Accepted Manuscript: for final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

1 Simultaneous chemical vapor deposition and thermal strengthening 2 of 3 4 Peter Sundberg1, Lina Grund Bäck1, Robin Orman2, Jonathan Booth2, Stefan Karlsson1* 5 1RISE Research Institutes of Sweden, Built Environment Division, Glass Section, SE-351 96 Växjö, 6 Sweden 7 2Johnson Matthey Technology Centre, Blounts court, Sonning Common, Reading, RG4 9NH, United 8 Kingdom 9 *Corresponding author: [email protected] 10 11 Keywords: Chemical vapour deposition, thermal strengthening, crack resistance, contact angle, 12 hardness 13 14 Declaration of interest: none. 15 16 Abstract

17 In the current paper we present a concept combining metal organic chemical vapor deposition with 18 thermal strengthening process of flat glass. As the flat glass is heated to be thermally strengthened, 19 which takes a few minutes, there is an opportunity for performing a surface modification. We describe 20 the application of transparent and amorphous Al2O3 thin films during the thermal strengthening process. 21 Al2O3 was chosen due to the following desirable properties: increased surface mechanical properties 22 and increased chemical durability, the latter has not been investigated in the current paper. The residual 23 surface compressive stresses after performed strengthening of the coated were quantified to 24 be in the range of 80-110 MPa. The Al2O3 content in the surface was measured using the Surface 25 Ablation Cell employed with Inductively Coupled Plasma Atomic Emission Spectroscopy and found to 26 be at least doubled at the surface and having an increased Al2O3 content at least 0.5 µm underneath 27 the glass surface. During the surface reaction, sodium is migrating to the surface giving a hazy salt layer 28 on the glass which can easily be washed off with water. The applied coatings are transparent and 29 provide increased surface hardness and crack resistance at low indentation loads. At higher indentation 30 loads the interaction volume is larger and displays the same effect on the surface mechanical properties 31 as for thermally strengthened glass. The contact angle with water compared to annealed is 32 significantly increased from 5° to 45° due to the different surface chemistry and surface structure.

33 1. Introduction

34 Glass is an essentially strong material. It is evident from its strength calculated from the bonds forming 35 the vitreous network or its strength under compressive loads [1]. However, due to unavoidable surface 36 defects, glass is brittle for tensile loads. The two most common methods of strengthening glass involve 37 pre-stressing the glass surface with compressive stresses by essentially either physical or chemical 38 means. Physical strengthening is more commonly called thermal strengthening (or tempering). 39 Tempered glass is made by uniformly heating glass to a temperature of up to 700°C and immediately 40 cooling down [2]. The hot glass undergoes a rapid cooling process by a uniform and simultaneous blast 41 of air on both surfaces. This gives a solid surface that contracts lesser than the interior that still has a 42 temperature above the which upon further cooling results in a permanent parabolic 43 profile [3]. Chemical strengthening of glass most often means exchanging smaller ions in glass 44 surface for larger ions by a molten salt bath [4] or by a spraying process [5]. It generally gives higher 45 strength than thermally tempered glass but requires more time (hours instead of minutes), the upscaling 46 is complicated, and therefore comprise a higher cost. Thermal strengthening of glass has in general two 47 major drawbacks, the frangibility (the explosive behavior upon fracture) [1] and spontaneous fracture 48 due to NiS inclusions [6]. Thinner thermally strengthened glass is possible to be made by using

1(7) For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

1 increased rapidity in the cooling [7]; however, at some point thermally strengthened glass also needs 2 increased surface mechanical properties. Therefore, it can be very favorable to combine thermal and 3 chemical strengthening in the same process. The traditional chemical strengthening is though not 4 possible to combine with thermal strengthening [8].

