Influence of amorphous phase separation on the crystallization behavior of glass-ceramics in the BaO-TiO2-SiO2 system Emilie Boulay, Céline Ragoen, Stéphane Godet ULB, 4MAT Department, CP194/3, 87 Av. Buyl, 1050 Brussels, Belgium Good morning everybody. I’m Emilie Boulay from belgium, Université Libre de Bruxelles. + Céline: part of the study during her  « Final Year Project ». Today I’m going to talk you about the influence of APS on crystallization behavior of glass ceramics in the BTS system. Crystallization 2012 Goslar, Germany 25th September 2012

Outline Influence of phase separation on crystallization Motivations The BaO-TiO2-SiO2 system Results & Discussion Conclusions Perspectives Here is the outline of the presentation: A review of repertoried possible influence of APS on crystallization in glass system in general The motivation to use such influence I will then introduce the particular system that we have studied Results and discussions And finally conclusions and outlooks. Crystallization 2012 - Goslar 2

Amorphous phase separation (APS) APS role Amorphous phase separation (APS) Well known in glass systems APS = matrix + droplets due to liquid immiscibility To first introduce, let me remind what APS is: It is a well known phenomenon in glass system involving the formation of two phases (a dispersed phase an a matrix) due to a liquid immiscibility. Crystallization 2012 - Goslar 3

Amorphous phase separation (APS) APS role Amorphous phase separation (APS) Well known in glass systems APS = matrix + droplets due to liquid immiscibility  Shift in composition + creation of interfaces Interfaces In the droplets The influence of APS is at least two-fold. First, induces a shift in composition of the parent glass and and second it creates amorphous/amorphous interfaces. This may in turn have an effect on the subsequent crystallization. Indeed, it can promote crystallization from the a/a interfaces, inside the droplets or inside the matrix. Shift In the matrix I. Gutzow, J. Schmelzer, “The vitrous state: thermodynamics, structures, rheology and crystallization, Springer, 1995 Crystallization 2012 - Goslar 4

