. 17 (2007) 4652. [15] T.H. Yang, S. Nori, H.H. Zhou, J. Narayan, Appl. Phys. Lett. 95 (2009) 102506. [16] D.H. Kim, H.S. Kwok, Appl. Phys. Lett. 65 (1994) 3188. [17] K.R. Speck, H.S.-W. Hu, M.E. Sherwin, R.S. Potember, Thin Solid Films 165 (1988) 317. [18] S.W. Lu, L.S. Hou, F.X. Gan, Adv. Mater. 9 (1997) 244. [19] T.J. Hanlon, J.A. Coath, M.A. Richardson, Thin Solid Films 436 (2003) 269. [20] J.H. Cho, Y.J. Byun, J.H. Kim, Y.J. Lee, Y.H. Jeong, M.P. Chun, et al., Ceram. Int. 38 (2012) S589. [21] G. Xu, P. Jin, M. Tazawa, K. Yoshimura, Jpn. J. Appl. Phys. 43 (2004) 186. [22] C.G. Granqvist, Thin Solid Films 193–194 (1990) 730. [23] N.R. Mlyuka, G.A. Niklasson, C.G. Granqvist, Appl. Phys. Lett. 95 (2009) 171909. [24] P. Jin, G. Xu, M. Tazawa, K. Yoshimura, Jpn. J. Appl. Phys. 41 (2002) 278. [25] P. Jin, G. Xu, M. Tazawa, K. Yoshimura, Appl. Phys. A 77 (2003) 455. [26] S. Saitzek, F. Guinneton, L. Sauques, K. Aguir, J.R. Gavarri, Opt. Mater. 30 (2007) 407. [27] N.R. Mlyuka, G.A. Niklasson, C.G. Granqvist, Sol. Energy Mater. Sol. Cells 93 (2009) 1685. [28] Z. Chen, Y.F. Gao, L.T. Kang, J. Du, Z.T. Zhang, H.J. Luo, et al., Sol. Energy Mater. Sol. Cells 95 (2011) 2677. [29] L.T. Kang, Y.F. Gao, H.J. Luo, ACS Appl. Mater. Interfaces 10 (2009) 2211. [30] L.T. Kang, Y.F. Gao, H.J. Luo, Z. Chen, J. Du, Z.T. Zhang, ACS Appl. Mater. Interfaces 3 (2011) 135. [31] S.Y. Li, G.A. Niklasson, C.G. Granqvist, J. Appl. Phys. 108 (2010) 063525. [32] J.C. Valmalette, J.R. Gavarri, Sol. Energy Mater. Sol. Cells 33 (1994) 135. [33] J.Q. Shi, S.X. Zhou, B. You, L.M. Wu, Sol. Energy Mater. Sol. Cells 91 (2007) 1856. [34] X.B. Chen, S.X. Zhou, B. You, L.M. Wu, J. Sol–Gel Sci. Technol. 58 (2011) 490. [35] G.M. Pharr, W.C. Oliver, J. Mater. Res. 7 (1992) 1564. [36] Y.F. Lu, S.X. Zhou, L.M. Wu, J. Disper. Sci. Technol. 33 (2012) 497. [37] C. Piccirillo, R. Binions, I.P. Parkin, Thin Solid Films 516 (2008) 1992. [38] C.S. Blackman, C. Piccirillo, R. Binions, I.P. Parkin, Thin Solid Films 517 (2009) 4565. [39] Z.T. Zhang, Y.F. Gao, Z. Chen, J. Du, C.X. Cao, L.T. Kang, et al., Langmuir 26 (2010) 10738. [40] M.A. Sobhan, R.T. Kivaisi, B. Stjerna, C.G. Granqvist, Sol. Energy Mater. Sol. Cells 44 (1996) 451. [41] S.X. Zhou, L.M. Wu, Macromol. Chem. Phys. 209 (2008) 1170. [42] G.D. Chen, S.X. Zhou, G.X. Gu, H.H. Yang, L.M. Wu, J. Colloid. Interf. Sci. 281 (2005) 339. [43] A. Sellinger, P.M. Weiss, A. Nguyen, Y.F. Lu, R.A. Assink, W.L. Gong, et al., Nature 394 (1998) 256.