Improvement on contamination resistance to volatile organics and moisture of sol–gel silica antireflective coating for 351 nm laser system by structural modulation with fluorinated compounds

Abstract

A new silica antireflective (AR) coating with excellent spectral stability and laser-induced damage threshold (LIDT), which are crucial characteristics in high-powered laser systems, was prepared using the sol–gel method. First, aided by 2,2,3,3,4,4,5,5,6,6,7,7-dodecafluoro-1,8-octanediol (simplified as DFDO), cross-linking among silica particles was facilitated. Thus high and adjustable inter-particle porosity was obtained and hence the refractive indices of the coatings were modulated. Next, poly(2,2,3,4,4,4-hexafluorobutyl) acrylate (simplified as PHFBA) was utilized to occupy micro-pores inside the coating in order to protect them from contamination. Consequently, the volatile organic and water vapor resistance of resultant DFDO/PHFBA/silica coatings was improved dramatically. By characterizing aggregating manner and stacking condition between particles, the modification process of the proposed AR coating was clarified in details.

---

1. Introduction

Inertial confinement fusion (ICF) is a cutting-edge technology in today's scientific fields. Its favorable application potential as a clean and efficient energy source is very attractive.¹ However, to realize ICF, a high-powered laser system with stable supply of short-wavelength laser beams is indispensable, where the properties of the optic components are crucial. Due to certain intrinsic characteristics that ensure high LIDT and transmittance, sol–gel coatings are widely employed in these transmission optic components.^2–8 Specific porosity in these sol–gel coatings is necessary to obtain desirable refractive index, in which the micro-pores among particles play a major role. However, the sol–gel coating can easily adsorb organic contaminants and moisture in the environment, due to its porous structure and enormous surface hydroxyls.^9,10 This adsorption significantly deteriorates the spectral performance of the component, and results accordingly in huge aggravation of the entire system.^11–13 In order to keep the system running effectively over a long period of time, maintenance of spectral stability in these components is an urgent problem. Previous research have reported progresses on fabricating silica AR coating with superhydrophobicity and mechanical stability,^14,15 which emphasizes the stability of silica AR coating in practical application.

Surface modification with fluorinated materials is considered a preferable choice to improve the stability of silica AR coating, due to some advantageous intrinsic properties of fluorine atom.¹⁶ Fluorine atom possesses the highest electronegativity (3.98) and the smallest ion radius (0.133 nm) of all the elements. The high electronegativity produces a strong C–F bond due to the well matching of both 2s and 2p electronic orbits in F atom and C atom, and the more F atoms bond on one C atom, the more stable this bonding becomes. Moreover, the high electronegativity and small ionic radius bring F atom the lowest polarizability which makes the dispersion force between polyfluorinated molecules particularly weak and also ensures fluorocarbons with outstanding simultaneous hydrophobicity and lipophobicity. Additionally, the fluoropolymers possess very low surface energy due to the dense covering of F atoms over the surface, and this contributes a lot to its hydrophobicity and lipophobicity as well. Lastly, fluoropolymers have a low refractive index which is very crucial to obtain AR coatings with high performance.

In this study, a novel approach to prepare silica AR coating with improved contamination resistant capability, while displaying high transmittance and LIDT at 351 nm via fluorine modification was presented. First, a cross-linking network comprised of silica particles was facilitated using fluorinated alcohol, with which the micro-pores are well maintained during coating formation. Thus, the refractive index of sol–gel coating was successfully modulated. Next, by filling these micro-pores with fluorinated polymer, they became well protected from contaminants. In combination with its lowered surface energy, contamination resistance of this sol–gel coating can be improved dramatically.

2. Experiment

2.1. Preparation of PHFBA

PHFBA was polymerized in bulk assisted by an ultrasonic process. The molecular weight of the polymer was characterized by gel permeation chromatography (GPC) (HLC-8320GPC), and the number-average molecular weight of the resultant polymer is 60 305.

2.2. Preparation of silica sol

Colloidal silica sol was prepared by the hydrolysis and condensation of TEOS with different amounts of DFDO in de-ionized water and anhydrous ethanol. In the sols, the weight ratios of DFDO to SiO₂ were 0, 1.0%, 3.0% and 5.0% respectively. The reaction was catalyzed by ammonium hydroxide in an appropriate molar ratio. The final concentration of SiO₂ was 3.3% by weight. After being stirred at room temperature for 6 h, the resultant sol was aged at 25 °C for 10 days.

