You are on page 1of 16

Received: 20 December 2018 

|  Revised: 25 March 2019 


|
  Accepted: 23 April 2019

DOI: 10.1111/jace.16526

ORIGINAL ARTICLE

Effect of CaF2, B2O3 and the CaO/SiO2 mass ratio on the


viscosity and structure of B2O3‐containing calcium‐silicate‐based
melts

Gi Hyun Kim   | Il Sohn

Department of Materials Science and


Engineering, Yonsei University, Seoul,
Abstract
Korea The relationship between the viscosity and structure of B2O3‐containing calcium‐
silicate‐based mold fluxes and the effects of fluidizers including CaF2, CaO, and
Correspondence
Il Sohn, Department of Materials Science B2O3 on the viscosity and their correlation with the structural aspects were studied
and Engineering, Yonsei University, 50 using a rheometer with Fourier transformation infrared and Raman spectroscopy.
Yonsei‐ro, Seodaemun‐gu, Seoul, Korea,
The viscosity decreased with increasing CaF2 addition up to 28 wt% at a fixed CaO/
03722.
Email: ilsohn@yonsei.ac.kr SiO2 ratio of 0.3, which was related to depolymerization. Furthermore, CaF2 addi-
tion also affected the apparent activation energy for viscous flow, which decreased
Funding information
with increasing CaF2 content to 105.1 from 151.1 kJ/mol. At higher C/S ratios, the
Graduate School of YONSEI University
Research Scholarship Grants; National viscosity decreased in the presence of greater Ca2+ and O2− supplied from CaO,
Research Foundation of Korea, Grant/ which subsequently increased the activation energy to 149.7 from 122.0  kJ/mol.
Award Number: NRF-2018R1A2B2006609
With regard to the B2O3‐melt, polymerization of the network structure was observed
by comparing the B2O3‐free to 4.4 wt% B2O3 content. However, the viscosity was
relatively constant with increasing B2O3 addition. However, the viscosity decreased
due to greater simplification of the network structure above 4.4 wt% B2O3. The break
temperature decreased with greater B2O3 addition as the crystallization was inhib-
ited. Furthermore, the apparent activation energy decreased as depolymerization of
the network structures occurred above 4.4 wt% B2O3.

KEYWORDS
B2O3, CaF2, CaO/SiO2 mass ratio, mold flux, network structure, viscosity

1  |   IN T RO D U C T ION for application in high‐Al‐content steels exhibiting transfor-


mation induced plasticity and twinning induced plasticity
The development of continuous casting mold fluxes has been steels.14,15 The use of B2O3 is often employed to lower both
an active topic of research aiming for greater productivity the liquidus and solidus temperatures of the melt and expand
and improved quality in steel products. Optimized fluxes the solid‐liquid co‐existing region of the flux to enhance
ensure sufficient and uniform lubrication between the par- lubrication.1,7,13,16,17
tially solidified shell and water‐cooled copper mold, uniform To ensure uniform heat transfer across the copper mold,
heat transfer across the solidified flux film, and insulation optimal use of B2O3 in continuous casting mold flux requires
from excessive radiative heat loss above the top surface of a uniform distribution of amorphous and crystalline phases
the flux film.1‒13 In particular, the wider use of B2O3 fluid- dispersed within the flux layer. Often known for its glass‐
izers in the constituents that compose calcium‐silicate and forming ability in the glass industry, B2O3 in mold fluxes has
calcium‐aluminate‐based mold fluxes has been implemented been known to inhibit crystallization by forming complex

J Am Ceram Soc. 2019;00:1–16. wileyonlinelibrary.com/journal/jace © 2019 The American Ceramic Society     1 |


|
2      KIM and SOHN

boro‐silicate structural units, which may not be optimal for Thus, to ensure that a flux can retain the cuspidine
steel grades requiring lower heat removal.4‒6,8,14,16 (Ca4Si2O7F2) primary crystal phase during crystallization,
In general, fluoride‐containing compounds such as CaF2 while having a lower melting temperature and optimal vis-
and NaF have been widely used as fluidizers in commercial cosity, a fundamental understanding of the CaF2, C/S ratio,
continuous casting mold powders. In the typical silicate‐based and B2O3 effects on the thermo‐physical properties and struc-
mold fluxes, CaF2 and NaF can depolymerize the silicate ture is necessary but has not yet been obtained in the contin-
structural units, leading to a lower viscosity.18‒20 According to uous casting for high‐carbon steels with a high casting speed
Sakamaki et al,21 additional NaF interacts with Si‐O to form between 3.0 and 3.5 m/min to diminish the defects and cracks
Si‐F bonds, and bridged oxygen anions (O0) become non‐ on the surface of the slabs relative to the results of using a
bridged oxygen anions (O−) in the Na2O‐NaF‐SiO2 system. commercial mold flux.9
According to Hayashi et al,22 CaF2 addition was found to affect In this study, the viscosity of B2O3‐containing calcium‐
the silicate network structures through Ca2+, as Ca‐F+ pairs silicate commercial melts was studied and correlated with the
indirectly link to the silicate network structures, which can re- structure for temperatures up to 1673 K (1400°C). The effect
duce the resistance for shearing, and the viscosity is decreased of CaF2, C/S ratio, and B2O3 on the viscosity was observed in
in CaO‐CaF2‐SiO2(‐FeOx) systems with higher CaF2 content. detail considering the post‐experimental compositional vari-
In addition, according to Takeda et al23 and Sasaki et al,24,25 ation. Structural analysis of as‐quenched fluxes from 1673 K
Na‐F and Ca‐F+ complexes are formed, and F− anions are not (1400°C) was carried out with FTIR and Raman spectroscopy.
directly linked to the silicate network structures and do not
directly depolymerize the silicate network structures in Na2O‐
2  |   EXPERIM ENTAL M ETHOD
NaF‐SiO2 and CaO‐CaF2‐SiO2 since F− anions tend to move to
AND PROCEDURES
higher‐field‐strength alkali and alkali earth cations. However,
it can be speculated that the bonding preference of F− anions
2.1  |  Materials preparation
towards the alkali and alkali earth cations tends to better release
the O2− to interact with the O0 in the silicate structural units. Commercial mold fluxes (Flux 1, 4, 5, and 6) and reagent
In the case of B2O3‐containing melts, Kim and Morita26 grade chemicals (Flux 2, 3, 7, and 8) were used in the prepara-
and Park and Sohn27 characterized the network‐forming abil- tion and comparison of the calcium‐silicate containing B2O3
ities of B2O3 in high‐temperature melts by forming 4‐coordi- multi‐component fluxes. The post‐experimental composition
nated boron structural units. According to Kim and Morita,28 was the average of three separate measurements using X‐ray
B2O3 could be incorporated into the silicate network at low fluorescence (S4 Explorer, Bruker AXS GmbH, Germany)
C/S ratios forming complex boro‐silicate structures, but at and inductively coupled plasma‐optical emission spectros-
higher basicities of 1.15, a higher B2O3 content increased the copy (700 Series, Agilent, USA) as presented in Table 1.
amount of complex independent meta‐borate structures in the In the post‐experimental composition, the volatilization of
CaO‐SiO2‐B2O3 ternary system. According to Zhou et al1 and fluoride was observed within 1 wt%. Flux 1, 4, 5, and 6 are
Zhou and Wang,2 B2O3 in F‐free mold fluxes resulted in a de- commercial mold fluxes used for high‐carbon steels, and the
crease in the viscosity and the crystallization temperature to others in the present work are synthesized fluxes using rea-
form the calcium‐boro‐silicate (Ca11Si4B2O22) phase that re- gent grade chemicals that mimic the major components of
placed the cuspidine (Ca4O7Si2F2) phase, as also observed by commercial fluxes. CaCO3, Na2CO3, and Li2CO3 were fully
Yang et al3 Similar results were also observed by Yan et al14 calcined and mixed with the other constituents. The mixed
and Fox et al,17 who reported that the addition of network‐ powders were pre‐melted at 1673  K (1400°C) and crushed
forming B2O3 resulted in prolonged incubation times and for the actual experiments, following similar procedures de-
lower crystallization temperature in CaO‐Al2O3‐based melts. scribed previously.9,18,30‒33
The effect of B2O3 additions to the structure of Ti‐bearing
molten fluxes was studied by Li et al29 using a combination
2.2  |  Viscosity measurements
of Fourier transform infrared (FTIR), Raman spectroscopy,
nuclear magnetic resonance (NMR), and X‐ray photoelectron Pulverized pre‐melted flux (100  g) was placed in a
spectroscopy. The results showed a tendency of the non‐ring Pt‐10%Rh crucible (OD: 42 mm, ID: 46 mm, H: 65 mm),
[BO3] to convert into the [BO4]‐tetrahedral, which promoted which was inserted into a tight‐fitting Al2O3 crucible (OD:
polymerization of the structural units and subsequently in- 52 mm, ID: 46 mm, H: 70 mm). UHP Ar (99.9999%) gas
hibited crystallization. However, the pronounced amount of of 300 sccm was injected into the reaction tube of a verti-
two‐dimensional ring [BO3] structures resulted in decreased cal resistance furnace, as shown in Figure 1A. Figure 1B
strength between the structural units and tended lower the shows the viscosity measurements of Flux 2 taken during
flux viscosity with B2O3 addition even with the existence of the cooling cycle with respect to time. Sequential measure-
complex [BO4]‐tetrahedral structural units. ments were taken during the holding period of the target
KIM and SOHN
|
     3

