Study on the strength of titanium doped hollow glass microspheres

Study on the strength of titanium doped hollow glass microspheres

Journal of Non-Crystalline Solids 459 (2017) 18–25 Contents lists available at ScienceDirect Journal of Non-Crystalline Solids journal homepage: www...

3MB Sizes 8 Downloads 96 Views

Journal of Non-Crystalline Solids 459 (2017) 18–25

Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/locate/jnoncrysol

Study on the strength of titanium doped hollow glass microspheres Fang Li, Zhanwen Zhang, Jing Li, Dawei Pan, Jianhong Feng, Ruiting Shi, Bo Li ⁎ Research Center of Laser Fusion, China Academy of Engineering Physics, Mianyang, Sichuan 621900, China

a r t i c l e

i n f o

Article history: Received 5 November 2016 Received in revised form 19 December 2016 Accepted 20 December 2016 Available online xxxx Keywords: Composition Oxygen vacancy Titanium Strength Hollow glass microsphere

a b s t r a c t The composition, structure and strength of titanium doped hollow glass microspheres (HGMs) were studied. Results showed that the strengths of titanium doped glass spheres were determined by the composition, structure and geometric qualities of glass shells. The Young's modulus of HGMs reduced with the increasing titanium concentrations due to the decreased geometric quality of shells and increased with the increase of the fabrication temperature because of the greater loss of alkali oxides at higher temperature. Even though the calculation on the falling process showed that class B HGM had longer average residence time in furnace than class A HGM and consequently greater loss of alkali oxides in glass, the strength of class B HGM was reduced as a result of the generations of oxygen vacancy and phase separation. The strength of class B HGM was improved by compensating oxygen atom by heat treatment in air. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Hollow glass microspheres (HGMs) are common fuel (normally deuterium-tritium, DT) containers in the research of inertial confinement fusion (ICF) [1–4]. Titanium is a particular diagnostic dopant in glass shell. It helps to diagnose the implosion uniformity, temperature, and mixing of the ablation layer and fuel during the implosion process. In addition, the dopant captures suprathermal electrons and shields them from preheating the fuel. Moreover, it may modify the density profile at ablation front to reduce the growth rate of Rayleigh-Taylor (RT) instability [5–7]. HGM is also a promising hydrogen storage container because of its high efficiency, safety, lightness, cheapness, and simplicity [8–10]. The gas filling and outgassing of HGMs are currently performed by heating to between 300 °C and 400 °C. However, the temperature-dependent gas diffusion velocity brings inconvenience for the gas filling and release processes and hinders the development of hydrogen energy economy. It has been found that titanium doped HGMs with the same doping level show different absorptions of visible light in our previous work [11]. The class A HGMs is colorless and transparent, while the class B HGM is colored in blue. The visible light absorption of class B HGM is produced by Ti3 +, which is generated by the volatilization of oxygen atoms in the inertial atmosphere in furnace. The class B HGMs shows significant photo activity, which has never been reported. It releases gas upon exposure to light and stops outgassing rapidly as soon as the light is off. The photo-enhanced gas diffusion in titanium doped HGMs

⁎ Corresponding author. E-mail address: [email protected] (B. Li).

http://dx.doi.org/10.1016/j.jnoncrysol.2016.12.029 0022-3093/© 2016 Elsevier B.V. All rights reserved.

is hopefully to develop a revolutionary new method for storing, transporting, and delivering hydrogen. During the gas filling and release processes, the glass shells are subjected to a pressure gradient across the shell and the gas filling rate is proportional to the pressure gradient. Thus, the strength is undoubtedly important for HGMs to be used as hydrogen or DT container. The present study was undertaken to investigate the effects of glass composition and structure on the strength of titanium doped HGMs. HGMs with different doping levels were fabricated at different temperatures and their strengths were measured. The differences of the composition, structure and strength between the two classes of HGMs were investigated. Class B HGM was found to be faded by heat treatment in air [11], which beyond question is a result of the change of structure or chemical state in glass. Effect of heat treatment condition on the structure and strength of HGMs were investigated. Titanium doped HGMs before and after heat treatment were characterized using X-ray photoelectron spectroscopy (XPS), ultraviolet-visible absorption spectrometry (UV–Vis), X-ray diffraction (XRD), inductively coupled plasma optical emission spectrometry (ICP-OES) and scanning electron microscopy (SEM).

