Formation of nanoscopic CaF2via a fluorolytic sol–gel process for antireflective coatings

The synthesis of nanoscopic calcium fluoride was performed by the fluorolytic sol–gel process. Antireflective coatings of CaF2 were prepared from sols obtained by the reaction of CaCl2 with HF and subsequent dip coating. The addition of tetramethyl orthosilicate (TMOS) or tetraethyl orthosilicate (TEOS) after fluorination promotes the formation of transparent sols. The formation and crystallisation of CaF2 nanoparticles was studied by 19F liquid and solid state NMR spectroscopy, dynamic light scattering (DLS) and X-ray powder diffraction (XRD). The morphology of a CaF2-film was analysed by high resolution scanning electron microscopy (HR-SEM) and the mechanical stability of a CaF2-film was evaluated by the Crockmeter test using both felt and steel wool. The refractive index for a CaF2-film was measured by ellipsometry. The synthesis of CaF2 nanoparticles derived from CaCl2 is a good way to achieve porous antireflective coating layers.


Introduction
Calcium uoride and other metal uorides exhibit superior optical properties. Their optical transmittance ranges from low UV up to the high IR region. For example, calcium uoride is a dielectric material with an optical window ranging from 0.13 to 10 mm (ref. 1) exhibiting a broad spectrum of optical applications like IR, UV, microscope, astronomical instrumentation and spectroscopy. Especially the use of calcium uoride as ceramic material for laser applications is advantageous because not only the mechanical and optical properties compared with single crystals of calcium uoride are better but it can be produced in large volumes. [2][3][4] Furthermore, calcium uoride and magnesium uoride own a low refractive index (CaF 2 : n ¼ 1.44 at 500 nm MgF 2 : n ¼ 1.38 at 500 nm). 1,5 Thus, calcium uoride and magnesium uoride are promising candidates for antireective (AR) coatings. If the refractive index of the coating is lower than the refractive index of the substrate, the reection of the light will decrease. The complete anti-reection for single coatings depends on the mean geometric refractive index of the AR-layer with n 1 2 ¼ n s n 0 and its thickness with n 1 d ¼ l/4, where n 1 , n S and n 0 are the refractive indices of the coating, substrate and air, respectively, d the thickness of the lm and l the wavelength of the light. 6 For example, common glass mainly built up by SiO 2 (n S ¼ 1.55 at 500 nm) usually is surrounded by the medium air (n 0 ¼ 1.0). Hence, the optimum refractive index is n 1 ¼ 1. 22. For destructive interference, the antireective coating is one quarter of the wavelength of light. In this case, the thickness of the AR-layer should be 102 nm. Thus, CaF 2 is a good candidate to decrease the reection of the incident light of glass. A comparison between CaF 2 and MgF 2 shows similar optical properties. Only the solubility in water reveals that CaF 2 (16 mg L À1 ) is less soluble than MgF 2 (76 mg L À1 ). 7 This is a factor, which could be interesting for the chemical resistance of an AR-layer against outdoor conditions like rain or moisture. Due to its very low solubility, CaF 2 -lms could be more weatherproof than MgF 2 -lms. It is well known, that porous antireective alkaline earth metal uoride lms can be prepared by spin coating from sols of calcium triuoroacetate. 8 That is, calcium triuoroacetate sols in water/isopropanol are rst prepared by reacting tri-uoroacetic acid and calcium acetate. From the sols obtained this way a surface layer of calcium triuorocacetate can be created by spin coating. The following annealing initiates the thermal decomposition of the triuoroacetate resulting in a porous calcium uoride layer as reported in J. Sol-Gel Sci. Technol. 9 The disadvantage of this method is the thermal decomposition of the metal triuoroacetates into toxic and corrosive reaction products like CF 3 COF, COF 2 and HF. Another point is that a mixture of metal acetate and triuoroacetic acid may yield metal oxide uoride phases leading to an increase of the refractive index due to the oxygen content inside the lm.
