Effect of Al O / (B O + Na O) Ratio On Cao-Al O - Based Mold Fluxes: Melting Property, Viscosity, Heat Transfer, and Structure

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Effect of Al2O3/(B2O3 + Na2O) Ratio on CaO-Al2O3-

Based Mold Fluxes: Melting Property, Viscosity, Heat


Transfer, and Structure
JIAN YANG, QI WANG, JIANQIANG ZHANG, OLEG OSTROVSKI, CHEN ZHANG,
and DEXIANG CAI

CaO-Al2O3-based mold fluxes, which are under development for the continuous casting of
high-Al steel, contain fluxing compounds, such as Na2O and B2O3. The reaction between [Al]
and the fluxing agents in mold fluxes leads to an increase in Al2O3 and a decrease in B2O3 and
Na2O concentrations, changing the properties of mold fluxes. The effect of the Al2O3/
(B2O3 + Na2O) ratio on the melting properties, viscosity, heat transfer, and structure of the
CaO-Al2O3-based mold fluxes is presented in this work. The increase of the Al2O3/
(B2O3 + Na2O) ratio in the fluxes raised the melting temperature and high-temperature
viscosity of mold fluxes but decreased the heat transfer rate across the flux disks. It also
enhanced the degree of polymerization by promoting the formation of 3-D aluminate structure,
which accounted for the change of the viscous behavior.

https://doi.org/10.1007/s11663-019-01711-z
 The Minerals, Metals & Materials Society and ASM International 2019

I. INTRODUCTION et al.[6] suggested that CaO-Al2O3-based mold fluxes


with 9 mass pct Na2O and 16 mass pct B2O3 in CaO-
HIGH-AL steel attracts great attentions in the Al2O3-based mold fluxes exhibited the best performance
automobile industry due to its superior mechanical in the casting of high-Al steel. Blazek et al.[7] also
properties and light weight.[1–3] However, when the reported that a high concentration of B2O3 up to
conventional CaO-SiO2-based mold fluxes are used in 15 mass pct in CaO-Al2O3-based mold fluxes is neces-
the continuous casting of high-Al steel, [Al] from the sary to maintain an appropriate portion of glassy flux
liquid steel reacts with SiO2 from the mold fluxes, layer during casting practice. Although these fluxing
inevitably changing the flux properties and affecting the agents are indispensable in CaO-Al2O3-based mold
castability of high-Al steel.[4–6] To mitigate the chemical fluxes, the reaction between fluxing agents and high-Al
reaction between [Al] and SiO2, CaO-Al2O3-based mold steel can occur at high temperature through the follow-
fluxes with a low concentration of SiO2 were pro- ing reactions[4,12]:
posed.[6–9] The reactivity of CaO-Al2O3-based mold
fluxes is expected to be lower than that of the 2½AlþðB2 O3 Þ ! 2½B + (Al2 O3 Þ
½1
CaO-SiO2-based oxides due to the low activity of DG0 ¼ 252:3þ0:029  T (kJ)
SiO2.[10]
CaO-Al2O3 oxides have high liquidus temperature
and viscosity.[11] The addition of fluxing agents, e.g.,
B2O3 and Na2O, is required to ameliorate the melting 2½Alþ3ðNa2 OÞ ! 6½Na + (Al2 O3 Þ
½2
properties and viscosity of CaO-Al2O3-based mold DG0 ¼ 560:5 þ 0:045  T (kJ)
fluxes. Although Li2O is also an effective fluxing agent,
its high cost restricts its heavy use in the industry. Cho The Gibbs energy of Reactions 1 and 2 at 1773 K
(1500 C) was calculated using FactSage 7.2, which
proves that the reactions are thermodynamically feasible
to occur at the casting temperature. Wu et al.[12]
JIAN YANG, QI WANG, JIANQIANG ZHANG, and OLEG observed that Na2O and B2O3 from CaO-Al2O3-based
OSTROVSKI are with the School of Materials Science and mold fluxes were both reduced significantly during the
Engineering, University of New South Wales, Sydney, NSW, 2052, reaction with [Al] at high temperature. Notable reduc-
Australia. Contact e-mail: j.q.zhang@unsw.edu.au CHEN ZHANG tion of B2O3 during the pilot casting of high-Al steel was
and DEXIANG CAI are with the Baosteel Group Corporation
Research Institute, Shanghai, 201900, P.R. China. also reported by some researchers.[6–9] The reaction
Manuscript submitted 21 May, 2019. causes the accumulation of Al2O3 and reduction of
Article published online October 17, 2019.

