Relationship between Thermal Conductivity and Structure of the CaO–BO_1.5–AlO_1.5 System

Abstract

During the continuous casting process in steel making, the mold flux plays an important role in establishing adequate heat flow. Therefore, it is important to optimize the thermal conductivity of the flux system to control this process. Boron oxide (B₂O₃) is one of the components of the mold flux system and its structural complexity is well known. With the aim of revealing the relationship between the thermal conductivity and flux structure, the authors previously studied BO_1.5-containing mold flux systems. In this study, the thermal conductivity of the CaO–BO_1.5–AlO_1.5 flux system was measured above 1500 K for various compositions using the transient hot-wire method. The composition dependence of the flux thermal conductivity was investigated in terms of its local structure, as analyzed using Raman spectrometry and MAS-NMR. The non-additive change in the thermal conductivity of the CaO–BO_1.5–AlO_1.5 system, which is known as the borate anomaly, is attributed to the relative fraction of the BO_1.5 structural unit or the three-dimensional (3D) structural network involving the ^[4]A–O–^[3]B bond. The results obtained by Raman spectrometry revealed that the complexation of the flux structure by the 3D AlB₃O₇ structure can increase the thermal conductivity at a high BO_1.5 content. The formation probability for this structure was calculated based on the MAS-NMR results. Thus, the increase in thermal conductivity can be adequately explained by the formation of the AlB₃O₇ structure. For practical purposes, the effect of substituting SiO₂ for AlO_1.5 on thermal conductivity was also investigated with fixed BO_1.5 and CaO concentrations.

1. Introduction

The mold flux used in the continuous casting process employed in steel production plays an important role as a heat insulator and lubricant between the molten steel and casting mold. Therefore, the thermal conductivity of the mold flux is an important property for sustaining an adequate heat flow. In addition to collecting reliable data on thermal conductivity, the ability to predict the thermal conductivity from the flux composition is important during the process designing. The authors have invested the relationship between the thermal conductivity of molten oxides and its local structure, which is determined by its composition, for various systems.^{1,2,3,4,5,6,7)} The conventional mold flux contains CaO and SiO₂ as the main components. In addition, fluoride is vital component to lower the viscosity of the flux. Addition of fluoride is also important to control the heat flow during casting by the precipitation of the cuspidine phase. However, environmental pollutions due to the evaporation of fluoride is inevitable during the process. In order to develop an environmentally sound mold flux, BO_1.5 is one of the promising substitutes to sustain the low viscosity of fluoride-free mold flux system.

With this background, the authors have previously investigated borate-containing systems, including the CaO–BO_1.5–SiO₂ system.^3,4,5,6,7)

Heat is considered to be transferred by phonons. This behavior is expressed by the following equation, which is analogous to the kinetic theory of gas:   

$$\lambda =1/3\mathrm{\hspace{0.17em} } Cvl$$(1)

where C is the specific heat, v is the mean particle velocity, and l is the mean free path of collision.⁸⁾ Because C and v are considered to be approximately constant above the Debye temperature, the thermal conductivity is mainly determined by the change in l.⁹⁾ In a thermal conductivity measurement using the hot-wire method, the negative effect of the increasing temperature on the thermal conductivity of the melts is observed. This effect is mainly attributed to the decrease in the phonon mean free path with the increase of temperature, in response to the accompanying structural changes, and the increased phonon-phonon interaction. Regarding the relationship between the thermal conductivity and the melt components, the polymerized network structures of acidic oxides, such as SiO₂ and BO_1.5, are expected to have a positive effect on the thermal conductivity, and are regarded as network formers. Moreover, CaO is considered to be a source of non-bridging oxygen (NBO); thus, these structures decrease the thermal conductivity and act as network modifiers. In the CaO–BO_1.5–SiO₂ system,^7,10) the change of thermal conductivity has been successfully explained from the viewpoint of local structures. The MAS-NMR of ²⁹Si and ¹¹B have revealed a higher polymerization degree leading to a higher thermal conductivity. Additionally, the structural groups detected using Raman spectrometry, such as the chain/ring-type borate structures, also affect the thermal conductivity.