temperature during cooling. Using the rotating spindle

CaO/SiO2
method, the torque of the Pt‐10%Rh bob was measured with
a Brookfield rheometer (LV‐II+; Brookfield Engineering

0.8
0.8
1.0
0.8
0.4
0.3
0.3
0.3
Laboratories, Middleboro, MA) which was calibrated by
standard silicon oils (9.5, 47.8, and 96.7 mPa/s) at a rota-
tion speed of 100  rpm. A reference B‐type thermocouple

2.6 (±0.016)
2.5 (±0.010)
2.3 (±0.031)

2.7 (±0.14)
2.5 (±0.13)
2.6 (±0.02)
2.8 (±0.19)
2.4 (±0.11) (Pt‐30%Rh: Pt‐6%Rh) was used to calibrate the furnace
temperature control thermocouple connected to a propor-
Li2O

tional integral derivative controller to within ±3 K of the


target temperature of 1673  K (1400°C). The temperature
T A B L E 1   Chemical composition of the flux samples after experiments using X‐ray fluorescence and inductively coupled plasma‐optical emission spectroscopy

4.4 (±0.0014) was lowered in 25 K intervals for 10 minutes and held for

7.2 (±0.08)
4.2 (±0.22)
4.6 (±0.26)
2.4 (±0.05)
2.4 (±0.03)
2.7 (±0.04)

30 minutes to obtain thermal equilibrium within the molten


flux. The temperature was subsequently lowered below
B2O3

the break temperature of the flux until a high volumetric


0

ratio of the solid particle region was achieved. The viscos-


ity measurements are expressed as the average of the col-
13.4 (±0.46)
12.9 (±0.49)

14.3 (±1.2)
13.6 (±1.2)
12.5 (±2.9)
14.0 (±1.3)
28.0 (±1.4)

lected data for 5 minutes; thermal equilibrium was reached


after 25  minutes. Typical standard errors were less than
CaF2

±0.007 Pa/s, as is typical for this method.30


0

2.3  |  Structural analysis using FTIR and


14.3 (±0.19)
16.2 (±0.19)
16.4 (±0.19)
17.2 (±0.20)
15.5 (±0.19)
14.6 (±0.19)
16.7 (±0.19)

15.3 (±0.2)

Raman spectroscopy
Na2O

Sufficient B2O3 additions result in the formation of independ-


ent and incorporated borate structural units within the domi-
nant silicate structural units found in calcium‐silicate‐based
0.7 (±0.0095)
0.7 (±0.0096)

2.6 (±0.018)
2.2 (±0.015)
2.2 (±0.015)
2.4 (±0.016)
4.6 (±0.014)
0.6 (±0.009)

fluxes. The schematic of the various borate structural units


adapted from previous reports is depicted in Figure 232,34
and the structural change from silicate network structures
FeO

to boro‐silicate network structures by B2O3 additions is de-


picted in Figure 3.35 The borate structures can be classified
3.3 (±0.069)
3.3 (±0.045)
3.2 (±0.044)
3.1 (±0.044)
3.1 (±0.044)
3.6 (±0.072)
3.7 (±0.073)
3.9 (±0.049)

broadly into three‐dimensional and two‐dimensional struc-


tural units, where the three‐dimensional structural units are
MgO

more complex. Within the present study, structural analysis


of as‐quenched glassy fluxes from 1673  K (1400°C) was
conducted using a combination of FTIR and Raman spectros-
1.7 (±0.057)
2.1 (±0.057)
2.1 (±0.056)
1.9 (±0.057)
3.3 (±0.126)
3.3 (±0.079)

copy. The reference data to identify the IR‐ and Raman‐active


2.2 (±0.12)
3.4 (±0.14)

structural units was obtained from the literature.9,27,28,32,35‒37


Al2O3

The structural analyses were performed at least three times


to estimate the standard errors; the standard errors of the de-
convoluted area of Raman spectroscopy are the differences
30.5 (±0.42)
33.0 (±0.21)
27.9 (±0.19)
31.2 (±0.20)
37.5 (±0.22)
54.6 (±0.51)
45.4 (±0.49)
31.9 (±0.21)

between the observed spectra and the fitted spectra by Peakfit


v4 program (AISN software Inc., United Kingdom).30
SiO2

3  |  RESULTS AND DISCUSSIO N


10.4 (±0.012)

23.9 (±0.16)
26.6 (±0.13)
27.8 (±0.13)
24.5 (±0.13)
16.0 (±0.11)
17.6 (±0.12)
14.6 (±0.14)

3.1  |  Effect of CaF2 on the viscosity


The effect of CaF2 on the viscosity in the CaO‐SiO2‐Al2O3‐
CaO

MgO‐FeO‐Na2O‐CaF2‐Li2O‐B2O3 multi‐component flux at a


C/S ratio of 0.3 and various temperatures is shown in Figure
4A,B. Higher concentrations of CaF2 lowered the viscosity of
Flux

the flux. Relative to the CaF2‐free flux, the flux with CaF2 had
8
7
6
5
4
3
2
1
|
4      KIM and SOHN

F I G U R E 1   (A) A schematic of the experimental apparatus and (B) a relation between viscosity and temperature cycle for viscosity
measurements in the present study