2794—VOLUME 50B, DECEMBER 2019 METALLURGICAL AND MATERIALS TRANSACTIONS B


Na2O and B2O3 in the course of casting, deteriorating Diffraction System, Netherlands). The XRD patterns
the in-mold performances of CaO-Al2O3-based mold indicate that the as-quenched mold fluxes were amor-
fluxes. phous, as shown in Figure 1.
A number of recent investigations have been carried
out to demonstrate the properties of CaO-Al2O3-based
mold fluxes with different CaO/Al2O3 ratios,[13] B. Experimental Procedure
B2O3,[14,15] CaF2,[16] and Li2O concentrations,[17,18] in
Melting properties of mold fluxes were investigated
the non-dynamic conditions where no flux-steel reaction
using a hot-stage microscopy. The as-quenched flux
is considered. However, the properties of CaO-Al2O3-
powders were sieved by 0.074 mm meshes and well
based mold fluxes, e.g., viscosity, melting property, and
mixed with a small amount of ethanol as a binder. Then,
heat transfer, are susceptible to the dynamic change of
the powders were pressed to pellets (F3 9 3 mm).
flux composition in contact with high-Al steel.[7,8]
Pellets were continuously heated in a horizontal tube
Therefore, it is necessary to understand the impacts of
furnace at a heating rate of 15 K/min; their appearance
the reduction of the fluxing agents on the properties of
change was monitored by a video camera. To charac-
CaO-Al2O3-based mold fluxes, which is essential for the
terize the melting properties of mold fluxes, the soften-
continuous casting of high-Al steel. In this study, the
ing, hemispherical, and fluidity temperatures were
effect of the increase of Al2O3/(B2O3 + Na2O) ratio on
defined as the temperatures at which the height of flux
viscosity, solidification behavior, heat transfer, and
pellet dropped to 75, 50, and 25 pct of its original
structure of CaO-Al2O3-based mold fluxes was eluci-
height, respectively. The details of melting property
dated by studying the designed CaO-Al2O3-based mold
measurement were described elsewhere.[19] Melting tem-
fluxes with varied Al2O3/(B2O3 + Na2O) ratios. This
peratures were measured in accordance with YB/T
work provides a further understanding of the flux
186-2014 Standard (‘‘Method of the test for melting
property variation during the high-Al steel casting as a
temperature of continuous casting mold powder’’). The
result of the flux composition change caused by the
melting properties of fluxes were represented by the
flux-steel reaction.
average values of these characteristic temperatures,
measurement of which was repeated four times for each
sample.
II. EXPERIMENTAL METHODS Viscosity of mold fluxes was investigated using a
rotation viscometer (model ZC-1600, China), whose
A. Sample Preparation
schematics was described elsewhere.[19] Approximately
Reagent-grade CaCO3, SiO2, Na2CO3, Al2O3, B2O3, 140 g pre-melted flux was melted at 1673 K (1400 C) in
MgO, CaF2, and Li2CO3 were pre-mixed and melted at a graphite crucible under the protection of N2. After
1673 K (1400 C) in a graphite crucible for 1200 sec- holding the molten flux at 1673 K (1400 C) for
onds, during which carbonates were decomposed to 1200 seconds, a Mo bob was slowly submerged into
oxides. After the molten flux was homogenized, it was the homogenized melt and rotated at a rate of 12 rpm
quenched in distilled water. The as-quenched flux was for the measurement of viscosity. After the flux viscosity
then ground into powders using a ring mill. The was steadily recorded at 1673 K (1400 C), the viscosity
chemical compositions of as-quenched fluxes were was measured in the process of flux continuous cooling
determined using X-ray fluoroscopy (XRF, PANalytical with a cooling rate of 5 K/min until the Mo bob stopped
AXIOS-Advanced WDXRF spectrometer, Netherland), rotation. The viscosity was measured in the continuous
except B2O3 and Li2O whose contents were analyzed cooling process in accordance with YB/T 185-2017
using inductively coupled plasma-optical emission spec- Standard (‘‘Test method for viscosity of continuous
troscopy (ICP-OES, Thermo Scientific IRIS Intrepid II, casting mold powder’’), with exception of the cooling
MA) and atomic absorption spectroscopy (AAS, rate in the present study which was slightly lower than
Thermo Scientific Solaar S4 AA spectrometer, MA), the suggested value (6 K/min) to secure more stable ex-
respectively. The measured compositions of mold fluxes perimental conditions.
are listed in Table I. Concentrations of Na2O and B2O3 The structure of as-quenched mold flux from 1673 K
decreased from 9.2 and 14.5 mass pct in Flux 1 to 4.0 (1400 C) was measured using Raman spectroscopy
and 6.6 mass pct in Flux 4, respectively, while the Al2O3 (Renishaw inVia Raman Microscope, UK). The pulver-
concentration increased from 18.6 to 32.5 mass pct ized mold flux was illuminated by Ar-ion laser beam.
accordingly. The free carbon concentration in mold The excitation wavelength of Ar-ion laser was 514 nm
fluxes analyzed using an infrared carbon sulfur analyzer with a beam spot size of 1.5 lm. The measurement was
(CS600 Carbon/Sulfur Determinator, Leco Corpora- conducted in the Raman shift range from 400 to 1700
tion, MI) by following YB/T 190.6-2014 Standard cm1. The obtained Raman spectra were deconvoluted
(‘‘Continuous casting mold powder—Determination of using WiRE 4.4 software. The coordination status of Al
dissociation carbon content’’) was all below 0.1 was further examined using a nuclear magnetic reso-
mass pct. The X-ray diffraction (XRD) patterns of nance (NMR) spectrometer (300 MHz Bruker Avance
as-quenched mold fluxes were examined using an X-ray III NMR, Germany) to identify 27Al MAS-NMR at a
diffractometer (PANalytical X’pert Multipurpose X-ray resonance frequency of 78.1 MHz (7.0 T). The flux

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 50B, DECEMBER 2019—2795


Table I. Chemical Compositions of As-quenched Mold Fluxes (Mass Percent)

Sample CaO Al2O3 SiO2 B2O3 Na2O MgO Li2O F Al2O3/(B2O3 + Na2O) ratio
1 38.8 18.6 7.5 14.5 9.2 1.9 3.6 5.9 0.8
2 38.5 23.2 7.1 11.9 7.4 1.9 3.6 6.4 1.2
3 39.2 28.1 6.9 8.7 5.6 2.0 3.6 5.9 2.0
4 38.8 32.5 6.5 6.6 4.0 1.9 3.5 6.2 3.1

Fig. 2—Incident radiation intensity profile in IET measurement.

Fig. 1—XRD patterns of as-quenched mold fluxes.