5 There are also other methods to improve the strength of glass as well; these can also be categorized 6 as chemical methods to strengthen glass and involves gases that react with the glass surface. Some of 7 these methods are summarized in Tab. 1. To be able to thermally strengthen we require a heating cycle 8 as shown in Fig. 1. At temperatures as high as 600-700 °C there is an opportunity for enhancing the 9 surface mechanical properties. Rancoule describes benefits and problems of adding gaseous SO2 10 during the tempering process of float glass [17]. The formation of sodium sulfate is said to “benefit for 11 the contact surface longevity” i.e. being a lubricant [10] but can also lead to defects [17]. The yield of 12 the sodium sulfate depends on moisture, the temperature and the concentration of the gas during the 13 process. No details of the SO2 additions were given in the cited references. In the current paper we 14 investigated combining chemical vapor deposition for providing amorphous Al2O3 thin films during the 15 thermal strengthening process [18, 19]. Amorphous Al2O3 thin films provide various desirable properties 16 including improved chemical durability and surface mechanical properties [18-20]. 17 18 2. Experimental

19 Conventional soda-lime-silicate float glass in the size of 70×70 mm2 with thicknesses of 4 and 2 mm 20 were used with a typical composition as given in ref [21]. The general experimental procedure is given 21 in Fig. 2. The flat glass was heated in the furnace up to 640-680 °C. At higher temperatures the glasses 22 became non-flat and at lower temperatures the strengthening level became too low due to insufficient 23 difference between the temperature of the surface and the interior upon cooling. For general guidance 24 about thermal strengthening in experimental details please have a look in Barr’s Handbook [22]. The 25 Al2O3-precursor was Al-(acac)3, from Strem Chemicals, which is a commonly used precursor for 26 Metalorganic Chemical Vapor Deposition (MOCVD) [19]. The Al2O3-precursor was partly dissolved in 27 isopropyl alcohol (IPA) at elevated temperature (about 70 °C) using a magnetic stirrer with the 28 proportions of approximately 3.5:96.5 weight ratio of Al2O3-precursor and IPA respectively. The solution 29 was then sprayed manually and as homogeneous as possible on to the grating made of stainless steel 30 using a Preval sprayer at the constant spray pressure 4.4 bar and driven by the propellant gases, 31 dimethyl ether, propane and isobutane. The spray system is an aerosol-based spray system that allows 32 custom-made solutions to be turned into sprayable material. The gratings were then dried in a heating 33 cabinet at 105 °C (Thermoscientific’s Heratherm oven OMH100) making the IPA to evaporated and only 34 leaving deposited Al2O3-precursor. The gratings’ difference in weight was recorded before and after the 35 spray and drying process so that the amount of applied Al2O3-precursor on the grating could be 36 determined. The amounts of applied Al2O3-precursor on the gratings, in dried state, were in the range 37 of 100-200 mg. The size of the gratings was 10x11 cm2 for each side of the flat glass so approximately 2 38 0.5-1 mg/cm Al2O3-precursor was deposited on the grating. The grating together with an untreated float 39 glass sample were then put into a hot (640-680 °C) muffle furnace (Naber Industrieofen N2OH) with 40 normal air atmosphere as illustrated in Fig. 2A, so that the flat glass on both sides was located <1 cm 41 from the grating. In the furnace the glass and the grating heats up, the glass surface reaches the furnace 42 temperature after approximately 5 minutes, the Al2O3-precursor vaporizes creating an Al-(acac)3/air 43 atmosphere. The Al2O3-precursor starts to evaporate above its boiling temperature 189 °C [19]. As the 44 Al2O3-precursor reaches the hot glass surface it immediately decomposes and forms an Al2O3 thin film. 45 Please note that all Al2O3-precursor do not end up forming a thin film on the glass. The glass and the 46 grating were let in the furnace for 20 minutes in total as was predetermined from thermal strengthening 47 tests. The reaction was verified to be completed by weighing the grating afterwards. After 20 minutes 48 the sample and grating were removed and placed into a stream of air having an ambient temperature, 49 see Fig 2B. The air flow generators were leaf blowers from Ryobi (3 kW) providing an air flow of 16 50 m3/min. The rapidly cooled glass reach room temperature within 10-20 seconds. The weight change of 51 the samples was not recorded as the weight change of the glass could not only have been prescribed