In this paper, we present an easy synthesis approach toward nanoscaled CaF 2 -sols via the uorolytic sol-gel process to fabricate nally antireective CaF 2 -lms by dip-coating. For excellent AR-layers, it is necessary to start from a transparent sol with particle sizes of about 10 to 20 nm or smaller. The choice of the solvent is an important factor because only volatile solvents like e.g. methanol, ethanol, and isopropanol lead to homogeneous layers.
Here, we report a way to use the cost-efficient synthesis of calcium uoride sols with commercially available calcium chloride precursors by reaction with anhydrous HF. The inuence of TMOS to the calcium uoride sol is investigated as well as the optical and mechanical properties of the calcium uoride lms obtained from thus prepared CaF 2 -sols.

Precursor synthesis
All chemicals for the synthesis of calcium uoride are commercially available and need no drying or further processing. The 19.84 M HF-solution was prepared by dissolving anhydrous HF in ethanol.
A series of under-stoichiometric calcium uoride sols with uorine to calcium ratios F : Ca < 2 was prepared by dissolution of 1.33 g anhydrous CaCl 2 (96% ABCR, 12 mmol) in ethanol (99.8%, ROTH, 0.4 M solution). Under vigorous stirring at ambient conditions, the required amount of HF-solution was added dropwise to the solution.
The stoichiometric calcium uoride sol (Ca : F ¼ 1 : 2) for coating on glass slides (Borosilicate) was prepared by the reaction of 8.88 g of CaCl 2 (80 mmol) with 8 mL of HF (160 mmol; 2 eq.) in 192 mL ethanol. Aer uorination 0.59 mL of TMOS (5 mol% of Ca) respectively 0.88 mL of TEOS (5 mol% of Ca) was added. The inuence of TMOS was studied by 19 F NMR.

Deposition of CaF 2 coating lms
A typical procedure for the deposition of calcium uoride coating lms was as follows. The freshly prepared calcium uoride sols were deposited on borosilicate glass by dipcoating. Before the coating experiment the glass slides were cleaned in an alkaline solution (RBS®50) for 15 min. The substrates were rinsed with deionized water and dried by blowing compressed air. Aer the coating, the substrates were calcined in a vented air-oven (Barnstead Thermolyne type F48000) at 500 C for 15 min followed by slow cooling down to room temperature.

Materials characterization
The calcium uoride sols were characterized by dynamic light scattering (DLS) and 19 F liquid NMR spectroscopy. The DLS measurements were performed using a Zetasizer Nano ZS (Malvern instruments, Worcestershire, UK) using disposable PMMA cuvettes. Hydrodynamic diameters were calculated from the correlation functions by the Malvern Nanosizer Soware. The viscosity was determined simultaneously to DLS measurements with a microviscometer from Anton Paar (AMVn, Graz, Austria) at 25 C. The 19 F NMR spectra of the sols with varying uorine content were carried out using a Bruker AVANCE II 300 (liquid state NMR spectrometer with a Larmor frequency of 282.4 MHz). The 19 F isotropic chemical shis are given with respect to the CFCl 3 standard.
The Transmission Electron Microscope (TEM) analysis has been carried out using a Philips CM200 LaB 6 microscope operating at 200 kV. A few drops of the solution containing the nanoparticles have been deposited on a carbon-coated copper grid and let them dry prior to the inspection.
The calcium uoride xerogels were characterized by an X-ray powder diffractometer from Seifert (XRD 3003 TT) and by 19 F solid state NMR spectroscopy. The static 19 F solid state NMR spectra were recorded with a Bruker AVANCE 400 (solid state spectrometer, Larmor frequency of 376.4 MHz) with a 4 mm Bruker probe and the 19 F MAS NMR spectra were recorded with a 2.5 mm Bruker probe. The 19 F isotropic chemical shis are given with respect to C 6 F 6 as secondary standard with d iso ¼ À166.6 ppm against CFCl 3 . Rotor-synchronized Hahn spin-echo experiments (rs-echo) were performed which react sensitive on homonuclear dipolar couplings. This means that uorine species with an effective spin exchange (short spin-spin relaxation times T 2 , i.e. well-bridged species) are not or with decreased intensity detected aer applying longer dipolar evolution times, whereas those with longer spin-spin relaxation times (not well-incorporated, terminal) are detected. The number of rotor periods before echo detection (L0) is given in the gure captions. The percentage of chloride in the xerogels was determined by elemental analysis.