 
1 X dT
q¼ k ½3
n i dx i
powders were contained in a ZrO2 rotor with a spinning
rate of 12 kHz and the recycling time of 1 second. The where q is the heat flux, n presents the total number of
flip angle in the NMR measurement was p/6 (rad). The thermocouple pairs, k is the thermal conductivity of
NMR spectra were well fitted using ‘‘DMFIT’’ pro- copper. The heat fluxes measured at an incident radia-
gram,[20] which simulates the quadrupolar line shapes tion of 1.6 MW/m2 were compared to reflect the heat
based on GIM model.[21] transfer ability of fluxes during continuous casting.
The heat transfer rate across the mold fluxes was The cross-section microstructure of the flux film after
measured using infrared emitter technique (IET, Central the IET experiment was examined using a scanning
South University, China). The details of IET system electron microscope (SEM, Hitachi S3400, Japan). A
were described elsewhere.[22–24] The flux disk was pre- part of the flux disk was pulverized, and the crystalline
pared using 13 g as-quenched flux powders which were phases in the flux disk were determined using XRD. The
melted at 1673 K (1400 C) and held for 1200 seconds scanning range in the XRD analysis was 2h = 10 to
before pouring into a copper cylinder (F 40 mm) and 80 deg with a scanning speed of 0.021 deg/s. The XRD
pressing to a disk. The pressed mold flux disk was patterns were analyzed using HighScore Plus 4.2.
immediately placed in a muffle furnace in which the
temperature gradually decreased from 1073 K to 298 K
(800 C to 25 C) with a slow cooling rate of 1 K/min to III. RESULTS
minimize the internal stress within the disk. The
fabricated flux disk was ground using a diamond A. Melting Property
grinding wheel, and then carefully polished using The softening, hemispherical, and fluidity tempera-
sandpapers from 300 to 1200 grits to the thickness of tures of fluxes are given in Figure 3. The hemispherical
4 ± 0.01 mm. In the heat flux measurement, the mold temperature of fluxes increased from 1199 ± 2.5 K to
flux disk was placed on the copper base of the IET 1268 ± 4.2 K, 1293 ± 7.7 K, and 1388 ± 7.1 K
system. The radiation was emitted in accordance with a (926 ± 2.5 C to 995 ± 4.2 C, 1020 ± 7.7 C, and
stepwise power profile presented in Figure 2. The 1115 ± 7.1 C) as Al2O3/(B2O3 + Na2O) ratio
incident thermal radiation was increased to 1.6 MW/ increased from 0.8 in Flux 1 to 3.1 in Flux 4. The
m2, which is close to the radiation released from the steel softening and fluidity temperatures of fluxes also
strand in the continuous casting. The temperatures increased accordingly. It indicates that the reduction
recorded by the embedded thermocouples were used for of fluxing agents (B2O3 and Na2O) increased all these
the heat flux calculation according to the Fourier’s characteristic temperatures of CaO-Al2O3-based mold
law[25]: fluxes.

2796—VOLUME 50B, DECEMBER 2019 METALLURGICAL AND MATERIALS TRANSACTIONS B


B. Viscosity g ¼ A expðEa =RTÞ ½4
The viscosity of Fluxes 1 through 4 during the cooling where g is the flux viscosity, A is a constant, Ea is acti-
process is demonstrated in Figure 4. The viscosities at vation energy for viscous flow, R is the ideal gas con-
different temperatures are also compared in Table II. It stant, T is the absolute temperature. Functions of ln g
should be noted that only the viscosity at 1673 K against 1/T are plotted in Figure 5. The break temper-
(1400 C) was measured at equilibrium. Viscosity mea- atures of Fluxes 1 and 2 were determined to be 1221
sured during the continuous cooling process with a K and 1253 K (948 C and 980 C), respectively. The
cooling rate of 5 K/min could deviate from those activation energies for viscous flow in Newtonian fluid
obtained at equilibrium. However, the measurement of are listed in Table II. The activation energies of Fluxes
viscosity of different fluxes under the same experimental 1 and 2 were close to one another but much lower
conditions provided sufficient data for the analysis of than the Ea values of Fluxes 3 and 4. The fitting for
the effect of the flux composition on the viscosity and Fluxes 3 and 4 in Figure 5 was confined within Newto-
solidification behavior. As shown in Figure 4, Fluxes 1 nian fluid range, as the experimental data gradually
and 2 exhibited a sharp increase in viscosity at apparent deviated from linearity at lower temperatures due to
break temperatures. In contrast, the viscosity of Fluxes the transition from Newtonian to non-Newtonian
3 and 4 steadily increased as temperature decreased fluid.
without the occurrence of apparent break point during
continuous cooling. Increasing the ratio of Al2O3/
(B2O3 + Na2O) changed the solidification characteris- C. Structure
tics of flux and increased the flux viscosity at given The deconvoluted Raman spectra of Fluxes 1 through
temperatures in most of the studied temperature range. 4 are shown in Figure 6. The assignments of the
The relationship between viscosity and temperature for deconvoluted Raman bands are listed in Table III. In
the Newtonian fluid follows Arrhenius equation: the low frequency region (< 800 cm1), the deconvo-
luted Raman bands were related to aluminate structures.
The peaks around 500 cm1 were assigned to Al-F
stretching vibration in AlF6[26–28]; the distinct bands
centered within 540 to 550 cm1 were assigned to
Al-O-Al linkage in AlO4 tetrahedra,[28–34] which is the
major bond that constitutes 3-D aluminate network.
Other minor peaks were related to the Al-O stretching
vibration in AlO6 units,[28,29] and Al-O stretching
vibration in AlO4 units with 1 or 2 non-bridging
oxygens (NBOs).[28–31] In the medium frequency range
(800 to 1100 cm1), the deconvoluted Raman peaks
were assigned to silicate structures, including Al-O-Si
linkage,[31–33] Si-O stretching vibration in SiO44 (Q0),
Si2O76 (Q1) and SiO32 (Q2).[31–38] In high frequency
range, the Raman bands were assigned to different
borate structures. The peaks around 1200 cm1 corre-
Fig. 3—Softening, hemispherical, and fluidity temperatures of Fluxes
1 through 4. sponded to the B-O stretching vibration in BO3 units in
2-D groups, such as pyroborate[39–41] and boroxol
rings[42,43]; while the peaks located from 1300 to 1500
cm1 were assigned to B-O stretching vibration in
BO3 units attached to large borate groups based on
either BO3 triangular units or BO4 tetrahedral
units.[40–45] The area fractions of these Raman bands
in Fluxes 1 to 4 are compared in Figure 7. With the
reduction of Na2O and B2O3 and the increase of Al2O3
content, the bands for Al-O-Al and Al-O-Si linkages
became more prominent; while the bands for both 2-D
and 3-D borate structures located in the high frequency
range (> 1100 cm1) were suppressed. The Si-O
stretching vibrations in Si2O76 (Q1) and SiO32 (Q2)
were also significantly depressed. It might come from the
interaction with aluminate network through the forma-
tion of Al-O-Si linkages, which consumed the NBOs in
both silicate and aluminate structures. The proportions
of AlO6 and AlF6 gradually became lower with the
increase of Al2O3 content in fluxes. Therefore, the
Fig. 4—Viscosity of Fluxes 1 through 4 during continuous cooling
accumulation of Al2O3 promoted the formation of
process. polymerized aluminate structures. Moreover, the