2.3. Modification of silica sol

PHFBA was used as a modifier. Assisted by ultrasonication, the PHFBA was dissolved in a mixed solvent of ethanol and sec-butyl alcohol (at a 1 : 1 volume ratio). The polymer solution was then added into the silica sol and stirred for another 2 h. The weight ratios of PHFBA to silica were 1.3%, 4.0%, 6.3%, 8.7% and 11.3%, respectively. All the sols were carefully filtered through 0.22 nm PVDF filters before coating.

2.4. Preparation of AR coating

The fused silica and silicon substrates were cleaned ultrasonically. Next, the modified silica sol was deposited on these well-cleaned substrates by dip-coating at a withdrawal rate of 70–110 mm min⁻¹. The viscosity of the sols ranged from 1.50 to 1.72 mPa s⁻¹, and different withdrawal rates were attempted to ensure that the transmittance peak of prepared sol–gel coatings located at 351 nm. In order to compare the coating refractive indices before and after modification, the unmodified silica sol samples were also deposited on well-cleaned silicon substrates by dip-coating at the same withdrawal rate. All coatings were ultimately heated at 160 °C for 24 h before characterizations.

2.5. Characterization of spectral stability

All coated samples were stored under the condition of 20 °C and 40% RH initially. Parts of these samples were then instantly transferred into an enclosure of 25 °C and 75% RH. Samples without such treatment, and those treated for 4 days and 8 days, were selected for transmittance measurement via an UV-Vis spectrophotometer (Lambda 950), in order to evaluate the spectral stability of the coatings against water vapor. In addition, parts of the stored samples were placed into a vacuum chamber connected to an oil pump by valve. The pressure of the vacuum was set at 0.15 MPa, and the main organic ingredients inside this chamber were verified as cyclohexane, cyclopentane, alkanes and alkenes by the GC-MS (HP6890/5973 GC-MS) analysis. Samples without such treatment, and those treated for 3 days, 7 days, and 10 days, were also selected for transmittance measurement, in order to estimate the contamination resistance of the coatings to volatile organics.

2.6. Measurement of coating absorption

The absorption of coated sample was measured using laser calorimeter at a wavelength of 351 nm according to ISO 11551 standard. The samples were respectively placed into a thermo-insulated chamber for more than 1 h until the temperature became stable, then a quasi-cw laser irradiated the sample center for 120 s. During this procedure, the samples experienced both a heating up and cooling down process. By utilizing a set of resistors, the temperature variance at a distance of 7 mm from the surface center was measured, and then fitted to the heat transfer equation in order to evaluate the samples' absorption.

2.7. Measurement of laser damage threshold

The LIDT measurement was performed according to ISO 11254-1 standard. 10 × 10 sites with intervals of 2 mm were performed on each sample. The laser power during measurement was set to 10 J cm⁻² initially, and increased about 1 J cm⁻² row-by-row. The pulse duration was 3 ns, with a diameter of 430 μm on the sample. Fluences were consistently given in beam normal with a precision of ± 10% and incident angle was 0°. About 100 sites were tested to obtain a damage probability curve in entirety. From these results, the LIDT, i.e. the highest fluence value that caused no damage, was determined. The scatter light and visible changes of sample surface were recorded in real-time by photo diode and CCD camera during the measurement procedure. An additional off-line morphologic analysis was performed with laser confocal microscope (VK-X100).

2.8. Characterization of sol–gel silica coating

The connectivity between silica particles in the sols was characterized using a transmission electron microscope (TEM) (FEI Tecnai G² F20). The thickness and refractive index of the AR coatings was determined via ellipsometry (SENTECH SE850 UV). Atomic force microscopy (AFM) (MFP-3D) was used to characterize the surface morphology of the AR coatings. A scanning electron microscope with focused ion double beam (FIB-SEM) (Helios Nanolab 650) was used to characterize the particle morphology and stacking manner of the silica AR coatings. The pore volume of the coating was obtained via dynamic BET specific surface area measurement (Quantachrome autosorb-1 analyzer).