2. Experimental details 2.1. Preparation and heat treatment of titanium doped HGMs HGMs were fabricated by dried gel method in a vertical high-temperature furnace [12–14]. In this work, they were fabricated from gel particles with size ranging from 200 μm to 250 μm, under an inert atmosphere formed by a mixture of 83% helium (He) and 17% argon (Ar) and

F. Li et al. / Journal of Non-Crystalline Solids 459 (2017) 18–25

pressure at 1.0 atm. Letter q was used to denote the titanium doping level via the atom ratio of Ti to Si. HGMs were heat treated in a VMK 1400 (Linn High Therm, Germany) muffle furnace under different conditions. It was raised at 10 °C/min to the target temperature and kept for a certain time. The heat treatment temperature (Th) ranges from 300 °C to 650 °C and heat treatment times (th) are 2 h, 4 h or 6 h. The atmosphere in furnace is air.

19

[15]. It means that the failure pressure of HGMs was determined by the shell aspect ratio (Ra, the shell diameter-to-wall thickness) and the Young's modulus (E) [16,17].  2 8E tw Pb ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi    D 3 1−ν 2

ð1Þ

2.2. Characterization of the titanium doped HGMs Micrographs of HGMs were obtained using the VHX digital microscope (KEYENCE, Japan). The diameter (D) and wall thickness (tw) of spheres were measured by optical measuring microscope (Nikon Co., MM-400) and interferometric microscope (Veeco Co., WYKDNT1100), respectively. For all statistics in this work, about 600 spheres for each batch of HGMs were counted. UV–Vis spectra were recorded on a Lambda 950 spectrophotometer (PerkinElmer, USA) to study the visible light absorption of HGMs. The samples were prepared by adhering spheres to a quartz substrate with ethanol. XPS (Thermo-VG Escalab 250, USA) was used to analyze the chemical state of titanium in glass. Before tests, spheres were etched by argon ions for 30 s. The region of interest was 250 × 250 μm. X-ray diffraction was used to study the effect of heat treatment on the crystallization of HGMs. It was performed on a Regarku Smart Lab X-ray diffractometer with Cu Kα radiation. The 2θ range for the measurement was 10–90° and θ increment was 0.01°. A SEM (MERLIN VP compact, Carl Zeiss, Germany) FEG electron microscope was used to investigate the morphology of the glass shell. The composition tests of gels and HGMs were carried out on an iCAP 6500 ICP-OES (Thermo, USA). 2.3. Strength test of the titanium doped HGMs Failure of thin-walled glass spheres, normally with aspect ratio N 55, under uniform compressive load appears to occur by elastic buckling

where ν is the Poisson's ratio and it was assumed to be 0.21 [18]. Pb is the buckling pressure and it is determined by measuring the pressure required to implode a set of shells. The Young's modu1us of HGMs is ca1culated from the slope of the line produced by Pb versus R−2 a . The pressure test was carried out in an in-house apparatus by filling argon in pressure chamber at 0.1 MPa/min until the tested sphere cracked [11]. About 45 spheres were tested for each batch of HGMs. 3. Results 3.1. Generation and elimination of oxygen vacancy in titanium doped HGMs Fig. 1 shows the micrographs (×30) of HGMs fabricated at 1400 °C and 1500 °C keeping the other preparation conditions the same. For convenience, they are denoted as 1400-HGMs and 1500-HGMs, respectively. Samples used for characterizations and strength tests were 1500HGMs unless otherwise specified. Apparently, the yield of class B HGMs in 1500-HGMs is much higher than that in 1400-HGMs. The percentage of class B HGMs is denoted as B%. B% of 1400-HGMs and 1500-HGMs with q ranging from 5% to 20% were counted (Fig. 2). The random error of the method was given by the standard derivation across five counts, which was represented by error bars in the Fig. B% of 1400-HGMs with 5% q, 8% q, 10% q, 15% q and 20% q were 4.1%, 4.6%, 5.6%, 13.4% and 3.4% and those of 1500-HGMs were 11.8%, 18.8%,

Fig. 1. Micrographs (×30) of titanium doped HGMs.

20

F. Li et al. / Journal of Non-Crystalline Solids 459 (2017) 18–25

Fig. 2. Statistical percentages of class B HGMs in 1400-HGMs and 1500-HGMs.