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 50B, DECEMBER 2019—2797


Table II. Values of Viscosity and Activation Energy of Mold Fluxes

Viscosity (Pa s)

Sample 1673 K (1400 C) 1623 K (1350 C) 1573 K (1300 C) 1523 K (1250 C) 1473 K (1200 C) Ea (kJ/mol)
1 0.05 0.06 0.10 0.13 0.18 123.5
2 0.08 0.12 0.17 0.21 0.26 125.5
3 0.07 0.11 0.19 0.27 0.38 158.3
4 0.11 0.20 0.29 0.42 0.66 167.1

The MAS-NMR spectra of 27Al in Fluxes 1 through 4


are illustrated in Figure 8(a). The total NMR intensity
of 27Al peaks was normalized to the maximum peak
height in each spectrum for comparison. With the
accumulation of Al2O3, the IVAl band became slightly
broader; it also notably shifted to the higher frequency
when the composition changed from Flux 1 to Flux 2,
but a further accumulation of Al2O3 did not significantly
affect the peak position of IVAl species. VIAl site became
more distinguishable from the background with a
slimmer appearance. Three species, including VIAl,
V
Al, and IVAl, were well resolved using GIM model,
which is exemplified in Figure 8(b).[46–48] The relative
intensities of these species are listed in Table IV. The
relative intensity of IVAl species increased; while those of
V
Al and VIAl species decreased from Fluxes 1 to 4. It
Fig. 5—Plots of ln g against 1/T.
seems that the depression and downfield shift (toward
high ppm direction) of VAl band made the appearance
of VIAl peak more distinct, but the relative intensity of
VI
Al peak became slightly lower as it became slimmer. It
indicates that the increase of Al2O3 and decrease of both
B2O3 and Na2O concentrations due to the flux-steel
reaction could gradually convert the five- and six-fold
coordinated aluminum into four-fold coordinated
aluminum.

D. Heat Transfer
The responding heat flux across the flux disk at an
incident radiation of 1.6 MW/m2 in IET experiments is
demonstrated in Figure 9. After approximately
2100 seconds, the heat flux of each sample was stabi-
lized. The average heat fluxes of Samples 1 to 4 were
765 ± 19.7, 664 ± 6.0, 561 ± 7.6, and 540 ± 6.5 kW/
m2, respectively. It can be seen that the accumulation of
Al2O3 and the decrease of Na2O and B2O3 concentra-
tions reduced the heat transfer rate across the flux.
The XRD patterns of the crystalline phases in the IET
disks are demonstrated in Figure 10. In Flux 1, the
dominant phase was Ca5(BO3)3F based on the major
peak intensity. Other major phases included Na2Al2SiO6
and Ca2Al2SiO7. Minor phases, such as LiAlO2 and
CaF2, were also detected in Flux 1. With the increasing
Fig. 6—Deconvoluted Raman spectra of Fluxes 1 through 4. Al2O3/(B2O3 + Na2O) ratio, the precipitation of
Ca5(BO3)3F was suppressed; while the major phase
connection between aluminate and silicate networks became Ca2Al2SiO7 in Fluxes 3 and 4. XRD patterns for
became more significant, increasing the degree of Fluxes 3 and 4 also included strong peaks for CaAl2O4,
polymerization of the flux structure. CaF2 and LiAlO2.

2798—VOLUME 50B, DECEMBER 2019 METALLURGICAL AND MATERIALS TRANSACTIONS B


Table III. Assignments of Deconvoluted Raman Bands

Flux

1 2 3 4 Assignments Ref
1
Raman Shifts (cm ) 491.3 487.6 498.0 501.6 Al-F stretching vibration in AlF6 26, 28
548.4 546.1 549.9 550.0 Al-O-Al linkage 28 to 34
605.0 592.1 600.9 600.8 Al-O stretching vibration in AlO6 units 28, 29
771.8 773.1 773.8 777.9 Al-O stretching vibration in AlO4 units with 1 or 2 NBOs 28 to 31
843.3 846.8 852.3 852.8 Al-O-Si linkage 31 to 33
870.2 873.6 881.8 880.6 Si-O stretching vibration in SiO44 31 to 38
920.4 920.5 920.9 919.8 Si-O stretching vibration in Si2O76
955.0 955.7 958.9 954.2 Si-O stretching vibration in SiO32
1211.0 1196.4 1188.2 1191.7 B-O stretching vibration in BO3 units 39 to 43

1335.3 1325.3 1335.7 1328.7 B-O stretching vibration in BO3 units attached to other 40 to 45
borate groups
1543.4 1588.8 — —

Fig. 7—Area fraction of Raman bands in Fluxes 1 to 4.

