Nanocrystallization in Oxyfluoride Glasses Controlled by Amorphous

Aug 27, 2015 - ... Shiyu Sun , Gang-Ding Peng , Elfed Lewis , Jing Ren , Jianzhong Zhang ... Martina Stoica , Martin Brehl , Christian Bocker , Andrea...
0 downloads 0 Views 8MB Size
Letter pubs.acs.org/NanoLett

Nanocrystallization in Oxyfluoride Glasses Controlled by Amorphous Phase Separation Changgui Lin,*,†,‡ Christian Bocker,‡ and Christian Rüssel‡ †

Otto-Schott-Institut für Materialforschung, Jena University, Fraunhoferstraβe 6, 07743 Jena, Germany Laboratory of Infrared Materials and Devices, Research Institute of Advanced Technology, Ningbo University, Ningbo 315211, China

Downloaded by UNIV OF CAMBRIDGE on September 6, 2015 | http://pubs.acs.org Publication Date (Web): August 31, 2015 | doi: 10.1021/acs.nanolett.5b02605



ABSTRACT: Transparent bulk glass-ceramics containing ZnF2, K2SiF6, and KZnF3 nanocrystals are successfully obtained from xKF−xZnF2−(100 − 2x)SiO2 oxyfluoride glasses for the first time to the best of our knowledge. The glass transition temperatures of heat-treated samples increase with time and approach values that resemble the temperatures chosen for thermal treatment. During nucleation and crystal growth, the residual glass around the crystals is depleted in fluoride which as glass component usually leads to a decrease in viscosity. The crystallization behavior notably depends on the glass composition and changes within a small range from x = 20 to 22.5 mol %. The occurrence of liquid/ liquid phase separation in dependence of the composition is responsible for the physicochemical changes. Two different microstructures of droplet and interpenetrating phase separation and their compositional evolution are observed by replica transmission electron microscopy technique in the multicomponent glassy system. This study suggests that the size and crystal phase of precipitated crystallites can be controlled by the initial phase separation. KEYWORDS: Nanocrystallization, oxyfluoride glass, glass-ceramic, phase separation, KZnF3, K2SiF6

O

above conception and suggest that amorphous phase separation (APS) in glasses would assist nanocrystallization.16−19 APS in glasses is a longstanding essential topic of glass research.20,21 It indeed provides unique opportunities not only for designing materials with hierarchical microstructure at different length scales and unique properties22−24 but also for the scientific understanding of glass structure theories.25,26 In general, APS exists in various multicomponent glasses and is recognized as the first stage of crystallization in many glass compositions. The phase incompatible with the host glass matrix would promote the precipitation of the crystalline nuclei with the assemblage of enriched elements. Such a behavior is a primary origin of controlled crystallization in glasses, which leads to a remarkable number of glass-ceramic materials with exceptional optical and mechanical properties. In principle, APS includes droplet and interpenetrating phase separation. The nanocrystallization mentioned above in oxyfluoride glasses were evidenced to be controlled by the droplet phase separation.27−29 It is clear that the droplet APS may initiate crystallization through the nucleation within them, and subsequent crystal growth is restricted within the size of the droplet because of the formation of diffusion barrier originated from the compositional difference between the droplet, the crystal, and the matrix phase. However, only few works were

xyfluoride glass-ceramic is a newly growing class of photonic materials comprised of fluoride crystallites (FCs) and oxide glasses. Because of the joint advantages of the low phonon energy of FCs and good durability and mechanical properties of oxide glasses, it has attracted considerable attention as an effective host matrix for active ions recently. In the numerous studies performed within the past few years,1−5 most of them dealt with two main issues: nanocrystallization mechanism and optical properties of active ions doped oxyfluoride glass-ceramics. The former aims at the controllable nanocrystallization of FCs in glass matrix; the latter focuses on the intriguing optical properties of oxyfluoride glassceramics doped with rare earth ions for the applications in telecommunications and optoelectronics, for example, lightemitting diodes, solid-state lasers, optical amplifiers, and so forth. Thus, a large number of oxyfluoride glass-ceramics were synthesized with the precipitation of different FC phases, such as CaF2,6,7 SrF2,8 PbF2,9 Sr2RF7 (R = Y, Gd, or La),10 LaF3,5,11 NaYbF4,12 and NaGdF4.13 More importantly, a mechanism that a diffusion barrier around each crystal is formed that hinders further crystal growth was proposed recently for guiding controllable nanocrystallization of FCs in glass matrices.14 It enlightens us about the preparation of nanosized FCs, because the nanometric size of the crystals can be controlled by the formation of diffusion barrier around each crystal.4,15 Intensive investigations of both experimental evidence and theoretical explanation have been conducted to successfully validate the © XXXX American Chemical Society