64.0%, 91.3% and 93.2%, respectively. At the same doping level, B% increases dramatically with the increasing fabrication temperature. At a given fabrication temperature, B% increases with the increasing doping level with q no more than 15%. However, the B% of HGMs with 20% q is close to or even lower than that of HGMs with 15% q. It has been found that the average tw of HGMs with 20% q is much greater than that of other HGMs, as a result of great loss of blowing agent in gel precursors [12,13]. It is difficult for oxygen atoms to escape from the thick wall of HGMs with 20% q at 1400 °C. Moreover, the class B HGMs has thinner wall than class A HGMs [11] and this also facilitates the volatilization of oxygen atoms. To sum up, the wall thickness of HGMs, the pressure in furnace, especially the fabrication temperature are significant factors for the formation of oxygen vacancy. Class A HGMs is colorless in vision, whereas class B HGMs seems blue. The difference of sphere appearance between the two classes of HGMs is a result of different absorption of visible light. Fig. 3 shows the UV–Vis absorption spectra of the two classes of 1500-HGMs. There is no absorption of visible light for all class A HGMs, whereas the class B HGMs shows a broad absorption band from about 400 nm to 1100 nm and the absorption strength increases with increasing titanium concentrations (essentially Ti3+ concentration). However, the absorption of visible light for class B HGMs with 20% q was not as strong as those with 15% q. The thick wall of HGMs with 20% q obviously restrains the volatilization of oxygen atoms. Since the oxygen vacancy is thermodynamically reversible, it can be eliminated by heat treatment in air. Fig. 4 shows the micrographs and UV–Vis spectra of the heat treated 1500-HGMs with 15% q. After being heat treated in air, the class B HGM turns to be two kinds of HGMs again, the colorless one and blue one, and they are denoted as class BA

and class BB, respectively. The compensation of oxygen atom has a temperature threshold. Few class BA spheres generated after being heat treated at 400 °C and the UV–Vis spectrum is quite similar to that of the class B HGMs. Once Th reaches 500 °C, the absorption intensity will decrease with the increases of Th. HGMs being heat treated at 650 °C for 2 h are all turned to be class BA, of which the UV–Vis spectrum is just alike to that of class A HGMs, showing no absorption to visible light. The increase of th also reduces the absorption strength of visible light, which is not as effective as the increasing Th. Heat treatment efficiency is signified by the percentage of class BA HGMs. Fig. 5(a) shows the influences of Th and th on the heat treatment efficiency. The error of this method (based on repetitive counts) was b0.6%. The temperature threshold to compensate oxygen atom is about 400 °C. The percentage of class BA HGM grows exponentially with the increasing Th when th keeps at 4 h. Extension of th at given temperature does not increase the heat treatment efficiency significantly. In addition, the yield of class BA HGMs is also influenced by the wall thickness of spheres. According to the statistical results shown in Fig. 5(b), the class BA HGMs tends to have thinner wall than the corresponding class BB HGMs. Meanwhile, the proportion of thick-walled HGMs generally increases with the increasing Th. It implies that the oxygen atom in thick-walled HGMs tends to be compensated at higher temperature. The results confirm that Th as well as the wall thickness determine the yield of class BA HGMs. 3.2. Strength of titanium doped HGMs Oxygen vacancy is one kind of structure defect and it is of significance with respect to the behavior of materials. HGMs being heat treated at 650 °C for 2 h are very fragile, falling apart readily during the transferring and measuring processes. These spheres are unvalued for hydrogen storage or ICF experiment. The Pb of titanium doped HGMs for were tested. Fig. 6(a) shows the fitted results of Pb versus R−2 a 1500-HGMs with different doping levels. It is noteworthy that the measured Pb displays a statistical distribution due to the diversities of the geometric parameters (like wall uniformity and sphericity) of HGMs. In addition, flaws in different degrees in glass shells also result in uncertainties on the measured Pb. The systematic error of the Pb measurement method was mainly derived from the increasing pressure velocity, about 0.01 MPa. The slopes of the five lines in Fig. 6(a) for 0%, 8%B, 8%A, 15%B and 15%A were 2.850, 1.970, 2.316, 1.676, and 1.967, respectively, and the calculated average Young's modulus from Eq. (1) were 60.3, 41.7, 49.0, 35.5 and 41.1 GPa, respectively. Table 1 shows the calculated Young's modulus of HGMs fabricated at different temperatures and those being heat treated at different conditions. Compared with the corresponding Young's modulus of 0%, 8%B, 8%A, 15%B and 15%A HGMs fabricated at 1400 °C (52.1, 36.7, 41.4, 34.0 and 34.7 GPa, respectively [11]), the strengths of these HGMs increase 15.73%, 13.71%, 18.38%, 4.35% and 18.43%, respectively, with the addition of 100 °C fabrication temperature. 15%A HGMs being heat treated at 400 °C and 500 °C have close Young's modulus to the initial 15%A. The Young's modulus 15%A HGMs being heat treated at 600 °C reduces a little. The same class of HGM, for 15%A and 15%B, after being heat treated has close Young's modulus. The Young's modulus of 15%BA increases about 8% after being heat treated, whereas that of 15%BB reduces slightly. 3.3. Structures of the titanium doped HGMs

Fig. 3. UV–Vis absorption spectra of the two classes of 1500-HGMs with q ranging from 5% to 20%.

