Chemical strengthening of flat glass by
vapour deposition and in-line
alkali metal ion exchange
Stefan Karlsson
1, Sharafat Ali
2och Michael Strand
21
Glafo – the Glass Research Institute
2Linnaeus University, Faculty of Technology
Glafo – the Glass Research Institute
Chemical strengthening of flat glass by
vapour deposition and in-line alkali
metal ion exchange
Stefan Karlsson
1, Sharafat Ali
2, Michael Strand
2 1Glafo – the Glass Research Institute
3
Abstract
Glass is a common material in the everyday life. It is widely used in a variety of applications e.g. architectural, automotive, containers, drinking vessels, displays, insulation and optical fibers due to its universal forming ability, transparency, chemical durability, form stability, hardness and relatively low price. Flat glass is a wide market of the glass industry and generally ninety percent of all flat glass produced worldwide is manufactured using the float forming process. There is a large market strive for thinner and stronger glass in order to reduce costs, save energy, reduce environmental footprint, find new applications and to improve the working environment for labour working with mounting flat glass.
This study comprises the modification of flat/float glass surface by a novel route; exchange of ionic species originating from in-line vapour deposition of salt compared to the conventional route of immersing the glass in molten salt baths. The aim of this work is to develop a novel process in order to improve the mechanical strength of flat/float glass by introducing external material to the surface in a process with the obvious potential to be automatic in industrial processes. Chemical strengthening has been performed by applying potassium chloride to the glass surface by vapour deposition and thermally activated ion exchange. The method presented here is anticipated to be used in production in the future and would make it possible to produce larger quantities of chemically strengthened flat glass to a considerably lower cost.
Key words: Chemical strengthening, In-line ion exchange, Flat glass, Vapour deposition
Glafo - glasforskningsinstitutet Glafo – the Glass Research Institute Glafo-rapport 2014:3P00070 Växjö 2014
Table of Contents
Abstract 3
Table of Contents
4
Preface
5
Sammanfattning 6
1
Introduction 7
2
Experimental Procedure and Characterization
9
2.1 Glass Composition and Experimental Procedure 9
2.2 Characterization 10
3
Results 11
3.1 Deposit of salt 11
3.2 Concentration profiles and Effective Diffusion Coefficients 12
3.2.1 Concentration profiles 12
3.2.2 Effective Diffusion Coefficients 14
3.3 UV-Vis-NIR Transmission/Reflection and Refractive Index 14
3.4 Contact Angle 17
3.5 Infrared Reflectance Spectroscopy (IR-RS) 19
3.6 Nanoindentation and Scratch resistance 21
3.6.1 Indentation Hardness and Young’s modulus 21
3.6.2 Scratch Resistance 23
4
Conclusions and Future Work
24
5
Preface
The project was initiated from an idea to combine the two separate processes, vapour deposition and chemical strengthening of glass, in order to give an in-line process which can be automated in a flat glass manufacturing line. Chemical strengthening of glass is currently a topic that attracts much attention due to its commercial success as display glasses. It is though still a very expensive process performed separately to the flat glass manufacturing. The purpose of the project was to validate the idea on how to improve the chemical strengthening process of glass and show the possibilities to lower the costs of the process. Results of the project were presented as a poster at FFAG6 (6th International workshop on Flow and Fracture of Advanced Glasses) in Weimar, Germany (7th-10th of October, 2014).
We wish to acknowledge ÅForsk for providing funding for this project (Grant No. 13-364). The funding granted to Stefan Karlsson via the VINNMER programme (Vinnova) co-funded by Marie Curie Actions FP7-PEOPLE-2011-COFUND (GROWTH 291795) is gratefully acknowledged. We also wish to acknowledge Pilkington Sverige AB for providing float glass for the experiments in the project. Furthermore, we wish to acknowledge people that has performed measurements within the frame of the project: René Limbach (Otto Schott Institute of Materials Research, Friedrich Schiller University of Jena) for performing indentation and scratch/wear resistance measurements, Maria Lang (Glafo – the Glass Research Institute) for performing SEM/EDS measurements and Matilda Schander (Glafo – the Glass Research Institute) for performing contact angle measurements. At last we wish to acknowledge valuable discussions with Professor Lothar Wondraczek (Otto Schott Institute of Materials Research, Friedrich Schiller University of Jena).
Växjö 2014-11-28
The project is a collaboration between Glafo – the Glass Research Institute and Linnaeus University.
Sammanfattning
Glas är ett vanligt material i vardagen. Det används frekvent i en mängd olika
applikationer t.ex. arkitektoniska, fordon, flaskor/burkar, dricksglas, displayer, isolering och optiska fibrer på grund av den allsidiga formningsförmågan, transparensen, kemisk beständigheten, formstabiliteten, hårdheten och det relativt låga priset. Planglas är en stor marknad för glasindustrin och generellt tillverkas nittio procent av allt planglas i världen med floatprocessen. Det finns ett stort marknadsdriv efter tunnare och starkare glas för att reducera kostnaderna, spara energi, minska miljöpåverkan, hitta nya tillämpningar och för att förbättra arbetsmiljön för arbetskraft som arbetar med montering av planglas.
Denna studie omfattar kemisk härdning av planglas genom en ny metod; utbyte av joner genom in-line ångdeposition av salt på planglas. Traditionellt tillverkas kemiskt härdat glas genom nedsänkning av glas i smält saltbad. Syftet med projektet är att utveckla ett ny metod för att förbättra den mekaniska styrkan av planglas genom att införa externt
material till ytan i en process med det uppenbar potential att bli automatisk i industriella processer. Kemisk härdning har utförts genom att applicera kaliumklorid på glasytan med ångdeposition och termiskt aktiverat jonbyte. Metoden som presenteras förväntas att användas i produktion i framtiden och gör det möjligt att producera större mängder av kemiskt härdat planglas till en betydligt lägre kostnad.