The cross-section surfaces of disks of Fluxes 1 and 2


clearly showed the glassy structure on the top and a
crystalline layer on the bottom; while Fluxes 3 and 4
exhibited a foggy morphology on the top and a
relatively dense crystalline layer on the bottom. The
microstructures of cross-sections of IET disks of Fluxes
1 and 3 examined using SEM are shown in Figure 11. In
Flux 1, the amorphous and crystalline phases had a clear
boundary. A densely sintered crystalline structure was
formed on the bottom mixed with some needle-shaped
crystals. An EDS analysis of the sintered phase revealed
that the Ca:F:O atomic ratio was close to 5:1:9 as listed
in Table V. This ratio matches well to Ca5(BO3)3F,
although boron is not listed in Table V as it cannot be
accurately determined using EDS. The size of needle-
shaped crystals was slightly larger adjacent to the liquid/
crystalline interface. In Flux 3, the foggy structure on
the high temperature side was found to be fine crystals Fig. 8—(a) Comparison of 27Al MAS-NMR spectra of Fluxes 1
with an average length of 40 lm dispersed in the liquid through 4; (b) deconvolution of Flux 3 based on GIM model.
phase (glass after solidification). It was a transition
region in which crystalline and liquid phases were decreased from the top to the bottom of the disk, the
mixed; while crystals were more heavily precipitated in crystalline phases became denser and partially sintered
the lower-temperature part of the disk. As temperature on the bottom. It can be suggested from the EDS

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 50B, DECEMBER 2019—2799


27
Table IV. Relative Fractions and Isotropic Chemical Shift of Al Species in Al MAS-NMR Spectra

Flux

1 2 3 4 Species
VI
Relative Fractions (Pct) 6.6 6.3 5.8 5.2 Al
V
18.4 14.6 13.3 12.3 Al
IV
75.0 79.1 80.9 82.5 Al
VI
Isotropic Chemical Shift (ppm) 10.40 10.16 10.61 10.29 Al
V
38.22 45.62 45.97 45.08 Al
IV
69.37 79.94 80.21 80.57 Al

EDS analysis. Based on the SEM images, the fractions


of crystalline phase after IET experiments, estimated by
ImageJ, were 37 and 39 pct for Fluxes 1 and 2,
respectively. Fluxes 3 and 4 had much higher fractions
of crystalline phases, 86 pct for Flux 3 and close to
100 pct for Flux 4, whose upper part was constituted by
fine crystals scattered in the amorphous matrix and
showed a foggy morphology by visual observation.

IV. DISCUSSION
A. Melting Properties of CaO-Al2O3-Based Mold Fluxes
B2O3 and Na2O are important fluxing agents in
CaO-Al2O3-based mold fluxes as they normally decrease
Fig. 9—Responding heat transfer across Fluxes 1 through 4 at an melting temperature.[14,49,50] Therefore, reducing the
incident radiation of 1.6 MW/m2 in IET experiments. concentrations of Na2O and B2O3 raised the melting
temperature of mold fluxes, reflected by the increased
softening, hemispherical, and fluidity temperatures.
When Na2O and B2O3 contents were reduced from 5.6
to 8.7 mass pct to 4.0 and 6.6 mass pct, respectively, the
hemispherical temperature was considerably increased
from 1293 K to 1388 K (1020 C to 1115 C). An
increase of melting temperature is detrimental to the
lubrication of steel strand during continuous casting.

B. Flux Viscosity and Structure


The increase of the Al2O3/(B2O3 + Na2O) ratio
increased the flux viscosity from 1273 K to 1573 K
(1000 C to 1300 C) (Figure 4 and Table II). Viscosity
is related to the flux structure. According to Raman and
27
Al MAS-NMR spectra, the degree of polymerization
was promoted with the increase of Al2O3 and the
decrease of Na2O and B2O3, i.e., the increase of Al2O3/
(B2O3 + Na2O) ratio. Based on the Raman spectra, the
increase of this ratio increased the portion of Al-O-Al
linkages in tetrahedral aluminates (AlO4), which was the
structural unit of 3-D aluminate network.[28–34] The
linkage of Al-O-Si also became more prominent. It was
related to the interaction between aluminate and silicate
networks through the formation of Al-O-Si linkages,
which could further add up the structural complexity.
Fig. 10—XRD patterns of crystalline phases in the flux disks after
The results were consistent with the further analysis on
IET experiments. aluminate structure using MAS-NMR. IVAl-O-IVAl
could build up 3-D aluminate network. Considering
analysis (Table V) and XRD pattern (Figure 10) that the existence of fluorine in the flux, F can provide the
the crystalline phase was Ca2Al2SiO7, although silicon lone pair electrons to form AlO3F or AlO2F2, which
content was lower than what is expected for this phase in were also believed to form 3-D structure in