Possible effects of APS on crystallization APS role Possible effects of APS on crystallization No influence Shift matrix Shift droplets Interfaces Interfacial Energy Diffusion zone Na2O-SiO2 [1] BaO-SiO2 [5] LaF3-Na2O-Al2O3-SiO2 [9] Li2O-SiO2 [10] Na2O-CaO-SiO2 [2][3] TiO2-SiO2 [6] BaO-SiO2-TiO2 [11] Li2O-SiO2 [4] MgO-Al2O3-SiO2-TiO2 [7] ZnO-Al2O3-SiO2 [7] ZnO-Al2O3-SiO2-ZrO2 [7] Interfaces do not promote crystallization [3] [4] [12] Li2O-SiO2-P2O5 [8] [1] Scherrer, G W et Uhlmann, D R. , 1976 [8] ] Harper, H et McMillan, P W. , 1972 [2] Hammel, J J. , 1966 [9] Bhattacharyya, S, et al., 2009 [3] Boulay, 2011 [10] Tomozawa, M, et al., 1990 [4] K. Nakagawa, T. Izumitani, 1969 [11] Hijiya, H et al., 2008 [5] James, P F et Ramsden, A H. , 1984 [12] Li Z., 1985 [6] Katsumata, K, et al., 2004 [7] Jiazhi, L et Chih-yao, F. , 1986 A literature review was conducted in order to determine the possible effects of APS on crystallization. As it is presented in the table, influences depends on the system and some contradictions were found. Interface influence was found to be the most contradicted. [1] Surface crystallization with and without APS [2] Interfacial energy is too low [3] Numerous experiments were conducted in the lab but no bulk crystallization was obtained [4] Number of crystals and number of droplet are farly different [5]: James montre les courbes de nucléation pour la compo stoechio, une compo non stoech sans APS une compo avec APS. La compo stoech a la vitesse de nucléation la plus rapide. Il montre qu’a HT, lorsque l’APS a pu se former, la vitesse de nucleation de la compo avec APS augmente bcp plus que celle sans… mais aucune des deux ne rejoint la compo stoech. -> inlfuence de la compo mais PAS des interfaces [6] APS modify the crystal form [7]: the phase shift favors the bulk crystallization [8] idem [9] La3F crystals form inside the droplets and size can be controlled with the silica contained into the droplets [10] A depleted zone low in silica just around the droplets favor nucleation [11] Activation energy is constant with grain size, interfacial energy allows heterogenous nucleation [12] The very small interfacial energy between the two liquid phases regarding the interfacial energy between liquid phase and crystalline phase is in the range of 5 and 50-200 ergs/cm2 in SLS (Hammel) [1] Scherrer, G W et Uhlmann, D R. Effects of phase separation on crystallization behavior. Journal of non-Crystalline Solids. 1976, Vol. 21, pp. 199-213. [2] Hammel, J J. Direct Measurements of Homogeneous Nucleation Rates in Glas-Forming System. The Journal of Chemical Physics. 1966, Vol. 46, 6. [3] Boulay, E., Microstructural optimization of inorganic glasses through amorphous phase separation: application to crystallization, ULB [4] K. Nakagawa, T. Izumitani, Relationship between phase separation and crystallization in Li2O,2,5SiO2 glass and a lithium silicate containing a large amount of titanium oxide, Physics and Chemistry of glasses, Vol. 10, N°5, oct 1969. [5] James, P F et Ramsden, A H. The effect of amorphous phase separation on crystal nucleation kinetics in BaO-SiO2 glasses, Part I: General survey. Journal of Materials Science. 1984, Vol. 19, p. 1406. [6] Katsumata, K, et al., et al. Preparation of phase-separated textures and crystalline phases from two-liquid immiscible melts in the TiO2-SiO2 system. [éd.] Elsevier. Materials Research Bulletin. 2004, Vol. 39, pp. 1131-1139. [7] Jiazhi, L et Chih-yao, F. Prospects on the relationship between liquid-phase separation and crystallization in glass. Journal of non-crystalline solids. 1986, Vol. 87, pp. 387-391. [8] Harper, H et McMillan, P W. The formation of glass-ceramic microstructures. Physics and chemistry of glass. 1972, Vol. 13, 4, pp. 97-101. [9] Bhattacharyya, S, et al., et al. Nano-crystallization in LaF3-Na2O-Al2O3-SiO2 glass. Journal of Crystal Growth. 2009, Vol. 311, pp. 4350-4355. [10] Tomozawa, M, McGahay, V et Hyde, J M. Phase separation of glasses. Journal of Non-Crystalline Solids. 1990, Vol. 123, pp. 197-207. [11] Hijiya, H, Kishi, T et Yasumori, A. Effect of phase separation on crystallisation of glasses in the BaO-TiO2-SiO2 system. Journal of the Ceramic Society of Japan. 2009, Vol. 117, 1, pp. 120-126. [12] Li Z. The effect of liquid-liquid phase separation of glass on the properties and crystallization. Nasa technical memorandum, 1985. Obstacle to growth SrO-SiO2-TiO2 Le mettre en rem. Crystallization 2012 - Goslar 5

Possible effects of APS on crystallization APS role Possible effects of APS on crystallization No influence Shift matrix Shift droplets Interfaces Interfacial Energy Diffusion zone Na2O-SiO2 [1] BaO-SiO2 [5] LaF3-Na2O-Al2O3-SiO2 [9] Li2O-SiO2 [10] Na2O-CaO-SiO2 [2][3] TiO2-SiO2 [6] BaO-SiO2-TiO2 [11] Li2O-SiO2 [4] MgO-Al2O3-SiO2-TiO2 [7] ZnO-Al2O3-SiO2 [7] ZnO-Al2O3-SiO2-ZrO2 [7] Interfaces do not promote crystallization [3] [4] [12] Li2O-SiO2-P2O5 [8] Systematic study of prior amorphous phase separation effect on crystallization in the BaO-TiO2-SiO2 system: Interfaces: debated in literature Photoluminescence properties Crystallization 2012 - Goslar 5

The BaO-TiO2-SiO2 system This glass system exhibits APS by the presence of a large miscibility gap in the silica corner The exact miscibility gap location is unknown Crystal phase near immiscibility: fresnoite (2BaO.TiO2.2SiO2) ? * Hijiya et al., “Effect of phase separation on crystallization of glasses in the BaO-TiO2-SiO2 system”, 2009 Crystallization 2012 - Goslar 6