2(7) For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

1 to be due to the thin film deposition but also sodium migration to the surface (see results section). The 2 thermal strengthening level of the tempered samples was measured using SCALP-03, GlasStress Ltd 3 [23]. All samples were cleaned with deionized water prior to all type of characterizations and in all 4 characterizations was the air-side investigated. 5 6 The UV-Vis transmittance spectra were measured using the spectrophotometer Lambda 25 from Perkin 7 Elmer. A 1 nm slit width, a 2 nm data interval and a scan rate of 480 nm/min was employed for the 8 spectral range 350 to 1100 nm. 9 10 The Scanning Electron Microscopy (SEM) images were acquired using a Zeiss ultra 55 Field Emission 11 Electron Microscope equipped with in-lens secondary electron and backscattered detectors. The 12 compositional analysis and the low-resolution general imaging were acquired using an accelerating 13 voltage of 20 kV, a working distance of 7.3 mm. 14 15 The surface chemical composition of a treated glass was studied using Surface Ablation Cell (SAC) [24] 16 employing Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) to quantify the glass 17 constituents dissolved. The experimental procedure is principally the same as outlined in [25] but 18 adjusted to give a better depth resolution using a less concentrated acid mixture.

19 The surface hardness was measured using an Anton Paar Nanoindenter (model NHT2) and the 20 indentation hardness was determined using the Oliver and Pharr method [26]. For each reported 21 indentation hardness value at least 20 indents were made within an area of 1 mm2 (apart from 1 mN 22 where 30 were made); outliers were removed giving an <6% error of the resulting values. The crack 23 resistance (CR) was measured using CSM Instruments Micro-Combitester and was determined 24 according to the description in [27] performing at least 15 indents. The uncertainty of the CR method is 25 lower than the determination of indentation toughness, for a discussion see ref [28]. The contact angle 26 with deionized water was recorded using a pocket goniometer (PGX+) from FIBRO System AB. The 27 optical profilometry measurements were performed using a Veeco instrument operating in the phase 28 shifting interferometry (PSI) mode with a magnification of 2.7x. The vertical resolution of the PSI 29 measurements is about 0.1 nm and the error of the root mean square data about 10%.

30 3. Results

31 The applied Al2O3 films are transparent, see Fig. 3, and believed to have an amorphous structure in 32 accordance of what is reported by others [12, 29]. During the thermal strengthening some of the alkali- 33 species leaves the glass due to its volatile nature at these temperatures (640-680 °C). We suppose that 34 this process is enhanced by the deposition of the Al2O3 thin film similar to the results shown by Fonné 35 et al. [30]. The outward diffusion of the alkali is shown by as a white haze on the glass, see left inset of 36 Fig. 4A. The white haze is easily washed off with deionized water and it was chemically quantified to 37 contain mostly Na2O. As shown in Fig. 4A the Al2O3 concentration is approximately increased in the 38 surface by a factor of 2 compared to bulk glass concentration. The profile of Al2O3 show a diffusion-like 39 profile and the diffusion coefficient was estimated using Green’s function [25] having an effective 40 diffusion coefficient in the range of.2×10-13 to 3×10-14 cm2/s. The polariscope photo taken of two in-situ 41 CVD thermally strengthened glasses with deposited Al2O3 thin film, hereinafter referred to Al-CVD, show 42 a similar pattern as commercially tempered glasses. The Na2O concentration profile also shows 43 indication of having a bit lower concentration at the outmost surface, see Fig. 4B.

44 Repeated Scattered Light Polariscope (Scalp) measurements showed that the surface compressive 45 stresses of the Al-CVD thermally strengthened glasses was in the level of 80 to 110 MPa. Conventional 46 thermally strengthened glass is in similar range [31]. The surface hardness, as measured by 47 nanoindentation, was determined by measuring the hardness and the maximum contact depth (hm) for 48 different loads, see Fig. 5. The trend of the hardness curves (the three upper curves) with load reveals 49 discrepancies at low loads (1-5 mN). The annealed float glass shows increasing hardness with

3(7)

For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

1 increasing load. The thermally strengthened and the Al-CVD thermally strengthened show a minimum 2 in the hardness at the load of 5 mN. At the loads 1 and 3 mN the Al-CVD thermally strengthened glass 3 show a significantly higher hardness compared to the thermally strengthened glass. This can be 4 explained by the fact that the Al2O3 thin film gives a higher hardness than the glass. We believe that the 5 reason for a small discrepancy between the annealed float glass and the thermally at low loads is due 6 to the fact that some Na2O evaporates during the thermal strengthening process.