The high resolution Ti 2p XPS spectra of 15% q HGMs before and after heat treatment are shown in Fig. 7. The binding energy peaks at 457.4 and 462.6 eV are assigned to the 2p3/2 and 2p1/2 core levels of Ti3 + and those at 458.4 and 463.5 eV are related to 2p3/2 and 2p1/2 core levels of Ti4 +. After being heat treated, 15%BB has the same 2p3/2 and 2p1/2 peak values with 15%B and these peaks correspond to Ti3+. In addition, a weak shoulder at about 458.4 eV appears in their spectra. This implies

F. Li et al. / Journal of Non-Crystalline Solids 459 (2017) 18–25

21

Fig. 4. UV–Vis absorption spectra of 15% q HGMs heat treated under different conditions. The inserted micrographs are the heat treated HGMs at different conditions.

that there is still little proportion of Ti4 + in classes B and BB HGMs, whereas the peaks of 15%BA shift to the higher energy side. Apparently, Ti3+ in class BA HGM is oxidized to Ti4+ by heat treatment in air accompanied with the compensation of oxygen atoms. The peak energy of 15%BA is the same with that of 15%A. As is known, Ti4+ is colorless and Ti3+ is blue. The appearance of HGMs is determined by titanium valance. Thus the class BB HGM seems similar to class B HGM. Just as the micrographs and UV–Vis spectra in Fig. 3 show, the class BA HGM has

Fig. 5. Statistical (a) percentage of class BA HGMs and (b) wall thickness of HGMs after being heat treated under different temperatures and times.

similar appearance and absorption of visible light to that of class A HGMs as well. An oxide glass is in the solid state at a temperature lower than the glass transition temperature, and atoms or molecules are static so that

Fig. 6. Buckling pressure vs aspect ratio of (a) titanium doped HGMs fabricated at 1500 °C with different titanium doping level and (b) 15% q HGMs being heat treated at different conditions. The error bars of the Pb data do not exceed the size of the symbols. Lines in diagram are intended as visual guides. the Pb vs. R−2 a

22

F. Li et al. / Journal of Non-Crystalline Solids 459 (2017) 18–25

Table 1 Young's moduli (GPa) of the initial and heat treated HGMs with different titanium concentrations. Numbers in parentheses give uncertainty of the last digit. Initial HGMs

Heat treated HGMs (fabricated at 1500 °C)

Sample

1400-HGMs

1500-HGMs

Heat treatment condition

15%BA

15%BB

15%A

0% 8%A 8%B 15%A 15%B

52.1 (8) 41.4 (6) 36.7 (6) 34.7 (5) 34.0 (5)

60.3 (9) 49.0 (7) 41.7 (6) 41.1 (6) 35.5 (5)

400 500 600 600 600

/ 38.6 (6) 38.4 (6) 38.1 (6) 38.5 (6)

/ 34.3 (5) 32.8 (5) 33.6 (5) 32.2 (5)

40.5 (6) 40.8 (6) / 39.4 (6) /

°C-4 °C-4 °C-2 °C-4 °C-6

phase separation or nucleation of the glass usually cannot take place. However, it has been found that the chemical, physical and mechanical properties of heat treated TiO2-containing glass are different from those of the quenched glass, when crystallization heat treatment has been carried out [19]. Fig. 8 shows the XRD patterns of 15%A and 15%B and the heat treated HGMs. There is still no visible crystallization peak in their XRD patterns of 15%A and that being heat treated at 500 °C. However, when the Th reaches 600 °C, several peaks attributed to SiO2 (JCPDS card No. 75-0923) appear. This implies that a few SiO2 crystals start to form at 600 °C. There are weak crystallization peaks in class B HGM, which belong to titanium oxide (JCPDS card No. 53-0619). An interesting finding is that the peak intensity of class BB HGMs being heat treated at no more than 600 °C is enhanced markedly, whereas these peaks disappear in class BA HGMs in spite of the same heat treatment conditions. Once the Th gets to 650 °C, many crystallization peaks related to SiO2 and TiO2 appear. It reveals that this batch of HGM has high crystallinity. Fig. 9 shows the SEM graphs of HGMs before and after heat treatment. The surface of 0% HGM seems very uniform and homogeneous and there is no characterized morphology can be observed. The morphologies of 15%A and that being annealed are similar to that of 0%. Even though there seems no distinguishable separation zone in class BA HGMs, 15%BA HGMs (with Th b 650 °C) are not as uniform as the undoped glass. In 15%B and 15%BB HGMs, there are dendritic separation domains and these domains grow with the increase of Th. When the Th is up to 650 °C, many holes with dozens of nanometers appear in glass and the separated domain is in flake with hundreds of nanometers. The holes may be responsible for the especially low strength of 650 °C-2 h HGMs.