7
1
Introduction
Glass is a fantastic material with several desirable properties: transparency, high hardness, good chemical durability, form stability, forming ability, relatively low price and with the possibility of recycling. Glass is used in a wide variety of applications, e.g. windows, containers, displays, thermal insulation, optical lenses, bioactive glasses and telecommunications. The fact that glass is a relatively hard material depends on the nature of the intrinsic bonds of the vitreous network. The mechanical strength on the other hand depends on entirely different factors, mainly the distribution of defects in the glass surface and surrounding environmental factors causing surface defects, see Fig. 1. As glass has found many applications due to its universal process-ability and optical properties the use of glass is frequently limited by its brittle nature and unreliable mechanical fracture.
Figure 1: The effect of surface flaws on the tensile strength of soda-lime-silicate glass.
There are many different ways to increase the strength of glass [1]; most of them involve the modification of the surface [2]. A method which has received much attention recently is chemical strengthening. It is based on substituting smaller ions in the glassy matrix with larger ions from a molten salt e.g. Na+ is substituted for K+, see Fig. 2. The larger ions are literally squeezed into the sites of the smaller ions generating compressive stresses in the glass surface which counteracts applied tensile stresses from the environment.
Figure 2: Schematic ion exchange process for the modification of glass surfaces.
The most commonly described route of chemical strengthening is K+-Na+ ion exchange, which was discovered by Kistler [3] and Acloque [4] independently. Chemical strengthening of glass has recently been reviewed extensively by several authors [2, 5-8].
1 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 log( /MP a )
log(effective flaw depth/mm)
Theoretical strength
Processed flat glass
micro-cracks visual cracks
Despite the fact that it was discovered more than half a century ago, chemical strengthening has failed to reach wide markets until recently when large specialty glass companies launched chemically strengthened flat glass e.g. Corning® Gorilla® Glass [9], Schott XensationTM [10], AGC DragontrailTM [11] and NEG CX-01TM [12]. Today, chemically strengthened flat glass is found in most touch display smartphones available in the market [13]. Despite the considerably higher cost of chemically strengthened flat glass it was worthy to use it as cellphone displays. However, for larger sheets and other types of applications as for instance in large displays, substrate for photovoltaic cells, car windows, buildings and furniture’s, the cost for chemically strengthened glass is still too high.
Chemical strengthening of glass is complex and many factors affect the ion exchange process as well as the resulting strength [14].
(i) The temperature effect on the interdiffusion coefficient. (ii) The time of exchange.
(iii) The interface between glass and salt. (iv) The glass composition.
(v) The exchanging pair of ions.
(vi) The temperature influence on relaxation.
With the present knowledge many of these factors have been optimized for specific products with the cost limitations present. There are though other possibilities to reduce the price of chemical strengthening even further – by optimizing the large-scale process. Optimization of industrial processes are most often performed by automation of the process [15]. Chemically strengthened flat glass is today manufactured by immersing prefabricated flat glass into a molten salt bath.
In this project we suggest a novel process approach: vapour deposition and in-line alkali metal ion exchange, see Fig. 3. A similar approach has previously been studied by Sil’vestrovich et al [16, 17], however, there the vapour deposition and the ion exchange process were separated into two different steps while in the current work we combine them into one step. The aim of the work is to present a novel process route to reduce the costs of chemically strengthened glass. There are also alternative approaches, where the glass is spray-coated with a salt mixture and subsequently dried as well as heat treated [18].
Figure 3: Overview of the in-line vapour deposition of salt for chemical strengthening of flat glass.
9
2
Experimental Procedure and
Characterization
2.1
Glass Composition and Experimental Procedure
The glass used was a conventional soda-lime-silicate float glass (4 mm) with the chemical composition analyzed using different wet chemical methods e.g. Atomic Absorption Spectroscopy (AAS). The dissolution of the sample and quantification followed BS 2649 except for TiO2 and SO3. Due to safety reasons other acids replaced perchloric acid. The normalized chemical composition of the float glass used in the experiments is given in Tab. 1. For experiments, individual specimens of 2 x 10 cm were prepared. The specimens were cleaned with ethanol (99.5%) and deionised water, thereafter they were dried before weighing of the specimens. Reagent grade quality KCl delivered by Scharlau Chemie SL was mixed with water to obtain different concentrations (10, 15 and 20 g per 100 ml H2O). The experimental process parameters and specimen overview are given in Tab. 2. Concentrations of 10, 15 and 20 g per 100 ml H2O was run for 2, 4 and 6 h at constant temperature. The as-received glass is denoted “Ref” in the report.
Table 1: Normalized chemical composition of the float glass used in the experiments.
wt% mol% SiO2 72.80 71.32 Na2O 13.39 12.72 K2O 0.04 0.03 Al2O3 0.12 0.07 MgO 4.07 5.94 CaO 9.25 9.71 Fe2O3 0.10 0.04 SO3 0.21 0.16 TiO2 0.02 0.01 Sum 100 100
Table 2: Process parameters of in-line vapour deposition on float glass.
Notation KCl Conc. (g/100 ml H2O) Furnace Temperature – (start at 700°C) Time (h) Thermocouple Temperature (at sample) KCl20 10 800 °C 2 ~450°C KCl16 15 800 °C 2 ~450°C KCl21 20 800 °C 2 ~450°C KCl19 10 800 °C 4 ~450°C KCl15 15 800 °C 4 ~450°C KCl24 20 800 °C 4 ~450°C KCl18 10 800 °C 6 ~450°C KCl17 15 800 °C 6 ~450°C KCl23 20 800 °C 6 ~450°C
The experiment setup is given in Fig. 4. An aerosol generator from PALAS Particle Technology model AGK 2000 was used, it was consistently run at 2 bar pressurized air in the experiments. The tube furnace was a Carbolite Tube Furnace model 12/65/550 with a
max temperature of 1200 °C. The temperature distribution was measured with thermo elements – in the middle of the furnace it was 800 °C (at start 700 °C) while at the middle of the specimen ~450 °C, the whole sample was covered within 450±60 °C.