F I G U R E 2   A schematic of borate structural units

a significantly lowered viscosity, which can be correlated with estimated viscosities for Flux 1 in the fully liquid state (below
the depolymerization of the silicate structures with higher the break temperature, the viscosity is not affected by the pre-
CaF2 addition, which has been previously elucidated,9,18,20 cipitated crystals because the crystals occupy a smaller volu-
where CaF2 can react with the complex silicate network struc- metric fraction) were determined using an Arrhenius‐type
tures resulting in the depolymerization of silicate rings and equation. The Einstein‐Roscoe equation considers the solid
chains to form simple silicate network structures such as di- phase volumetric fraction within the liquid phase below the
mers and monomers. Figure 4B shows that a 28 wt% CaF2‐ crystallization temperature estimated by FactSage®.38
containing flux exhibited a slower increase in viscosity during The Einstein‐Roscoe equation assumes that solid crystals
cooling and a break temperature of 1223 K (950°C). Below are insoluble spherical particles according to the following
the break temperature, the viscosity calculated by FactSage® Equation (1), where the constants a and n are 1.35 and 2.5,
significantly increased by the precipitated crystal phase; the respectively39‒43:
KIM and SOHN
|
     5

𝜂 = 𝜂0 × (1 − a × f )−n (1) viscosity of the fully liquid phase (Pa·s), and f is the volumetric
fraction of the spherical solid particles.
where η is the dynamic viscosity of the solid‐liquid phase Numerous works39‒43 have examined the effect of solid
containing spherical solid particles (Pa·s), η0 is the dynamic particles on the viscosity under solid‐liquid co‐existing mix-
tures by stirring the fluid with a spindle at various rotation
speeds. The present approach assumed a crystalline solid
dispersed in a Newtonian fluid when the volume fraction of
solid particles was lower than 0.1.

3.2  |  Effect of the CaO/SiO2 mass ratio


on the viscosity
Figure 5A,B show the effect of the C/S ratio on the viscos-
ity at comparable contents of other constituents in the flux.
Increasing the C/S ratio from 0.4 to 0.8 had a significant ef-
fect in lowering the viscosity. The addition of CaO provides
free oxygen ions (O2−) in the flux at various temperatures,
which can be used to modify the complex silicate and boro‐
silicate network structures connected by the bridged oxygen
atoms (O0) resulting in higher amounts of O−, as suggested
by Waseda and Toguri44. This modification of the network
structures is clearly evident for acidic fluxes (Flux 4), where
the C/S ratio is 0.4. At a higher C/S ratio above 0.8, the addi-
tional O2− has a smaller effect on the viscosity. The combined
effects of the higher C/S ratio and pre‐existing fluidizers in-
cluding Li2O, Na2O, and CaF2 limited the change in viscos-
ity above a C/S ratio of 0.8. This trend correlated well with
the modification of the structural units comprising the flux,
which is discussed later in detail.
In terms of the crystallization behavior and its correlation
with the structure, the excess O2− from CaO can modify the
complex network flux structures and simplify the structural
F I G U R E 3   A schematic of effect of B2O3 addition in the silicate
units comprising the melt. A more simplified unit structure
network structures
requires less driving forces for atom re‐organization and

F I G U R E 4   Viscosity of the CaO‐SiO2‐(3.3‐3.4)Al2O3‐(3.6‐3.9)MgO‐(0.6‐0.7)FeO‐(14.6‐16.7)Na2O‐(2.4‐2.7)B2O3‐(2.3‐2.5)Li2O multi


component fluxes (wt%) with respect to (A) CaF2 content and (B) temperature at a fixed CaO/SiO2 ratio of 0.3
|
6      KIM and SOHN

F I G U R E 5   Viscosity of the CaO‐SiO2‐(1.9‐2.1)Al2O3‐(3.1‐3.2)MgO‐(2.2‐4.6)FeO‐(15.5‐17.2)Na2O‐(12.5‐14.0)CaF2‐(4.2‐4.6)B2O3‐


(2.5‐2.8)Li2O multi‐component fluxes (wt%) with respect to (A) CaO/SiO2 ratio and (B) temperature

crystallization. In addition, the lower viscosity enhances up to 4.4 wt% had a negligible influence on the viscosity,
ionic mass transport in the molten oxide system increasing which is attributed to the existence of other highly basic flu-
the mass transfer coefficient and the kinetics.1‒5,8,10,11 The idizers of Li2O and Na2O composing the flux in excess of 17
flux for a C/S ratio of 1.0 (Flux 6) has a relatively high wt%. In the present compositional range of this work, B2O3
break temperature, which is discussed later in detail. enhanced the formation of boro‐silicate network structures29
but did not increase the viscosity due to the competitive ef-
fects between the degree of polymerization (DOP, Q3/Q2) of
3.3  |  Effect of B2O3 on the viscosity
the network structures and the presence of [BO3]‐trigonal and
Figure 6A and C show the effect of B2O3 on the viscosity at [BO4]‐tetrahedral structural units.35 With greater B2O3 con-
comparable contents of other constituents in the fluxes at a tent of 8.8 wt% (Flux 8), the viscosity in the fully liquid re-
fixed C/S ratio of 0.8. Comparison of the B2O3‐free (Flux 7) gion was lower than that of other fluxes. The lower viscosity
and 4.4 wt% B2O3 (Flux 5) fluxes showed negligible effects observed with B2O3 addition of 8.8 wt% can be explained by
on the viscosity within the temperature range of this work. the greater proportion of independent borate structural units
By comparing the viscosity measurements of the B2O3‐con- compared to the silicate structural units relative to the flux
taining and B2O3‐free fluxes as a function of temperature with 4.4 wt% B2O3, which is discussed later in detail.
in Figure 6B, the addition of B2O3 is found to significantly
extend the minimum temperature at which stable viscosity
3.4  |  Effect of CaF2 on the structural
measurements can be made. This extension corresponds to a
units of high‐temperature melts
decrease in the break temperature of the melt with B2O3 ad-
ditions and subsequent extension of the solid‐liquid region. Using FTIR and Raman spectroscopy, the silicate structural
Above the break temperature of the melt, the addition of B2O3 units between 1200 and 800  cm in terms of wavenumber

F I G U R E 6   Viscosity of the CaO‐SiO2‐(1.7‐2.2)Al2O3‐(3.1‐3.3)MgO‐(2.2‐2.6)FeO‐(14.3‐17.2)Na2O‐(2.6‐2.7)Li2O multi‐component fluxes