2800—VOLUME 50B, DECEMBER 2019 METALLURGICAL AND MATERIALS TRANSACTIONS B


CaO-Al2O3-based mold fluxes[33]; while a higher coor-
dination number, e.g., VIAl, is considered to form simple
aluminate structures, such as AlO6 and AlF6. Neuville
et al.[47] suggested aluminum in the five-fold coordina-
tion was amphoteric, whose role is not specified in terms
of a network former or modifier; while Min et al.[51]
considered it as a network modifier. Overall, an increase
of IVAl and a decrease of VIAl led to a higher degree of
polymerization.
When the degree of polymerization of flux structure
became higher with the increase of Al2O3/(Na2O+
B2O3) ratio, the viscous behavior of fluxes was also
changed. Viscosity of Fluxes 1 and 2 was low at high
temperature until the occurrence of break point during
the cooling process. However, Fluxes 3 and 4, whose
viscosity continuously increased during cooling process,
did not exhibit apparent break points. In CaO-SiO2-
based mold fluxes, the absence of break point is
normally observed in acidic fluxes containing high
concentrations of SiO2 or B2O3.[19,52] These fluxes were
highly polymerized, which restrained the occurrence of
sharp viscosity change caused by the crystal precipita-
tion during continuous cooling process. In the present
study, the flux contained a high fraction of 3-D Fig. 11—Microstructure of IET disks of (a) Flux 1 and (b) Flux 3.
(The images marked as 1 to 4 are magnified views of Areas 1 to 4 in
aluminate structures based on the four-fold coordinated cross-section images; the crosses in the images were examined by the
aluminum also resulted in a high resistance to viscous point analysis using EDS).
flow. It retarded the mass transportation and cluster
agglomeration that were essential for the rapid nucle-
Table V. Composition of the Crystals Determined by EDS
ation. Therefore, a sharp change of viscosity of Fluxes 3 (Atomic Percent)
and 4 was not observed during cooling.
Fluxes 1 and 2 featured high B2O3 and Na2O Ca Al Si Na O F Mg
contents and low Al2O3 content. B2O3 was found to
lower the flux viscosity of CaO-Al2O3-based mold Flux 1-Point 29.72 7.94 2.07 4.56 48.97 5.82 0.93
fluxes, which is consistent with the literature.[43] Flux 3-Point 19.19 17.87 4.51 — 55.87 1.70 0.86
Although minor BO3 triangular units connect to larger
borate groups, which form 3-D borate structure, most
BO3-triangular units in the present study formed 2-D
structures, e.g., boroxol rings and pyroborates, which The microstructure of flux disks after IET experi-
required a lower activation energy for the viscous flow ments (Figure 11) indicates that, in Flux 1, the crys-
than 3-D aluminate structures. Na2O is a network talline layer was densely formed in the lower part of the
modifier at high temperature.[23,53] It reformed the flux disk; while the liquid layer (glassy layer after solidifica-
network into simpler structures through the breakage tion) was clearly observed in the upper part; in Flux 3,
of bridging oxygens in fluxes. Therefore, a high crystals were not only heavily precipitated in the lower
concentration of Na2O in Fluxes 1 and 2 lowered the part of the disk, but also dispersed in the liquid phase.
flux viscosity at high temperature. However, once the However, the structure of Fluxes 3 and 4 was more
temperature decreased below the crystallization tem- polymerized as shown in the structural analysis, which
perature, crystallization occurred rapidly as a low should lower the long-range mass transportation at high
degree of polymerization in the flux led to a high temperature and, therefore, the tendency for crystalliza-
diffusion rate. tion in the liquid state. This paradox originated from the
difference of their experimental methods. Unlike viscos-
ity measurement in which liquid phase solidified into the
glassy or crystalline phase during the continuous cooling
C. Crystallization and Heat Transfer of Mold Fluxes process, the crystallization in IET experiments occurred
The crystallization of mold fluxes is found to decrease through the devitrification of flux film, which was glassy
the heat flux across the flux film due to the scattering in the beginning of measurement, during heating and
effect at the grain boundaries, and the formation of isothermal stages. Such crystallization mechanism can
pores, cracks, and air gaps with high thermal resis- also be expected in the continuous casting,[54–56] apart
tances.[23,25] Measurement of the steady heat transfer from the direct crystallization from the liquid phase
rate at the isothermal stage in IET experiments (Fig- during infiltration. Zhou et al.[57] found that crystalliza-
ure 9) showed that the heat flux across the flux disks tion from the glassy state was faster than that from the
decreased with the increase of Al2O3/(B2O3 + Na2O) liquid state. In this way, it is likely that crystallization
ratio. occurred from the glassy flux disk of Fluxes 3 and 4