Figure 4: Experimental setup of the vapour deposition and in-line ion exchange.
2.2
Characterization
The thickness of the deposited salt layer was analysed using an optical profilometer, NPFLEX 3D Surface Metrology System from Bruker Corporation equipped with an objective of 5x magnification. The salt layer was removed in the middle of the sample so that an edge was created, see Fig. 5 below. The edge was scanned and the approximate thickness ranges were estimated from the image data.
Figure 5: Optical profilometer 3D vision of the created edge of the salt layer.
Scanning Electron Microscopy (SEM) equipped with Energy Dispersive Spectroscopy (EDS) was used for analyzing concentration profiles of the samples. The SEM was of the model JEOL JXA 840A and the EDS, Oxford Instruments 6506. The SEM/EDS instrument was calibrated using a cobalt standard. The samples were casted in epoxy resin and coated with carbon. The profiles were measured at the cross section of the samples approximately in the middle of the samples i.e. where the treatment temperature was approximately 450 °C, see Fig. 4. The count time was 30 s at 60 µA/20 kV. The displayed concentration profiles in section 3.2 are given as the average of three different line scan analyses at three different locations. Effective Diffusion Coefficients were calculated from the average concentration profiles according to Eq. 1, Green’s function [19].
√
11
Contact angle was measured with PGX+. The samples were cleaned with water and dishwashing detergent prior to the measurements. 4 µl of deionised water were dropped on the glass and the delay was 3 s before measuring the contact angle. The samples were measured in two different ways, (i) 10 measurements at the same spot (hot end) and (ii) over the whole sample (as many measurements that could fit).
Transmission and Reflectance was measured with an Agilent Technologies UV-VIS-NIR model Cary 5000, 200-3200 nm and 200-2500 nm respectively. Scan rate was 600 nm/min. At 800 nm the detector was changed, therefore a jump in transmission/reflection was recorded. The transmission and reflectance was measured at the hot-end of the glass samples. The Refractive Index of the glass was calculated from the transmission spectra according to Eq. 2 and from the reflectance spectra according to Eq. 3. Where ns is the refractive index of the glass and Tmax is where it was a minimum in the absorption (520 nm). It shall be noted that there exist other equations to calculate the refractive index from the transmission spectra, therefore the results shall be viewed as comparable to each other and not comparable to other published results.
(2)
√
(3)
Infrared Reflectance Spectroscopy (IR-RS) was measured with an instrument from Shimadzu Scientific Instruments, IRAffinity-1. The range was 400-4000 cm-1 with a resolution of 4 cm-1. 64 number of scans were used in the measurements. The spectra were transformed according to Kramers-Kronig transformation. As a baseline an aluminum mirror was used. The IR-RS spectra was measured at the hot end of the glass samples.
Nanoindentation was measured with Agilent Technologies G200. The tip was a Berkovich tip (TB 21071ISO). Ten measurements was made for each sample. The indentation hardness and Young’s modulus are reported as average values from the range 400 – 1800 nm. The nanoindentation was measured at the cold end of the glass samples. The scratch/wear resistance was measured with the same nanoindenter instrument (Agilent Technologies G200). The tip was a cone shaped tip (CONE 0021077 ISO). In total, 100 wear cycles with constant load (25 mN) were made with a speed of 50 µm/s at five different locations. The average of the five different locations are reported in the
results. The scratch/wear resistance was measured at the cold end of the glass samples.
3
Results
3.1
Deposit of salt
The weight of the deposited KCl salt and the thickness of the salt layer from the vapour deposition can be seen in Tab. 3. The thickness of the salt layer was determined by optical profilometer measurements. The deposit of the KCl salt in grams on the glass samples has been plotted versus the max thickness of the deposited layer, see Fig. 6. Fairly linear trends can be seen for the different concentrations, this proves that the salt deposition works with the experimental setup. Similar trends was also observed when plotting series of treatment time versus the max thickness of the deposited layer. The trend when plotting deposit of salt in grams respectively max thickness of the deposited layer versus treatment time are as well fairly linear. On the other hand when plotting
de co Ta lay (g Fi
3.
Th de ch co deA
3. eposit of sa oncentration able 3: Depo yer. KCl Conc. g/100 ml H2O 10 15 20 10 15 20 10 15 20 igure 6: Max.2
C
C
he deposited escribed by E hemical poten onsidered to eposited salt salt glassB
A
2.1 C alt in grams then the trenosit results ex O) Time ( 2 2 2 4 4 4 6 6 6 x thickness of
Concentra
Coefficien
d salt reacts w Eq. 4. Ion e ntial or the G be driven by contains K+ s glassA
B
Concentratio s respectivel nds are more xpressed as w h) Deposit sam 0. 0. 0. 0. 0. 0. 0. 0. 0. f deposited laation prof
nts
with the glass exchange is Gibb’s Free E y the concen ions which s salt on profiles ly max thic exponential weight of cha t of salt on mple (g) 0028 0031 0037 0054 0086 0287 0107 0170 0330 layer in µm v
files and E
s surface via driven by th Enthalpy gra ntration gradi substitutes th ckness of th . ange with sa Thickness layer 1-4 µm 1-4 µm 4-16 µ 2-14 µ 2-20 µ 30-45 µ 25-35 µ 30-40 µ 50-60 µ vs deposit ofEffective
a an ion exch hermodynam adient [20]. I ient between he Na+ ions in he deposited lt and thickn of salt r m m m m m µm µm µm µm salt in gramsDiffusion
hange proces mics of the re n the presen the salt and n the glass su (4) d layer vers ness of salt s.n
ss which can eaction i.e. t nt case it can d the glass. T urface. sus be the be The13
The results show that both the treatment time and the concentration matter, however, the treatment time matters more. In general, the samples treated for 2 h gave an approximate depth of 4 µm while those treated for 4 and 6 h gave depths exceeding 6 µm. The concentration profile in Fig.8 treated with the concentration 15g KCl per 100 ml H2O during 4 hrs have a somewhat different profile than the others, we believe there might have been a discrepancy in the experiment. In general, the results show that for the ion exchange kinetics the optimal concentration is 10<x<20 g KCl per 100 ml H2O and that the optimal treatment time is 2<x<6 hrs. Sil’vestrovich et al [16] found somewhat larger penetration depths of K+ ions. However, this can be explained by the different in the experimental procedure i.e. the two step procedure by Sil’vestrovich et al compared to our one step procedure in the present work.