(wt%) with respect to (A) B2O3 contents and (B) temperature at a fixed CaO/SiO2 ratio of 0.8 and (c) comparisons of other work1
KIM and SOHN      7
|
and Raman shift can be observed. The non‐bridged oxy- similar trends to the FTIR. The effect of CaF2 on the fraction
gen per tetrahedral cation (NBO/T) and in particular the of structural units is shown in Figure 9A. The average NBO/T
ratio of Q3/Q2 can be correlated with the DOP, as described and the DOP (Q3/Q2) are represented in Figure 9B.
previously.45‒51 From the FTIR analysis, the existence and Major structural unit changes could be identified for the
estimated depth of the trough at specific wavenumbers can NBO/T = 4 and 3, the Q3/Q2, bending vibration in boro‐sil-
be assigned to the silicate and borate structural units. From icate structural units and metaborate and BO2O− linked to
the Raman analysis, a more quantitative analysis through [BO4]‐tetrahedral structures in borate structural units.
deconvolution of the spectra and subsequent area fraction With CaF2 addition up to 14 wt%, the NBO/T = 3, DOP,
analysis can reveal the amount of structural units present and bending vibrations of the Si‐O‐Si bonds decreased, but
in the fluxes.45 the BO2O− linked to [BO3]‐trigonal structural units increased
Figure 7 shows the change in structural units with CaF2 ad- due to excess Ca2+ from CaF2 addition at low C/S ratios.34,52
ditions according to the FTIR analysis at comparable amounts Thus, the modification of the relatively larger silicate struc-
of the other constituents. The flux containing 28 wt% CaF2 tural units and generation of the simple borate structural units
(Flux 1) had a deeper trough at NBO/T = 4 with respect to decreased the viscosity.
the other silicate structural units in comparison to the flux With greater CaF2 addition up to 28 wt%, the DOP slightly
containing 14 wt% CaF2 (Flux 2) and CaF2‐free flux (Flux decreased, and the simpler structural units of NBO/T = 4 in-
3). With CaF2 addition, the bands between 1200 and 800 cm, creased. In addition, the borate structural units were further
which are assigned to Si‐O stretching vibrations became nar- transformed from BO2O− linked to [BO3]‐trigonal structural
rower, which indicates that the larger silicate structural units units to BO2O− linked to [BO4]‐tetrahedral structural units with
gradually disappeared. Specifically, the peaks of NBO/T = 3 [BO3] resulting in the incorporation of [BO4]‐tetrahedral units
and 1 became weaker, but the peaks of NBO/T  =  4 and 2 into the [SiO4]‐tetrahedral structures linked with BO2O−‐trigo-
become stronger with increasing CaF2. At 28 wt% CaF2, the nal structures having O− similar to alkali cations since Ca2+
peak of NBO/T = 0 at approximately 1150 cm is compara- could compensate the B4+ in [BO4]‐tetrahedral units at higher
tively less pronounced and the peaks of the stretching vibra- CaF2 contents, producing BO2O− structural units.34,52‒55 This
tion in [BO3]‐trigonal structural units are stronger than for the finding is consistent with the 11B NMR findings of Park et al9
14 wt% CaF2 and CaF2‐free fluxes. The FTIR results suggests that non‐ring type [BO3]‐trigonal and [BO4]‐tetrahedral struc-
changes of the NBO/T = 4, 3, 2, 1 and [BO3] structural units tural units were more prevalent at greater CaF2 content.
with CaF2 addition. To supplement the qualitative structural According to Umesaki et al,56 the average NBO/T can be
analysis of the FTIR spectra, semi‐quantitative Raman spectra calculated using the mole fraction (Xi) of tetrahedral struc-
for structural analysis were also obtained tural units with non‐bridging oxygen per cation (T). The aver-
The typical Raman spectra for flux 6, shown in Figure 8, age NBO/T values are 1.30, 0.95, and 0.92 at 0 wt%, 14 wt%,
can be deconvoluted and analyzed by integrating the area. The and 28 wt% CaF2, respectively, which are corresponded with
corresponding Raman analysis, shown in Figure 9A, shows the DOP as shown in Figure 9B.

F I G U R E 7   Fourier transform infrared results of as‐quenched F I G U R E 8   Observed Raman spectra and the deconvoluted
samples from 1673 K (1400°C) with respect to CaF2 contents from 0 to spectra for Flux 6
28 wt% at a fixed CaO/SiO2 ratio of 0.3
|
8      KIM and SOHN

F I G U R E 9   (A), The structural fractions and (B) the average non‐bridged oxygen per tetrahedral cation (NBO/T) and the degree of
polymerization from the integration of the deconvoluted Raman spectra with respect to CaF2 contents for Flux 1, 2, and 3 at a fixed CaO/SiO2 ratio
of 0.3

Thus, CaF2 affected both the silicate and borate structures


in B2O3 containing multi‐component calcium‐silicate‐based
fluxes, which is consistent with the aforementioned viscosity
trends.

3.5  |  Effect of the CaO/SiO2 mass


ratio on the structural units of high‐
temperature melts
Using FTIR and Raman spectroscopy, the effect of the
C/S ratio on the structural units of the flux was analyzed
as shown in Figures 10 and 11, respectively. Considering
the dominant chemical composition in the fluxes, the pro-
nounced structural units are the silicate structural units,
where SiO2 comprises approximately 30‐40 wt% with B2O3
addition not exceeding 5 wt%. According to Kim et al,57
F I G U R E 1 0   Fourier transform infrared results of as‐quenched
the ratio of Na2O/B2O3 could affect the formation of bo- samples from 1673 K (1400°C) with respect to CaO/SiO2 ratio from
rate structural units, which could be transformed to ring and 0.4 to 1.0
non‐ring [BO3] and [BO4]‐tetrahedral structures, as previ-
ously illustrated within Figure 2. In the boro‐silicate flux
system of the present work, borate structural units could typically assigned to the stretching vibrations of the B‐O
be incorporated into the silicate structures, resulting in a in [BO3]‐trigonal structural units, is diminished with an in-
complex boro‐silicate structural unit. With higher C/S ra- creasing C/S ratio. This finding is consistent with the results
tios, the trough corresponding to NBO/Si  =  0 disappears, of Zhang et al,13 who found that excess O2− transformed
while a greater depth can be observed for the NBO/Si = 4 the [BO3]‐trigonal structural units to BO2O− having O− and
trough, as shown in Figure 10. The O2− mainly supplied by [BO4]‐tetrahedral structural units.58,59
the CaO depolymerizes the complex silicate structural units Figure 11A,B shows the effect of the C/S ratio on the
to simpler silicate structural units (NBO/=4), subsequently structural units after deconvolution and subsequent area inte-
decreasing the DOP. The silicate stretching band also shifts gration of the Raman spectra. In the silicate structural units,
towards lower wavenumbers with the relative disappearance NBO/T = 4 increased and the DOP decreased, correspond-
of NBO/T = 1 and 0 at higher C/S ratios, as the NBO/T = 0 ing to a depolymerization with higher C/S ratios. The aver-
structural units are modified with the interaction of O2−. age NBO/T values are 2.47, 2.26, and 1.74 at 0.4, 0.8, and
For the borate stretching bands observed between 1600 1.0 of C/S ratios, respectively, which are corresponded with
and 1200  cm, the prominent peak at 1350  cm, which is the DOP in Figure 11A. With respect to the borate structural
KIM and SOHN
|
     9

F I G U R E 1 1   (A), The structural fractions and (B) the average non‐bridged oxygen per tetrahedral cation (NBO/T) and the degree of
polymerization from the integration of the deconvoluted Raman spectra with respect to CaO/SiO2 ratio for Flux 4, 5, and 6

units, the fraction of BO2O− vibration linked to [BO4]‐tetra- [BO3]‐trigonal structural units can be converted into [BO4]‐
hedral structural units gradually increased with higher C/S tetrahedral structural units by reaction with O2− supplied
ratios as the excess O2− interacted with the O0 and trans- from alkali and alkali‐earth oxides including Na2O, Li2O, and
formed the [BO3]‐trigonal to [BO4]‐tetrahedral structures in- CaO. Ring and non‐ring type [BO3] structural units are two‐
corporated into silicate structures. As a result, the fraction of dimensional trigonal structures and [BO4] structural units are
BO2O− vibration linked to [BO3]‐trigonal structures slightly three‐dimensional tetrahedral structures.
decreased. According to the 11B magic angle spin‐NMR re- Figure 12 shows the effect of B2O3 on the structural units
sults of Park et al9 and Raman results of Kim et al,57 the of the melt analyzed by FTIR between the wavenumbers of
borate structural units can transform to either the non‐ring 1600 and 600 cm. The wavenumber region between 1600 and
[BO3], ring [BO3], or [BO4]. With an increasing C/S ratio, 1200 cm can be assigned to the stretching vibrations of the B‐O
the majority of non‐ring [BO3] and [BO4] (1B, 3Si and 0B, bonding in [BO3]‐trigonal related structures. The stretching vi-
4Si) increased. According to Park et al,9 there are 2 types brations of Si‐O bonding in [SiO4]‐tetrahedral related structures
of [BO4]‐tetrahedral structural units: (1B, 3Si) and (0B, appear within the wavenumbers between 1200 and 800 cm.
4Si). The [BO4] (1B, 3Si) unit corresponds to [BO4] bonded With B2O3 addition to 4.4 wt%, the stretching vibration
around one [BO4] and three [SiO4] and the [BO4] (0B, 4Si) bands between 1150 and 750 cm became narrower due to the
unit corresponds to [BO4] bonded around four [SiO4]. As
previously mentioned, the [BO4]‐tetrahedral structural units
are incorporated into the silicate structures according to the
11
B NMR results. In the Raman analysis of the present com-
positional range, the variation of BO2O− vibrations linked
to [BO4]‐tetrahedral structural units was consistent with the
results of Kim et al28 with a higher C/S ratio. As mentioned
by Park et al9 and in other reports,28,57 borate structures can
be incorporated into silicate structures depending on the
number of O−. Therefore, the variation of BO2O− vibrations
linked to [BO4]‐tetrahedral structural units increased at the
expense of [BO3]‐trigonal structural units according to the
present Raman results.