With this background, the objective of this study is to clarify the relationship between the thermal conductivity and the local structures of the CaO–BO_1.5–AlO_1.5 flux systems. In the casting process using the CaO–BO_1.5–SiO₂ mold flux, Al in molten steel would reduce SiO₂ and BO_1.5 to introduce Si and B into steel. The effect of this reaction is severe for steel with high Al content, such as transformation induced plasticity (TRIP) steel. According to a review by Wang,¹¹⁾ mold fluxes at high AlO_1.5 concentrations have been developed and their physical/chemical properties have been investigated.^12,13,14) However, the effect of the flux composition on thermal conductivity and the structural change in the AlO_1.5-containing borate system still remains unknown. Therefore, the thermal conductivity of the system was measured above 1500 K using the transient hot-wire method for various compositions. Then, the composition dependency of the flux thermal conductivity was investigated in terms of its local structure, which was analyzed using Raman spectroscopy and MAS-NMR. The effect of substituting SiO₂ for AlO_1.5 on the thermal conductivity was also investigated.

2. Experimental Procedures

2.1. Thermal Conductivity Measurement

The thermal conductivity of the flux was measured above 1500 K using the transient hot-wire method. The reagent grade powders were mixed and pre-melted at 1773 K. During the measurement, a Pt-10%Rh crucible (internal diameter: 32 mm, outer diameter: 38 mm, height: 70 mm) was used to hold the sample. The sample compositions are summarized in Table 1. For the CaO–BO_1.5–AlO_1.5 system (Table 1(a)), the sample compositions were selected to cover the inside of the liquidus at 1673 K, as calculated by FactSage 7.0. Although BO_1.5 content more than 25 mol% is too high as a mold flux in practical use, sample compositions were selected to collect the data relevant to the structure and thermal conductivity in wide-ranged compositions. Because the authors have investigated the thermal conductivity of the CaO–BO_1.5–SiO₂ system at a BO_1.5 concentration lower than 25 mol%,⁷⁾ the samples were prepared to investigate the effect of substituting SiO₂ for AlO_1.5 in this composition region (Table 1(b)). By fixing the BO_1.5 and CaO concentration at 25 mol% and 36–44 mol% respectively, 50% and 100% of AlO_1.5 was replaced with SiO₂.

Table 1. Nominal sample compositions.

(a) The CaO–BO1.5–AlO1.5 system
Sample No.	Nominal Composition (mol%)			CaO/AlO1.5	CaO/BO1.5
	CaO	BO1.5	AlO1.5
1	12.8	78.0	9.2	1.39	0.16
2	15.7	78.5	5.8	2.71	0.20
3	14.8	73.7	11.5	1.29	0.20
4	13.8	69.1	17.1	0.81	0.20
5	31.1	62.3	6.6	4.71	0.50
6	29.0	58.1	12.9	2.25	0.50
7	27.0	54.0	19.0	1.42	0.50
8	49.5	43.0	7.5	6.60	1.15
9	45.7	39.7	14.6	3.13	1.15
10	42.1	36.6	21.3	1.98	1.15
11	38.7	33.3	28.0	1.38	1.16
12	33.3	33.3	33.3	1.00	1.00
13	43.5	25.0	31.5	1.38	1.74
14	40.1	25.0	34.9	1.15	1.60
15	35.9	25.0	39.1	0.92	1.44
16	25.0	25.0	50.0	0.50	1.00
17	49.3	15.0	35.7	1.38	3.29
18	45.5	15.0	39.5	1.15	3.03
19	40.7	15.0	44.3	0.92	2.71
20	20.0	60.0	20.0	1.00	0.33

(b) The CaO–BO1.5–AlO1.5–SiO2 system
Sample No.	Nominal Composition (mol%)				CaO/(AlO1.5 + SiO2)
	CaO	BO1.5	AlO1.5	SiO2
13’	43.5	25.0	15.8	15.8	1.38
14’	40.1	25.0	17.5	17.5	1.15
15’	35.9	25.0	19.6	19.6	0.92

Figure 1 shows the schematic of the experimental setup. The sample was set in an electric resistance furnace (heating element: SiC). Then, the furnace was heated to 1773 K. After the sample had completely melted, the Pt-13% Rh thin wire was immersed in the sample. During the measurement, a constant current of 1.0 A was supplied to the wire by the galvanostat, and the voltage change was monitored using the four-terminal method. The thermal conductivity was calculated from the linear relationship between ΔV and lnt, as follows:   

$$\lambda =\frac{Q}{4\pi }/\frac{d\mathrm{\Delta }V}{dlnt}$$(2)

where Q (W/m), ΔV (V), and t(s) represent the heat generation rate of the hot-wire per unit length, the voltage change, and time, respectively. The theoretical details have been presented in previous papers.^2,15)

Fig. 1.