Figure 7: Concentration profiles of series treated with 10 g KCl per 100 ml H2O.
Figure 8: Concentration profiles of series treated with 15 g KCl per 100 ml H2O.
0 2 4 6 8 10 Depth (µm) 0 1 2 3 4 5 C onc entrati on (w t% ) Ref 10g 2h 10g 4h 10g 6h 0 2 4 6 8 10 Depth (µm) 0 1 2 3 4 5 C o nc en tr at io n ( w t% ) Ref 15g 2h 15g 4h 15g 6h
Figure 9: Concentration profiles of series treated with 20 g KCl per 100 ml H2O.
3.2.2 Effective Diffusion Coefficients
Effective Diffusion Coefficients were calculated from the concentration profiles according to Green’s function (c.f. Eq. 1). For the samples treated for 2 h the concentration profile values between 0.5 - 4 µm were used and for the others values between 0.5 - 6 µm were used (c.f. Fig. 7-9). The calculated Effective Diffusion Coefficients are given in Tab. 4. It can be seen that the Effective Diffusion Coefficients ranges between 1.2.10-12 and 4.2.10-12. It is considerably lower Effective Diffusion Coefficients than previously reported results [19], however, there are many parameters that effect the ion exchange kinetics. An important factor is believed to be the thickness of the deposited salt layer, it is very thin compared to any other reported study. The concentration of K+ ions in a thinner salt layer will decrease more rapidly than with a thicker as the ion exchange process proceeds, it might even be depleted. This might be the reason why the optimal treatment time is less than 6 hrs.
Table 4: Effective Diffusion Coefficients calculated with Green’s function from concentration profiles.
KCl Concentration (g/100 ml H2O)
Treatment time (h)
Effective Diffusion Coefficient (cm2.s-1) 10 2 1.3 . 10-12 10 4 1.7 . 10-12 10 6 2.1 . 10-12 15 2 1.7 . 10-12 15 4 4.2 . 10-12 15 6 2.0 . 10-12 20 2 1.5 . 10-12 20 4 1.5 . 10-12 20 6 1.2 . 10-12
3.3
UV-Vis-NIR Transmission/Reflection and Refractive
Index
The transmission spectra of the modified samples were recorded between 200 and 3200
0 2 4 6 8 10 Depth (µm) 0 1 2 3 4 5 C onc entrati on (w t% ) Ref 20g 2h 20g 4h 20g 6h
or ex K2 co to at m Eq ch re of sh ot ex Fi Fi rdinary featu xchange treat 2O increases onsequently m Na2O that h the wavelen minimum. Fur q. 2, the refl hange of RI eview paper t f Δn ≤ 0.01 hall be noted ther. It is pos xtent, howev igure 10: Tra igure 11: Tra ures more th tments. The s the refracti more light w has a polariza ngth where th rthermore, th flectance spe (Δn) and in that the refra
[21], it is in d that the ca ssible that th er, not know
ansmission sp ansmission sp han a small shift in tran ive index (R will be reflec ability of 0.4 he transmissi he RI can als ectra is show ncrease of R active index n good agreem alculated RI he generated wn to what de spectra of all spectra in the l loss of tra nsmission at RI) of the gla cted. K2O ha 43 Å3 . The R ion is at max so be calcula wn in Fig. 12 RI are shown change for s ment with th from T% an d compressiv egree. l samples inc e range 300-ansmission a 800 nm is d ass due to it as a polarizab RI can be cal ximum (520 n ated from th 2. The calcu n in Tab. 5 substituting he obtained nd R% cann ve stresses af cluding an un 550 nm. as expected due to the de ts higher pol bility of 1.33 lculated acco nm) i.e. the a he reflectance ulated RI val . Ramasvam Na+ with K+ results show not be compa ffect the RI v ntreated refe d from the i etector chang larizability a 3 Å3 compar ording to Eq. absorption is e spectra usi lues as well my reports in + is in the ord wn in Tab. 5 ared with ea values to som erence sampl 15 ion ge. and red . 1, s at ing as n a der . It ach me le.
Fi Ta C ( Th in ca co 4h se re m tw ap igure 12: Ref able 5: Calcu oncentration (g/ 100 ml) REF 10 10 10 15 15 15 20 20 20 he refractiv ncreasing tim alculated fro oncentration h), some tren een. It shall t eceived sam measurements wo samples pproximately flectance spe ulated Refrac n Treatmen time (h) REF 2 4 6 2 4 6 2 4 6 e index wa me respective om T% ver is similar. T nds of increa though be no mple (REF) s and a small (10g 2h an y the range (3 ectra in the r ctive Index (R nt C RI (n) 1.47838 1.48875 1.48404 1.47918 1.48398 1.49152 1.48298 1.48098 1.4835 1.48415 as assumed ely. Howeve rsus time do The plot in F asing RI. Wh oted that ano , see Tab l surface scra nd 15g 4h) 300-1000 nm range 200-25 (RI) from T% Calculated fr Δn 8 - 5 0.010 4 0.006 8 0.001 8 0.006 2 0.013 8 0.005 8 0.003 1 0.005 5 0.006 to increase er, as can b oes not sho Fig. 14 show hen plotting v other sample . 5. Refle atch or dirt c ) that show m), see Tab. 5 500 nm. % and R% at from T% Increase RI (% - 0.701 0.383 0.054 0.379 0.888 0.311 0.176 0.347 0.390 e with incr be seen in th ow this feat however, ap vs concentra e (10g, 2h) s ectance mea can change th w lesser RI 5 and Fig. 12 520 nm. C e of %) RI (n 1.52 1 1.52 3 1.52 4 1.52 9 1.52 8 1.51 1 1.53 6 1.52 7 1.53 0 1.53 reasing conc he plot of F ture. The p part from on
tion those tre show lower R asurements he result con only show 2. Calculated fr n) Δn 235 218 -0.002 252 0.002 257 0.002 286 0.005 96 -0.004 336 0.010 269 0.003 312 0.008 320 0.009 centration a Fig. 13 the plot of RI ne sample (15 rends cannot RI than the are sensiti nsiderably. T w lower RI rom R% Increase o RI (%) -0.114 0.112 0.147 0.337 -0.259 0.66 0.226 0.50 0.559 and RI vs. 5g, be as-ive The in of 4 2 7 7 9 1 6 8 9
Fi Fi
3.