7 Using microindentation it is possible to study the crack initiation behavior. We have used the 8 methodology of CR which is based on statistical data and a sigmoidal curve-fitting, see Fig. 6. The 9 annealed and the thermally strengthened glass gave quite similar statistics at lower loads but at higher 10 loads these were significantly different. The combined Al-CVD thermally strengthened glass clearly 11 show a significant difference at lower loads but show a similar value for 100% PCI as the thermally 12 strengthened glass. The reported CR values are given from the fitted curve at 50% PCI. The significant 13 differences at low loads give the resulting difference of CR: 0.7 N for annealed, 0.8 N for thermally 14 strengthened and 1.3 N for combined Al-CVD thermally strengthened float glass. The significant 15 differences at low loads are caused by the deposited Al2O3 thin films. However, the thin films are so thin 16 that at higher loads the bulk glass interaction overshadows the effect of the thin film. This resembles the 17 results presented in Fig. 4 as well.

18 Al2O3 thin films are known to give a higher contact angle than soda-lime-silicate (SLS) glass e.g. see 19 following ref for superhydrophobic Al2O3 thin films [32, 33]. The contact angle of float glass i.e. SLS 20 glass is about 5° immediately after its been cleaned [34] but generally increases with time. The silicate 21 network often forms Si-OH species at the surface which gives the hydrophilic behavior. We investigated 22 the wetting behavior of annealed float glass and Al-CVD thermally strengthened glass, see Fig. 7. 23 Analogous to its normal behavior the float glass shows a contact angle vs. water of about 5° while the 24 Al-CVD thermally strengthened glass shows an average contact angle of 45°. The high difference for 25 the latter is likely to be due to the surface topography and surface chemistry in addition to the general 26 error of contact angle measurements. It can easily be explained by the different surface chemistry and 27 possibly surface topography provided with the Al2O3 thin film. Two exemplary images of the surface 28 topography are provided in Fig. 8 and Fig. 9. They show quite similar pattern, so the surface topography 29 cannot be ruled out. The root mean square (Rq) data give a value of 0.8 nm for the Al-CVD toughened 30 glass compared to 1.3 nm error of the root mean square data is about 10% over the surface taken as 31 an average of two measurements of two samples. However, surface topography cannot be ruled out to 32 have an impact on the contact angle data shown in Fig. 6, which is one of the reasons for the deviation 33 of the contact angle data. The heat-treatment could possibly affect the contact angle as well but a 34 previous study with heat-treatments in the range of 400-600 °C showed that SLS glass still gives a low 35 contact angle [34], about 5°.

36 4. Discussion

37 There are many questions still to be answered and more samples to be investigated e.g. the 38 homogeneity of the thin films has not been investigated so far. The method is not optimized but providing 39 a concept approach to improve thermally strengthening process. However, the present results give a 40 proof of the innovation of a process that potentially can increase the surface mechanical properties of 41 thermally strengthened glass. On the other hand, it can possibly be used for giving the thermally 42 strengthened glass other functional properties such as chemical barrier coatings for providing an inert 43 surface [35], conductive coatings [36, 37], photoluminescence [38], photocatalytic [39] etc. There are 44 plenty of metalorganic substances that can be suitable for being adopted into the presented concept 45 process [11, 40]. Another interesting, but outstanding aspect, is that the applied Al2O3 thin films may 46 also impose somewhat increased surface compressive stresses due to the slightly lower thermal 47 expansion coefficient of Al2O3.