h h h h h

precursors and HGMs were determined by ICP-OES (shown in Table 2). The error in the fractions of oxides was b 0.1 mol%. The ultimate SiO2 concentrations in the 1500-HGMs are higher than those in 1400HGMs and class B HGMs has higher SiO2 concentrations than the corresponding class A HGMs. In class A HGMs, the losses of Li2O, Na2O and K2O are 27.4%, 43.2% and 53.1% at 1400 °C, respectively, and they increase to 49.4%, 66.8% and 70.9% at 1500 °C. The loss of the three kinds of alkali oxides in class B HGMs is much higher than that in class A HGMs in spite of the same fabrication conditions. The average equivalent spherical diameters (De) of gel particles for class A HGMs and class B HGMs are about 260 μm and 230 μm. Gel particles fall in a 0.6 m feeding zone and then the furnace body, about 2.5 m in height, under the forces of gravity (Fg), gas buoyancy (Fb) and flow resistance (Ff) in furnace. F g ¼ mg m g ρp ρg u2 F f ¼ C D AD 2 F b ¼ ρg

ð2Þ ð3Þ ð4Þ

3.4. Compositions of the titanium doped HGMs As is known, glass composition changes in a large degree due to the volatilization of alkali and boron when the refining temperature is higher than 1200 °C [17,20]. The chemical compositions of gel

Fig. 7. High resolution Ti 2p XPS spectra of 15% q HGMs.

Fig. 8. XRD patterns of (a) 15%B and that heat treated under different conditions (b) 15%A and that heat treated under different conditions.

F. Li et al. / Journal of Non-Crystalline Solids 459 (2017) 18–25

23

Fig. 9. SEM graphs (×30 K) of the initial HGMs and HGMs being heat treated.

where u is the velocity of gel particle relative to the gas, ρg is the gas density, CD is the resistance coefficient and AD is the section area of gel particle. According to the Neuton's second law, the differential equation of motion is given as following F g − F b −F f ¼ m

du dt

ð5Þ

Substituting Eqs. (2)–(4) into Eq. (5) yields

du ¼ dt

! ρp −ρg C D AD g− ρ u2 ρp 2m g

ð6Þ

For the equivalent spherical gel particle, it can be denoted as du ¼ dt

! ρp −ρg 3C D ρg 2 g− u ρp 4De ρp

ð7Þ

Since Ff changes with the change of velocity, the particles may go through an inconstant acceleration motion and then a uniform motion. The relative velocity when the forces reach equilibrium is defined as the free falling velocity (ut). CD is a function of Reynolds number (Re), so CD and ut vary in different Re range. CD ¼

24 ðRe b2Þ Re

ð8Þ

24

F. Li et al. / Journal of Non-Crystalline Solids 459 (2017) 18–25

Table 2 Compositions (in mol%) of gel precursors and HGMs with different titanium oxide concentrations. The uncertainty in the fractions of oxides was b0.1 mol%. Sample 0%

5% q 8% q

10% q 15% q

20% q Loss

CD ¼

ut ¼

18:5 Re 0:6

Gel 1400-HGMs 1500-HGMs Gel 1500-HGMs-A Gel 1400-HGMs-A 1500-HGMs-A 1500-HGMs-B Gel 1500-HGMs-A Gel 1400-HGMs-A 1500-HGMs-A 1500-HGMs-B Gel 1500-HGMs-A 1400-HGMs-A 1500-HGMs-A 1500-HGMs-B

SiO2

B2O3

Li2O

Na2O

K2O

TiO2

69.7 79.9 86.8 67.3 83.2 66.1 75.1 81.0 83.9 65.2 79.4 63.1 71.2 76.2 78.4 61.2 73.1 0 0 0

1.6 0.2 0.2 1.6 0.2 1.5 0.2 0.2 0.2 1.5 0.2 1.5 0.2 0.2 0.2 1.4 0.2 87.7 89.0 88.2

9.5 7.8 5.7 9.1 5.6 8.9 7.4 5.5 4.0 8.8 5.5 8.5 7.1 5.4 4.1 8.3 5.2 27.4 49.4 63.0

16.4 10.6 6.4 15.9 6.2 15.6 10.0 6.2 5.0 15.4 6.3 14.9 9.6 6.2 5.2 14.4 6.1 43.2 66.8 73.4

2.8 1.5 0.9 2.7 0.9 2.6 1.4 0.9 0.7 2.6 1.0 2.5 1.4 0.9 0.8 2.4 0.9 53.1 70.9 76.7