3. Results and discussion

3.1. Refractive index and coating structure

Refractive indices and thickness of the silica coatings is presented in Table 1, indicating that the refractive index of DFDO/silica coating is lower than that of pure silica coating, likely attributed to the differences in the coating porosity. In pure silica sol, typical clusters only contain a few silica particles, as shown in Fig. 1(a). As demonstrated in Fig. 1(b) and (c), particles in the DFDO/silica sols tend to form cross-linked cluster by connecting them together. Such cross-linked structure is similar to the schematic representation shown in Fig. 2(d). During the formation of DFDO/silica coating, compression of the cross-linked structure is less likely to occur due to its irregular shape, thus the arrangement of the coating is much looser, so that the particles in pure silica coating stacked more densely compared to DFDO/silica.

Table 1 Refractive indices (at 351 nm) and thickness of silica coatings

Weight ratio of DFDO to silica/%	Silica coatings		Weight ratio of PHFBA to silica/%	Silica coatings
	Refractive index (n)	Thickness/nm		Refractive index (n)	Thickness/nm
			1.3%	1.247	62.6
0	1.233	71.2	4.0%	1.263	67.6
			6.3%	1.271	69.0
1.0	1.219	71.4	—	—	—
3.0	1.187	76.1	—	—	—
			1.3%	1.241	59.1
			4.0%	1.252	61.4
5.0	1.173	81.5	6.3%	1.259	63.5
			8.7%	1.261	68.0
			11.3%	1.264	72.1

---

Fig. 1 TEM images of (a) pure silica sol, (b) silica sol containing 3.0 wt% of DFDO, and (c) silica sol containing 5.0 wt% of DFDO.

Fig. 2 Schematic representation of (a) clusters in pure silica sol, (b) pure silica AR coating, (c) pure silica AR coating modified by PHFBA, (d) clusters in DFDO/silica sol, (e) DFDO/silica AR coating, and (f) DFDO/silica AR coating modified by PHFBA.

In the pure silica sol, the free energy of particles is much higher due to substantial hydroxyl groups over its surface. Along with the solvent evaporation process, these hydroxyl groups undergo condensation reaction, resulting in shrinkage of the coating. At the same time, this high surface free energy also leads to high surface tension which creates concave menisci in the pores of the coating, resulting in compressive force around the perimeter of the pores.¹⁷ Conversely, as far as DFDO/silica particles are concerned, surface hydroxyl groups can be shielded by fluorine atoms that are induced from the DFDO compound. This shield effectively avoids condensation reaction between hydroxyl groups and lowers the surface free energy of particles, which reduces the surface tension quite efficiently and hence greatly decreases the coating contraction force. These two mechanisms produce DFDO/silica coating that is more porous than pure silica coating, due to looser particle stacking and less coating contraction force.

In addition, as shown in Table 1, as the weight ratio of DFDO to silica increases from 1.0 wt% to 5.0 wt%, the refractive indices of the DFDO/silica decrease from 1.219 to 1.173. Comparing Fig. 1(b) and (c), it shows that particles incline to form cross-linking structure with much larger size as the DFDO/silica weight ratio increases from 3.0 wt% to 5.0 wt%, and hence the particle arrangement is much looser. So, the porosity of the coating intends to be well maintained and the coating refractive index stays low. Fig. 2(a) and (b) demonstrate the coating formation process of pure silica. The subparticles with enormous surface hydroxyl groups aggregate together, thus forming particles which are the basic unit of sol–gel coating. Next, these particles stack layer-by-layer over the substrate surface during the dip-coating process. After the solvent evaporates, the particles form a stacked structure and the sol–gel coating is basically fabricated. Fig. 2(d) and (e) reveal the formation process of DFDO/silica, in which a much looser structure can be inferred compared with pure silica coating, due to the irregular shape of cross-linked particles and the reduction of surface tension caused by the shielding of hydroxyl groups. To this effect, the porosity difference derived from structural distinction between these two coatings originates in the refractive index variation.

As far as the PHFBA modification, regardless of DFDO, the refractive indices of silica coatings increase gradually with the increasing of PHFBA. The refractive index of 5.0 wt% DFDO-included silica coating varies from 1.241 to 1.259, and coating without DFDO varies from 1.247 to 1.271 when the concentration of PHFBA increases from 1.3 wt% to 6.3 wt%. This is because that PHFBA polymer mainly fills in the space among particles after being added into the silica sols, which results in porosity reduction. This is verified by the FIB-SEM low voltage image shown in Fig. 3, in which the filling in of PHFBA in the micro-pores among silica particles is clearly identified.

Fig. 3 Low voltage FIB-SEM image of PHFBA modified DFDO/silica coating.