Received: July 2, 2015 Revised: August 12, 2015

A

DOI: 10.1021/acs.nanolett.5b02605 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Downloaded by UNIV OF CAMBRIDGE on September 6, 2015 | http://pubs.acs.org Publication Date (Web): August 31, 2015 | doi: 10.1021/acs.nanolett.5b02605

Nano Letters

Results and Discussion. In contrast to the most frequently studied oxyfluoride glass-ceramic compositions that based on Wang and Ohwaki’s rule,34 here SiO2 glassy phase was selected as the sole target matrix to maintain and support possible FCs. It would be an excellent fit for the present silica fiber system. As a start, the nanocrystallization study was conducted on 15KF− 15ZnF2−70SiO2 (KZS70) glass through the thermal treatment at 530 °C for different periods of time. As shown in Figure 1,

devoted to study the nucleation and crystallization in interpenetrating phase-separated glasses. Especially, there is no report of nanocrystallization in oxyfluoride glasses with interpenetrating phase separation. In interpenetrating phaseseparated oxide glasses, it is suggested that anhedral crystallization would happen to generate an internal microstructure that inherits the interlocking morphology of the original phase separation30 and result in better mechanical properties than that of other glass-ceramics with standard microstructures. Hence, it is of special scientific interest and technological potential to get insight into the relationship between APS and nanocrystallization behavior in oxyfluoride glasses. It would help to achieve fluoride nanocrystals embedded oxyfluoride glass-ceramics with functional microstructures. Herein, we introduce an effective approach to precipitate an important FC group of ZnF2, K2SiF6, and KZnF3 in transparent bulk oxyfluoride glass-ceramics through APS-assisted crystallization. ZnF2, K2SiF6, and KZnF3 could provide octahedral sites for the substitution of transition metal ions,31−33 such as Si4+ for Mn4+ and Zn2+ for Mn2+ or Ni2+, respectively, make them of great interest for transition metal ions based photonic applications. These crystals were studied intensively in crystals that were obtained by single crystal growth or hydrothermal methods. Nevertheless, their processing ability is very poor for further photonic or optoelectronic applications. An effective way is to engineer such crystals directly in the well-adopted silica glass. This article is a first report on the preparation of innovative oxyfluoride glass-ceramics containing ZnF2, K2SiF6, and KZnF3 nanocrystals. We show that controllable precipitation of different fluoride nanocrystals can be achieved via the self-organized phase separation with droplet and interpenetrating microstructures. The relationship between composition, APS, and nanocrystallization behavior is also explored and discussed in KF−ZnF2−SiO2 glasses. Experimental Procedure. Oxyfluoride glasses with the nominal compositions of xKF−xZnF2−(100 − 2x)SiO2 (in mol %, x = 15, 20, 22.5, and 25) were prepared from high purity raw materials of SiO2, KF, and ZnF2. These samples will be referred to as KZS(100 − 2x) hereinafter. Batches of 100 g were melted in a covered Pt−Pd crucible in the temperature range from 1450 to 1500 °C. Bulk samples were obtained by casting onto cold brass plate and annealing the samples subsequently at 450 °C. Then glass slices with the thickness of 2 mm were cut and optically polished on both sides. Thermal analysis measurements were carried out using a dilatometer (Netzsch DIL 402 PC) with the heating rate of 5 K·min−1 from which the glass transition temperature, Tg was determined. Glass-ceramic samples were obtained by annealing the glass slices at the temperatures above Tg for different periods of time. X-ray diffraction (XRD) was performed on powder samples that were crushed to the size smaller than 100 μm. The diffraction patterns were recorded using an X-ray diffractometer (D5000, Siemens, Germany) with Cu−Kα irradiation. Transmission electron microscopy (TEM) at 200 kV (HITACHI H8100, Japan) was used in order to study the microstructure of some selected samples. Bulk samples were mechanically polished to 120 μm thickness, dimpled to about 10 μm thickness, and finished by gentle ion beam milling (3 kV, 1.5 mA). TEM-replica samples were prepared by etching the fractured surface with diluted HF (5%) and coated afterward by Pt−Ir−C in slanted shadowing.21