Technical interests of fresnoite BaO-TiO2-SiO2 Technical interests of fresnoite Fresnoite exhibits blue/white photoluminescence (PL) under ultraviolet excitation PL effect can be optimized by heat treatments on the stoichiometric composition Excitation at 254 nm Response at 470 nm Photoluminescence Well chosen composition inside the BTS phase diagram allows fresnoite to be formed. This crystalline phase exhibits b/w PL under ultra violet excitation. The effetct can be enhanced on the stoichiometric composition by optimizing hte termal treatment. T. Komatsu, “Effect of heat treatment temperature on the optical properties of Ba2TiSi2O8 nanocrystallized glasses”, 2005 Crystallization 2012 - Goslar 7

Possible enhancement of optical properties BaO-TiO2-SiO2 Possible enhancement of optical properties No stoichiometric compositions show also PL effect (=254 nm) Hijiya (2008) suggested phase separation may have an influence on crystallization and PL effect Stoich. Non stoich. As it can be seen, non stoichiometric composition show also PL effect. Hijiya suggested that APS may have en effect on crystallization and further on the PL effect: APS seems to avoid an oriented growth of fresnoite and allows a finer crystallization to be formed, enhancing the PL effect. We decide to investigate this influence on crystallization. SiO2↑ APS [211] intensity vs [002]: orientation PL effect Hijiya et al., “Effect of phase separation on crystallization of glasses in the BaO-TiO2-SiO2 system”, 2009 8

Material investigated – Fresnoite-SiO2 line Results & Discussions Material investigated – Fresnoite-SiO2 line Mixing powder + melting (1500-1560°C, 3H) Air quenched + annealed (600C, 10H) Quenched after melting FRES NoAPS APS FRES: stoichiometric composition Quenched after melting NoAPS: non stoichiometric composition outside the miscibility gap Quenched after melting To study the effect of APS and the possible enhancement regarding the stoichiometric composition, we decided to compare 3 compositions: FRES: stoichiometric composition exhibiting bulk crystallization NoAPS: non stoichiometric composition outside the miscibility gap exhibiting strong surface crystallization APS: non stoichiometric composition inside the miscibility gap The crystallization mechanisme was investigated by DSC, the different microstructures by SEM and the crystals orientation by XRD, EBSD and diffaction with TEM APS: non stoichiometric composition inside the miscibility gap Water-quenched after melting Crystallization 2012 - Goslar 9

Crystallization mechanism by DSC Effect of quenching rate Results & Discussions Crystallization mechanism by DSC Effect of quenching rate Effect of composition Microstructure by SEM Morphologies at early and final stages of crystallization Morphological orientation: Large scale: XRD Small scale: EBSD (FEG – SEM) and ACOM (TEM) Crystallization 2012 - Goslar 10

APS air-quenched versus APS water-quenched Results & Discussions APS air-quenched versus APS water-quenched No prior APS (APS) Prior APS (APS) Crystallization mechanism: As said previously, APS shows phase separation directly after melt air-quenching. In order to see if this prior APS has an influence on result, we compare it with a sample quenched in water. Except a small shift for the crystallization peak, we did not observe any difference. Crystallization 2012 - Goslar 11

Cristallization mechanism – Effect of composition Results & Discussions Cristallization mechanism – Effect of composition FRES APS G>850µm 10°C/min 200<G<850µm 112<G<200µm 25<G<112µm 25<G<112µm 112<G<200µm 200<G<850µm G>850µm 25<G<112µm 40K/min If APS has an effect on the crystallization mechanism and allows bulk crystallization to be generate from the interfaces, APS should have a behavior comparable to FRES. We observe a higher shift with here the increasing granulometry which is a clear sign of surface crystallization. We also observe a higher dispersion with the heating rate. This behavior is typical for surface crystallization mechanism. 30K/min 20K/min 10K/min 5K/min 5K/min 10K/min 20K/min 30K/min 40K/min Crystallization 2012 - Goslar 12

Cristallization mechanism – Effect of composition Results & Discussions Cristallization mechanism – Effect of composition NoAPS APS 10°C/min 25<G<112µm 112<G<200µm 200<G<850µm G>850µm 25<G<112µm 112<G<200µm 200<G<850µm G>850µm 25<G<112µm We compare now with the non stoichiometric composition and see … Peaks are clearly more spread in APS case and peaks at low heating rate are assymmetric. 40K/min 30K/min 20K/min 10K/min 5K/min 5K/min 10K/min 20K/min 30K/min 40K/min Crystallization 2012 - Goslar 13