The surfaces of calcium uoride lms were examined by high-resolution scanning electron microscopy (HR-SEM) from Zeiss (Supra 40). The refractive index n (500 nm) and thickness t of the CaF 2 -layers was determined by spectroscopic ellipsometry. The ellipsometric parameters psi and delta were measured with a SENpro instrument (Sentech, Berlin, Germany) with the soware "SpectraRay3". The calculated parameters psi and delta are determined with a Sellmeier model. The reection and absorption of the CaF 2 -lm was measured with an UV-Visspectrometer (Shimadzu UV-3100, Kyoto, Japan), in the range of 300-1400 nm.
The mechanic stability of the CaF 2 -lms was tested by a crockmeter from Erichsen (scratchmarker 249) using felt and steel wool with a neness of 0000. The stamp with a contact area of 4.5 cm 2 was pressed on the sample with a force of 4 N.

Results and discussion
The uorolytic sol-gel synthesis consists of the reaction between a suitable metal precursor and anhydrous hydrogen uoride in a suitable organic solvent, preferentially methanol or ethanol. 10 Several calcium precursors like Ca(OMe) 2 , Ca(OEt) 2 , Ca(OAc) 2 and CaCl 2 were investigated. In most of the cases, metal alkoxides like Si(OMe) 4 , Ti(O i Pr) 4 and Al(O i Pr) 3 or Mg(OMe) 2 , respectively, are used in the classical oxide sol-gel route. [11][12][13][14] In case of the uorolytic sol-gel synthesis, the synthesis of transparent CaF 2 sols starting from calcium alkoxides is not recommended because of the insolubility of the calcium alkoxide in methanol and ethanol at room temperature. Thus, Ca(OMe) 2 and Ca(OEt) 2 always form crystalline white precipitates of their Ca(OR) 2 $ 4 ROH solvates. By subsequent de-solvation ne crystalline powders with a threedimensional polymeric structure of the CdI 2 type are formed. 15 Unfortunately, it is impossible to obtain transparent CaF 2 -sols under these circumstances; however, starting from calcium acetate suspended in an ethanol/acetic acid solution results in the formation of transparent CaF 2 -sols but the stability of these sols is poor. That means, aer a few days the CaF 2 -sol is completely transformed to a gel. The reason for strong gelation tendency is esterication of acetic acid with the corresponding alcohol and the release of water. The more water is formed in the system the faster gelation occurs (eqn (1)).
In spite of the gelation, coatings from these sols are possible for a short time only. The optical properties of such coatings are good but the mechanical properties are very poor due the high porosity in the lm. 16 Thus, calcium acetate as precursor for calcium uoride coatings is not recommended. Therefore, we decided to use calcium chloride as precursor instead because of its good solubility in alcohol and the absence of an esterication reaction. Following the general synthesis approach of MgF 2 -sols via the uorolytic sol-gel route by reacting MgCl 2 in EtOH with anhydrous HF, 17 we started from CaCl 2 in EtOH and reacted it with ethanolic HF solution to form nanoscopic calcium uoride sols (eqn (2)).