Figure 1. XRD patterns of 15KF−15ZnF2−70SiO2 (KZS70) samples heat treated at 530 °C for 0 (base), 5, 10, 15, 20, 30, and 50 h, respectively. The standard JCPDF cards of no. 85-1382 K2SiF6 and 711971 ZnF2 are also collected. The diffraction peak marked by inverted triangle might belong to K4Si8O18 crystal phase (JCPDF card no. 701056). The inset displays the photos of base glass and sample heat treated at 530 °C for 10 h.

the KZS70 glass has two orders of appearance of crystallizing phases. The sole precipitation of K2SiF6 (JCPDF no. 85-1382) occurs during the first 10 h of the thermal treatment, and the further annealing process leads to the precipitation of ZnF2 (JCPDF no. 71-1971) and some other diffraction peaks that might be ascribed to K4Si8O18 crystallites (JCPDF no. 701056). The obtained glass-ceramics maintain a high transparency (the inset photographs in Figure 1), thus indicating the successful prevention of the collapse of the SiO2 glassy network and demonstrating the promising prospect for future photonic applications. Also, the high transparency of the obtained glassceramics should be due to the nanosize of the K2SiF6 crystal phase. TEM characterizations of the KZS70 sample crystallized at 530 °C for 20 h are shown in Figure 2. Nanocrystals with sizes in the range from 20 to 30 nm are precipitated with different morphologies as indicated by the arrows in Figure 2a− c. The small crystal size of below 50 nm avoids light scattering and results in the high transparency as shown in the inset of Figure 1. In the micrograph of Figure 2, cubic (spherical) and rodlike crystallites are clearly observed. They might be attributed to the FCs of cubic K2SiF6 and tetragonal ZnF2, respectively. This conclusion is also supported by the mass thickness contrast. The K2SiF6 crystals possess the smaller calculated density of 1.19 g·cm−3 in comparison to the glass (2.35 g·cm−3) and therefore appear bright. The ZnF2 crystals scatter more electrons due to larger density of 3.11 g·cm−3 and B

DOI: 10.1021/acs.nanolett.5b02605 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Downloaded by UNIV OF CAMBRIDGE on September 6, 2015 | http://pubs.acs.org Publication Date (Web): August 31, 2015 | doi: 10.1021/acs.nanolett.5b02605

Figure 2. TEM images of KZS70 sample crystallized at 530 °C for 20h with different magnification factors from (a−e).