48 5. Conclusions

4(7) For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

1 In the current paper we present a concept of combined MOCVD with thermal strengthening process. As 2 the glass is heated there is an opportunity for applying a coating immediately prior being rapidly cooled 3 in the thermal strengthening process. In the current paper we have investigated the application of Al2O3 4 thin films in the thermal strengthening process and its beneficial effect on the surface mechanical 5 properties. The residual surface compressive stresses of the coated glasses were quantified to be in 6 the range same range as fully tempered glass. The Al2O3 content was quantified being at least doubled 7 at the surface and having an increased Al2O3 content at least 0.5 µm into the glass surface. During the 8 surface reaction, sodium migrates to the surface giving a hazy salt layer on the glass which can easily 9 be washed off with water. The applied Al2O3 coatings are transparent and are assumed to be 10 amorphous. They provide increased surface hardness and crack resistance, especially at low 11 indentation loads which would statistically give a positive impact on the mechanical strength. At higher 12 indentation loads the interaction volume is larger and therefor they display the same result as for 13 conventionally thermally strengthened glass. Because of the different surface chemistry and surface 14 structure the contact angle with water is significantly increased from 5° to 45° compared to annealed 15 float glass.

16 Acknowledgement

17 The authors wish to acknowledge Solar-ERA.NET as well as Swedish Energy Agency (contract no. 18 38349-1) and Technology Strategy Board (contract no. 620087) for providing funding for this research. 19 We also wish to acknowledge fruitful discussions with Prof. Lothar Wondraczek, Otto Schott Institute of 20 Materials Research, University of Jena and Dr. Paul A. Bingham, Sheffield Hallam University.

21 References

22 1. Wondraczek, L., J.C. Mauro, J. Eckert, U. Kühn, J. Horbach, J. Deubener, and T. Rouxel, Towards 23 Ultrastrong Glasses. Adv. Mater., 2011. 23(39): p. 4578-4586. DOI: 10.1002/adma.201102795. 24 2. Gardon, R., Thermal Tempering of Glass, in Glass Science and Technology vol 5 Elasticity and 25 Strength in Glasses, D.R. Uhlmann and N.J. Kreidl, Editors. 1980, Academic Press, New York. p. 26 145-216. 27 3. Karlsson, S. and L. Wondraczek, Strengthening of Oxide Glasses, in: Encyclopedia for Glass 28 Science, Technology, History and Culture, P. Richet (Ed.), John Wiley & Sons Inc.: Hoboken, New 29 Jersey, Accepted, In Press. 30 4. Karlsson, S., B. Jonson, and C. Stålhandske, The Technology of Chemical Glass Strengthening - 31 a Review. Glass Technol. – Eur. J. Glass Sci. Technol. Part A, 2010. 51(2): p. 41-54. 32 5. Patschger, M. and C. Rüssel, Strengthening of a soda-lime-silica glass by using an 33 adherent potassium salt coating. Glass Technol. – Eur. J. Glass Sci. Technol. Part A, 2016. 57(1): 34 p. 6-14. DOI: 10.13036/1753-3546.57.1.017. 35 6. Karlsson, S., Spontaneous Fracture in Thermally Strengthened Glass - A Review and Outlook. 36 Ceram.-Silikáty, 2017. 61(3): p. 188-201. DOI: 10.13168/cs.2017.0016. 37 7. Monnoyer, F. and D. Lochegnies, Heat transfer and flow characteristics of the cooling system 38 of an industrial glass tempering unit. Appl. Therm. Eng., 2008. 28(17–18): p. 2167-2177. DOI: 39 10.1016/j.applthermaleng.2007.12.014. 40 8. Zaccaria, M. and M. Overend. The mechanical performances of bi-treated glass. In: Challenging 41 Glass 4 & COST Action TU0905. Lausanne, Switzerland: Taylor & Francis Group. 2014. 42 9. Bartoue, R.D., F. Caillaud, J. Dodsworth, and J. Osele, Maximizing Ceramic Furnace Roll 43 Performance. US Glass Magazine, 1999(June): p. 33-37. 44 10. Bartoue, R.D., Maximizing Ceramic Furnace Roll Performance, in :Glass Processing Days: 45 Tampere, Finland, 1999, p. 337-341. 46 11. Sheth, N., J. Luo, J. Banerjee, C.G. Pantano, and S.H. Kim, Characterization of surface structures 47 of dealkalized soda lime silica glass using X-ray photoelectron, specular reflection infrared,