3.6  |  Effect of B2O3 on the structural


units of high‐temperature melts
As previously illustrated, molten B2O3 can form 3 types F I G U R E 1 2   Fourier transform infrared results of as‐quenched
of structural units, namely, ring [BO3], non‐ring [BO3], samples from 1673 K (1400°C) with respect to B2O3 contents from 0
and [BO4] tetrahedral structural units. B2O3 comprising of to 8.8 wt% at a fixed CaO/SiO2 ratio of 0.8
|
10      KIM and SOHN

presence of borate structural units relative to the B2O3‐free the boro‐silicates with initial B2O3 addition, which is depicted
flux. In addition, the bending vibrations of Si‐O‐Si and the in Figure 3. The average NBO/T values are 2.18, 2.26, and 2.04
stretching vibration in NBO/T = 3 and 2 became less promi- at 0 wt%, 4.4 wt%, and 8.8 wt% B2O3, respectively, which are
nent, but that at NBO/T = 4 and 1 became more pronounced, corresponded with the DOP in Figure 13B. With greater addi-
and the stretching vibrations of [BO3] appeared at the wave- tion of B2O3 from 4.4 to 8.8 wt%, the DOP decreased, but the
numbers of 1450, 1370, and 1230 cm. amount of O− in BO2O− linked to the [BO3] and [BO4] struc-
With additional B2O3 from 4.4 to 8.8 wt%, the band between tural units increased, which led to a modification of the network
1150 and 750  cm became narrower and the stretching vibra- structures and a lowered viscosity.
tions of [BO3] became more pronounced at the wavenumbers Therefore, complex structural changes were observed
of 1450, 1370, and 1230 cm, which led to higher depolymer- from the addition of 4.4 wt% B2O3 relative to the B2O3‐free
ization and subsequently lower viscosity. This result correlates state by linking between Si and B but the viscosity was al-
well with that of Zhang et al,13 who found that higher B2O3 most constant. As more B2O3 is added, the network structures
addition resulted in greater propensity toward the dissociated become slightly depolymerized with greater amounts of the
simple network structures from more complex structures. O− in borate structures resulting in lower viscosity.
From the Raman analysis of Figure 13A,B, a similar trend to
the FTIR results can be observed. From 0 to 4.4 wt% B2O3 ad-
3.7  |  Activation energy for viscous
dition, the DOP was increased and the [BO4]‐tetrahedral struc-
flow and the break temperature
tural units could be observed, but the Si‐O‐Si bending vibration
decreased. The lower amount of Si‐O‐Si bending vibrations Figure 14 shows the temperature dependence of Flux 4, 5,
suggests that some of the borate structural units were incorpo- and 6 with the natural logarithm of the viscosity as a function
rated into the silicate structures, breaking the Si‐O‐Si bonding of the reciprocal temperature following Equation (2)
forming boro‐silicate structural units. In addition, NBO/T = 4
and [BO3] and [BO4] structural units with BO2O− increased ( )
Ea
even though the DOP‐related Q3/Q2 ratio increased. Therefore, 𝜂 = 𝜂0 × exp
RT (2)
the viscosity was unchanged by the competition between the
increased DOP by incorporated borate structural units and the
modification of the network structures by O− in BO2O− with where η is the viscosity, Ea is the activation energy for the vis-
[BO3] and [BO4] structural units. In the boro‐silicate system, cous flow, R is the ideal gas constant (8.314 kJ/mol/K), and T
the effect of B2O3 on the structural aspect was elucidated using is the temperature.
FTIR, Raman, and NMR spectroscopy.35,47 It was concluded The shaded symbols are the measured viscosities, the open
that additional B2O3 was incorporated into the silicate network symbols are the calculated viscosities using FactSage® consid-
structures by connecting the 4Si atoms (0B, 4Si) to the [BO4]‐ ering the fraction of solid crystalline particles co‐existing in the
tetrahedral structures, inhibiting the formation of independent liquid according to the Einstein‐Roscoe Equation (1), and the
[BO4]‐tetrahedral structures and resulting in higher DOPs for half‐shaded symbols are the break temperature of the fluxes.

F I G U R E 1 3   (A), The structural fractions and (B) the average non‐bridged oxygen per tetrahedral cation (NBO/T) and the degree of
polymerization from the integration of the deconvoluted Raman spectra with respect B2O3 contents for Flux 5, 7, and 8 at a fixed CaO/SiO2 ratio of
0.8
KIM and SOHN
|
     11

the open‐times symbols, which corresponds to the high solid


fraction, intersects at the break temperature. The obtained acti-
vation energies for viscous flow within the fully liquid region
and below the break temperature are provided in Table 2, and
the break temperatures of fluxes are provided in Table 3.
The break temperature determines the lubrication ability
of flux below the meniscus within the solid‐liquid co‐existing
temperature region. During continuous casting, the infiltrated
liquid flux layer between the partially solidified steel shell and
the water‐cooled copper mold solidifies. If a lower break tem-
perature can be achieved, complete solidification of the flux
can be extended within the mold and improve the lubrication
ability of the flux traversing the mold.8,9,13 Mills et al8 and
Fox et al17 suggested that the break temperature is similar to
the crystallization temperature; however, the break tempera-
F I G U R E 1 4   Temperature dependence and estimated break ture more closely correlates with the bulk crystallization and
temperature of Flux 4, 5, and 6 according to the Arrhenius type not the initial crystallization temperature of the flux often ob-
relationship for viscous flow served in situ using high‐temperature confocal laser scanning
microscopy.9 Thus, the trend of the break temperature typi-
In the open symbols, the dynamic viscosity of the fully liquid cally follows the bulk crystallization temperature of the fluxes.
phase and the volumetric fraction of the precipitated crystal- A higher CaF2 lowered the activation energy for the vis-
line phases can be calculated using viscosity calculation and cous flow for Flux 1, 2, and 3. Flux 2 and 3 had the higher
equilibrium calculation in FactSage®. The linear regression of viscosity than Flux 1 and they exhibited a large increase in

T A B L E 2   Activation energy (kJ/mol)