as-casted sample, only diffuse humps can be observed, indicating its amorphous character. Thermal treatment within the first 20 h results in the appearance of diffraction peaks that all are attributable to cubic pervoskite KZnF3 (JCPDF card no. 89-4110). When the time of heat treatment is longer than 20 h, a new crystalline phase ZnF2 appeared, accompanying by another undefined SiO2-related phase. TEM images in Figure 4 reveal the microstructure of the KZS50 sample crystallized at 520 °C for 15 h. It illustrates monodispersed cubic nanoparticles that appear dark and clearly as cubes in Figure 4a,b. The measured sizes of these particles are in the range from 30 to 50 nm, as shown in Figure 4c,d. Furthermore, it can be seen that most of these particles possess a perfect cubic structure. The slightly bigger size of the KZnF3 crystallites, compared with that of K2SiF6 in KZS70, strongly increases the light scattering and leads to the turbidity inside the annealed sample as shown in the inset of Figure 3. Selected area electronic diffraction (SAED) results in bright spots as shown in Figure 4e. This confirms the nanocrystalline character of these cubic particles. It should be noted that the calculated density of the KZnF3 crystal (2.26 g·cm−3) is smaller than the glass KZS50 (2.70 g·cm−3) and the dark appearance of the crystals in the bright field images is not due to mass thickness contrast but diffraction contrast. So some crystals appear dark (strong diffraction and overlay of diffraction pattern resulting in bright halo) and some bright (marked by circle in Figure 4d) in dependence of their orientation to the incident electron beam. Although the crystallization behavior of these two samples is widely different, one similarity is their Tg determined by dilatometer measurements that shows the same dependency upon the heat-treated time as presented in Figure 5. Glasses and thermally treated samples of KZS70 and KZS50 were drilled out and cut into rods with the size of 6 mm × 18 mm for dilatometer measurements (the inset of Figure 5a). All dilatometric curves have similar profile (the inset of Figure 5a) and exhibit a kink allowing to determine the characteristic temperatures of glass transition temperature, Tg. With the prolongation of thermal treatment, the Tg of the KZS70 samples (Figure 5a) increases rapidly from 466 to 519 °C within the initial 10 h and does not further increase after thermal treatment of longer time. Similar phenomena were also observed in the KZS50 glass and the crystallized samples as shown in Figure 5b. After crystallization treatment at 520 °C, the Tg of KZS50 samples shows a steep increase from 460 to

therefore appear dark. It should be noted that the images contain also diffraction contrast visible by the bright halo of some dark appearing crystals that is the overlay of the diffraction pattern of strongly diffracting crystals. The images in Figure 2d,e show lattice fringes of large interplanar spacing of approximately 1.15 nm, which is in good agreement with the spacing of 100 planes in K4Si8O18 crystals (1.1487 nm according to JCPDF card no. 70-1056). This is further evidence for the precipitation of K4Si8O18 crystallite that gives rise to the small protuberance at 2θ = 19.6° in the XRD pattern recorded after the thermal treatment longer than 20 h (see Figure 1). Crystallization was furthermore achieved in 25KF−25ZnF2− 50SiO2 (KZS50) glassy samples through thermal treatment at 520 °C for different periods of time. Figure 3 shows the XRD patterns of KZS50 samples before and after annealing. In the

Figure 3. XRD patterns of 25KF-25ZnF2-50SiO2 (KZS50) samples heat treated at 520 °C for 0 (base), 5, 10, 15, 20, 25, 30, 40, and 50 h, respectively. The standard JCPDF cards of no. 89-4110 KZnF3 and 711971 ZnF2 are also collected. The inset displays the photos of base glass and sample heat treated at 520 °C for 10 h. C

DOI: 10.1021/acs.nanolett.5b02605 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Downloaded by UNIV OF CAMBRIDGE on September 6, 2015 | http://pubs.acs.org Publication Date (Web): August 31, 2015 | doi: 10.1021/acs.nanolett.5b02605

Nano Letters

Figure 4. TEM images of KZS50 sample crystallized at 520 °C for 15 h with different magnification factors from (a−d) and the SAED pattern (e). (b) Figure is a dark-field image.