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 50B, DECEMBER 2019—2801


during the stepwise heating process, with sufficient ACKNOWLEDGMENTS
incubation time, in IET experiments, although an
apparent break temperature was not exhibited in vis- Financial supports from Baosteel-Australia Joint
cosity measurement. Crystal phases in Fluxes 3 and 4, Research and Development Centre (BAJC) (BA16006)
such as Ca2Al2SiO7 whose melting temperature was and Australian Research Council (ARC) Industrial
reported to be 1863 K (1590 C),[58] may not be com- Transformation Hub (IH140100035) are greatly
pletely melted at the experimental temperature but acknowledged. The authors also would like to
dispersed in the liquids, even though the incident acknowledge Dr Aditya Rawal of the NMR Facility
radiation intensity reached 1.6 MW/m2. The major within the Mark Wainwright Analytical Centre at the
crystalline phase in Fluxes 1 and 2 was Ca5(BO3)3F, University of New South Wales for NMR support.
whose melting point is 1665 K (1392 C),[59] which is
much lower than Ca2Al2SiO7. Therefore, the crystals
formed in Fluxes 1 and 2 during the heating process may REFERENCES
be melted at high temperature. Moreover, according to 1. O. Grässel, L. Krüger, G. Frommeyer, and L.W. Meyer: Int. J.
Figure 3, the characteristic melting temperatures Plast, 2000, vol. 16, pp. 1391–1409.
became higher from Fluxes 1 to 4, also implying that 2. D.R. Steinmetz, T. Jäpel, B. Wietbrock, P. Eisenlohr, I. Gutierrez-
the solid phases were more likely to sustain at high Urrutia, A. Saeed-Akbari, T. Hickel, F. Roters, and D. Raabe:
Acta Mater., 2013, vol. 61, pp. 494–510.
temperature during the heating process with the increase 3. K. Sato, M. Ichinose, Y. Hirotsu, and Y. Inoue: ISIJ Int., 1989,
of Al2O3/(B2O3 + Na2O) ratio. Therefore, the crys- vol. 29, pp. 868–77.
talline phases in Fluxes 3 and 4 could exist in a wider 4. M.-S. Kim, S.-W. Lee, J.-W. Cho, M.-S. Park, H.-G. Lee, and
temperature range, i.e., a larger region in flux disk, than Y.-B. Kang: Metall. Mater. Trans. B, 2013, vol. 44B, pp. 299–
Fluxes 1 and 2 in IET experiments. 308.
5. Y.-B. Kang, M.-S. Kim, S.-W. Lee, J.-W. Cho, M.-S. Park, and
For the flux disk, as the total amount of crystalline H.-G. Lee: Metall. Mater. Trans. B, 2013, vol. 44B, pp. 309–16.
phase in the flux increased, it is expected that the heat 6. J.-W. Cho, K. Blazek, M. Frazee, H. Yin, J.H. Park, and S.-W.
flux decreased in line with the increasing total amount of Moon: ISIJ Int., 2013, vol. 53, pp. 62–70.
crystalline phases with the increase of Al2O3/(B2O3 + 7. K. Blazek, H. Yin, G. Skoczylas, M. McClymonds, and M.
Frazee: Iron Steel Technol., 2011, vol. 8, pp. 232–40.
Na2O) ratio. A low heat transfer rate across the flux film 8. Q. Liu, G. Wen, J. Li, X. Fu, P. Tang, and W. Li: Ironmak.
during continuous casting slows down the solidification Steelmak., 2014, vol. 41, pp. 292–97.
of steel shell, which may increase the likelihood of 9. J. Yang, J. Zhang, O. Ostrovski, C. Zhang, and D. Cai: La Metall.
breakout. Ital., 2019, vol. 1, pp. 12–19.
10. K. Morita, K. Kume, and N. Sano: Scand. J. Metall., 2002,
vol. 31, pp. 178–83.
11. B. Hallstedl: J. Am. Ceram. Soc., 1990, vol. 73, pp. 15–23.
12. T. Wu, S.P. He, Y.T. Guo, and Q. Wang: Characterization of
V. CONCLUSIONS minerals, metals, and materials 2014, TMS, San Diego, 2014,
pp. 265–70.
Major findings of the study of the effect of Al2O3/ 13. W. Yan, W. Chen, Y. Yang, C. Lippold, and A. McLean: Iron-
(B2O3 + Na2O) ratio in the CaO-Al2O3-based mold mak. Steelmak., 2015, vol. 42, pp. 698–704.
fluxes on the melting properties, viscous behavior, 14. J. Yang, J. Zhang, O. Ostrovski, C. Zhang, and D. Cai: Metall.
Mater. Trans. B, 2019, vol. 50B, pp. 291–303.
structure, and heat transfer are summarized as follows: 15. D. Xiao, W. Wang, and B. Lu: Metall. Mater. Trans. B, 2015,
(1) The increase of Al2O3/(B2O3 + Na2O) ratio in- vol. 46, pp. 873–81.
16. B. Lu and W. Wang: Metall. Mater. Trans. B, 2015, vol. 46B,
creased the softening, hemispherical, and fluidity pp. 852–62.
temperatures of CaO-Al2O3-based mold fluxes, 17. J. Qi, C. Liu, and M. Jiang: J. Non-Cryst. Solids, 2017, vol. 475,
which deteriorates the infiltration of flux and lubri- pp. 101–07.
cation during casting; 18. L. Zhou, H. Li, W. Wang, D. Xiao, L. Zhang, and J. Yu: Metall.
(2) The flux viscosity generally increased from Fluxes 1 Mater. Trans. B, 2018, vol. 49B, pp. 2232–40.
19. L. Wang, Y. Cui, J. Yang, C. Zhang, D. Cai, J. Zhang, Y. Sasaki,
to 4 in the high temperature region. Fluxes 1 and 2 and O. Ostrovski: Steel Res. Int., 2015, vol. 86, pp. 670–77.
exhibited break temperatures at 1221 K and 1253 K 20. D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calvé, B.
(948 C and 980 C), respectively; while apparent Alonso, J.-O. Durand, B. Bujoli, Z. Gan, and G. Hoatson: Magn.
break temperature was not observed in Fluxes 3 and Reson. Chem., 2002, vol. 40, pp. 70–76.
21. G.L. Caër and R.A. Brand: J. Phys., 1998, vol. 10, pp. 10715–74.
4; 22. J. Yang, J. Zhang, Y. Sasaki, O. Ostrovski, C. Zhang, D. Cai, and
(3) Raman and 27Al MAS-NMR spectra indicated that Y. Kashiwaya: Metall. Mater. Trans. B, 2017, vol. 48B, pp. 2077–
the increase of Al2O3/(B2O3 + Na2O) ratio led to 91.
the transition from 2-D to 3-D aluminate structure. 23. J. Yang, J. Zhang, Y. Sasaki, O. Ostrovski, C. Zhang, D. Cai, and
Moreover, the intensities of silicate and borate Y. Kashiwaya: Metall. Mater. Trans. B, 2016, vol. 47B, pp. 2447–
58.
structures were suppressed in the fluxes. The higher 24. J. Yang, J. Zhang, Y. Sasaki, O. Ostrovski, C. Zhang, D. Cai, and
degree of polymerization accounted for the higher Y. Kashiwaya: ISIJ Int., 2016, vol. 56, pp. 574–83.
viscosity of Fluxes 3 and 4; 25. K. Gu, W. Wang, L. Zhou, F. Ma, and D. Huang: Metall. Mater.
(4) IET experiments demonstrated that heat flux across Trans. B, 2012, vol. 43B, pp. 937–45.
26. N. Ma, J. You, L. Lu, J. Wang, M. Wang, and S. Wan: Inorg.
the mold flux disks decreased with the increasing Chem. Front., 2018, vol. 5, pp. 1861–68.
ratio of Al2O3/(B2O3 + Na2O), which is related to 27. J. Yang, J. Zhang, O. Ostrovski, C. Zhang, and D. Cai: Metall.
the devitrification of mold fluxes. Mater. Trans. B, 2019, vol. 50B, pp. 1766–72.