Although the refractive indices of both pure silica and DFDO/silica increased after being modified by PHFBA, different results were still produced. Before PHFBA modification, porosity of the DFDO/silica coating is higher than that of pure silica, as shown in Fig. 2(b) and (e). After modification with same weight ratio of PHFBA, the total volume of the remaining pores in the DFDO/silica coating is still greater than that of the pure silica coating. To verify this assumption, dynamic BET specific surface area measurement was preformed. The measured overall pore volumes (at P/P₀ = 0.99) of pure silica coating, DFDO/silica coating, PHFBA/silica coating, and DFDO/PHFBA/silica coating were 391 cm³ g⁻¹, 520 cm³ g⁻¹, 141 cm³ g⁻¹, and 183 cm³ g⁻¹ respectively, which corresponds with previous assumption. Moreover, the particle surfaces in DFDO/silica sols are shielded by fluorine atoms also contained in PHFBA. According to the Hansen solubility parameters principle, affinities between different polymers and affinities to surfaces can improve dispersion and adhesion, which infers that the PHFBA compound mixes with particles much more easily in the DFDO/silica due to its similar solubility parameters as DFDO. As a result, PHFBA distributes well in the micro-pores among these particles during coating formation. In contrast, the particle surface in the pure silica system is covered with substantial hydroxyl groups, and hence the PHFBA compound does not mix as well. The comparative status of PHFBA occupancy is depicted in Fig. 2(c) and (f), where porosity of pure silica and DFDO/silica coating after PHFBA modification is detailed.

The microstructure of the coatings was also verified through AFM and FIB-SEM microscopy. Fig. 4(a)–(d) represented the morphology of pure silica coating, DFDO/silica coating, pure silica coating only modified by PHFBA, and DFDO/silica coating modified by PHFBA, respectively. By comparing the surface morphology of pure silica coating and DFDO/silica coating shown in Fig. 4(a) and (b), it seems that the DFDO/silica coating shows a slightly looser stacking structure. Furthermore, as shown in Fig. 4(c) and (d), the spacing between clusters in the DFDO/silica coating is enlarged, while the pure silica coating still exhibits a dense arrangement, after being modified by PHFBA. These results correspond with both the TEM results and the schematic representation. As far as texture shown in Fig. 4(a) and (c), the PHFBA modified pure silica coating possesses smoother surface morphology than that of pure silica, possibly attributed to the coverage of PHFBA compound over its densely arranged surface, which improves the planarization of the coating surface also as supposed in Fig. 2(c).

Fig. 4 Surface morphology, FIB-SEM images and water contact angles of (a) pure silica coating, (b) 5 wt% of DFDO-included silica coating, (c) pure silica coating modified by 6.3 wt% of PHFBA, and (d) 5 wt% of DFDO-included silica coating modified by 6.3 wt% of PHFBA.

Additionally, the water contact angle analysis results from Fig. 4(a) and (b) show 36° and 91° angles for pure silica coating and DFDO/silica coating respectively, indicating lower surface energy in the DFDO/silica coating than that in the pure silica coating. This is primarily attributed to the shielding of surface hydroxyl groups by fluorine atoms in the DFDO/silica coating. The contact angle of silica coating varies from 36° to 65° after modification with PHFBA. Meanwhile, the contact angle of DFDO/PHFBA/silica coating is almost the same as that of DFDO/silica coating. This rather prominent contact angle increase in the PHFBA/silica coating is due to the existence of fluorinated compounds over the coating surface. Surface shielding decreases the coating free energy, as proven by other researchers,¹⁸ although the exact mechanism varies. Moreover, a number of nanoscaled bumps are found on the surface of DFDO/PHFBA/silica coating as shown in the AFM image of Fig. 4(d). As the coating is formed by large clusters with irregular shape, an uneven coating arrangement in the stacking process is generated, which finally forms these bumps.

3.2. Water vapor and vacuum contamination resistance

The optical spectral stability of AR coatings before and after modification was also examined in this study. The transmission spectra of pure silica coating, DFDO/silica coating, and DFDO/PHFBA/silica coating treated for different periods at 75% RH and 25 °C condition are depicted in Fig. 5(a)–(c). The peak transmittance of pure silica coating drops from 99.9% to 99.2% after being treated for 8 days. Under the same treatment, the peak transmittance of DFDO/PHFBA/silica coating only decreases from 99.8% to 99.6%. The peak position of both coatings is affected by the treatment in a moist environment, in which the peak position of DFDO/PHFBA/silica coating exhibits a red-shift of about 30 nm and pure silica coating exhibits a 50 nm shift. This indicates a considerable spectral stability improvement under moisture in DFDO and PHFBA modified silica coatings.