during crystal growth. In the case of a phase-separated glass, nucleation and crystal growth preferentially occurs in the phase enriched in those compounds that decrease the viscosity, that is, predominantly in network modifiers and fluoride (including K, Zn, and alkaline earth elements). These locations possess good diffusion ability due to low viscosity. It results in the precipitation of K2SiF6 and KZnF3 nanocrystals during the first crystallization stage in the glasses KZS70 and KZS50, respectively. In a second step, ZnF2 crystals were formed that further results in a change of the chemical composition in the residual glassy matrix. Finally, network former, that is, SiO2enriched particles, were crystallized. The viscosity of the whole residual glass increases toward 1012 to 1013 Pa·s during this process, that is, the Tg of the residual glass matrix would reach the temperature at which the thermal treatment is carried out as indicated by Figure 5 and then the crystallization process is frozen in. Thus, the diffusion barrier (glass matrix with rigid structure) formed during nucleation and subsequent crystal growth contributes to the self-limited growth of K2SiF6 and KZnF3 nanocrystals, which is in analogy to that in other systems.14,16,17,35 The phase separation discussed here will be further evidenced hereinafter. To further study the effect of glass composition, the crystallization behavior has been investigated explicitly in the xKF−xZnF2−(100 − 2x)SiO2 glasses. In Figure 6, the XRD patterns illustrate the dependency of glass composition on the crystallization behavior after the thermal treatments at Tg + 70 K for 10 h. Samples of KZS70 and KZS60 that possess high SiO2 concentrations show the precipitation of K2SiF6 crystals after thermal treatment, while KZnF 3 crystallites are precipitated in the compositions KZS55 and KZS50. Even though there is only 5 mol % difference in the SiO 2 concentration, the crystallization behavior of KZS60 is clearly different from that of KZS55. Figure 7 presents the TEM images of the freshly fractured surfaces (etched in 5% HF for 5−10 s) of KZS70, KZS60, KZS55, and KZS50 as-casted glasses using replica technique. It is seen that different structures of nanoscale phase separation occurred with the compositional variation in KF-ZnF2−SiO2 glasses. KZS70 shows a clear droplet phase separation (Figure 7a), while the glass KZS50 possesses an interpenetrating structure (Figure 7d). Figure 7b shows the electron micrograph of the KZS60 glass containing the microheterogeneities that are

Figure 5. Glass transition temperature, Tg, as a function of the heattreated time: (a) KZS70 and (b) KZS50 samples after thermal treatment at 530 and 520 °C, respectively. The inset in panel a is a typical dilatometric curve of KZS70 base glass with a heating rate of 5 °C/min and typical photograph of the sample rod of 6 mm × 18 mm for the dilatometric measurements. The inset in panel b is the parameters obtained through fitting the Tg values with the equation of Tg = A + B(1 − e−C·t).

536 °C within the first 20 h and remains constant within the error limit. The listed parameters in the inset of Figure 5b are obtained by a fit according to the equation of Tg = A + B(1 − e−C·t),13 where A = 460 °C, B = 75 °C, and C = 0.1460 h−1. In the same way, the parameters A, B, and C also can be obtained for KZS70 samples from the fit process in Figure 5a, which are equal to 466 °C, 55 °C, and 0.2522 h−1, respectively. The limiting value of Tg increases to 535 °C for KZS50 glass (thermally treated at 520 °C), whereas this value is 521 °C for the KZS70 glass after thermal treatment at 530 °C. It is worthy to be emphasized that the resulting Tg increases with the thermal treatment and approaches a limiting value that is in line with the heat-treated temperature within an error limit. The crystallization behavior of KZS70 and KZS50 is distinct from each other as described above. Nevertheless, it still follows the main principle of crystallization in nonisochemical systems that crystal growth velocities should always depend on time because the chemical composition near the crystals changes D

DOI: 10.1021/acs.nanolett.5b02605 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Downloaded by UNIV OF CAMBRIDGE on September 6, 2015 | http://pubs.acs.org Publication Date (Web): August 31, 2015 | doi: 10.1021/acs.nanolett.5b02605