5(7) For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

1 attenuated total reflection infrared and sum frequency generation spectroscopies. J. Non- 2 Cryst. Solids, 2017. 474: p. 24-31. DOI: 10.1016/j.jnoncrysol.2017.08.009. 3 12. Vohs, J.K., A. Bentz, K. Eleamos, J. Poole, and B.D. Fahlman, Chemical Vapor Deposition of 4 Aluminum Oxide Thin Films. J. Chem. Educ., 2010. 87(10): p. 1102-1104. DOI: 5 10.1021/ed100391p. 6 13. Hessenkemper, H. and H. Landermann-Hessenkemper, Alkaline glasses with modified glass 7 surfaces and process for the production thereof. 2007, US Patent 20070141349 A1. 8 14. Dériano, S., T. Rouxel, S. Malherbe, J. Rocherullé, G. Duisit, and G. Jézéquel, Mechanical 9 strength improvement of a soda-lime–silica glass by thermal treatment under flowing gas. J. 10 Eur. Ceram. Soc., 2004. 24(9): p. 2803-2812. DOI: 10.1016/j.jeurceramsoc.2003.09.019. 11 15. Yashchishin, I.N., L.V. Zhuk, and O.I. Kozii, Influence of Gas-Thermal Nitridation of Optical Lead 12 Silicate Glass on Its Surface Properties. Glass Phys. Chem., 2001. 27(5): p. 470-473. DOI: 13 10.1023/A:1012412418521. 14 16. Kozii, O.I., I.N. Yashchishin, and L.O. Bashko, Nitriding of Industrial Glass Surface. Glass Ceram., 15 2004. 61(9): p. 328-330. 16 17. Rancoule, G., Glass handling in tempering furnaces. Glass Worldw. Magazine, 17 2011(July/August). 18 18. Gordon, R., Chemical vapor deposition of coatings on glass. J. Non-Cryst. Solids, 1997. 218(0): 19 p. 81-91. DOI: 10.1016/S0022-3093(97)00198-1. 20 19. Gordon, R.G., K. Kramer, and X. Liu. Chemical Vapor Deposition and Properties of Amorphous 21 Aluminum Oxide Films. in MRS Proceedings. 1996. 446(383) p. 1-6. DOI: 10.1557/PROC-446- 22 383 23 20. Zhang, W., Z. Zhao, L. Shen, J. Zhang, and S. Yi, Surface modification of ion-exchanged float 24 aluminosilicate glass during deposition of amorphous alumina coatings by e-beam 25 evaporation. J. Non-Cryst. Solids, 2016. 447: p. 80-84. DOI: 10.1016/j.jnoncrysol.2016.05.034. 26 21. Karlsson, S., S. Ali, R. Limbach, M. Strand, and L. Wondraczek, Alkali salt vapour deposition and 27 in-line ion exchange on flat glass surfaces. Glass Technol. – Eur. J. Glass Sci. Technol. Part A, 28 2015. 56(6): p. 203-213. DOI: 10.13036/1753-3546.56.6.203. 29 22. Barr, J., The Glass Tempering Handbook - Understanding the Glass Tempering Process, Self 30 Published, 2015. 31 23. Aben, H., J. Anton, and A. Errapart, Modern Photoelasticity for Residual Stress Measurement 32 in Glass. Strain, 2008. 44(1): p. 40-48. DOI: 10.1111/j.1475-1305.2008.00422.x. 33 24. Sundberg, P., S. Karlsson, D. Brochot, C. Strubel, and J. Simons, Surface Analysis of float glass 34 using Surface Ablation Cell (SAC) Part 1: Initial collaboration and comparison with SIMS. Glass 35 Technol. – Eur. J. Glass Sci. Technol. Part A, 2010. 51(1): p. 13-21. 36 25. Karlsson, S., B. Jonson, P. Sundberg, and C. Stålhandske, Surface Analysis of float glass using 37 Surface Ablation Cell (SAC) Part 2: Determination of the diffusion characteristics of K+-Na+ Ion 38 Exchange. Glass Technol. – Eur. J. Glass Sci. Technol. Part A, 2010. 51(2): p. 55-62. 39 26. Oliver, W.C. and G.M. Pharr, Measurement of hardness and elastic modulus by instrumented 40 indentation: Advances in understanding and refinements to methodology. J.Mater. Res., 2004. 41 19(1): p. 3-20. DOI: 10.1557/Jmr.2004.19.1.3. 42 27. Limbach, R., S. Karlsson, G. Scannell, R. Mathew, M. Edén, and L. Wondraczek, The effect of 43 TiO2 on the structure of Na2O-CaO-SiO2 glasses and its implications for thermal and mechanical 44 properties. J. Non-Cryst. Solids, 2017. 471: p. 6-18. DOI: 10.1016/j.jnoncrysol.2017.04.013. 45 28. Kato, Y., H. Yamazaki, S. Yoshida, and J. Matsuoka, Effect of densification on crack initiation 46 under Vickers indentation test. J. Non-Cryst. Solids, 2010. 356(35-36): p. 1768-1773. DOI: 47 10.1016/j.jnoncrysol.2010.07.015. 48 29. Valant, M., U. Luin, M. Fanetti, A. Mavrič, K. Vyshniakova, Z. Siketić, and M. Kalin, Fully 49 Transparent Nanocomposite Coating with an Amorphous Alumina Matrix and Exceptional 50 Wear and Scratch Resistance. Adv. Funct. Mater., 2016. 26(24): p. 4362-4369. DOI: 51 10.1002/adfm.201600213.