Schematic illustration of experimental setup. (Online version in color.)

The measurement temperature changed from 1773 K to 1473 K in 50 K intervals. The measurement was repeated three times at each temperature with a sufficient time interval (approximately 5 min) to eliminate the effect of convection caused by the temperature increase during the measurement. Before each measurement, the resistance of the hot wire at each temperature was determined by applying a constant current of 50 mA, for which the temperature increase was negligible.

2.2. Structural Analysis

Structural analysis was conducted for the quenched glass sample under the assumption that the melt structure was kept in the quenched glass. Reagent grade powders mixed for each composition was melted at 1773 K in a Pt crucible (diameter: 20–34 mm, height: 38 mm). Then, the melt was quenched on the water-cooled copper plate. Part of the quenched sample was tested by X-ray diffraction (XRD) to confirm its amorphous structure. Subsequent structural analysis was conducted on samples that were proved to be amorphous glass.

Raman spectrometry (NRS-5100, JASCO) was used with an excitation provided by a laser with a wavelength of 532 nm operated at 5 mW. The resulting Raman signal was collected in the wavenumber range of 400 to 1600 cm⁻¹. Although Raman spectroscopy provides information on the molecular vibrations, the structural information cannot be obtained directly from the Raman spectra. Thus, to obtain the required structural information, the spectral separation procedure was preceded by Gaussian deconvolution. Peaks were assigned to certain structural units according to previous reports.^{14,16,17,18,19,20)} The area fractions of each peak to the total area of each spectrum were calculated and considered as the relative fractions of each structure.

The ¹¹B and ²⁷Al magic angle spinning (MAS) NMR spectra of the samples were collected by the FT-NMR spectrometer (ECA-500; JEOL, Tokyo, Japan). The Larmor frequency was 160.4 MHz for ¹¹B and 130.3 MHz for ²⁷Al. The applied magnetic field was 11.74 T. The crushed glass samples were spun in ZrO₂ rotors at 15 kHz. All spectra were collected using RF pulses with a 30° flip angle. In the measurement of the ¹¹B spectrum, the pulse intensity, recycle delay, and repetition times were 21.3 kHz, 1.25 s, and 256, respectively. In the case of ²⁷Al, they were 17.3 kHz, 0.17 s, and 4096, respectively. The ¹¹B and ²⁷Al chemical shifts were referenced to the saturated H₃BO₃ solution (+19.49 ppm) and to 1 mol L⁻¹ of Al(NO₃)₃ solution (0 ppm), respectively.

3. Results and Discussion

3.1. Thermal Conductivity of the CaO–BO_1.5–AlO_1.5 System

Measured thermal conductivity showed negative effect of temperature as always observed in the measurement by hot-wire method.^{1,2,3,4,5,6,7)} Figure 2 shows the results of the thermal conductivity measurement at 1673 K for the CaO–BO_1.5–AlO_1.5 system, on the iso-thermal section of the ternary phase diagram. The compositions used in the measurement are plotted in the diagram. Several compositions, which are indicated by the black triangles, exhibited crystallization at the free surface of the samples when they were quenched. Therefore, Raman spectrometry and MAS-NMR measurement were not conducted for these samples. Although the thermal conductivities of all samples were expected to be measured above their liquidus, the results for these compositions may have been influenced by their relatively high crystallization temperature. The results obtained for these compositions are not considered in the following discussion. Based on the obtained values, the dotted lines indicate the approximate composition regions for which the flux expressed a similar thermal conductivity. Among the measured thermal conductivity, samples no.1 (λ = 0.48±0.1) and no.16 (λ = 0.45±0.1) showed the largest standard error. Samples no.7 (λ = 0.59±0.06) and no.11 (λ = 0.38±0.07) also showed relatively large error. For other samples, standard error was less than ±0.05 W m⁻¹ K⁻¹. Though the thermal conductivity became lower with the rise in temperature, the composition dependency of the thermal conductivity sustained similar trend as shown in Fig. 2 at above 1623 K. As shown in Fig. 2, the thermal conductivity of the CaO–BO_1.5–AlO_1.5 system became relatively large when the AlO_1.5 concentration increased. Additionally, the effect of the AlO_1.5 addition was more drastic for a lower CaO/BO_1.5 ratio. When the CaO concentration was higher than 40 mol%, the thermal conductivity decreased to approximately 0.3 W m⁻¹ K⁻¹, which is a relatively low value amongst the measured compositions. Moreover, the thermal conductivity was less dependent on the composition at this region. These results revealed the poor polymerization of the melt structure with a high CaO content or high basicity.

