- Email: [email protected]
Preparation and characterization of rapidly quenched BaOB2O2Fe2O3 glass
764
Materials Science and Engineering, A133 (1991 ) 764-766
Preparation and characterization of rapidly quenched BaO-BzO3-Fe203 glass Th. Schubert, R. Seliger and M. Jurisch Central Institute of Solid State Physics and Materials Research, 8027 Dresden (F.R.G.)
Abstract The optimal technological conditions, the corresponding heat transfer coefficients and the average cooling rate were evaluated for the preparation of amorphous BaO-B203-FezO3 glass by a twin-roller technique.
1. Introduction Due to their outstanding magnetic properties, M-type hexaferrite powders BaFel2Ol9 are of potential interest both for permanent magnets and magnetic recording media [1-3]. Compared with other preparation methods discussed in the literature [4-6] the so-called glass-crystallization process [7, 8] has the advantage in supplying strain-free single crystalline particles with the desired size distribution and a defined aspect ration of the crystallites. The glass-crystallization process consists of a controlled crystallization of an appropriate glass prepared from a BaO-B203-Fe20 3 flux melt by rapid quenching. The ferrite particles formed are extracted by dissolving the accompanying borate phase by dilute acetic acid. Presupposition for a narrow size distribution of the particles are uniformly cooled glass flakes or ribbons. Whereas many theoretical and experimental studies on rapid quenching of metallic systems have been reported so far, only a little is known about the rapid quenching behaviour of oxidic melts. In the present paper the results of an investigation of the factors determining the ribbon formation from oxidic glass melts by a twin-roller technique are presented.
nized at temperatures TO-~ 1300 °C in a platinum crucible equipped with a circular nozzle and a special gate to start the process. An instrumented twin-roller facility (see ref. 9 for details) was used to convert the melt into an amorphous glass. The effects of melt flow rate, roll supporting force, melt superheating and surface velocity of the rolls on the geometrical and structural characteristics of the ribbons or flakes have been studied. In order to get an estimation of the heat transfer conditions during rapid quenching the surface temperature of the rolls, the average temperature of the ribbons formed and the thermal expansion of the rolls have been measured. The surface temperature of the rolls during the process was recorded by an infrared pyrometer at an angle of 120 ° behind the kissing point of the roll gap. The gradual thermal expansion of the rolls was followed by a rod sliding on the roll and connected with an inductive transducer. This rod touches the roller surface at short time intervals in order to minimize errors caused by heating up the sensor. The temperature of the flakes and ribbons, respectively, leaving the rolls was estimated from calorimetric measurements.
2. Experimental
3. Results
A mixture of the glass-forming component BeO3-BaO and of the barium hexaferrite forming component BaO-FezO 3 are melted and homoge-
3.l. Features o f ribbon f o r m a t i o n
0921-5093/91/$3.50
A continuous ribbon formation was only realized if the melt-jet velocity vj was equal or nearly
765
equal to the surface velocity v d of the rolls. This velocity was in the range from 0.5 to 1.2 m s L The roll rotation speed was shown to be the key parameter for thickness control of the ribbons. The influence of % on the average thickness d r and structure of the ribbons or flakes prepared is represented in Fig. 1. As can be seen, d, decreases with increasing surface velocity according to the relation d , - v d ,.5 which is also valid for metallic systems without lateral constraints of the melt puddle [9]. This dependence has been explained by a velocity-dependent heat transfer coefficient h [9]: h - vu°5 Ribbons or flakes produced were X-ray amorphous for d~ < 100/~m. As was demonstrated by TEM studies of as-quenched material (see Fig. 1) ribbons having an average thickness lower than 50 /,m are amorphous whereas for d~> 50 y m some borate nucleii and/or hexaferrite crystallites were observed (Fig. 2).