The transparency of the CaF 2 -sol is opaque. By addition of 5 mol% tetramethyl orthosilicate (TMOS) or tetraethyl orthosilicate (TEOS) aer uorination the former opaque sol turned into a transparent CaF 2 -sol. The reaction of CaCl 2 in ethanol with two equivalents of ethanolic HF solution leads to the formation of nanoscopic CaF 2 particles. For good optical and mechanical properties of an AR-layer the sol should be transparent to achieve a homogenous layer on the substrate. Beside sol-transparency as an indication for homodispersed nano-particle distribution, volatility of the solvent is a further pre-condition to obtain homogeneous coatings. The higher the volatility of the solvent is, the faster a homogenous lm is formed. Although methanol is more volatile than ethanol, it is toxic compared to ethanol. Hence, the latter was used in all experiments. The formation of calcium uoride nanoparticles was investigated by 19 F NMR spectroscopy. The amount of HF to calcium chloride was varied from 0.1 : 1 to 2 : 1 in order to follow species that might be formed in the course of reaction. The 19 F NMR spectra of the sols with different amount of HF are listed in Fig. 1. For a better illustration of the spectra without the known background signal in the 19 F liquid NMR, the samples were measured statically by solid state NMR. The liquid 19 F NMR spectra are included in the ESI. † Fig. 1 shows the typical signal for calcium uoride at À108 ppm. Additionally to the CaF 2 signal a second broad signal appears at about À95 ppm and À93 ppm in the 19 F liquid NMR, respectively. We speculated, it may stand for another calcium uoride species like calcium chloride uoride (CaClF) or calcium alkoxide uoride (CaF 2Àx OR x ), which could be a solvated species. For further information of this species 19 F solid state NMR and XRD measurements of the xerogels were performed. Furthermore, the reaction of CaCl 2 with two equivalents of HF exhibits a broad signal in the static solid state 19 F NMR as well as in the liquid 19 F NMR at about À180 ppm (line width $2000 Hz). Samples with under-stoichiometric amounts of uorine also contain a signal between À174 ppm and À180 ppm in the liquid NMR. This signal corresponds to unreacted HF adsorbed to the particle surface. 18,19 A similar observation was made in the synthesis of MgF 2 via the uorolytic sol-gel route, with either Mg(OMe) 2 or MgCl 2 as reaction. Thus, we speculate that this corresponds to HF adsorbed at the precursor. 17 During ageing of the MgF 2 -sol the signal vanishes.
The addition of TMOS or TEOS to the CaF 2 -sol vanishes this signal completely in a short time. The formation of SiOR 4Àx F x and SiF 6 2À species (the second broad signal around À128 ppm could be CaSiF 6 (Fig. 2)) aer TMOS/TEOS addition are further evidences for unreacted HF in the sols at this stage of nanoparticles surface causes agglomeration, but consequently, removal of this leads to clear sols. Due to the line width of $4000 Hz the sol consists not only of calcium uoride particles but also of calcium hexauorosilicate or other uorosilicate  particles. The other small signals with line width of about 100 Hz are probably dissolved SiOR 4Àx F x species. Both species evidently are formed from unreacted HF and added Si(OMe) 4 . All as-prepared CaF 2 -sols with 5 mol% TMOS or TEOS are transparent and possess low viscosity. Without addition of TMOS or TEOS the sol remains turbid and tends to form a gel. Aer a few days the particles are aggregated, that a settle to the bottom of the ask is observed, which is referred to as sedimentation of the sol. According to Fig. 3 the mean particle diameter of a sol with 5 mol% TMOS was determined by dynamic light scattering. The intensity weighted particle distribution shows two size classes with mean particle diameters of 20 nm and 320 nm. Due to the Stokes-Einstein relation bigger particles scatter much more than smaller particles because the intensity of scattering is proportional to the sixth magnitude of its diameter (Rayleigh approximation). It follows that small particles are signicantly under-estimated. Hence, calculating the volume weighted particle distribution from intensity weighted one shows that in fact just one class of particles with a mean particle diameter of approximately 10 nm is present. In