Activation energy
for viscous flow of various silicate base
Flux (kJ/mol) Remark Reference
mold fluxes
1 105.1 (±3.5) High CaF2  
2 122.5 (±1.5)    
3 151.1 (±0.9) CaF2‐free  
4 122.0 (±3.0) C/S = 0.4  
5 149.7 (±2.9) C/S = 0.8  
6 132.5 (±7.0) C/S = 1.0  
7 151.8 (±3.8) B2O3‐free  
8 121.2 (±10.5) High B2O3  
CaO‐SiO2‐4.2%‐4.5% Al2O3‐Na2O‐(B2O3)‐Li2O‐(F)
7.4Na2O‐1.0Li2O‐7.3F, 193.6 (±12.0) B2O3‐free Zhou L et
C/S = 1.25 al1
10.9Na2O‐10.0B2O3‐2.0Li2O, 152.7 (±2.4) F‐free  
C/S = 1.25
11.9Na2O‐10.0B2O3‐2.9Li2O, 160.7 (±2.4) F‐free  
C/S = 1.25
9.9Na2O‐8.0B2O3‐2.0Li2O, 154.9 (±3.5) F‐free  
C/S = 1.15
7.9Na2O‐6.0B2O3‐2.0Li2O, 228.6 (±4.6) F‐free  
C/S = 1.15
CaO‐SiO2‐8.0% Na2O‐B2O3‐2.0% Li2O, C/S = 1.15
3.9B2O3 382.3 (±34.4) F‐free Lu B et al5
5.8B2O3 183.0 (±8.9) F‐free  
7.9B2O3 173.8 (±16.3) F‐free  
The bold texts indicate which compositions are changing below.
|
12      KIM and SOHN

T A B L E 3   Break and crystallization


Break temperature
temperatures (K) of various silicate base
Flux (K) Remark Reference
mold fluxes
1 1223 High CaF2  
2 No crystallization    
3 No crystallization CaF2‐free  
4 1323 C/S = 0.4  
5 1248 C/S = 0.8  
6 1383 C/S = 1.0  
7 1373 B2O3‐free  
8 1248 High B2O3  
20% CaO‐40% SiO2‐10% Al2O3‐10% MgO‐20% Na2O‐1% Li2O‐B2O3
1.5, 4, 8B2O3 1405, 1301, 1123 F‐free Fox AB et
al17
CaO‐SiO2‐4.2%‐4.5% Al2O3‐Na2O‐(B2O3)‐Li2O‐(F)
7.4Na2O‐1.0Li2O‐7.3F, 1531 B2O3‐free Zhou L et
C/S = 1.25 al1
10.9Na2O‐10.0B2O3‐2.0Li2O, 1480 F‐free  
C/S = 1.25
11.9Na2O‐10.0B2O3‐2.9Li2O, 1462 F‐free  
C/S = 1.25
9.9Na2O‐8.0B2O3‐2.0Li2O, 1495 F‐free  
C/S = 1.15
7.9Na2O‐6.0B2O3‐2.0Li2O, 1507 F‐free  
C/S = 1.15
CaO‐SiO2‐8.0% Na2O‐B2O3‐2.0% Li2O, C/S = 1.15
3.9, 5.8, 7.9B2O3 1464, 1505, 1456 F‐free Lu B et al5
The bold texts indicate which compositions are changing below.

viscosity with decreasing temperature, as shown in Figure viscosity, which also decreased the activation energy be-
4. Flux 1 had smaller network structures due to higher CaF2 cause the smaller network structures had lower resistance
content, which led to a slight increase in the viscosity with for viscous flow. However, the break temperature increased.
decreasing temperature. In the CaO‐SiO2‐MgO ternary phase diagrams containing
As the C/S ratio increased from 0.4 (Flux 4) to 0.8 (Flux 2 wt% Al2O3, 16 wt% Na2O, and 14 wt% CaF2 which cal-
5), the activation energy increased, but the viscosity de- culated using FactSage® and presented in Figure 15. In
creased. In the FactSage® calculation, Flux 4 and 5 were the phase diagram, higher C/S ratio increased a liquidus
expected to form cuspidine (Ca4Si2O7F2) as the precipi- temperature of the system at above a region having lower
tated crystalline phase during cooling. In addition, Flux liquidus temperature. It is speculated that the liquidus tem-
5 with greater CaO concentrations had a higher thermo- perature increased by the precipitated crystals from cus-
dynamic driving force for forming cuspidine (Ca4Si2O7F2) pidine (Ca4Si2O7F2) to merwinite (Ca3MgSi2O8) having
than Flux 4 according to Park et al.9 During cooling, Flux higher crystallization temperature with higher C/S ratio.
5 had a higher activation energy for viscous flow by higher Therefore, the break temperature increases with increasing
crystallization tendency than Flux 4 because of increased C/S ratio.
ionic mass transfer and kinetics associated with the lower According to Lee et al,60 the activation energy for vis-
viscosity. It was also observed that the viscosity in solid‐ cous flow was dominated by the equilibrium of the silicate
liquid region below the break temperature was signifi- anion structural units as shown in Figure 16 which also rep-
cantly increased by the precipitated crystals, as shown in resented the effect of the precipitated crystalline phases on
Figure 5B. determining the activation energy for viscous flow in this
With a higher C/S ratio from 0.8 (Flux 5) to 1.0 (Flux study using the FactsSage® calculation as shown in Figure
6), the activation energy decreased. The depolymerized 15. The C/S ratio led to change the activation energy for vis-
network structures by higher C/S ratio led to the lower cous flow by the precipitated crystalline phase which altered
KIM and SOHN      13
|

F I G U R E 1 5   CaO‐SiO2‐MgO ternary phase diagram containing 2 wt% Al2O3, 16 wt% Na2O, and 14 wt% CaF2 calculated by FactSage®

crystalline phase changed from cuspidine (Ca4Si2O7F2) to


merwinite (Ca3MgSi2O8), which decreased the activation
energy for viscous flow by higher CaO contents. Therefore,
the precipitate crystalline phase changed from cuspidine
(Ca4Si2O7F2) to merwinite (Ca3MgSi2O8) with the higher
C/S ratio in this system, which also altered the activation en-
ergy for viscous flow.
From the results of Flux 5 and 7, no appreciable effect
with B2O3 up to 4.4 wt% on the activation energy could be
observed, but B2O3 did lower the break temperature, which
lowered the crystallization temperature and extended the
solid‐liquid co‐existing region. As the B2O3 content in-
creased up to 8.8 wt%, the activation energy decreased due
to the depolymerized network structures by the O− in bo-
rate structural units. The break temperature also decreased
F I G U R E 1 6   Activation energy (kJ/mol) for viscous flow and because the increased B2O3 content (with its low melting
the C/S ratio temperature) inhibited the crystallization tendency.2,16,61,62

with the C/S ratio according to the phase diagram as shown


in Figure 15. When the precipitated crystalline phase was not 4  |  CONCLUSIONS
changed, the activation energy for viscous flow increased by
the higher thermodynamic driving force for forming the cus- The influence of CaF2, CaO/SiO2 mass ratio, and B2O3 on
pidine (Ca4Si2O7F2) with higher CaO contents. With higher the viscous behaviors, structures, break temperatures, and
C/S ratio, the precipitated crystalline phase the precipitate activation energy for the viscous flow was investigated and
|
14      KIM and SOHN