Co re re Th is di co de av co tre igure 13: Ca igure 14: Ca.4
C
ontact angle esults are giv espectively gi he standard d however log iscussed. It is onsiderably l egree it is aff verage contac oncentration ends. alculated Ref alculated RefContact A
data were co en in Tab. 6 ives relativel deviation is c gic due to all s, however, c ower compa fected by the ct angle valu and treatmen fractive Index fractive IndexAngle
ollected in tw and 7. Plotti ly similar tre considerably l parameters certain that s ared to the tre e thermal trea ues measured nt time respe x (RI) from T x (RI) from R wo different w ing the conta ends for both y larger when that are affec omething ha eated sample atment itself. d at the same ectively, as c T% vs. Treatm R% vs. Treat ways, c.f. Ex act angle vs c h ways to me n measuring octing the pro as happened a es, however, . In Fig. 15 a location (ho an be seen th ment time (h) tment time (h) xperimental S concentration asure the con over the who ocess as previ as the referen it is not know and 16 are plo ot end) versu
here are not a
h). h). Section. The n and time ntact angle. ole sample, th viously nce values ar wn to what ots shown fo us any clear 17 e his re or
Ta Co (g Ta Co (g Fi 10 able 6: Conta oncentration / 100ml) REF 10 10 10 15 15 15 20 20 20 able 7: Conta oncentration / 100 ml) REF 10 10 10 15 15 15 20 20 20 igure 15: Av 00 ml). act angle da n Treatmen Time (h) F RE 0 0 0 5 5 5 0 0 0 act angle da n Treatmen Time (h) F RE 0 0 0 5 5 5 0 0 0 erage contac ta – 10 meas nt ) Average EF 2 4 6 2 4 6 2 4 6 ata – as many nt ) Average EF 2 4 6 2 4 6 2 4 6 ct angle at th surements at e Contact an y measureme e Contact An he same loca
the same loc
gle (°) Sta Dev 25.34 38.31 38.76 28.11 48.06 40.36 32.04 29.04 40.85 40.38 ents as possi ngle (°) Sta Dev 19.03 43.35 43.03 35.23 45.14 42.01 37.48 35.01 39.79 38.88
tion (hot end
cation (hot e ndard viation 1.19 2.68 2.06 1.86 1.46 2.36 2.54 3.71 3.02 2.48
ible over the
andard viation 0.82 4.73 3.51 4.84 6.04 6.75 4.09 10.07 3.66 6.19 d) vs. concen end). whole samp ntration (g / le.
Fi
3.
IR co Fi Pe sp A no na lo Th al co eq th igure 16: Av.5
I
R Reflectance omplete spec igure 17: Ov enetration of pectra but aff s can be see ot a large c arrowing tha wer wavenu he feature of [23, 24] orresponds to quilibrium is he chemical e erage contacnfrared R
e Spectrosco ctra is given i verview of the f K+ by ion fects the pea en in Fig. 18 change but s at is changing umbers and a f peak narrow and in alu o the chemica shifted to th equilibrium ct angle at thReflectanc
opy (IR-RS) w in Fig. 17. e IR-RS spec exchange do ak of Si-O st and Tab. 8, still a meas g (see Fig. 1 also a small wing has prev umino-borosi al equilibrium he right i.e. t to the left i. he same locace Spectr
was measure ctra, 400-400oes not crea tretching (~1 , the treated surable chan 18) but also t increase in viously been ilicate glass m Q2+Q4 ⇌ towards mor .e. peak broa
tion (hot end
oscopy (I
ed for all sam00 cm-1.
ate new abso 1090 cm-1) b samples exh nge. Furtherm the ~1090 cm the absorptio n shown in S s by Stavro 2Q3. For K+ re Q3 while f adening. Leb d) vs. treatme
R-RS)
mples, an ove orption bands y narrowing hibits peak n more, it is n m-1 peak is s on coefficien LS float glas u et al [22 ion exchang for Ag+ ion e boeuf et al [2 ent time (h). erview of the s in the IR-R g the peak [2 narrowing, it not only pe shifted towar nt can be see ss by Ingram 2], the featu ge the chemi exchange shi 25] studied 19 e RS 2]. t is eak rds en. m et ure cal ifts K+io al th IR vi vi hy cm re atm ar stu Fi an Ta C (g Th n exchange so a decreas he present stu R spectroscop ibration, ~11 ibration at ~ ydronium ion m-1 as well a eplacement o mosphere co round 1090 c
udy, see Fig.