Nano Letters

glass, crystallization would preferentially happen in the regions containing fluoride due to the low viscosity. With the dissipation of these regions in the glass matrix, Tg of the obtained glass-ceramics increases indicating the viscosity around the crystals is increasing. Thus, the crystal growth is inhibited by the formation of diffusion barriers around the crystals. The Tg will first increase and in a later stage not change any more as shown in Figure 5. That is, the crystals can only grow to some extent, resulting in the nanosize of the precipitated crystals. This is a kind of self-organization that is not so different from a homogeneous crystallization of fluorides as, for example, observed for the cases of Na2O/K2O/BaO/ Al2O3/SiO2/MF2 (M = Ca or Ba) glasses,14,16,17,35 in which the precipitation of alkaline earth fluoride nanocrystals occurs. In particular, a diffusion barrier of SiO2 shell was clearly observed in the Na2O/K2O/MO/Al2O3/SiO2/BaF2, showing the mechanism of the self-limited crystal growth in such glasses.17 On the other hand, different crystal phases could be designed and controlled through the different APS structure inspired by the compositional variation (see Figure 6). In multicomponent systems without stable phase separation, droplet or interpenetrating structures are always formed during the undercooling of glassy melts between the liquidus temperature and Tg. It has been well explained by the thermodynamics and kinetics of immiscibility depending on the composition and the resulting bonding conditions. The phase-separated structures are thermodynamically more favorable than the nonphaseseparated glass, although they are not attributed to the thermodynamically stable state, which below the liquidus temperature is the crystalline state. The above-mentioned experimental results reveal a principle of APS controlled crystallization behavior in glasses that can be traced as follows. The results clearly show that nucleation occurs via two different mechanisms depending on the glass composition in the xKF− xZnF2−(100 − 2x)SiO2 system. One of the microstructural features that can be observed in Figure 7a is the spherical APS embedded in another glassy matrix. In this particular case of the droplet phase-separated glasses with x < 20 mol %, the spherical structures rise above the matrix in the TEM replica micrograph (which can be proved by dust particles on top of surface, not shown in the micrograph). This means that the etching rate of the droplet phase is smaller in comparison to the matrix. Usually the SiO2 network is attacked stronger by HF, that is, the droplets containing fluoride and network modifiers (including K and Zn) become visible in the SiO2 matrix. Because of a large amount of Si in these droplet regions, K2SiF6 is first precipitated in these regions. During annealing, the viscosity of the matrix phase should hence increase with the enrichment in SiO2. This should also lead to decreasing diffusion coefficients in the matrix and hence to a deceleration of the crystal growth of K2SiF6. Thus, within the droplet regions ZnF2 crystallites start to be precipitated with the dissipation of Si-containing compounds. It can be also noticed that the glasses with the compositions of x > 20 mol % exhibit typical spinodally separated structures of two interpenetrating glassy phases (Figure 7d). Such glasses possess interconnected transporting channel for the mobile ions. Thus, because of the high mobility of the constituent elements, KZnF3 is preferentially precipitated in the interpenetrating phaseseparated KZS50 glass and also in the homogeneous KZS55. The nanocrystallization mechanism of the interpenetrating phase-separated glass is much more complicated than that of

Figure 6. XRD patterns of 15KF−15ZnF2−70SiO2 (KZS70), 20KF− 20ZnF2−60SiO2 (KZS60), 22.5KF−22.5ZnF2−55SiO2 (KZS55), and 25KF−25ZnF2−50SiO2 (KZS50) samples crystallized at Tg + 70 K for 10 h. The standard JCPDF cards of no. 894110 KZnF3 and 851382 K2SiF6 are also shown for indexing the diffraction peaks.

Figure 7. Replica TEM image of KF−ZnF2−SiO2 base glasses: (a) KZS70, (b) KZS60, (c) KZS55, and (d) KZS50.

near the resolution limit of the used replica technique, and homogeneous microstructure occurred in the sample KZS55 (Figure 7c). With the compositional variation, that is, decreasing SiO2 concentration, KZS60 show a decreased droplet size (barely discernible) and KZS55 appears homogeneous within the limits of the replica technique. The transition structures of phase separation from droplet to interpenetrating are displayed in Figure 7. The shadow-like circular areas in Figure 7c are artifacts caused by the droplet residues after the etching and dissolution process. Only the structure of the carbon membrane can be seen. Two main themes in the nanocrystallization mechanism of glasses are the controlled precipitation of nanosized crystals and the designed crystal phases for various functional purposes. In the present case, both of them can be achieved through the initial APS. On one hand, for the phase-separated oxyfluoride E