/ / / 3.4 3.9 5.3 5.9 6.2 6.2 6.5 7.6 9.5 10.5 11.1 11.3 12.3 14.5 1.2 4.1 6.1

ð2bRe b500Þ

  De 2 ρp −ρg g 18μ

ðRe b2Þ

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  u  ud ρp −ρg gRe 0:6 t ð2bRe b500Þ ut ¼ 0:27 ρg

ð9Þ

ð10Þ

ð11Þ

where μ is the gas viscosity, ρp is the gel density and it is 1120 kg/m3 [14]. The gases in the two phases are air at about 40 °C and He\\Ar mixture at 1400 °C, respectively, of which the densities are 1.128 kg/m3 and 0.0727 kg/m3 and viscosities are 1.91 ∗ 10−5 Pa·s and 10 ∗ 10−5 Pa·s, respectively. Re is represented as

Re ¼

ρg uDe μ

ð12Þ

The Re of the two phases are in the ranges of Re b 2 and 2 b Re b 500, respectively, which are confirmed by trial and error. The ut1 of 260 μm and 230 μm gel particles are 0.963 m/s and 0.837 m/s and ut2 are 0.412 m/s and 0.323 m/s, respectively, which are calculated by Eqs. (10)–(12). Integrating Eq. (7) from 0 to ut1 based on Eqs. (8)–(12), the results show gel particles reach uniform motion within 0.6 m. Once they fall in the furnace body, the Ff increases greatly. The gel particles with 260 μm and 230 μm in sizes decelerate in 0.31 s and 0.24 s to ut2 and then go through a 5.92 s and 7.63 s uniform falling, respectively, in the rest of the furnace body. So, the average residence times in furnace body of the gel particles for class A HGMs and class B HGMs are 6.23 s and 7.88 s, respectively. The longer residence time during the refining process takes the main responsibility for the more serious loss of oxides in class B HGMs and the extended time during the cooling process may lead to the crystallization of glass. This is the main reason for the differences of the composition and structure between the two classes of HGMs. The loss of boron oxide in all HGMs reaches up to 90%. Since the volatilization of boron oxide occurs under 1000 °C [21], increasing the refining temperature or residence time does not cause the additional loss of boron oxide.

4. Discussion 4.1. Oxygen vacancy and phase separation in glass The class B HGMs contains oxygen deficiency related TiO2 (i.e., TiO2 – x), which is formed by the volatilization of oxygen atoms and resulting Ti3+. 0 1 2TiO2 ⇔V ••o þ 2TiTi þ 3OO þ O2 ðg Þ 2

ð13Þ

The oxygen vacancy is one kind of thermodynamic defect, thus its density is the function of thermodynamic parameters (temperature, pressure, etc.). The fraction of class B spheres has been found to increase with decrease in inert gas pressure [11]. It implies that the low gas pressure, namely low oxygen pressure, avails the evaporation of oxygen atoms. At thermal equilibrium, the oxygen vacancy density is only determined by temperature and it can be computed by the principle of minimum free energy.

 

ΔG f V ••o ¼ exp − kT

ð14Þ

where [V•• o ] is the density of oxygen vacancy and ΔGf is the activation energy of formation of oxygen vacancy, which is approximately considered not to change with temperature. Eq. (14) reveals that the concentration of oxygen vacancy increases exponentially with the increase of temperature. Thus, the concentration of oxygen vacancy in 1500-HGMs is denser than in 1400-HGMs and B% at 1500 °C is higher as well. Oxygen vacancy is one kind of structure defects and it is of significance with respect to the behavior of materials. Fortunately, the lost oxygen atoms can be compensated by heat treatment in air. Beyond all question, the influential factors of the generation of oxygen vacancy are also responsible for its elimination. The results (see Fig. 5) confirm that Th as well as the wall thickness determined the yield of class BA HGMs. Titanium oxide is known as nucleation in glass and it is prone to separate from the silicate network during the heat treatment process due to the high field of the cation. Vekey et al. [22] suggests that TiO2 produces glass-in-glass immiscibility, and that one of the phases then crystallizes during the heat treatment. Thus phase separation in class BB HGMs is in anticipation. With regard to the class BA HGMs, it seems that the compensation of oxygen atoms hinders the crystallization and phase separation in glass. But if the Th is high enough, crystallization and phase separation still occur in class BA HGMs. The SEM graphs and XRD patterns manifest that phase separation is always accompanied with crystallization. 4.2. Effect of titanium doping on the strength of HGMs The connection between the physico-mechanical character of the HGMs and the chemical composition of glass is quite complex [23]. Recent research has found that TiO2 has a tightening effect on the glass network and the Young's modulus increases with the TiO2 concentration [24]. Thus, titanium doping should have improved the Young's modulus of HGMs. However, the fact is not as expected. Young's modulus decreases with the increase of titanium concentration, which holds for both the two kinds of HGMs. The strength of HGMs in relation to volume compression is largely determined by the geometric parameters of the shells and it also depends on the defects in the shape and surface [23]. The reduced strengths of titanium doped HGMs are mainly caused by the poor geometric quality. Because titanium oxide reduces the surface tension and increases the viscosity of glass melt in furnace, the geometric quality, like sphericity, surface finish and wall thickness uniformity of HGMs decrease with the increase of titanium concentrations [11,14]. Moreover, the increased viscosities of high doping glasses also lead to fewer losses of alkali oxides and consequently lower SiO2