Fig. 2.

Measured thermal conductivity, λ (W m⁻¹ K⁻¹) of the CaO–BO_1.5–AlO_1.5 system at 1673 K.

3.2. Relationship between Thermal Conductivity and Structural Units Analyzed Using Raman Spectrometry

The abovementioned changes in the thermal conductivity with different composition are discussed based on the Raman spectrometry results. Figure 3 shows examples of the Gaussian deconvolution of the Raman spectra obtained in the CaO–BO_1.5–AlO_1.5 system. The representative examples provided here include the comparison of the spectrum during the addition of AlO_1.5 at CaO/BO_1.5=0.20 (Fig. 3(a)), the increase in the CaO/BO_1.5 molar ratio for AlO_1.5 fixed at approximately 20 mol% (Fig. 3(b)), and the further addition of AlO_1.5 at a CaO/BO_1.5 ratio of approximately 1.0 (Fig. 3(c)). The peaks were assigned to the structure as shown in Table 2.^14,16)

Fig. 3.

Raman spectrum of representative compositions. Bolded solid line, blokenline and fine solid line represent the original spectrum, the fitted band and the deconbolution band, respectively.

Table 2. Reference peak positions of Raman spectra and assignments for the CaO–BO_1.5–AlO_1.5 system. O° represents the bridging oxygen atoms (BO) and O^– represents the non-bridging oxygen atoms (NBO).

Reference position (cm−1)	Assignments
480–550	Al–O0 stretching vibration in [AlO4]-tetrahedral structure,14) Al–O–B stretching vibration in aluminate network16)
600–650	B–O–B stretching vibration in ring-metaborates16)
700–720	Al–O– or Al–O–B stretching vibration in aluminate network,16) [4]Al–O–[3]B bending vibration14)
770	B–O0 stretching vibration in six-membered borate rings with 1 or 2 [BO4]-tetrahedral structure (di-triborates, triborates, tetraborates, pentaborates)14,16)
860	Symmetric B–O0 stretching vibration in [BO3] pyroborate14)
890–940	B–O– stretching vibration in orthoborates16)
960–980	[4]Al–O–[3]B stretching vibration in aluminoborate network16)
1200–1300	B–O– stretching vibration in pyroborates,16) B–O– stretching vibration in [BO3] (various borate group)14)

The structural changes throughout the composition range considered in this study were mainly characterized by the structural groups shown in Fig. 4.^{14,16,17,18,19,20)} Orthoborate (Fig. 4(a)), which is the 3-coordinated B (^[3]B) with 3 NBO, works as a network modifier and decreases the thermal conductivity at a high CaO content (high CaO/BO_1.5 molar ratio). The Borate groups containing a tetrahedral ^[4]B (Figs. 4(b)–4(d)) were expected to increase the complexation of the melt by their three-dimensional (3D) structures, which would also increase the thermal conductivity. When AlO_1.5 is added to a CaO–BO_1.5 system, ^[4]Al replaces ^[4]B and forms an alumino-borate network. The resulting 3D network structure, such as AlB₃O₇¹⁷⁾ (Fig. 4(e)), or a chain structure, such as AlBO₄¹⁸⁾ (Fig. 4(f)), is expected to increase the thermal conductivity.

Fig. 4.