igure 18: IR-nd the shift of able 8: Peak Concentratio g / 100 ml H2 Ref 10 10 10 15 15 15 20 20 20 he treatment - The p to the loweri in SLS flat e in the abso udy. Doremu py. Two pea 100 cm-1 and ~950 cm-1, w ns exchanges as increasing f Si-O-A for ontains water cm-1 is incre . 18. -RS spectra b of the peak at k broadness o n O) Treatm time ( Ref 2h 4h 6h 2h 4h 6h 2h 4h 6h also induces eak at ~495 flexion vibr ing in the a glass and co orption coeff us [26] studie
aks are obse d ~1050 cm where A is a
s with the alk the absorpti r Si-O-H-O-H r vapour and easing and s between 900 t ~1090 cm-1. of the peaks a ent h) Peak b ~1090 s other chang cm-1 shows ration modes absorption co ould also re ficient with e ed the hydra erved in non m-1, and anoth an alkali ion kali ions and ion band of H2. During th d this might e shifted towar 0 and 1300 cm . at ~1090 cm -broadness a 0 cm‐1 (cm‐1) 186 182 181 181 183 182 179 182 181 181 ges in the IR peak narrow s zone of SiO oefficient ex port an incr exchange tim ation of alkal n-hydrated g
her one assi n. As the w d decreases th ~1100 cm-1 he experimen explain that th rds lower w m-1, demonst -1 , ~785 cm-1 t ) Peak bro ~785 cm -RS spectra. wing, see Tab O4 tetrahedra
xcept for the
ease in Q3 u me which can li-silicate gla lass for Si-O gned to Si-O water reacts w he absorption and ~1050 c nts in the pre he absorption avenumbers rating the pe and ~495 cm oadness at m‐1 (cm‐1) 65 57 57 56 59 58 58 64 63 58 b. 8. This are [25]. The pe e sample tre units, howev nnot be seen ass surfaces O-Si stretchi O-A stretchi with the gla n band at ~9 cm-1 due to t esent study, t on coefficient in the prese eak narrowin m-1. Peak broad ~495 cm‐1 69 66 64 64 66 66 65 68 67 67 rea is attribut eak also sho eated with 1 ver, n in by ing ing ass, 950 the the t at ent ng dness at 1 (cm‐1) ted ws 0g
Fi str
3.
3. Th en m gi wa Ta su ap io - The m - The p toward coeffi accord - The p coeffi - The m shift to - The sm is the [26, 2 treatm the pre igure 19: IR retching, mo.6
N
6.1 In he deposition nriched in K mechanical pr ive informati as measured ab. 9 both t urfaces are in pproximately n exchange t minima at ~70 peak at ~785 ds lower wa cient except dance with th peaks at ~232 cient. The pe minima at ~2 o higher wav mall peak at area of OH 27], see Fig. ment. This sh esent study. Reflectance olecular adsoNanoinden
ndentation H n of KCl on t K2O. Presum roperties. Th ion on inden d ten times to the indentati ncreased due y on the same treatment ha 00 cm-1 show 5 cm-1 show avelengths. Tfor the sam he study of L 20 cm-1 and eak at ~2320 2373 cm-1 sh venumbers. 3791 cm-1 sh H-stretching, 19. In this a ows that the
spectra, 350 orbed water a
ntation an
Hardness anthe glass and mably the s herefore the tation hardne get a statisti on hardness e to the trea e level for al s taken place ws a lowering ws peak narr The peak al mples treated Lebeouf et al ~2353 cm-1 0 cm-1 also sh hows a decre hows an incr molecular a area there ar glass also re 00-3800 cm-1 and hydroge
nd Scratc
nd Young’s d the ion exc surface modnanoindenta ess and You ically valid a s and the Yo atment. The ll samples, in e. g in the abso rowing (see lso shows lo d with 10g d [25]. both shows hows shift to ease in the ab rease in the a adsorbed wa re also many eacts with wa 1 , displaying en bonding ar
ch resistan
modulus change proce dification sh ation measur ung’s modulu average value oung’s modu standard de ndicating tha rption coeffi Tab. 8) and owering in during 2 and increase in lower waven bsorption co absorption co ater and hydy other chan ater during th the changes rea.
nce
ss gives a m hould chang rements was us, see Tab. 9e. Given from ulus of the m viation of th at a relatively icient. d a small sh the absorpti d 4h. This is the absorpti numbers. oefficient and oefficient. Th drogen bondi nges due to t the treatment in the OH-modified surfa ge the surfa s performed 9. Each samp m the results modified gla he samples a y homogeneo 21 hift ion in ion d a his ing the t in ace ace to ple s in ass are ous
Table 9: Hardness and Young’s modulus data from Nanoindentation measurements determined as average values from the range 400 – 1800 nm.
Concentration (g / 100 ml H2O)
Treatment
time (h) Hardness (GPa)
Young’s modulus (GPa)
Temperature (° C)
Avr. StDev. Avr. StDev. Avr. StDev.