Determination of Avrami parameters Results & Discussions Determination of Avrami parameters n (Ozawa’s method) x= crystallized fraction n= Avrami’s parameter (growth dimension) m= Avrami’s parameter (growth direction) Granulo [µm] Avrami parameter « n » FRES NoAPS APS 25<G<112 2,0±0,1 ~ 2 112<G<200 200<G<850 G>850 1,3±0,3 ~ 1 The determination of the n parameter was carried out using the Ozawa method. For 4 granulometries, the crystallized fraction was calculated at given temperature. According to the following equation, the n parameter can be calculated from a linear regression. An average n value of 3 for FRES indicates clearly a 3D growth of crystal. The Values for the APS the are more spread between 2 and 1. The same also applies for NoAPS but with a smaller spread Crystallization 2012 - Goslar 14

Activation energy Eact (Matusita’s method) Results & Discussions = heating rate [K/min] n= Avrami’s parameter (growth dimension) Tp=Max crystallisation peak Eact=activation energy [kJ/mol] m= Avrami’s parameter (growth direction) R= gas constant Eact (Matusita’s method) The activation energies for FRES, NoAPS and APS were calculated using the Matusita’s method. We observe that FRES has a constant E act, as it is the case with bulk crystallisation. E act of APS is lower than NoAPS but tendency are roughtly the same. For all composition, E act is nearly constant at higher granulometry. Pour les grandes granulos, on voit que NoAPS et APS ont approximativement la même valeur, par contre les résultats obtenus diffèrent pour les petites granulometries où Eact n’est pas constante pour APS -> l’interpretation est conséquemment tout à fait différente. Crystallization 2012 - Goslar 15

Activation energy Eact (Matusita’s method) Results & Discussions = heating rate [K/min] n= Avrami’s parameter (growth dimension) Tp=Max crystallisation peak Eact=activation energy [kJ/mol] m= Avrami’s parameter (growth direction) R= gas constant Eact (Matusita’s method) Hijiya et al., “Effect of phase separation on crystallization of glasses in the BaO-TiO2-SiO2 system”, 2009 The activation energies for FRES, NoAPS and APS were calculated using the Matusita’s method. We observe that FRES has a constant E act, as it is the case with bulk crystallisation. E act of APS is lower than NoAPS but tendency are roughtly the same. For all composition, E act is nearly constant at higher granulometry which is in aggreement with previous publication. This is not the case for small granulometry. Pour les grandes granulos, on voit que NoAPS et APS ont approximativement la même valeur, par contre les résultats obtenus diffèrent pour les petites granulometries où Eact n’est pas constante pour APS -> l’interpretation est conséquemment tout à fait différente. Crystallization 2012 - Goslar 15

Crystallization mechanism by DSC Effect of quenching rate Results & Discussions Crystallization mechanism by DSC Effect of quenching rate Effect of composition Microstructure by SEM Morphologies at early and final stages of crystallization Morphological orientation: Large scale: XRD Small scale: EBSD (FEG – SEM) and ACOM (TEM) This DSC results clearly point toward both APS and NoAPS undergo SURFACE crystallization. Let’s now turn to the corresponding microstructures and their evolution during heat treatments. Crystallization 2012 - Goslar 10

Crystal morphologies – Final stage Results & Discussions Crystal morphologies – Final stage It changes and becomes finer Finer crystallization should mean PL enhancement … 1000°C 72H FRES NoAPS 5 µm 5 µm SiO2   + APS Regarding the crystal morphologies: The following SEM images present microstuctures for long treatments (final stage of crystallization). As seen with DSC, they are completly different. Those composition samples are different regarding the location in the miscibility gap and their composition APS 5 µm Crystallization 2012 - Goslar 16

Crystal morphologies – Final stage Results & Discussions Crystal morphologies – Final stage It changes and becomes finer Finer crystallization should mean PL enhancement … 1000°C 72H FRES NoAPS 5 µm 5 µm SiO2   + APS Regarding the crystal morphologies: The following SEM images present microstuctures for long treatments (final stage of crystallization). As seen with DSC, they are completly different. Those composition samples are different regarding the location in the miscibility gap and their composition APS More study on early and final stages 5 µm Crystallization 2012 - Goslar 16