F. Li et al. / Journal of Non-Crystalline Solids 459 (2017) 18–25

concentration. As a result, the Young's modulus decreases with the increasing titanium concentration. The strength difference between the two classes of HGMs with the same doping level is determined by both the glass composition and defects in glass. As is known, alkali borosilicate glass has lower strength than high silica glass, due to the reduced integrity of glass network. The silica concentration in class B HGMs is higher than that in the corresponding class A HGMs. From this point of view, the strength of class B HGMs should have been higher than that of class A HGM. However, the defects in class B HGMs, such as oxygen vacancy, phase boundary and hole, reduce the buckling pressure of spheres. Thus, the strength of class B HGMs is a result of the interaction of the two factors. HGMs prepared at higher temperature contained denser oxygen vacancies and suffered greater loss of alkali metals. As a comprehensive result, the strength increases with the increase of fabrication temperature. Heat treatment in air compensates oxygen atoms in class BA HGMs, thus whose strength is improved. In class BB HGMs, the oxygen vacancy was not eliminated and meanwhile the increased phase boundaries supply more stress concentration centers. As a result, the strength of class BB HGMs decreases. 5. Conclusions Titanium doped HGMs with different doping levels and those being heat treated at different conditions were applied for buckling pressure tests in this work. The effects of composition and structure in the strength of titanium doped HGMs were studied. Results showed that class B HGMs was generated by the volatilization of oxygen atom and the resulting Ti3 + and oxygen atoms could be compensated by heat treatment in air. The efficiencies of formation and elimination of oxygen vacancy increased with the increasing temperature. The strength of HGMs was determined by glass composition, defects, like oxygen vacancy and phase boundary, and geometric property. Titanium doping reduced the Young's modulus of HGMs due to the decreased geometric quality of shells. The strength of 1500-HGMs was higher than that of 1400-HGMs because of greater loss of alkali oxides at higher fabrication temperature. The calculation on the falling process of gel particles showed class B HGMs had longer average residence time in furnace and consequently greater loss of alkali oxides than class A HGMs. In spite of this, the strength of class B HGMs was lower than that of class A HGMs. It was caused by the oxygen vacancy and phase separation in glass. The strength of class BA HGMs in which the oxygen vacancy was eliminated during heat treatment process could be greatly improved, whereas that of class BB HGMs reduced slightly due to the aggravating phase separation in glass. Acknowledgments The authors are grateful to the China Academy of Engineering Physics for financial support (2014B0302051). References [1] P.C. Souers, R.T. Tsugawa, R.R. Stone, Manufacture of D–T filled hollow glass microsphere, laser targets, Rev. Sci. Instrum. 46 (1975) 682–685.