Fig. 5 Evolution of coating transmission spectra. (a) Pure silica, (b) 5 wt% of DFDO-included silica and (c) 5 wt% of DFDO-included silica modified by 6.3 wt% of PHFBA coated on fused silica substrate in a closed container with 75% RH at 25 °C, (d) pure silica, (e) 5 wt% of DFDO-included silica and (f) 5 wt% of DFDO-included silica modified by 6.3 wt% of PHFBA coated on fused silica substrate in a vacuum chamber of 0.15 MPa with organic contaminants.

The contamination resistance against volatile organic vapor for the coatings was investigated as well. The transmission spectra of pure silica coating, DFDO/silica coating, and DFDO/PHFBA/silica coating that have been treated for different periods in a vacuum chamber at 0.15 MPa, are described in Fig. 5 (d)–(f). As shown in Fig. 5(d) and (f), the peak transmittance of DFDO/PHFBA/silica coating drops from 99.8% to 99.5%, and peak position shifts about 20 nm after being treated for 10 days. Meanwhile, the peak transmittance of pure silica coating decreases from 99.9% to 98.1%, and its peak position shifts about 40 nm after the same treatment. Reduction of peak transmittance in coatings is dominated by the adsorption of organic contaminants in micro-pores, which increases the refractive index of the coatings. In addition, the red-shift of peak position is assigned to the increase in coating thickness, which is also due to the adsorption of volatile organic contaminants.

The hydroxyl groups on the particle surfaces are shielded by fluorine atoms in coatings modified by DFDO and PHFBA, which makes the surface free energy lower than that of pure silica coating. And meanwhile, PHFBA also acts as filler for space among particles after occupying the micro-pores in the coating, which prevents the coating from contamination and hence improves overall contamination resistance. Surface modification that provides contamination resistance is much more easily realized in this process compared with the method of fabricating closed ordered mesopores.¹⁸ HMDS modification has also been reported a favorable choice to increase the stability of silica coating by replacing hydroxyl groups with methyl,¹⁹ but the antireflective capability of such modified silica coating cannot be well maintained compared with that proposed in this paper.

Fig. 5 also shows that the spectral stability of DFDO/silica coating is stronger than that of pure silica coating, but worse than that of DFDO/PHFBA/silica in both water vapor and organic contamination environments. Although the water contact angle of DFDO/silica coating is a bit higher than that of DFDO/PHFBA/silica coating, the results of transmission spectra still indicate that the fill in of PHFBA successfully prevents the penetration of water and organic contamination, and indeed improves the lifetime of silica AR coating.

3.3. Absorption

Component absorption is a crucial consideration for a successful and effective high-powered laser system. In this paper, bare substrate, substrate coated with pure silica and substrate coated with DFDO/PHFBA/silica were respectively measured according to the process described in the Experimental section. The green line in inset of Fig. 6 indicates the quasi-cw laser irradiation and the red line in inset of Fig. 6 represents temperature variance at a distance of 7 mm from the surface's center. This temperature variance was fitted using the heat transfer equation (see the blue line in inset of Fig. 6) and absorption was calculated as shown in Fig. 6. The absorption of DFDO/PHFBA/silica AR coating is obviously larger than that of pure silica AR coating, and absorption of pure silica AR coating is nearly the same as that of bare substrate, which indicates that the existence of DFDO and PHFBA compounds is responsible for the generation of extra absorption.

Fig. 6 Absorption result of fused silica substrates without coating, with pure silica AR coating and DFDO/PHFBA/silica AR coating.

3.4. Laser damage threshold

LIDT at 351 nm is one of the most important requirements for high-powered laser system as well. Laser damage tests for the fused silica substrates, substrates coated with pure silica and DFDO/PHFBA/silica were performed according to the measurement procedure described in the Experimental section. Fig. 7 shows that LIDT of fused silica substrates with DFDO/PHFBA/silica AR coating is about 19 J cm⁻². Meanwhile, the substrates with or without pure silica AR coating exhibit an LIDT at the level of 19–21 J cm⁻², indicating that the incorporation of DFDO and PHFBA does not reduce LIDT. If fused silica substrates that possess higher LIDT are utilized, LIDT for the DFDO/PHFBA/silica AR coating can be improved accordingly. In addition to absorption measurement, test results indicate that although the absorption of the DFDO/PHFBA/silica coating is slightly higher than that of pure silica at 351 nm, the difference in LIDT between these two coatings is hardly observable.