correlated using a rheometer, FTIR and Raman spectros- TiO2‐Al2O3‐MgO‐Li2O fluorine‐free mould fluxes with different
copy, and FactSage® calculations. The limited modifications CaO/SiO2 ratios. ISIJ Int. 2016;56:574–83.
of the viscosities and structures of the fluxes were achieved 4. Zhou L, Wang W, Lu B, Wen G, Yang J. Effect of basicity and
B2O3 on viscosity, melting and crystallization behaviors of low flu-
with 14 wt% CaF2, CaO/SiO2 ratio of 0.8, and 4.4 wt% B2O3,
orine mold fluxes for casting medium carbon steels. Met Mater Int.
respectively. According to the structural analyses, the net-
2015;21:126–33.
work structures were highly depolymerized at above 14 wt% 5. Lu B, Wang W, Li J, Zhao H, Huang D. Effects of basicity and
CaF2 and 4.4 wt% B2O3, otherwise, the structures occasion- B2O3 on the crystallization and heat transfer behaviors of low flu-
ally depolymerized above a CaO/SiO2 ratio of 0.8. Unlike orine mold flux for casting medium carbon steels. Metall Mater
other fluidizers such as CaO, Na2O, Li2O, and CaF2, B2O3 Trans B. 2013;44:365–77.
transformed from [BO3] to [BO4] by alkali and alkali earth 6. Wei J, Wang W, Zhou L, Huang D, Zhao H, Ma F. Effect of Na2O
cations and free oxygen ions (O2−), which randomly con- and B2O3 on the crystallization behavior of low fluorine mold
fluxes for casting medium carbon steels. Metall Mater Trans B.
nected to [SiO4]‐tetrahedral structures and produced BO2O−
2014;45:643–52.
structural units containing non‐bridging oxygen (O−) and
7. Wang Z, Shu QF, Chou K. Estimation of liquidus temperature for
[BO4]‐tetrahedral units. The process corrupts the uniform- B2O3‐and TiO2‐containing fluoride free mould fluxes from activa-
ity of structures and influences the viscous behaviors, break tion energy for viscous flow and DTA measurements. Can Metall
temperature, and activation energy for the viscous flow. The Q. 2013;52:405–12.
break temperature and the activation energy for the viscous 8. Mills KC, Fox AB, Li Z, Thackray RP. Performance and properties
flow were related to not only the network structures but also of mould fluxes. Ironmak Steelmak. 2005;32:26–34.
the primary crystalline phases, as elucidated by the Einstein‐ 9. Park J‐Y, Kim GH, Kim JB, Park S, Sohn I. Thermo‐physical prop-
erties of B2O3‐containing mold flux for high carbon steels in thin
Roscoe equation and FactSage® calculations. Generally,
slab continuous casters: structure, viscosity, crystallization, and
CaF2 and B2O3 additions decreased the viscosity, the break wettability. Metall Mater Trans B. 2016;47:2582–94.
temperatures, and the activation energy for the viscous flow 10. Zhou L, Wang W, Ma F, Li J, Wei J, Matsuura H, et al. A kinetic
if the crystallization phases were the same in the solid‐liquid study of the effect of basicity on the mold fluxes crystallization.
co‐existing region. With a higher CaO/SiO2 ratio, the vis- Metall Mater Trans B. 2012;43:354–62.
cosity decreased, and the break temperature and the activa- 11. Hanao M. Influence of basicity of mold flux on its crystallization
tion energy influenced the crystallization behavior. Finally, rate. ISIJ Int. 2013;53:648–54.
12. Yang J, Zhang J, Sasaki Y, Ostrovski O, Zhang C, Cai D, et al.
the addition of B2O3 extended the solid‐liquid co‐existing
Crystallization behavior and heat transfer of fluorine‐free mold
region by inhibiting the precipitation of the primary crystal
fluxes with different Na2O concentration. Metall Mater Trans B.
phases. 2016;47:2447–58.
13. Zhang L, Wang W, Xie S, Zhang K, Sohn I. Effect of basicity and
B2O3 on the viscosity and structure of fluorine‐free mold flux. J
ACKNOWLEDGMENTS Non Cryst Solids. 2017;460:113–8.
This research was supported by the Graduate School of 14. Yan W, Chen W, Yang Y, Lippold C, McLean A. Evaluation of
B2O3 as replacement for CaF2 in CaO‐Al2O3 based mould flux.
YONSEI University Research Scholarship Grants in 2017
Ironmak Steelmak. 2016;43:316–23.
and National Research Foundation of Korea, Project No.
15. Shi C‐B, Seo M‐D, Cho J‐W, Kim S‐H. Crystallization character-
NRF‐2018R1A2B2006609. istics of CaO‐Al2O3‐based mold flux and their effects on in‐mold
performance during high‐aluminum TRIP steels continuous cast-
ing. Metall Mater Trans B. 2014;45:1081–97.
ORCID
16. Klug JL, Pereira MMSM, Nohara EL, Freitas SL, Ferreira GT,
Gi Hyun Kim  https://orcid.org/0000-0002-7101-9353 Jung D. F‐free mould powders for low carbon steel slab casting‐
technological parameters and industrial trials. Ironmak Steelmak.
Il Sohn  https://orcid.org/0000-0002-8571-9182 2016;43:559–63.
17. Fox AB, Mills KC, Lever D, Bezerra C, Valadares C, Unamuno I,
et al. Development of fluoride‐free fluxes for billet casting. ISIJ
R E F E R E NC E S
Int. 2005;45:1051–8.
1. Zhou L, Wang W, Zhou K. Viscosity and crystallization behav- 18. Kim GH, Sohn I. A study of the viscous properties with NaF ad-
ior of F‐free mold flux for casting medium carbon steels. ISIJ Int. ditions in the CaO‐SiO2‐12 mass pct Na2O Based Slags. Metall
2015;55:1916–24. Mater Trans B. 2011;42:1218–23.
2. Zhou L, Wang W. The development of CaO‐SiO2‐B2O3‐based flu- 19. Kim H, Sohn I. Effect of CaF2 and Li2O additives on the viscosity
orine‐free mold flux for a continuous casting process. Metall Mater of CaO‐SiO2‐Na2O slags. ISIJ Int. 2011;51:1–8.
Trans E. 2016;3:139–44. 20. Park HS, Kim H, Sohn I. Influence of CaF2 and Li2O on the vis-
3. Yang J, Zhang J, Sasaki Y, Ostrovski O, Zhang C, Cai D, et al. In‐ cous behavior of calcium silicate melts containing 12 wt pct Na2O.
situ study of crystallisation behaviour of CaO‐SiO2‐Na2O‐B2O3‐ Metall Mater Trans B. 2011;42:324–30.
KIM and SOHN
|
     15