Ref Ref 6.84 0.03 77.3 0.3 30.8 <0.1 10 2 7.07 0.04 77.4 0.2 30.7 <0.1 10 4 6.95 0.05 77.5 0.2 30.8 <0.1 10 6 7.41 0.04 78.7 0.4 31.2 <0.1 15 2 7.03 0.04 77.8 0.2 31.1 <0.1 15 4 7.39 0.03 78.9 0.2 31.1 <0.1 15 6 7.41 0.04 79.1 0.3 31.2 <0.1 20 2 7.07 0.03 77.6 0.2 31.3 <0.1 20 4 7.31 0.04 78.2 0.3 31.6 <0.1 20 6 7.40 0.04 78.9 0.3 31.4 <0.1
In Fig. 20 the hardness data are plotted versus the treatment time. The series of 15 g and 20 g together with the reference sample show both relatively linear relationship. The sample treated with 10 g during 4h makes the trend of the 10 g series irregular. However, the other samples in the 10 g series coincide well with the trends of the other series. As can be deduced from Fig. 20, the concentration of the salt mixture itself does not affect the hardness much. The treatment time seem to be dominant for the hardness improvement. In Fig. 21 the Young’s modulus data are plotted versus the treatment time. In the plot one can see that the are some deviation for the samples treated at 4h while the others are fairly close to each other. The trends of the series are not linear but rather exponential, even the 10 g series have a relatively exponential trend. Similar to the hardness plot (Fig. 20), the concentration does not seem to have much to do with the increase of Young’s modulus as the treatment time have. However, it is though some small changes with concentration, it is in the order 15g>20g>10g (KCl per 100 ml H2O), see Figure 20. The same order is not seen in the hardness data, where the values are more close to each other, except for the series treated for 4h where we can see the same order, 15g>20g>10g (per 100 ml). Several studies on the hydration treatment of soda-lime-silicate glasses [28-30] observed small reductions (if any reduction depending on composition) in the hardness and the Young’s modulus due to the hydration. Kolluru et al [30] suggested that the reduction of the nanomechanical properties may also depend on the exposure history of the glasses. However, the studies of the nanomechanical properties of hydration treatments of SLS glass are well below (< 200 nm) the penetration depth of the indenter tip in our study (400 – 1800 nm) [28-30]. Fett et al [31] studied stresses in hydronium ion exchanged SLS glasses and found that very high compressive stresses are achieved due to the ion exchange. Furthermore, they found that the depth of the layer to a large extent depends on the temperature but also on the square root of the treatment time. Sil’vestrovich et al [16] also reported data on the indentation hardness, an increase from 5.20 GPa to 6.25 GPa was found for the aerosol method while the salt bath gave an indentation hardness of 7.1 GPa. The increase in the present study is considerably lower than what was reported by Sil’vestrovich et al, however, there are also differences in the experimental procedure which can explain this.
Fi Fi tre 3. Th Th we H2 ot su sa im igure 20: Ha igure 21: Yo eatment time 6.2 S he scratch/w he wear disp ear. In Fig. 2 2O. It can be ther and the r urface with th amples treate mproved scra ardness data ung’s modul e. Scratch Resi wear resistanc placement in 20 the result e seen that al reference sam he others foll ed at 4 and 6 atch/wear res of K+ ion ex lus data of K stance ce was invest
nto the glass ts are shown lready at 20 mple show th lowed in the 6h are relativ istance, see F xchanged flat K+ ion exchan tigated with s surface wa n for the sam wear cycles he largest av e order 2h<4h vely similar. Fig. 20. t glass as a fu
nged flat glas
a nanoinden as measured mples treated s the samples verage wear h/6h. The scr In general, function of tr ss as a functi nter with a co after every with 15 g K s start to dev displacemen ratch/wear re the modified reatment time ion of one formed t y 10th cycle KCl per 100 viate from ea nt into the gla
esistance of t d glasses sho 23 e. tip. of ml ach ass the ow
Figure 20: Average wear displacement as a function of number of wear cycles.
4
Conclusions and Future Work
The method of chemical strengthening of flat glass by vapour deposition of KCl salt and in-line K+ ion exchange has been demonstrated. It has been shown that it is possible to modify the glass surface. Potassium ions penetrate into the surface and the surface modification changes the properties of the glass surface. The method has the potential to be used in a production line of flat glass e.g. float line, down/up - drawn flat glass or fusion process. The process can also be used for chemically strengthen other glass compositions as well e.g. alkali alumina silicate compositions which are generally more effectively strengthened. Furthermore, the process of salt deposition on flat glass and in-line ion exchange is difficult to predict, it is a complex and challenging process. There are many parameters to control e.g. temperature, concentration, treatment time, pressure in aerosol generator, salt distribution on glass as well as salt adhesion to the glass.
Future work is advised to involve another glass composition and bending strength tests to quantify the actual strengthening effect of the treatment. It is advised to use two furnaces, one furnace for aerosol generation and another furnace for salt deposition so that a constant temperature over the whole samples is created. Also the pressure in the aerosol generator can be modified in order to create a more homogeneous salt layer. Potentially the method can find applications where the penetration depth of chemically strengthened glasses does not need to be so deep and for glass products which only allows low production costs.
5
References
1. Wondraczek, L., J.C. Mauro, J. Eckert, U. Kühn, J. Horbach, J. Deubener, and T. Rouxel, Towards Ultrastrong Glasses. Advanced Materials, 2011. 23(39): p. 4578-4586.
2. Karlsson, S., B. Jonson, and C. Stålhandske, The technology of chemical glass
strengthening - a review. Glass Technology - European Journal of Glass Science
and Technology Part A, 2010. 51(2): p. 41-54.
0 20 40 60 80 100
Number of Wear Cycles -120 -100 -80 -60 -40 -20 0 A v erage Wear D is p la ceme nt (nm) Ref 15g 2h 15g 4h 15g 6h
25
4. Acloque, P. and J. Tochon. Measurement of the mechanical strength of glass
after reinforcement. in Compte Rendu Symposium sur la Résistance Mécanique du Verre et les Moyens de l'Améliorer. 1961. Florence, Italy: Published by Union
Scientifique Continentale du Verre, Charlroi, Belgium 1962, p. 1044.
5. Varshneya, A.K., The physics of chemical strengthening of glass: Room for a new view. Journal of Non-Crystalline Solids, 2010. 356(44-49): p. 2289-2294.
6. Varshneya, A.K., Chemical Strengthening of Glass: Lessons Learned and Yet To Be Learned. International Journal of Applied Glass Science, 2010. 1(2): p.
131-142.
7. Gy, R., Ion exchange for glass strengthening. Materials science & engineering.
B, Solid-state materials for advanced technology, 2008. 149(2): p. 159.
8. Varshneya, A.K. and P.K. Kreski, The Chemistry of Chemical Strengthening of
Glass, in Processing, Properties, and Applications of Glass and Optical Materials 2012, John Wiley & Sons, Inc. p. 107-114.
9. [Anonymous], Corning' Gorilla Glass brings King Kong glass strength to high-tech toys. American Ceramic Society Bulletin, 2010. 89(8): p. 41-42.