Structural units considered in the CaO–BO_1.5–AlO_1.5 system.^{14,16,17,18,19,20)}

In Fig. 3, the peaks assigned to those structural units are indicated by the shaded area. When the CaO/BO_1.5 molar ratio was low, the structural change with the addition of AlO_1.5 (Fig. 3(a)) was characterized by the change in the relative percentage of the borate groups with the ^[4]B and AlB₃O₇ structures appearing at 770 cm⁻¹ and 710 cm⁻¹, respectively. When the CaO/BO_1.5 molar ratio increased with a similar AlO_1.5 content (Fig. 3(b)), the network forming structures relatively decreased and the orthoborate (appearing at 930 cm⁻¹) dramatically increased. The further addition of AlO_1.5 at a high CaO/BO_1.5 molar ratio (Fig. 3(c)) led to the decrease of the orthoborate and increase of the AlBO₄ chain structure appearing at 980 cm⁻¹. Moreover, with a high AlO_1.5 content, the relative area fraction of the bands around 450 cm⁻¹ became dominant, while that of the bands around 1300 cm⁻¹ decreased. This indicates that the matrix changed from borate-dominant at the BO_1.5-rich compositions to aluminate-dominant at the AlO_1.5-rich compositions.

Figure 5 shows the comparison between the abovementioned structural changes and the thermal conductivity. The compositions shown in Figs. 5(a)–5(c) correspond to those shown in Figs. 3(a)–3(c), respectively. The compositions closer to AlO_1.5:BO_1.5:CaO=1:3:1 (AlB₃O₇) and 1:1:1 (AlBO₄) yield larger peak intensities relating to these structural units (710–730 cm⁻¹ for AlB₃O₇ and 960–980 cm⁻¹ for AlBO₄, respectively). Because these units have relatively large 3D structures, they seem to elongate the phonon mean free path and increase the thermal conductivity. Moreover, the orthoborate, which has a Raman shift at 930 cm⁻¹, was one of the main structural units at a higher CaO/BO_1.5 molar ratio. The relatively low thermal conductivity obtained in this composition region was caused by the depolymerization of the aluminate or aluminoborate network, which was in turn caused by the orthoborate unit.

Fig. 5.

Changes in thermal conductivity and relative percentage of each structure.

From the analysis regarding the wide-ranging compositions of the CaO–BO_1.5–AlO_1.5 system, the AlB₃O₇ or AlBO₄ structures were expected to be effective with regard to increasing the thermal conductivity. As shown in Fig. 2, the thermal conductivity of CaO–BO_1.5–AlO_1.5 system apparently exhibited a non-additive relationship with BO_1.5. As an example, the change in the thermal conductivity with the BO_1.5 addition at CaO/AlO_1.5=1.38 is shown in Fig. 6(a). Although the basicity of the flux continuously decreased with the addition of BO_1.5, the thermal conductivity had a local maximum at a BO_1.5 concentration of approximately 50 mol%. This result can also be explained by considering the AlB₃O₇ and AlBO₄ structures. In the same manner as shown in Fig. 5, the change in the relative percentages of AlO_1.5-related structural units (AlB₃O₇ and AlBO₄) and those of the borate structural units are shown in Fig. 6(b). When the concentration of BO_1.5 was lower than 40 mol%, the relative percentage of the orthoborate unit was high, and this led to a low thermal conductivity. Although the relative percentage of the borate structure related to ^[4]B (770 cm⁻¹) did not change substantially, a local maximum was observed in the relative percentages of the AlO_1.5-related structural units. Considering this similarity to the change of thermal conductivity, it has been concluded that the AlB₃O₇ and AlBO₄ structures have a positive effect on the thermal conductivity of the CaO–BO_1.5–AlO_1.5 system.

Fig. 6.

Effect of BO_1.5 addition on thermal conductivity and relative percentage of each structure (CaO/AlO_1.5=1.38).

3.3. Results of MAS-NMR Analysis

The formation of the alumino-borate network was considered in a more quantitative manner based on the results of the MAS-NMR measurement. Table 3 summarizes the relative fractions of the x-coordinated element i to the total i (N_[x]i), as calculated from the relative fractions of the peak area. The molar percentages of the x-coordinated element i to the total cations X(^[x]i) can be calculated from these results, as follows:   

$$X\left({}^{[x]}\mathrm{B}\right)=C_{\mathrm{B}{\mathrm{O}}_{1.5}}\cdot N_{{}^{[x]}\mathrm{B}}$$(3)