21. Sakamaki T, Yagi T, Susa M. Form of fluorine in Na2O‐NaF‐SiO2 41. Migas P. Analysis of the rheological behaviour of selected semi‐
slags determined by infrared spectroscopy. Ironmak Steelmak. solid slag systems in blast furnace flow conditions. Arch Metall
2003;30:396–8. Mater. 2015;60:85–93.
22. Hayashi M, Nabeshima N, Fukuyama H, Nagata K. Effect of fluo- 42. Wu L, Ek M, Song M, Sichen D. The effect of solid particles on
rine on silicate network for CaO‐CaF2‐SiO2 and CaO‐CaF2‐SiO2‐ liquid viscosity. Steel Res Int. 2011;82:388–97.
FeOx glasses. ISIJ Int. 2002;42:352–8. 43. Zhen YL, Zhang GH, Chou KC. Viscosity of CaO‐MgO‐Al2O3‐
23. Takeda O, Ohnishi T, Sato Y. Viscosity measurement of SiO2‐ SiO2‐TiO2 melts containing TiC particles. Metall Mater Trans B.
Na2O melts with addition of NaF. ISIJ Int. 2014;54:2045–9. 2014;46:155–61.
24. Sasaki Y, Iguchi M, Hino M. The role of Ca and Na ions in the 44. Waseda Y, Toguri JM. The structure and properties of oxide melts:
effect of F ion on silicate polymerization in molten silicate system. application of basic science to metallurgical processing. Singapore:
ISIJ Int. 2007;47:638–42. World Scientific; 1998.
25. Sasaki Y, Urata H, Ishii K. Structural analysis of molten Na2O‐ 45. Shin S‐H, Cho J‐W, Kim S‐H. Shear thinning behavior of cal-
NaF‐SiO2 system by raman spectroscopy and molecular dynamics cium silicate‐based mold fluxes at 1623 K. J Am Ceram Soc.
simulation. ISIJ Int. 2003;43:797–802. 2014;97:3263–9.
26. Kim Y, Morita K. Thermal conductivity of molten Li2O‐B2O3 and 46. Wu T, Yuan F, Zhang Y. Viscosity measurements of CaO‐SiO2‐
K2O‐B2O3 systems. J Am Ceram Soc. 2015;98:3996–4002. CrO Slag. ISIJ Int. 2018;58:367–9.
27. Park S, Sohn I. Effect of Na2O on the high‐temperature thermal 47. Li Q, Yang S, Zhang Y, An Z, Guo Z. Effects of MgO, Na2O, and
conductivity and structure of Na2O‐B2O3 melts. J Am Ceram Soc. B2O3 on the viscosity and structure of Cr2O3‐bearing CaO‐SiO2‐
2016;99:612–8. Al2O3 slags. ISIJ Int. 2017;57:689–96.
28. Kim Y, Morita K. Relationship between molten oxide structure 48. Wang Z, Sohn I. Effect of substituting CaO with BaO on the vis-
and thermal conductivity in the CaO‐SiO2‐B2O3 system. ISIJ Int. cosity and structure of CaO‐BaO‐SiO2‐MgO‐Al2O3 slags. J Am
2014;54:2077–83. Ceram Soc. 2018;101:4285–96.
29. Li Z, Sun Y, Liu L, Zhang Z. Modification of the structure of Ti‐ 49. Jin Z, Yang H, Lv J, Tong L, Chen G, Zhang Q. Effect of ZnO on
bearing mold flux by the simultaneous addition of B2O3 and Na2O. viscosity and structure of CaO‐SiO2‐ZnO‐FeO‐Al2O3 slags. JOM.
Metall Mater Trans E. 2016;3:28–36. 2017;70:1430–6.
30. Kim G‐H, Sohn I. Effect of Al2O3 on the viscosity and structure 50. Wu T, Zhang Y, Yuan F, An Z. Effects of the Cr2O3 content on the
of calcium silicate‐based melts containing Na2O and CaF2. J Non viscosity of CaO‐SiO2‐10 pct Al2O3‐Cr2O3 quaternary slag. Metall
Cryst Solids. 2012;358:1530–7. Mater Trans B. 2018;49:1719–31.
31. Kim GH, Sohn I. Influence of Li2O on the Viscous behavior of 51. Diao J, Gu P, Liu D‐M, Jiang L, Wang C, Xie B. Viscosity and
CaO‐Al2O3‐12 mass% Na2O‐12 mass% CaF2 based slags. ISIJ Int. structure of CaO‐SiO2‐P2O5‐FetO system with varying P2O5 and
2012;52:68–73. FeO Content. JOM. 2017;69:1745–50.
32. Kim GH, Sohn I. Role of B2O3 on the viscosity and structure 52. Padmaja G, Kistaiah P. Infrared and Raman spectroscopic studies
in the CaO‐Al2O3‐Na2O‐based system. Metall Mater Trans B. on alkali borate glasses: evidence of mixed alkali effect. J Phys
2014;45:86–95. Chem A. 2009;113:2397–404.
33. Kim G‐H, Kim C‐S, Sohn I. Viscous behavior of alumina rich 53. Rinke MT, Eckert H. The mixed network former effect in glasses:
calcium‐silicate based mold fluxes and its correlation to the melt solid state NMR and XPS structural studies of the glass system
structure. ISIJ Int. 2013;53:170–6. (Na2O)x(BPO4)1–x. Phys Chem Chem Phys. 2011;13:6552–65.
34. Wright AC. Borate structures: crystalline and vitreous. Phys Chem 54. Sen S, Xu Z, Stebbins J. Temperature dependent structural
Glas. 2010;51:1–39. changes in borate, borosilicate and boroaluminate liquids: high‐
35. Sun Y, Zhang Z. Structural roles of boron and silicon in the CaO‐ resolution 11B, 29Si and 27Al NMR studies. J Non Cryst Solids.
SiO2‐B2O3 glasses using FTIR, Raman, and NMR spectroscopy. 1998;226:29–40.
Metall Mater Trans B. 2015;46:1549–54. 55. Quintas A, Caurant D, Majérus O, Charpentier T, Dussossoy J‐L.
36. Kim Y, Morita K. Thermal conductivity of molten B2O3, B2O3‐ Effect of compositional variations on charge compensation of AlO4
SiO2, Na2O‐B2O3, and Na2O‐SiO2 Systems. J Am Ceram Soc. and BO4 entities and on crystallization tendency of a rare‐earth‐
2015;1595:1588–95. rich aluminoborosilicate glass. Mater Res Bull. 2009;44:1895–8.
37. Kline J, Tangstad M, Tranell G. A Raman spectroscopic study of the 56. Umesaki N, Takahashi M, Tatsumisago M, Minami T. Raman
structural modifications associated with the addition of calcium oxide spectroscopic study of alkali silicate glasses and melts. J Non Cryst
and boron oxide to silica. Metall Mater Trans B. 2014;46:62–73. Solids. 1996;205–207:225–30.
38. Park JH. The effect of boron oxide on the crystallization behav- 57. Kim Y, Yanaba Y, Morita K. The effect of borate and silicate struc-
ior of MgAl2O4 spinel phase during the cooling of the CaO‐ ture on thermal conductivity in the molten Na2O‐B2O3‐SiO2 sys-
SiO2‐10 mass% MgO‐30 mass% Al2O3 systems. Met Mater Int. tem. J Non Cryst Solids. 2015;415:1–8.
2010;16:987–92. 58. Züchner L, Chan JCC, Müller‐Warmuth W, Eckert H. Short‐
39. Wright S, Zhang L, Sun S, Jahanshahi S. Viscosities of calcium range order and site connectivities in sodium aluminoborate
ferrite slags and calcium alumino‐silicate slags containing spinel glasses: I. Quantification of local environments by high‐res-
particles. J Non Cryst Solids. 2001;282:15–23. olution 11B, 23Na, and 27Al solid‐state NMR. J Phys Chem B.
40. Wright S, Zhang L, Sun S, Jahanshahi S. Viscosity of a CaO‐MgO‐ 1998;102:4495–506.
Al2O3‐SiO2 melt containing spinel particles at 1646K. Metall 59. Bertmer M, Züchner L, Chan JCC, Eckert H. Short and medium
Mater Trans B. 2000;31:97–104. range order in sodium aluminoborate glasses. 2. Site connectivities
|
16      KIM and SOHN

and cation distributions studied by rotational echo double reso-


nance NMR spectroscopy. J Phys Chem B. 2000;104:6541–53. How to cite this article: Kim GH, Sohn I. Effect of
60. Lee S, Min DJ. Anionic effect of chloride, fluoride, and sulfide CaF2, B2O3 and the CaO/SiO2 mass ratio on the
ions on the viscosity of slag melt. J Am Ceram Soc. 2017;100: viscosity and structure of B2O3‐containing calcium‐
2543–52. silicate‐based melts. J Am Ceram Soc. 2019;00:1–16.
61. Wang L, Cui Y, Yang J, Zhang C, Cai D, Zhang J, et al. Melting
https​://doi.org/10.1111/jace.16526​
properties and viscosity of SiO2‐CaO‐Al2O3‐B2O3 System. Steel
Res Int. 2015;86:670–7.
62. Sun Y, Liao J, Zheng K, Wang X, Zhang Z. Effect of B2O3 on the
structure and viscous behavior of Ti‐bearing blast furnace slags.
JOM. 2014;66:2168–75.

You might also like