APS crystal morphologies – Final stage Results & Discussions APS crystal morphologies – Final stage Longs treatments: fine microstructure with sometime the disappearance of APS Role of APS if same final microstructure ??? Surface: 950°C 24h Surface: 950°C 72h [HIJ] – 1200°C 24H A final stage, a fine crystallization is obtained, sometimes with the presence of APS. Result are similar with those of Hij who dealted essentially with final stage. Role APS or composition (SiO2 increases so does viscosity) Dendrites vers fine? Dendrites are not stable from a morphological point of view Crystallization 2012 - Goslar 17

APS crystal morphologies – Early stage Results & Discussions APS crystal morphologies – Early stage Complex crystallization microstructures with surface crystallization Role of APS as nucleation site? a/a interfacial energy too low… Conditions to become homogeneously fine : dendrite fragmentation? Energie interface Here we can see the complex crystallization mechanism. Morphologies can be very different (dendrites, fine crystallization), but surface is always crystallized before bulk. Crystallization 2012 - Goslar 18

Crystallization mechanism by DSC Effect of quenching rate Results & Discussions Crystallization mechanism by DSC Effect of quenching rate Effect of composition Microstructure by SEM Morphologies at early and final stages of crystallization Morphological orientation: Large scale: XRD Small scale: EBSD* (FEG – SEM) and ACOM** (TEM) *Electron BackScattered Diffraction ** Automated Crystallographic Orientation Mapping We observed the most finer microstructure with APS from SEM image even if their are not single at early stage. Because it seems that dendrites finally transform into « speroidized » structure, we interested about orientation of this crystal structure and see if it leads to a complete desoriented crystallization or not.. Crystallization 2012 - Goslar 10

Orientation - XRD APS: case 1 FRES APS: case 2 NoAPS Results & Discussions Orientation - XRD 002 001 211 002 211 001 APS: case 1 FRES Concerning the orientation, We clearly see a different between FRES and APS…. Again APS shows several case but most of the time, [002] is preferred. APS: case 2 NoAPS Crystallization 2012 - Goslar 19

Crystallographic orientation – EBSD and ACOM Results & Discussions Crystallographic orientation – EBSD and ACOM We conducted orientation study further by using more specific technique as EBSD and TEM. No orientation for FRES Orientated dendrite… or dendrites growing from each other The APS must be conducted further, the first results indicated large grain with one orientation. Droplets does not seem to be correlated with orientation changes. 15° FRES - ACOM NoAPS – EBSD APS - ACOM Crystallization 2012 - Goslar 20

Conclusions Mechanism: surface crystallization for APS and NoAPS Microstructures: evidence of surface crystallization for APS and NoAPS NoAPS: dendrites APS Short treatments: complex microstructures, no clear crystallization from interfaces Long treatments: single microstructure, sometimes without APS ! Morphological and crystallographic orientation Often [002] oriented growth for APS Amorphous droplets do not seem to have any influence  This systematic study shows that APS does not influence the crystallization mechanism Possible explainations: Amorphous/amorphous interfacial energy too low to promote crystallization Effect of viscosity through composition … Crystallization 2012 - Goslar 21

Perspectives Clarify APS’s role on morphology … By using compositions closer to the miscibilty gap boundary – avoid the « composition effect » By developing EBSD measurements on early and final crystallization for APS By finding mid-treatment to clarify the transformation into fine crystallization Why amorphous phase separation disappears with fresnoite dendrites ? … Crystallization 2012 - Goslar 22

Thank you for your attention Any questions? We acknowledge the financial support of FRIA. Crystallization 2012 - Goslar

APS crystal morphologies – Early stage Results & Discussions APS crystal morphologies – Early stage Complex crystallization microstructure with surface crystallization Role of APS as nucleation site? Infertial energy too low… Under which conditions it becomes homogeneously « speroïdized » ? Energie interface Here we can see the complex crystallization mechanism. Morphologies can be very different (dendrites, fine crystallization), but surface is always crystallized before bulk. Crystallization 2012 - Goslar 14