25

[2] J.D. Lindl, P. Amendt, R.L. Berger, S.G. Glendinning, S.H. Glenzer, S.H. Haan, S.W. Kauffman, R.L. Kauffmann, O.L. Landen, L.J. Suter, The physics basis for ignition using indirect-drive targets on the national ignition facility, Phys. Plasmas 11 (2004) 339–491. [3] V.A. Smalyuk, R. Betti, J.A. Delettrez, V.Y. Glebov, D.D. Meyerhofer, P.B. Radha, S.P. Regan, T.C. Sangster, J. Sanz, W. Seka, C. Stoeckl, B. Yaakobi, J.A. Frenje, C.K. Li, R.D. Seguin, F.H. Seguin, Implosion experiments using glass ablators for direct-drive inertial confinement fusion, Phys. Rev. Lett. 104 (2010) 165002. [4] R.L. Downs, M.A. Ebner, W.J. Miller, Hollow glass microsphere by sol-gel technology, in: L.C. Klein (Ed.), Sol-Gel Technology for Thin Films, Fibers, Preforms, Electronic, and Specialty Shapes, Noyes Publications, New Jersey 1988, pp. 330–335. [5] M.L. Hoppe, B.A. Vermillion, High-Z doping of glass shells, Fusion Sci. Technol. 55 (2009) 461–464. [6] X.M. Sang, X.J. Yang, Z.D. Cui, S.L. Zhu, J. Sheng, Preparation and characterization of nano-TiO2 doped polystyrene materials by melt blending for inertial confinement fusion, J. Macromol. Sci., Part B 43 (2011) 871–882. [7] K.E. Branham, H. Byrd, R. Cook, J.W. Mays, G.M. Gray, Preparation of soluble, linear titanium-containing copolymers by the free-radical copolymerization of vinyl titanate monomers with styrene, J. Appl. Polym. Sci. 78 (2000) 190–199. [8] S. Dalai, S. Vijayakshmi, P. Sharma, K.Y. Choo, Magnesium and iron loaded hollow glass microspheres (HGMs) for hydrogen storage, Int. J. Hydrog. Energy 39 (2014) 16451–16458. [9] S. Dalai, S. Vijayakshmi, P. Shrivastava, S.P. Sivam, P. Sharma, Effect of Co loading on the hydrogen storage characteristics of hollow glass microspheres (HGMs), Int. J. Hydrog. Energy 39 (2014) 3304–3312. [10] S. Shetty, M. Hall, Facile production of optically active hollow glass microspheres for photo-induced outgassing of stored hydrogen, Int. J. Hydrog. Energy 36 (2011) 9694–9701. [11] F. Li, J. Li, J.H. Feng, Z.W. Zhang, B. Li, Influence of titanium doping on the structure and properties of hollow glass microspheres for inertial confinement fusion, J. Non-Cryst. Solids 436 (2016) 22–28. [12] F. Li, J.H. Feng, J. Li, B. Li, Z.W. Zhang, Fabrication of titanium doped hollow glass microsphere for inertial confinement fusion targets by dried gel method, Appl. Mech. Mater. 672–674 (2014) 396–401. [13] F. Li, J.H. Feng, J. Li, B. Li, Z.W. Zhang, Preparation of titanium uniformly doped dried gel by sol–gel method, High Power Laser Part. Beam 27 (2014) 062002-1–7. [14] F. Li, L. Su, J. Li, J.H. Feng, B. Li, Z.W. Zhang, M.F. Liu, Influence of titanium doping on the shell-forming process and geometric properties of hollow glass microspheres, Int. J. Hydrog. Energy 41 (2016) 12605–12713. [15] Z.W. Zhang, B. Li, Y.J. Tang, H.L. Lei, Measurement of mechanism properties of hollow glass microspheres used for ICF experiments, High Power Laser Part. Beam 19 (2007) 1663–1666. [16] W.J. Miller, R.L. Downs, L.A. Scott, R.G. Schneggenburger, Strength of aluminum-containing glass shells, J. Vac. Sci. Technol. 18 (1981) 1290–1291. [17] J.H. Campbell, J.Z. Grens, J.F. Poco, Preparation and properties of hollow glass microspheres for use in laser fusion experiments, Lawrence Livermore National Laboratory, Livermore, Calif., Nov. 1983, Report UCRL-53516. [18] X.B. Qi, C. Gao, Z.W. Zhang, S.F. Chen, B. Li, S. Wei, Fabrication and characterization of millimeter-sized glass shells for inertial confinement fusion targets, Chem. Eng. Res. Des. 91 (2013) 2497–2508. [19] X. Zou, M. Yamane, J. Li, C. Wang, Effect of phase-separation on nucleation and growth of magnesium-aluminium-titanate crystals in MgO-Al2O3-SiO2-TiO2 glass, J. Non-Cryst. Solids 112 (1989) 268–271. [20] X.B. Qi, C. Gao, Z.W. Zhang, S.F. Chen, B. Li, S. Wei, Production and characterization of hollow glass microspheres with high diffusivity for hydrogen storage, Int. J. Hydrog. Energy 37 (2012) 1518–1530. [21] J.T. Wenzel, D.M. Sanders, Sodium and boron vaporization from a boric oxide and a borosilicate glass melt, Phys. Chem. Glasses 23 (1982) 47. [22] W. Vogel, Structure and Crystallization of Glasses, Vieweg & Sohn GmbH, Burgplatz 1, Braunschweig, 1971 73–100. [23] V.V. Budov, V.Y. Stetsenko, Choice of glass composition for producing hollow microspheres, Glas. Ceram. 45 (1988) 289–291. [24] G. Scannell, L. Huang, T. Rouxel, Elastic properties and indentation cracking behavior of Na2O-TiO2-SiO2 glasses, J. Non-Cryst. Solids 429 (2015) 129–142.