DOI: 10.1021/acs.nanolett.5b02605 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Downloaded by UNIV OF CAMBRIDGE on September 6, 2015 | http://pubs.acs.org Publication Date (Web): August 31, 2015 | doi: 10.1021/acs.nanolett.5b02605

Nano Letters

(6) Huang, L.; Yamashita, T.; Jose, R.; Arai, Y.; Suzuki, T.; Ohishi, Y. Appl. Phys. Lett. 2007, 90 (13), 131116. (7) Ye, S.; Zhu, B.; Chen, J.; Luo, J.; Qiu, J. R. Appl. Phys. Lett. 2008, 92 (14), 141112. (8) Reben, M. J. Non-Cryst. Solids 2011, 357 (14), 2653−2657. (9) Tikhomirov, V. K.; Furniss, D.; Seddon, A. B.; Reaney, I. M.; Beggiora, M.; Ferrari, M.; Montagna, M.; Rolli, R. Appl. Phys. Lett. 2002, 81 (11), 1937−1939. (10) Qiao, X.; Fan, X.; Wang, M.; Yang, H.; Zhang, X. J. Appl. Phys. 2008, 104 (4), 043508. (11) Xu, Y.; Zhang, X.; Dai, S.; Fan, B.; Ma, H.; Adam, J.-l.; Ren, J.; Chen, G. J. Phys. Chem. C 2011, 115 (26), 13056−13062. (12) Wei, Y.; Chi, X.; Liu, X.; Wei, R.; Guo, H. J. Am. Ceram. Soc. 2013, 96 (7), 2073−2076. (13) Herrmann, A.; Tylkowski, M.; Bocker, C.; Rüssel, C. Chem. Mater. 2013, 25 (14), 2878−2884. (14) Rüssel, C. Chem. Mater. 2005, 17, 5843−5847. (15) de Pablos-Martin, A.; Munoz, F.; Mather, G. C.; Patzig, C.; Bhattacharyya, S.; Jinschek, J. R.; Hoche, T.; Duran, A.; Pascual, M. J. CrystEngComm 2013, 15 (47), 10323−10332. (16) Almeida, R. m. P. F. d.; Bocker, C.; Rüssel, C. Chem. Mater. 2008, 20 (18), 5916−5921. (17) Bhattacharyya, S.; Bocker, C.; Heil, T.; Jinschek, J. R.; Hoche, T.; Rüssel, C.; Kohl, H. Nano Lett. 2009, 9 (6), 2493−2496. (18) Haas, S.; Hoell, A.; Wurth, R.; Rüssel, C.; Boesecke, P.; Vainio, U. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 81 (18), 184207. (19) Hoell, A.; Varga, Z.; Raghuwanshi, V. S.; Krumrey, M.; Bocker, C.; Russel, C. J. Appl. Crystallogr. 2014, 47 (1), 60−66. (20) Jiang, Z.; Zhang, Q. Prog. Mater. Sci. 2014, 61, 144−215. (21) Vogel, W. Glass Chemistry, 2nd ed.; Springer-Verlag: New York, 1994. (22) Karpukhina, N.; Hill, R. G.; Law, R. V. Chem. Soc. Rev. 2014, 43 (7), 2174−2186. (23) Beall, G. H.; Pinckney, L. R. J. Am. Ceram. Soc. 1999, 82 (1), 5− 16. (24) Pinckney, L. R.; Beall, G. H. J. Am. Ceram. Soc. 2008, 91 (3), 773−779. (25) Wright, A. C.; Thorpe, M. F. Phys. Status Solidi B 2013, 250 (5), 931−936. (26) Zachariasen, W. H. J. Am. Chem. Soc. 1932, 54 (10), 3841− 3851. (27) Bocker, C.; Russel, C.; Avramov, I. Int. Appl. Glass Sci. 2013, 4 (3), 174−181. (28) Bocker, C.; Wiemert, J.; Rüssel, C. J. Eur. Ceram. Soc. 2013, 33 (10), 1737−1745. (29) de Pablos-Martín, A.; Hém ono, N.; Mather, G. C.; Bhattacharyya, S.; Höche, T.; Bornhöft, H.; Deubener, J.; Muñoz, F.; Durán, A.; Pascual, M. J. J. Am. Ceram. Soc. 2011, 94 (8), 2420−2428. (30) Beall, G. Glass Technol.- Part A 2004, 45 (2), 54−58. (31) Brik, M. G.; Kumar, G. A.; Sardar, D. K. Mater. Chem. Phys. 2012, 136 (1), 90−102. (32) Song, E.; Ding, S.; Wu, M.; Ye, S.; Xiao, F.; Zhou, S.; Zhang, Q. Adv. Opt. Mater. 2014, 2 (7), 670−678. (33) Li, X.; Su, X.; Liu, P.; Liu, J.; Yao, Z.; Chen, J.; Yao, H.; Yu, X.; Zhan, M. CrystEngComm 2015, 17 (4), 930−936. (34) Wang, Y.; Ohwaki, J. Appl. Phys. Lett. 1993, 63 (24), 3268− 3270. (35) Bocker, C.; Bhattacharyya, S.; Höche, T.; Rüssel, C. Acta Mater. 2009, 57 (20), 5956−5963.