addition, the particle size was also investigated by transmission electron microscopy and results are included in the ESI. † For the identication of species with the characteristic 19 F NMR signal between À93 to À95 ppm 19 F solid state NMR measurements were performed. For an exact identication of the unknown species, the CaF 2 -sol without the addition of TMOS was dried at room temperature in air and calcined at 600 C in an electric furnace. The samples were compared with crystalline CaClF, which was synthesized by solid state reaction between CaCl 2 and CaF 2 in presence of NH 4 Cl as ux at 730 C for 3 h in an electric furnace. Wenz et al. reported an eutectic point at 645 C for a composition of 18.5 mol% CaCl 2 and 81.5 mol% CaF 2 . 20 Fig. 4 shows the 19 F MAS signal of the crystalline CaClF compound with a chemical shi of À83 ppm. The difference between the CaClF shi and the shi of the unknown species is about 12 ppm. Hence, the unknown species does not correspond to CaClF. The most plausible explanation is that this signal corresponds to an oxide uoride species, which probably is caused by a preferential oxygen donation in the second coordination sphere as was also found for the MgF 2 system. 18 For comparison, the annealed CaF 2 xerogel with one equivalent of HF shows a decrease of the signal at À95 ppm and an increase of the signal at À83 ppm in the spectrum unlike the un-annealed sample (Fig. 5). Thus, the formation of CaClF in the sol for the system CaCl 2 in EtOH can be ruled out, it will only be formed at high temperature. Furthermore, the 19 F signal at about À95 ppm also appears in the NMR spectra of CaF 2 -xerogels that have been prepared from other calcium precursors like CaBr 2 $ H 2 O and CaO which do not contain any chloride. The NMR spectra are listed in the ESI (Fig. S1-S3 †). In the 19 F ss-NMR spectrum of the CaF 2 -xerogel obtained from CaO with two equivalents of HF appears the corresponding signal as shoulder of the main signal of CaF 2 . The difference between the signal shapes in the CaO-HF system and the CaCl 2 /CaBr 2 -HF system might be caused by dipole-dipole coupling. The shorter the dipole-dipole coupling is, the more is the signal separated from the main signal. Another interesting fact is also the different Fig. 3 Hydrodynamic diameter of particles of CaF 2 -sol with 5 mol% TMOS measured by dynamic light scattering (intensity and volume weighted particle distribution).  relaxation time T 2 between the signal at À108 ppm and À95 ppm in the 19 F NMR spectra. According to Fig. 6 the inuence of the relaxation time T 2 of under-stoichiometric xerogels reveals that the intensity of the signal at À108 ppm decreases and the intensity of the signal at À95 ppm increases. Another timedependent NMR measurement was also performed with a CaF 2xerogel derived from CaCl 2 with H 2 O as solvent (Fig. 7). In the case of H 2 O as solvent the 19 F NMR spectra exhibit another signal at À83 ppm beside the main signal and the signal at À95 ppm. This signal at À83 ppm corresponds to CaClF, which has the same NMR shi like the reference sample of CaClF. It is obvious that the 19 F signal at about À95 ppm does not correspond to the CaClF species but rather to an oxide uoride species like CaF 2Àx OR x (R ¼ H, Et). An explanation of the formation of an oxide uoride species from CaCl 2 in EtOH or in H 2 O with ethanolic/aqueous HF-solution as uorine agent could be either the preferential oxygen donation in the second coordination sphere like mentioned before or the use of only 96% CaCl 2 as precursor, which includes impurities like Ca(OH) 2 or CaClOH. These impurities are caused by the manufacturing process and subsequent dehydration of CaCl 2 $ xH 2 O. 21 Thus, it could be possible that aer dehydration of CaCl 2 $ xH 2 O such oxygenated species are still present. If a certain amount of oxygen is incorporated in the CaF 2 -lm, the refractive index of the lm will increase, and thus, the optical properties of such lms decline. However, the percentage of oxygen is not only unwanted in CaF 2 -lms but also in ceramics due to the loss of laser oscillation as a result of the scattering at the grain boundaries, which occurs in the ceramics. 4 It is known that CaF 2 nanocrystals prepared by CaCl 2 in EtOH with NH 4 F as uorine agent can be also obtained via co-precipitation and hydrothermal synthesis. 