6(7)

For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

1 30. Fonné, J.-T., E. Burov, E. Gouillart, S. Grachev, H. Montigaud, and D. Vandembroucq, 2 Aluminum-enhanced alkali diffusion from float glass to PVD-sputtered silica thin films. J. Am. 3 Ceram. Soc., 2018. 101(4): p. 1516-1525. DOI: 10.1111/jace.15340. 4 31. Pisano, G. and G.R. Carfagni, Statistical interference of material strength and surface prestress 5 in heat-treated glass. J. Am. Ceram. Soc., 2017. 100(3): p. 954-967. DOI: 10.1111/jace.14608. 6 32. Tadanaga, K., N. Katata, and T. Minami, Formation Process of Super-Water-Repellent Al2O3 7 Coating Films with High Transparency by the Sol–Gel Method. J. Am. Ceram. Soc., 1997. 80(12): 8 p. 3213-3216. DOI: 10.1111/j.1151-2916.1997.tb03253.x. 9 33. Tadanaga, K., N. Katata, and T. Minami, Super-Water-Repellent Al2O3 Coating Films with High 10 Transparency. J. Am. Ceram. Soc., 1997. 80(4): p. 1040-1042. DOI: 10.1111/j.1151- 11 2916.1997.tb02943.x. 12 34. Suzuki, T., T. Sekine, K. Yamamoto, and K. Fukutani, Change in the surface OH group on soda 13 lime silicate glass and silica glass after heat treatment in nitrogen atmosphere. J. Non-Cryst. 14 Solids, 2017. 464: p. 89-91. DOI: 10.1016/j.jnoncrysol.2017.03.014. 15 35. Etchepare, P.-L., L. Baggetto, H. Vergnes, D. Samélor, D. Sadowski, B. Caussat, and C. Vahlas, 16 Amorphous Alumina Barrier Coatings on Glass: MOCVD Process and Hydrothermal Aging. Adv. 17 Mater. Interfaces, 2016. 3(8): p. 1600014. DOI: 10.1002/admi.201600014. 18 36. Lewis, B.G. and D.C. Paine, Applications and processing of transparent conducting oxides. MRS 19 Bull., 2000. 25(08): p. 22-27. DOI: 10.1557/mrs2000.147. 20 37. Ginley, D.S. and C. Bright, Transparent Conducting Oxides. MRS Bull., 2000. 25(08): p. 15-18. 21 DOI: 10.1557/mrs2000.256. 22 38. Sbrockey, N.M. and S. Ganesan, ZnO thin films by MOCVD. III-Vs Rev., 2004. 17(7): p. 23-25. 23 DOI: 10.1016/S0961-1290(04)00735-5. 24 39. Parkin, I.P. and R.G. Palgrave, Self-cleaning coatings. J. Mater. Chem., 2005. 15(17): p. 1689- 25 1695. DOI: 10.1039/B412803F. 26 40. Ajayi, O.B., M.S. Akanni, J.N. Lambi, C. Jeynes, and J.F. Watts, Compositional studies of various 27 metal oxide coatings on glass. Thin Solid Films, 1990. 185(1): p. 123-136. DOI: 10.1016/0040- 28 6090(90)90012-3.