2802—VOLUME 50B, DECEMBER 2019 METALLURGICAL AND MATERIALS TRANSACTIONS B


28. J.H. Park, D.J. Min, and H.S. Song: ISIJ Int., 2002, vol. 42, 45. G. Padmaja and P. Kistaiah: J. Phys. Chem. A, 2009, vol. 113,
pp. 38–43. pp. 2397–2404.
29. P. McMillan and B. Piriou: J. Non-Cryst. Solids, 1983, vol. 55, 46. A. Stamboulis, R.G. Hill, and R.V. Law: J. Non-Cryst. Solids,
pp. 221–42. 2004, vol. 333, pp. 101–07.
30. T.S. Kim and J.H. Park: ISIJ Int., 2014, vol. 54, pp. 2031– 47. D.R. Neuville, L. Cormier, and D. Massiot: Chem. Geol., 2006,
38. vol. 229, pp. 173–85.
31. E. Gao, W. Wang, and L. Zhang: J. Non-Cryst. Solids, 2017, 48. J.F. Stebbins, S. Kroeker, S. Keun Lee, and T.J. Kiczenski: J.
vol. 473, pp. 79–86. Non-Cryst. Solids, 2000, vol. 275, pp. 1–6.
32. J. Gao, G. Wen, T. Huang, B. Bai, P. Tang, and Q. Liu: J. Am. 49. G.-R. Li, H.-M. Wang, Q.-X. Dai, Y.-T. Zhao, and J.-S. Li: J.
Ceram. Soc., 2016, vol. 99, pp. 3941–47. Iron. Steel Res. Int., 2007, vol. 14, pp. 25–28.
33. J. Gao, G. Wen, T. Huang, B. Bai, P. Tang, and Q. Liu: J. 50. Y. Lu, G. Zhang, M. Jiang, H. Liu, and T. Li: J. Environ. Sci.,
Non-Cryst. Solids, 2016, vol. 452, pp. 119–24. 2011, vol. 23, pp. S167–69.
34. G.-H. Kim and I. Sohn: J. Non-Cryst. Solids, 2012, vol. 358, 51. D.J. Min and F. Tsukihashi: Met. Mater. Int., 2017, vol. 23, pp. 1–
pp. 1530–37. 19.
35. K. Zheng, Z. Zhang, L. Liu, and X. Wang: Metall. Mater. Trans. 52. Z. Wang, Q. Shu, and K. Chou: Steel Res. Int., 2013, vol. 84,
B, 2014, vol. 45B, pp. 1389–97. pp. 766–76.
36. J.-Y. Park, G.H. Kim, J.B. Kim, S. Park, and I. Sohn: Metall. 53. J. Wei, W. Wang, L. Zhou, D. Huang, H. Zhao, and F. Ma:
Mater. Trans. B, 2016, vol. 47B, pp. 2582–94. Metall. Mater. Trans. B, 2014, vol. 45B, pp. 643–52.
37. J.H. Park, D.J. Min, and H.S. Song: ISIJ Int., 2002, vol. 42, 54. Y.G. Maldonado, F.A. Acosta, A.H. Castillejos, and B.G. Tho-
pp. 344–51. mas: Iron Steel Technol., 2013, vol. 10, pp. 65–75.
38. Y. Sun and Z. Zhang: Metall. Mater. Trans. B, 2015, vol. 46B, 55. L. Zhou, H. Li, W. Wang, and J. Chang: Metall. Mater. Trans. B,
pp. 1549–54. 2018, vol. 49B, pp. 3019–29.
39. R.K. Brow, D.R. Tallant, and G.L. Turner: J. Am. Ceram. Soc., 56. R.J. O’Malley and J. Neal: Proc. METEC Congress, 1999, vol. 99,
1996, vol. 79, pp. 2410–16. pp. 13–15.
40. Y. Kim, Y. Yanaba, and K. Morita: J. Am. Ceram. Soc., 2017, 57. L. Zhou, W. Wang, D. Huang, J. Wei, and J. Li: Metall. Mater.
vol. 100, pp. 5746–54. Trans. B, 2012, vol. 43B, pp. 925–36.
41. Y. Kim and K. Morita: ISIJ Int., 2014, vol. 54, pp. 2077–83. 58. J.B. Ferguson and A. Buddington: Am. J. Sci., 1920, vol. 199,
42. E.I. Kamitsos, M.A. Karakassides, and G.D. Chryssikos: J. Phys. pp. 131–40.
Chem., 1987, vol. 91, pp. 1073–79. 59. K. Xu, P. Loiseau, G. Aka, and J. Lejay: Cryst. Growth Des., 2009,
43. G.H. Kim and I. Sohn: Metall. Mater. Trans. B, 2014, vol. 45B, vol. 9, pp. 2235–39.
pp. 86–95.
44. I.J. Hidi, G. Melinte, R. Stefan, M. Bindea, and L. Baia: J. Raman Publisher’s Note Springer Nature remains neutral with regard to
Spectrosc., 2013, vol. 44, pp. 1187–94. jurisdictional claims in published maps and institutional affiliations.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 50B, DECEMBER 2019—2803

You might also like