22 However, a certain amount of oxygen is present in the CaF 2 powder, which occurs from the evaporation of the precipitate. Apparently, the percentage of oxygen in the sample is induced by the drying process. Unfortunately, no 19 F MAS NMR spectra are listed in this publication, so that a comparison with the signal at À95 ppm of our shown spectra is not possible. The reections in the X-ray pattern of the xerogels in Fig. 8 reveal CaF 2 and CaClF. In the diffractograms as well as in the 19 F NMR spectra, it is notable that the CaClF species is formed at 600 C. Xerogels prepared from CaCl 2 as precursor with just one equivalent of HF show aer annealing reections of CaClF (rorisite, PDF 24-0815) and CaF 2 (uorite, PDF 35-0816). In contrast, un-annealed as well as annealed xerogels obtained from stoichiometric Ca 2+ to HF ratio show reections of CaF 2 only. Obviously, the nominal uorine stoichiometric content in the reaction of CaCl 2 with HF is crucial whether CaClF or CaF 2 is formed. The powder diffractograms show that CaF 2 is more crystalline in comparison to MgF 2 . The crystallite sizes of the un-annealed CaF 2 -xerogel derived from the Scherrer equation is   about 17 nm. In contrast, the crystallite sizes of a MgF 2 -xerogel annealed at 300 C is only about 11 nm. 17 Moreover, the chloride concentration of the xerogels, which is given in Table 1, reveals that chloride is still present. It is apparent that un-annealed as well as annealed xerogels with a stoichiometric Ca 2+ to HF ratio still have a small chloride percentage. It shows that the uorination of CaCl 2 is not complete. In case of one equivalent of HF the discrepancy of the experimentally found percentage of chloride in un-annealed and annealed xerogels is more pronounced. A plausible explanation for this effect is that the un-annealed samples contain a certain amount of water, which is caused by the hygroscopic nature of CaClF. 23 It is evident that the deviation from the experimental to the theoretical chloride percentage results from the moisture content of the environment. CaF 2 -sols were used for the deposition of CaF 2 -lms on borosilicate glass with subsequent thermal treatment at 500 C for 15 min. Films prepared from sols with stoichiometric uorine content (F À /Ca 2+ ¼ 2) were characterised regarding their optical and mechanical properties. The optical data varied depending on the CaF 2 -sol synthesis and post-treatment of the coated layer. E.g., the refractive index, determined by spectroscopic ellipsometry, of a 165.50 nm thick CaF 2 -lm on a siliconwafer and a post-calcination at 500 C for 15 minutes was 1.37 at 500 nm. For CaF 2 -coatings on glass the reection and absorption of the deposited CaF 2 -lm were determined by UV-Visspectroscopy. As can be seen in Fig. 9, the reectance at 600 nm reached almost nearly 0% and the absorbance of the layer between 400 and 900 nm is about 0% as well. Transmittance measurements of the CaF 2 lm and the glass substrate were not performed for this glass sample. However, AR-layers on oat glass samples from different batches showed transmission values at 600 nm repeatedly between 98.4 to 98.7%. Hence, we assume similar transmissions for this sample. Hence, the layer absorbance of CaF 2 which is given in Fig. 9 holds for the transmittance of the CaF 2 lm on glass. It is the difference between the absorbance of the glass substrate and the absorbance of the glass substrate with CaF 2 lm. To illustrate the morphology of the lm, the cross-section image of the CaF 2 -lm derived from high resolution scanning electron microscopy (HR-SEM) is shown in Fig. 10. The measured thickness of the lm is approximately 170 nm, and thus, corresponds very well to that determined by ellipsometry. This is in good agreement with the conditions of antireective coatings. The diameter of the particles is between 10 and 30 nm. This corresponds well with the crystallite sizes of the xerogel determined by XRD. Furthermore, the lm is very porous due to the pores, which completely proceed through the whole lm. Apparently, if chloride is still present in the lm, a certain amount of CaClF might be formed. Hence, the sintering process of a CaClFcontaining CaF 2 -lm might be rendered, which is reected by the high porosity as well mechanical abrasion (Fig. 11). Another fact causing creation of porosity could be also the release of the organic solvent together with HCl formed by the reaction of HF with CaCl 2 . Thus, these pores lled with air (n 0 z 1.0) are the reason for the low refractive index of the CaF 2 -lm. Similar porous coating lms are well known for SiO 2 -lms on glass. 