~ \ ' ~ \ c r y s t a l t i n e ports ( ~ 4 0 n m ) ~ , ~ \
60~1
- . ~ . ~ +
"
"b-~
/ nuclei (=3 - <1Onto) J
40,.0 ~,morph'oo~.,V,~, "~, I
20 0,4
continuous
ribbon I f°rmeti°n 0,6
dr i
i
v~-..+/ ,
0,8 1.0 2,0 m/s z,,O rotl-surfctce velocity vd
Fig. 1. D e p e n d e n c e of the r i b b o n thickness and s t r u c t u r e on the surface speed of the rolls.
3.2. Heattransferconditions A thermal model described in refs. 10 and 11 has been used to analyze the experimentally determined data of surface temperature, thermal expansion and average ribbon temperature. The effective (average) heat transfer coefficient describing heat transfer from the melt to the uncooled rolls was fitted to the experimental data. The rolls were considered thermally isolated outside the contact area. The thermal expansion of the rolls was calculated according to l 121: AR=2au(1 +
v)/R
with /?
u=f
7(r,z, ~k)rdr,
(I
where a is the linear thermal coefficient of thermal expansion, v the Poisson number, R the radius of the rolls, and T the temperature of the rolls. The average temperature of the ribbon iru was calculated using the heat capacity of the ribbons after leaving the gap and the specific heat of the material given in Table 1. The ribbon=roller contact angle q~k was estimated from single photographs of the melt puddle to be about 6 o. The experimentally determined roll surface temperature T~, the average ribbon temperature IPB and the thermal expansion of the rolls AR together with the theoretical results using the heat transfer coefficient given in the inserts are shown in Figs. 3-5 as a function of the process time. Best agreement with the experimental data was obtained using a heat transfer coefficient of h = 9 x 1()-~ W K ~ m -2. It is comparable with those characterizing twin roller quenching of metallic melts [9]. But it should be mentioned that
TABLE l Data used for the calculations
Fig. 2. T E M m i c r o g r a p h of an a s - q u e n c h e d r i b b o n (d r = 100 /~m).
Ribbon: thickness width density specific heat
d, - 65 ,urn b, = 13 m m p = 4.0 g c m -~ c B = 0 . 8 6 J n kg i
Roller:
Rbt,,j = p = c+,).~
radius b o d y length surface velocity density specific heat thermal conductivity thermal e x p a n s i o n
105 m m 70 m m 0.8 m s 7.5 g c m ~ 0.49 J i kg ' [13] 50 W ~ k m ~ [141 d = 8.4 x 10 /, K
766 140
°c
<1
100
•°=- 40 8.
E J~
~
h - 0,9-10~Wrn'2grd"+ -8,4 10"~ grd "1
Q: pm
meosured
colculeted
6o
~ " x curve
calculated
30
~
£ 20
d roll ~
/
br i]b ~ n
+
m 40
0
roU body
h - 0,9.10 ~'Wrn'2_~rd"1 CB-0,86 Jg-~grd-
20" 20
40
60
0
80 100 rolt rotOtions
20
40
60
80 100 rot! rototions
120
Fig. 5. Thermal expansion AR as a function of the process time.
Fig. 3. Increase of the roll surface temperature with the process time.
550, °C I~m
glass ribbons. This material is suitable for the manufacturing of ferrite powders with excellent recording properties. For optimum casting conditions a heat transfer coefficient of h = 9 x 103 W K-~ m -z was estimated leading to an average cooling rate of 5 x 104 K s- 1.