  

$$X\left({}^{[x]}\mathrm{A}\mathrm{l}\right)=C_{\mathrm{A}\mathrm{l}{\mathrm{O}}_{1.5}}\cdot N_{{}^{[x]}\mathrm{A}\mathrm{l}}$$(4)

where $C_{\mathrm{B}{\mathrm{O}}_{1.5}}$ and $C_{\mathrm{A}\mathrm{l}{\mathrm{O}}_{1.5}}$ are the BO_1.5 and AlO_1.5 concentrations (mol%), respectively. In Fig. 7, the calculated X(^[x]i) are plotted against the concentration of BO_1.5 at a fixed CaO/AlO_1.5 molar ratio of 1.38. From this result, it is clear that ^[4]Al decreased with the addition of BO_1.5, while ^[4]B increased. The network appeared to switch from aluminate-dominant to borate-dominant as concluded from the Raman spectrometry results. Moreover, the probability of formation for the AlB₃O₇ structural unit can also be inferred from these results. As shown in Fig. 4, at least one ^[4]Al tetrahedral, one ^[4]B tetrahedral, and two ^[3]B are necessary to construct the AlB₃O₇ structural unit (assuming that all of the oxygens coordinated to ^[3]B are bridging oxygens). In the measured compositions, ^[3]B was sufficient to form these structures, which indicated that the network was depolymerized by the NBO in the melt. When the structure is aluminate-dominant and contains sufficient ^[3]B, the probability of AlBO₄ formation is expected to be higher than that of AlB₃O₇ from the composition. As the composition become BO_1.5-richer, the probability of AlB₃O₇ formation increases. Naruse et al.¹⁷⁾ proposed that ^[4]B and ^[4]Al are expected to form AlB₃O₇ preferentially, and then the excess ^[4]Al and ^[4]B are used to form other aluminate or borate structural units. Under this assumption, the probability of AlB₃O₇ formation is determined by the X(^[4]B) in the aluminate-rich compositions and X(^[4]Al) in the borate-rich compositions. This expectation is graphically illustrated in Fig. 7. The relative percentage of the AlB₃O₇ structure expected in the Raman spectrometry has been successfully supported by the results of the MAS-NMR measurement, indicating the local maximum occurrs at approximately 60 mol% BO_1.5. Moreover, it is expected that the equimolar amount of ^[4]B and ^[4]Al in the melt would provide the maximum thermal conductivity. From the MAS-NMR measurement results, Fig. 8 is finally obtained by calculating X(^[4]B)/X(^[4]Al), which is considered as the molar ratio of ^[4]B to ^[4]Al for each composition. In this figure, the calculated value that is greater than unity indicates the borate-rich melt structure, while the value that is smaller than unity indicates the aluminate-rich structure. The compositions that have X(^[4]B)/X(^[4]Al) closer to unity are expected to have higher probability of AlB₃O₇ formation; that is, higher thermal conductivity. The distribution of this value shown in Fig. 8 is in good agreement with the thermal conductivity trend shown in Fig. 2, which provides quantitative evidence for the positive effect of the AlB₃O₇ structural unit on the thermal conductivity.

Table 3. Compositions and results of ¹¹B and ²⁷Al MAS-NMR measurements.

Sample No.	Relative Fractions of Each Coordination Number
	(to Total B)			(to Total Al)
	[3]B non-ring	[3]B ring	[4]B	[4]Al	[5]Al	[6]Al
1	0.43	0.30	0.27	0.66	0.26	0.08
2	0.45	0.32	0.23	0.51	0.41	0.08
3	0.45	0.28	0.27	0.53	0.39	0.08
4	0.38	0.26	0.36	0.56	0.38	0.06
5	0.38	0.26	0.36	0.59	0.28	0.13
6	0.41	0.30	0.29	0.62	0.27	0.12
7	0.38	0.39	0.22	0.63	0.29	0.09
8	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
9	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
10	0.56	0.36	0.08	0.72	0.24	0.04
11	0.54	0.39	0.06	0.78	0.11	0.11
12	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
13	0.42	0.50	0.08	0.64	0.22	0.14
14	0.53	0.39	0.07	0.80	0.14	0.06
15	0.53	0.41	0.06	0.78	0.19	0.03
16	0.41	0.51	0.09	0.81	0.17	0.02
17	0.55	0.41	0.04	0.70	0.30	0.00
18	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
19	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.
20	0.42	0.34	0.24	0.64	0.27	0.09

n.a.: Not analyzed.