the droplet phase-separated samples. Further works are undergoing to clarify it. Conclusions. For the first time, nanocrystallization of ZnF2, cubic K2SiF6, and KZnF3 was achieved in the glasses of ternary KF−ZnF2−SiO2 system. Through TEM observation, the sizes of the precipitated crystallites of K2SiF6 and KZnF3 were determined in the range from 20 to 30 nm and 30 to 50 nm, respectively. Transparency of the obtained glass-ceramics is achieved because of the nanocrystals of the size below 50 nm, which avoids light scattering. Nucleation and crystal growth of Zn and fluoride containing phases enriched in network modifiers lead to an increase in the viscosity of the residual glass matrix. Hence, Tg determined by dilatometry increases until the crystallization of K2SiF6 or KZnF3 fluoride crystallites is completed. It is deduced that the crystallization in floppy phases proceeds until the viscosity of the residual glassy matrix reaches 1012 Pa·s at which nucleation and crystallization are finally frozen in. This is a self-organized process that controls the precipitation of crystallites with sizes in the nanometer range. Furthermore, by studying the crystallization behavior of different glass compositions in the xKF−xZnF2−(100 − 2x)SiO2 system, we found a strong compositional dependency that a compositional threshold located between x = 20 and 22.5 mol % exists in the glassy system, and the glasses of compositions near this threshold show a totally different crystallization behavior. These features are considered to be originated from the compositional dependence of amorphous phase separation that two different microstructures of droplet and interpenetrating phase separation were observed by the replica TEM technique. These results strongly suggest that the precipitation of fluoride crystallites of nanometric size and different crystal phases (K2SiF6 and KZnF3) is not an “exceptional” case that occurs accidentally in some special glass systems but it is instead “controllable” by the amorphous phase separation of droplet and interpenetrating structure with a sufficiently elaborate glass design.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was partially supported by the Zhejiang Provincial NSF (Grant LY14E020001), and the Alexander von Humboldt Foundation.



REFERENCES

(1) Kiprianov, A. A.; Karpukhina, N. G. Glass Phys. Chem. 2006, 32 (1), 1−27. (2) Yonezawa, S.; Nishibu, S.; Leblanc, M.; Takashima, M. J. Fluorine Chem. 2007, 128 (4), 438−447. (3) Polishchuk, S. A.; Ignat’eva, L. N.; Marchenko, Y. V.; Bouznik, V. M. Glass Phys. Chem. 2011, 37 (1), 1−20. (4) de Pablos-Martín, A.; Durán, A.; Pascual, M. J. Int. Mater. Rev. 2012, 57 (3), 165−186. (5) Dejneka, M. J. MRS Bull. 1998, 23 (11), 57−62. F

DOI: 10.1021/acs.nanolett.5b02605 Nano Lett. XXXX, XXX, XXX−XXX