J h = 0,9 10t"Wm-2grd -1 c B-0,86 jg-1 grd-1 catcuteted
~ 450
References
" ~ ' ~ measured
I
o
400 0
20
40
60
80 rotl rototions
100
Fig. 4. Ribbon temperature after leaving the roll gap as a function of the process time.
the lower thermal conductivity of the glass causes a higher temperature gradient in these ribbons. In the thickness range of about 70/am the temperature difference between the surface and the centre of the ribbons was estimated to be about 100 K, but about 30 K for a typical metallic system Fe-B. Using a contact length of approximately 11 mm a cooling rate of 5 x 1 0 4 K s-1 is estimated. 3. Conclusions It has been demonstrated that the twin-roller technique is practicable for the continuous preparation of amorphous BaO-B203-FeeO 3
1 0 . Kubo, T. Ido and H. Yokoyama, IEEE Trans. Magn., MAG-18(1982) 1122. 2 T. Fujiwara, IEEE Trans. Magn., MA G-23(1984) 1480. 3 D. E. Speliotis, IEEE Trans. Magn., MAG-23 (1987) 3143. 4 D. Barb, L. Diamandesan, A. Rusi, Tarabasann, D. Mihaila, M. Morarin and V. Teodorescu, J. Mater. Sci., 21 (1986) 1118. 5 E Chou, X. Feng, J. Li and Y. Lin, J. Appl. Phys., 61 (1987) 3881. 6 Y. K. Hong, Y. J. Paig, D. G. Agresti and T. D. Shelfer, J. Appl. Phys., 61 (1987) 3872. 7 T. Ido, E. Eng, O. Kubo, H. Yokoyama and S. Keujo, Toshiba Rev., No. 154(1985) 10. 8 0 . Kubo, T. Ito, T. Nobura, K. Inometa and H. Yokoyama, German Patent, DE 3041 960 (1980). 9 M. Jurisch, H. Fiedler and R. Sellger, Proc. Rapid Solidi-
fication seminar, Dresden, Nov. 1988. 10 R. Sellger, Thesis, Academy of Sciences of GDR, Dresden (1986). 11 T. Flehmig, Thesis, Academy of Sciences of GDR, Dresden (1990). 12 T. Yamauchi, Trans. Iron Steellnst. Jpn., 28(1988) 23. 13 L. Parikov and Ju. Jurtschenko, Naukowa Dumka, Kiev (1985). 14 Landolt-B6rnstein, Zahlenwerte und Funktionen, IV Band Teil 2 (Springer-Verlag, 1965).
Materials Science and Engineering, A133 (1991 ) 764-766
Preparation and characterization of rapidly quenched BaO-BzO3-Fe203 glass Th. Schubert, R. Seliger and M. Jurisch Central Institute of Solid State Physics and Materials Research, 8027 Dresden (F.R.G.)
Abstract The optimal technological conditions, the corresponding heat transfer coefficients and the average cooling rate were evaluated for the preparation of amorphous BaO-B203-FezO3 glass by a twin-roller technique.
1. Introduction Due to their outstanding magnetic properties, M-type hexaferrite powders BaFel2Ol9 are of potential interest both for permanent magnets and magnetic recording media [1-3]. Compared with other preparation methods discussed in the literature [4-6] the so-called glass-crystallization process [7, 8] has the advantage in supplying strain-free single crystalline particles with the desired size distribution and a defined aspect ration of the crystallites. The glass-crystallization process consists of a controlled crystallization of an appropriate glass prepared from a BaO-B203-Fe20 3 flux melt by rapid quenching. The ferrite particles formed are extracted by dissolving the accompanying borate phase by dilute acetic acid. Presupposition for a narrow size distribution of the particles are uniformly cooled glass flakes or ribbons. Whereas many theoretical and experimental studies on rapid quenching of metallic systems have been reported so far, only a little is known about the rapid quenching behaviour of oxidic melts. In the present paper the results of an investigation of the factors determining the ribbon formation from oxidic glass melts by a twin-roller technique are presented.
nized at temperatures TO-~ 1300 °C in a platinum crucible equipped with a circular nozzle and a special gate to start the process. An instrumented twin-roller facility (see ref. 9 for details) was used to convert the melt into an amorphous glass. The effects of melt flow rate, roll supporting force, melt superheating and surface velocity of the rolls on the geometrical and structural characteristics of the ribbons or flakes have been studied. In order to get an estimation of the heat transfer conditions during rapid quenching the surface temperature of the rolls, the average temperature of the ribbons formed and the thermal expansion of the rolls have been measured. The surface temperature of the rolls during the process was recorded by an infrared pyrometer at an angle of 120 ° behind the kissing point of the roll gap. The gradual thermal expansion of the rolls was followed by a rod sliding on the roll and connected with an inductive transducer. This rod touches the roller surface at short time intervals in order to minimize errors caused by heating up the sensor. The temperature of the flakes and ribbons, respectively, leaving the rolls was estimated from calorimetric measurements.