Fig. 7.

Change in calculated X(^[x]i) with concentration of BO_1.5 at CaO/AlO_1.5=1.38. The calculated molar percentages of cations determining the maximum formation of the AlB₃O₇ structure are highlighted with the bold line.

Fig. 8.

Molar ratio of 4-coordinated B to 4-coordinated Al.

3.4. Effect of Substituting SiO₂ for AlO_1.5

The effect of substituting SiO₂ for AlO_1.5 on the thermal conductivity was also confirmed by comparing the results for the CaO–BO_1.5–AlO_1.5, CaO–BO_1.5–SiO₂–AlO_1.5, and CaO–BO_1.5–SiO₂ systems containing 25 mol% BO_1.5. Figure 9 shows the change of the thermal conductivity in response to the changes in the CaO concentration for each of these systems. As the CaO concentration increased from 36 to 40 and then to 44 mol%, the CaO/(AlO_1.5 + SiO₂) molar ratios increased from 0.92 to 1.15 and 1.38, respectively. For the CaO–BO_1.5–AlO_1.5 system, the composition corresponding to a CaO/AlO_1.5 molar ratio of 1.15 became maximum. From the Raman spectrometry results, a drastic increase in the orthoborate (930 cm⁻¹) and a decrease in the ^[4]Al–O–^[3]B stretching vibration in the aluminoborate network (980 cm⁻¹) were observed as a response to the increased CaO concentration. Therefore, the decrease of the thermal conductivity with a CaO content of more than 40 mol% can be explained by the corresponding de-polymerization of the aluminoborate network. With a CaO content of less than 40 mol%, the thermal conductivity increased with the addition of CaO. A slight increase was observed at 710 cm⁻¹, which was originated from the ^[4]Al–O stretching vibration in the aluminate network in this composition region. Because the thermal conductivities at BO_1.5=25 mol% are naturally low (owing to the depolymerized network) and less dependent on the compositions, the discussion herein includes uncertainty.

Fig. 9.

Change of thermal conductivity by substituting SiO₂ for AlO_1.5. The results for the CaO–BO_1.5–SiO₂ system were obtain by a previous study.⁷⁾

By comparing the results of other systems, the replacement of AlO_1.5 with more acidic SiO₂ yielded higher thermal conductivity. This can be easily explained as a result of introducing the silicate network. Raman spectrometry also revealed the increase in the relative fraction of silicate coordinating two/three bridging oxygen. However, the Raman spectrometry did not provide reliable results owing to the overlap of the bands used to determine the polymerization degree of the silicate network (900 cm⁻¹−1200 cm⁻¹) with the band used to detect the alumino-borate network (980 cm⁻¹). Thus, further investigation using ²⁹Si MAS-NMR should be conducted in future work.

4. Conclusion

The objective of this study is to clarify the relationship between the thermal conductivity and the structure of the CaO–BO_1.5–AlO_1.5 system. The thermal conductivity was measured using the transient hot-wire method. The composition dependency of thermal conductivity was discussed from the viewpoint of local structure, according to the analysis results obtained using Raman spectrometry and MAS-NMR.

The iso-thermal-conductivity curves were obtained at 1673 K in the CaO–BO_1.5–AlO_1.5 ternary diagram. The thermal conductivity was distributed in the range of 0.2–0.6 W m⁻¹ K⁻¹. Relatively high values (0.45–0.6 W m⁻¹ K⁻¹ at 1673 K) were obtained when the AlO_1.5 concentration increased, and this effect was more significant at a lower CaO/BO_1.5 molar ratio. The relative fraction of the 3D AlB₃O₇ network structure obtained by the Raman spectrometry was in good agreement with the increase of the thermal conductivity. Moreover, this result was quantitatively supported by the MAS-NMR results. When the concentrations of the BO_1.5 and AlO_1.5 were lower than 40 mol%, the thermal conductivity was relatively low (approximately 0.3 W m⁻¹ K⁻¹ at 1673 K), owing to the increase of the orthoborate unit.