29

30

31

7(7)

For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

Table 1: Overview of some reactive gases that have been applied to glass substrates with time and temperatures required to achieve some notable effect. Gas Time min Temp (°C) Depth Application Reference Traces of Float glass – lubricant, SO2 10-40 min ~ 680 [9, 10] Na2SO4 surface protection SO2 10-40 min 580-650 Dealkalisation - Bottles [11] Al-R3 30-60 min 200 70 nm Training / Al2O3 Coating [12] AlCl3 30-60 min 200 Training / Al2O3 Coating Surface protection – AlCl3 10-30 min 700-1000 [13] bottles, stemware NH3 30 hrs 600 200-400 nm soda-lime-silicate glass [14] lead (Pb) / soda-lime N2 1-20 hrs 375–425 100 nm [14-16] silicate glass

For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028

Figure Captions

Figure 1: Measured temperature in flat glass centre during the tempering process. The red box indicates the window for surface reactions to occur. The time domain of the reactive gas surface interaction starts after approximately a few minutes and continues until the quench begins.

Figure 2: Experimental setup for A) Heat-treatment and reactive gas atmospheric treatment and B) uniform rapid cooling of samples.

Figure 3: Transmittance spectra for thermally strengthened and Al-CVD thermally strengthened flat glass. The insets show SEM-images of the coated surface (the bottom material) and an EDS-spectra with three spot analysis at the coated surface showing an Al signal.

Figure 4: Surface concentration as measured using SAC-ICP-OES [24]: A) shows the Al2O3 and K2O concentration profile. The inset in A) shows the SiO2 profile. B) shows the concentration profile of CaO, MgO and Na2O. The inset in to the right shows a polariscope photo of two Al-CVD thermally strengthened samples and the inset to the left shows a sample immediately after the rapid cooling.

Figure 5: Surface hardness as measured using nanoindentation. The three upper curves show the hardness as a function of load while the three lower shows the maximum contact depth (hm) as a function of load. The error of the hardness and maximum contact depth measurements are 0.2 GPa and 8 nm respectively, as estimated from the average of the standard deviations.

Figure 6: Crack resistance (CR) diagram showing values for annealed (CR=0.7 N), thermally strengthened (CR=0.8 N) and combined Al-CVD thermally strengthened float glass (CR1.3 N). The data were fitted with sigmoidal curves giving the crack resistance load at 50% probability of crack resistance (PCI).

Figure 7: A few contact angle data vs. water showing the significant difference of the surfaces of free energy behavior of annealed and combined Al-CVD thermally strengthened float glass.

Figure 8: Surface topography images of a thermally strengthened float glass (left) and of an Al-CVD thermally strengthened glass (right).

1(1)

For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028 A) B) For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028 Flat glass Sample holder (grated) Flat glass

Reactive atmosphere Air flow generating fans Precursor on grating

Heating elements For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028 For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028 For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028 Probability of Crack Initiation (%) 100 20 40 60 80 0 0 For 1 final journal publication 2 see PCI Thermally Al-CVD Annealed https://doi.org/10.1016/j.tsf.2018.11.028 Load (%) and CR � (N) Strengthened loat PCI=0% Thermally PCI=25% glass 3 PCI=50% PCI=75%

CR= PCI=100% strengthened

0.7 CR=0.8 Load N 4 (N) N

CR=1.3 N 5 For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028 For final journal publication see https://doi.org/10.1016/j.tsf.2018.11.028