2. Experimental
3. Results
A mixture of the glass-forming component BeO3-BaO and of the barium hexaferrite forming component BaO-FezO 3 are melted and homoge-
3.l. Features o f ribbon f o r m a t i o n
0921-5093/91/$3.50
A continuous ribbon formation was only realized if the melt-jet velocity vj was equal or nearly
765
equal to the surface velocity v d of the rolls. This velocity was in the range from 0.5 to 1.2 m s L The roll rotation speed was shown to be the key parameter for thickness control of the ribbons. The influence of % on the average thickness d r and structure of the ribbons or flakes prepared is represented in Fig. 1. As can be seen, d, decreases with increasing surface velocity according to the relation d , - v d ,.5 which is also valid for metallic systems without lateral constraints of the melt puddle [9]. This dependence has been explained by a velocity-dependent heat transfer coefficient h [9]: h - vu°5 Ribbons or flakes produced were X-ray amorphous for d~ < 100/~m. As was demonstrated by TEM studies of as-quenched material (see Fig. 1) ribbons having an average thickness lower than 50 /,m are amorphous whereas for d~> 50 y m some borate nucleii and/or hexaferrite crystallites were observed (Fig. 2).
~ \ ' ~ \ c r y s t a l t i n e ports ( ~ 4 0 n m ) ~ , ~ \
60~1
- . ~ . ~ +
"
"b-~
/ nuclei (=3 - <1Onto) J
40,.0 ~,morph'oo~.,V,~, "~, I
20 0,4
continuous
ribbon I f°rmeti°n 0,6
dr i
i
v~-..+/ ,
0,8 1.0 2,0 m/s z,,O rotl-surfctce velocity vd
Fig. 1. D e p e n d e n c e of the r i b b o n thickness and s t r u c t u r e on the surface speed of the rolls.
3.2. Heattransferconditions A thermal model described in refs. 10 and 11 has been used to analyze the experimentally determined data of surface temperature, thermal expansion and average ribbon temperature. The effective (average) heat transfer coefficient describing heat transfer from the melt to the uncooled rolls was fitted to the experimental data. The rolls were considered thermally isolated outside the contact area. The thermal expansion of the rolls was calculated according to l 121: AR=2au(1 +
v)/R
with /?
u=f
7(r,z, ~k)rdr,
(I
where a is the linear thermal coefficient of thermal expansion, v the Poisson number, R the radius of the rolls, and T the temperature of the rolls. The average temperature of the ribbon iru was calculated using the heat capacity of the ribbons after leaving the gap and the specific heat of the material given in Table 1. The ribbon=roller contact angle q~k was estimated from single photographs of the melt puddle to be about 6 o. The experimentally determined roll surface temperature T~, the average ribbon temperature IPB and the thermal expansion of the rolls AR together with the theoretical results using the heat transfer coefficient given in the inserts are shown in Figs. 3-5 as a function of the process time. Best agreement with the experimental data was obtained using a heat transfer coefficient of h = 9 x 1()-~ W K ~ m -2. It is comparable with those characterizing twin roller quenching of metallic melts [9]. But it should be mentioned that
TABLE l Data used for the calculations
Fig. 2. T E M m i c r o g r a p h of an a s - q u e n c h e d r i b b o n (d r = 100 /~m).