10. Geithe, A., Touching Allowed, in The Schott Technology Magazine: Solutions2011. p. 28-29.
11. Press release (http://dragontrail.agc.com) 2011.01.20: AGC Unveils Damage-Resistant Dragontrail™ Glass for Smartphone, Tablet PC, TV Makers
worldwide. 2011 2011.10.28].
12. Press release: http://www.neg.co.jp/EN/pdf/20110418_cx01_en.pdf [Accessed 2014-11-25] - Nippon Electric Glass has started mass-producing specialty glass for chemical strengthening, which is used for cover glass of smart phones and mobile devices [2011-04-18]. 2011.
13. Lee, M.Y.M., Gases for Glass Production, Special Treatment and Coatings.
Gases & Instrumentation International Magazine, 2012.
14. Karlsson, S., Doctoral Thesis: Modification of Float Glass Surfaces by Ion Exchange, 2012, Linnaeus University: Växjö.
15. Karlsson, S., B. Jonson, and L. Wondraczek, Copper, silver, rubidium and
caesium ion exchange in soda-lime-silicate float glass by direct deposition and in line melting of salt pastes. Glass Technology - European Journal of Glass Science
and Technology Part A, 2012. 53(1): p. 1-7.
16. Sil'vestrovich, S.I., L.G. Samkova, E.V. Sobolev, V.V. Gorbachev, A.S. Bystrikov, and E.K. Grigorova, Ion-exchange modification of glass surface by
means of an aerosol method Soviet Journal of Glass Physics and Chemistry,
1986. 12(2): p. 180-187.
17. Sil'vestrovich, S.I., L.G. Samkova, V.D. Kazakov, and L.F. Korshunova,
Strengthening industrial glasses by ion exchange with solid-phase reagents.
Glass and Ceramics, 1984. 41(11): p. 471-474.
18. Patschger, M., Process Optimization for Chemical Strengthening Thin Glasses, in 23rd International Congress on Glass (ICG) 2013: Prague, Czech Republic.
19. Karlsson, S., B. Jonson, P. Sundberg, and C. Stålhandske, Surface Analysis of
float glass using Surface Ablation Cell (SAC) Part 2: Determination of the diffusion characteristics of K+-Na+ Ion Exchange. Glass Technology: European
Journal of Glass Science and Technology Part A, 2010. 51(2): p. 55-62. 20. Schaeffer, H.A., Transport Phenomena and Diffusion Anomalies in Glass.
Materiały Ceramiczne, 2012. 64(2): p. 156-161.
21. Ramaswamy, R.V. and R. Srivastava, Ion-exchanged glass waveguides: a review. Journal of lightwave technology, 1988. 6(6): p. 984.
22. Stavrou, E., D. Palles, E.I. Kamitsos, A. Lipovskii, D. Tagantsev, Y. Svirko, and S. Honkanen, Vibrational study of thermally ion-exchanged sodium
aluminoborosilicate glasses. Journal of Non-Crystalline Solids, 2014. 401(0): p.
23. Ingram, M.D., M.-H. Wu, A. Coats, E.I. Kamitsos, C.P.E. Varsamis, N. Garcia, and M. Sola, Evidence from infrared spectroscopy of structural relaxation during
field assisted and chemically driven ion exchange in soda-lime-silica glasses.
Physics and Chemistry of Glasses - European Journal of Glass Science and Technology Part B, 2005. 46(2): p. 84-89.
24. Ingram, M.D., J.E. Davidson, A.M. Coats, E.I. Kamitsos, and J.A. Kapoutsis,
Origins of Anomalous Mixed-Alkali effects in Ion-Exchanged glasses. Glass
Science and Technology, Glastechnische Berichte 2000. 73(4): p. 89.
25. Leboeuf, V., J.-P. Blondeau, D. De Sousa Meneses, and O. Véron, Potassium
ionic exchange in glasses for mechanical property improvement. Journal of
Non-Crystalline Solids, 2013. 377(0): p. 60-65.
26. Doremus, R.H., Infrared spectroscopy of surfaces of glasses containing alkali ions. Journal of Non-Crystalline Solids, 1980. 41(1): p. 145-149.
27. Husung, R.D. and R.H. Doremus, The infrared transmission spectra of four
silicate glasses before and after exposure to water. Journal of Materials
Research, 1990. 5(10): p. 2209-2217.
28. Tadjiev, D.R. and R.J. Hand, Surface hydration and nanoindentation of silicate
glasses. Journal of Non-Crystalline Solids, 2010. 356(2): p. 102-108.
29. Gonzalez Rodriguez, J.A. and R.J. Hand, Evolution of the modulus and hardness
of the tin and air sides of float glass as a function of hydration time. Glass
Technology - European Journal of Glass Science and Technology Part A, 2013. 54(1): p. 36-41.
30. Kolluru, P.V., D.J. Green, C.G. Pantano, and C.L. Muhlstein, Effects of Surface
Chemistry on the Nanomechanical Properties of Commercial Float Glass.
Journal of the American Ceramic Society, 2010. 93(3): p. 838-847.
31. Fett, T., J.P. Guin, and S.M. Wiederhorn, Stresses in ion-exchange layers of
soda-lime-silicate glass. Fatigue & Fracture of Engineering Materials &
Glafo PG Vejdes väg 15, 351 96 Växjö 010-516 63 50 info@glafo.se www.glafo.se Glafo-rapport 2014:3P00070 SP Sveriges Tekniska Forskningsinstitut
Vi arbetar med innovation och värdeskapande teknikutveckling. Genom att vi har Sveriges bredaste och mest kvalificerade resurser för teknisk utvärdering, mätteknik, forskning och utveckling har vi stor betydelse för näringslivets konkurrenskraft och hållbara utveckling. Vår forskning sker i nära samarbete med universitet och högskolor och bland våra cirka 9000 kunder finns allt från nytänkande småföretag till internationella koncerner.