Fig. 7 Laser damage threshold of fused silica substrates without coating, with pure silica AR coating, and with DFDO/PHFBA/silica AR coating.

4. Conclusion

A novel approach with two-step modification for sol–gel coating was proposed in this paper. The modification process first utilized fluorinated alcohol compound to facilitate the cross-linked between silica particles, which assisted in formation of micro-pores, and realization of total modulation on porosity, and thus modulation of the sol–gel AR coating's refractive index. Based on this modification, PHFBA polymers were utilized to fill in the micro-pores among particles in order to prevent them from contamination. With such surface modification and micro-pore protection, the resultant DFDO/PHFBA/silica AR coatings possessed a longer lifetime in humidity or coarse vacuum environments containing organic contaminants, which validated the improved stability of this coating. The absorption of the coating at 351 nm was a little higher than that of pure silica coating however, the LIDT of the coating still reached 19 J cm⁻² (3 ns). The LIDT value was very close to that of fused silica substrate, which indicated significant application potential to the ICF equipment.

Acknowledgements

This work was financially supported by the National Major Science and Technology Project of China (2013ZX04006011-101).

References

1. E. I. Moses, Fusion Eng. Des., 2010, 85, 983.
2. H. G. Floch, P. F. Belleville, P. M. Pegon, C. S. Dijonneau and J. R. Guerain, Proc. SPIE, 1995, 2633, 432.
3. I. M. Thomas, Appl. Opt., 1992, 28, 6145.
4. X. Dieudonné, K. Vallé and P. Belleville, Opt. Express, 2011, 19, 16356.
5. X. D. Wang, G. M. Wu, B. Zhou and J. Shen, Opt. Express, 2012, 20, 24482.
6. P. Falcaro, L. Malfatti, T. Kidchob, G. Giannini, A. Falqui, M. F. Casula, H. Amenitsch, B. Marmiroli, G. Grenci and P. Innocenzi, Chem. Mater., 2009, 21, 2055.
7. X. G. Li, M. Gross, B. Oreb and J. Shen, J. Phys. Chem. C, 2012, 116, 18367.
8. X. G. Li, M. Gross, K. Green, B. Oreb and J. Shen, Opt. Lett., 2012, 37, 2364.
9. X. Miao, X. Cheng, X. Yuan, H. Lv, Y. Ye, S. He and W. Zheng, High Power Laser Part. Beams, 2010, 22, 1740.
10. A. A. Hossin, D. Edward, C. John and J. Todd, Proc. SPIE, 2005, 5897, 44.
11. A. Fornier, C. Cordillot, D. Schirman, F. Y. Genin, A. K. Burnham, P. K. Whitman, M. D. Feit, A. M. Rubenchik and J. M. Yoshiyama, Proc. SPIE, 1997, 3244, 499.
12. S. Becker, A. Pereira, P. Bouchut, F. Geffraye and C. Anglade, Proc. SPIE, 2007, 6403, 64030J.
13. X. G. Li and J. Shen, J. Sol-Gel Sci. Technol., 2011, 59, 539.
14. E. J. Park, J. K. Sim, M.-G. Jeong, H. O. Seo and Y. D. Kim, RSC Adv., 2013, 3, 12571.
15. L. Ye, S. Zhang, Q. Wang, L. Yan, H. Lv and B. Jiang, RSC Adv., 2014, 4, 35818.
16. X. X. Zhang, S. Cai, D. You, L. H. Yan, H. B. Lv, X. D. Yuan and B. Jiang, Adv. Funct. Mater., 2013, 23, 4361.
17. S. Cai, Y. Zhang, H. Zhang, H. Yan, H. Lv and B. Jiang, ACS Appl. Mater. Interfaces, 2014, 6, 11470.
18. J. Sun, Q. Zhang, R. Ding, H. Lv, H. Yen, X. Yuan and Y. Xu, Phys. Chem. Chem. Phys., 2014, 16, 16684.
19. Y. Liu, J. Shen, X. G. Li, B. Zhou and Z. H. Zhang, Chin. J. Inorg. Chem., 2013, 29, 1339.

---