Ribbon: thickness width density specific heat
d, - 65 ,urn b, = 13 m m p = 4.0 g c m -~ c B = 0 . 8 6 J n kg i
Roller:
Rbt,,j = p = c+,).~
radius b o d y length surface velocity density specific heat thermal conductivity thermal e x p a n s i o n
105 m m 70 m m 0.8 m s 7.5 g c m ~ 0.49 J i kg ' [13] 50 W ~ k m ~ [141 d = 8.4 x 10 /, K
766 140
°c
<1
100
•°=- 40 8.
E J~
~
h - 0,9-10~Wrn'2grd"+ -8,4 10"~ grd "1
Q: pm
meosured
colculeted
6o
~ " x curve
calculated
30
~
£ 20
d roll ~
/
br i]b ~ n
+
m 40
0
roU body
h - 0,9.10 ~'Wrn'2_~rd"1 CB-0,86 Jg-~grd-
20" 20
40
60
0
80 100 rolt rotOtions
20
40
60
80 100 rot! rototions
120
Fig. 5. Thermal expansion AR as a function of the process time.
Fig. 3. Increase of the roll surface temperature with the process time.
550, °C I~m
glass ribbons. This material is suitable for the manufacturing of ferrite powders with excellent recording properties. For optimum casting conditions a heat transfer coefficient of h = 9 x 103 W K-~ m -z was estimated leading to an average cooling rate of 5 x 104 K s- 1.
J h = 0,9 10t"Wm-2grd -1 c B-0,86 jg-1 grd-1 catcuteted
~ 450
References
" ~ ' ~ measured
I
o
400 0
20
40
60
80 rotl rototions
100
Fig. 4. Ribbon temperature after leaving the roll gap as a function of the process time.
the lower thermal conductivity of the glass causes a higher temperature gradient in these ribbons. In the thickness range of about 70/am the temperature difference between the surface and the centre of the ribbons was estimated to be about 100 K, but about 30 K for a typical metallic system Fe-B. Using a contact length of approximately 11 mm a cooling rate of 5 x 1 0 4 K s-1 is estimated. 3. Conclusions It has been demonstrated that the twin-roller technique is practicable for the continuous preparation of amorphous BaO-B203-FeeO 3
1 0 . Kubo, T. Ido and H. Yokoyama, IEEE Trans. Magn., MAG-18(1982) 1122. 2 T. Fujiwara, IEEE Trans. Magn., MA G-23(1984) 1480. 3 D. E. Speliotis, IEEE Trans. Magn., MAG-23 (1987) 3143. 4 D. Barb, L. Diamandesan, A. Rusi, Tarabasann, D. Mihaila, M. Morarin and V. Teodorescu, J. Mater. Sci., 21 (1986) 1118. 5 E Chou, X. Feng, J. Li and Y. Lin, J. Appl. Phys., 61 (1987) 3881. 6 Y. K. Hong, Y. J. Paig, D. G. Agresti and T. D. Shelfer, J. Appl. Phys., 61 (1987) 3872. 7 T. Ido, E. Eng, O. Kubo, H. Yokoyama and S. Keujo, Toshiba Rev., No. 154(1985) 10. 8 0 . Kubo, T. Ito, T. Nobura, K. Inometa and H. Yokoyama, German Patent, DE 3041 960 (1980). 9 M. Jurisch, H. Fiedler and R. Sellger, Proc. Rapid Solidi-
fication seminar, Dresden, Nov. 1988. 10 R. Sellger, Thesis, Academy of Sciences of GDR, Dresden (1986). 11 T. Flehmig, Thesis, Academy of Sciences of GDR, Dresden (1990). 12 T. Yamauchi, Trans. Iron Steellnst. Jpn., 28(1988) 23. 13 L. Parikov and Ju. Jurtschenko, Naukowa Dumka, Kiev (1985). 14 Landolt-B6rnstein, Zahlenwerte und Funktionen, IV Band Teil 2 (Springer-Verlag, 1965).