With a fixed BO_1.5 content of 25 mol%, the CaO content increased from 36 to 44 mol% in the CaO–BO_1.5–AlO_1.5, CaO–BO_1.5–SiO₂–AlO_1.5, and CaO–BO_1.5–SiO₂ systems. Higher thermal conductivity was obtained when the SiO₂ was substituted for AlO_1.5 at the CaO contents of 36 and 44 mol%, because the SiO₂ had higher acidity than the AlO_1.5. This can be explained by the increase in the relative fraction of the silicate network. However, the structural analysis results obtained by Raman spectrometry were limited because the spectra representing the silicate and alumino-borate overlapped with each other. To clarify the CaO–BO_1.5–SiO₂–AlO_1.5 system, further investigations should be conducted using ²⁹Si MAS-NMR measurements.

Acknowledgment

The authors are grateful to Prof. S. Yamaguchi of the University of Tokyo (currently at National Institution for Academic Degrees and Quality Enhancement of Higher Education) and Prof. H. Inoue of the Institute of Industrial Science, the University of Tokyo, for their kind support in the usage of instruments for the structural analysis. They would like to specially thank Mr. K. Tanaka of the University of Tokyo for his guidance to use the facility. They also acknowledge Assistant Prof. Y. Kang of Dong-a University for his useful advice and discussion on the experimental work. They also thank Mr. S. Murai, Graduate Student of the University of Tokyo (currently at I-ARUMA’S Co., Ltd.) for his contribution in collecting data.

References

• 1)   Y.  Kang and  K.  Morita: ISIJ Int., 46 (2006), 420.
• 2)   Y.  Kang,  K.  Nomura,  K.  Tokumitsu,  H.  Tobo and  K.  Morita: Metall. Mater. Trans. B, 43 (2012), 1420.
• 3)   Y.  Kim and  K.  Morita: J. Am. Ceram. Soc., 98 (2015), 1588.
• 4)   Y.  Kim and  K.  Morita: Proc. 6th Int. Cong. on the Science and Technology of Steelmaking 2015, The Chinese Society for Metals, Beijing, (2015), 618.
• 5)   Y.  Kim,  Y.  Yanaba and  K.  Morita: J. Non-Cryst. Solids, 415 (2015), 1.
• 6)   Y.  Kim,  Y.  Yanaba and  K.  Morita: J. Am. Ceram. Soc., 100 (2017), 5746.
• 7)   Y.  Kim and  K.  Morita: ISIJ Int., 54 (2014), 2077.
• 8)   J. M.  Ziman: Electrons and Phonons: The Theory of Transport Phenomena in Solids, Oxford University Press, London, (1960), 288.
• 9)   W. D.  Kingery: Introduction to Ceramics, John Wiley & Sons, New York, (1967), 486.
• 10)   M.  Sakamoto,  Y.  Yanaba,  H.  Yamamura and  K.  Morita: ISIJ Int., 53 (2013), 1143.
• 11)   W.  Wang,  B.  Lu and  D.  Xiao: Metall. Mater. Trans. B, 47 (2016), 384.
• 12)   J.  Cho,  K.  Blazek,  M.  Frazee,  H.  Yin,  J. H.  Park and  S.  Moon: ISIJ Int., 53 (2013), 62.
• 13)   G. H.  Kim and  I.  Sohn: ISIJ Int., 52 (2012), 68.
• 14)   G. H.  Kim and  I.  Sohn: Metall. Mater. Trans. B, 45 (2014), 86.
• 15)   M.  Susa,  K.  Nagata and  K.  Goto: Trans. Jpn. Inst. Met., 29 (1988), 133.
• 16)   R. K.  Brow,  D. R.  Tallant and  G. L.  Turner: J. Am. Ceram. Soc., 80 (1997), 1239.
• 17)   A.  Naruse,  Y.  Abe and  A.  Takami: Yogyo-Kyokai-Shi, 79 (1971), 225 (in Japanese).
• 18)   K.  Fukumi,  J.  Fukunaga,  N.  Yoshida and  M.  Ihara: Yogyo-Kyokai-Shi, 92 (1984), 680 (in Japanese).
• 19)   G.  Padmaja and  P.  Kistaiah: J. Phys. Chem. A, 113 (2009), 2397.
• 20)   H.  Li,  Y.  Su,  L.  Li and  D. M.  Strachan: J. Non-Cryst. Solids, 292 (2001), 167.