25,26 Since CaF 2 exhibits a lower refractive index than SiO 2 , it was expected that CaF 2 -AR-layers with lower porosity but equally good refractive indices possess higher mechanical stability. Therefore, mechanical abrasion of the lms was tested too. For that purpose, the CaF 2 -lms annealed at 500 C were subjected to the Crockmeter test. Fig. 11 is showing the CaF 2 -lm before and aer abrasion with felt and steel wool. Aer 100 cycles with felt and 25 cycles with steel wool the lm has only a slight scratch track. Under these conditions, the lms are almost   robust against abrasion. Just the very surface near region seems to show some little damages without losing the optical performance. There is no visual difference in the abrasion with felt or steel wool. In conclusion, the mechanical stability of CaF 2 -lms is better than that of other CaF 2 -lms, which were derived from calcium acetate as mentioned above. All in all, the CaF 2 -lms show not only excellent optical performance, but show at the same time also good mechanical stability.

Conclusions
The synthesis of CaF 2 -sols containing nanoscopic particles was successfully performed by reaction of CaCl 2 as inexpensive precursor with anhydrous HF in ethanol. Addition of up to 5 mol% TMOS or TEOS aer uorination always causes a rapid formation of a transparent sol with small particle diameter in the range of 10 to 20 nm. This coating solution is stable for several weeks. No sedimentation was observed. TMOS not only causes a fast clearing up of the sol but also stabilizes the CaF 2particles in the sol. Apparently, addition of TMOS leads to a deagglomeration of the particles probably due to the reaction with the traces of un-reacted HF in the reaction system. We speculate that the electrostatic repulsion of the particles is changed due to the formation of Si(OR) 4Àx F x -specieseven the formation of SiF 4as intermediate product cannot be ruled out. Since these are strong Lewis acids, the formation of a CaSiF 6 species might be favoured. Further investigations are necessary to study the real rule of TMOS in the CaF 2 system. We also evidenced the formation of CaClF at 600 C beside the formation of an oxide uoride species during the uorination. Interestingly, the formation of CaClF is also favoured by using H 2 O as solvent with two equivalents of aqueous HF-solution and subsequent drying at room temperature. The observation for the oxide uoride species was not only recorded for CaCl 2 as precursor but also for other calcium precursors like calcium bromide or calcium oxide. The percentage of oxygen in the samples could be explained by dehydration of CaCl 2 $ xH 2 O. It is also known that CaF 2 nanocrystals derived from CaCl 2 in EtOH with NH 4 F, which are prepared by co-precipitation and solvothermal technique, possess impurities of oxygen too. That could be a problem for laser applications because it leads to optical scattering losses, and hence, prohibit laser oscillation. Hence, it is difficult to completely exclude oxygen in CaF 2 nanocrystals by sol-gel as well as other synthesis techniques. The CaF 2 -lms exhibit excellent antireective properties. The mechanical resistance of such lms compared to CaF 2 -lms, which were derived from calcium acetate, is signicantly better. Only corresponding MgF 2 -lms show marginal better mechanical resistance. It is worth noting that a further improvement of the mechanical properties of these antireective layers is essential for applications, where a mechanical resistance is indispensable. However, due to the lower solubility as compared to MgF 2 , CaF 2 is an interesting alternative candidate for the manufacture of antireective coatings. Fig. 11 Photographs of the CaF 2 -film on borosilicate glass substrate (20 Â 4 cm 2 ) before and after abrasion with 100 cycles of felt and 25 cycles of steel wool by Crockmeter test. The sample has been treated at 500 C for 15 min prior the analysis.