Process metallurgy

Viscosity estimation of slags

Georges Urbain

The viscosity of silicate melts and slags is a property difficult to measure accurately using sophisticated apparatus, it is both
time consuming and expensive. The use of models to estimate the viscosity is the first step to be carried out, mainly with
metallurgical slags involving numerous components. The proposed model is an experimental one based on accurate data for
pure binary and ternary melts. From this data, using a simple hypothesis on the behaviour of unusual compounds, it is possible
to estimate the viscosity of a complex melt. This estimation is used as a first approximation and can be amended, if necessary,
by only a few direct measurements.

Berechnung der Viskositaten von Schlacken. Die Viskositiitsmessungen von Schlacken und flussiqen Silikatschmelzen sind
zeitaufwendig und schwierig. Urn genaue Messungen zu erhalten, benotiqt man komplizierte Geriite, da man bei hohen Tem-
peraturen arbeiten rnuls, Fur viele Anwendungen und ganz speziell fur die metallurgischen Schlacken mit zahlreichen Kompo-
nenten konnen berechnete Werte genugen. Das Versuchsmodell fur diese Werte beruht auf Viskositiitsmessungen, die an
biniiren und terniiren Schmelzen vorgenommen wurden. Die berechneten Werte beziehen sich auf die Eigenschaften der Kom-
ponenten, fur die es keine direkten Messungen gibt. 1m allgemeinen ist der berechnete Wert ein erster Niiherungswert, der
allenfalls durch eine bestimmte Anzahl von direkten Messungen verbessert und vervollstiindigt werden muB.

For the moment, a coherent theory of the liquid state does The probabilities P; and P; are related to the variables of
not exist and theoretical models are not within our reach, state and to the structure of the liquid (degree of polymeri-
especially for associated liquids such as silicates or phos- zation). The "jumping" probability P; is deduced from the
phates. A pragmatic way is to use empirical models based on balance between the kT and the energy level of the potential
measured viscosities. Two models are favoured by geolog- well E w • Following a statistical approach proposed by H. D.
ists: those proposed in 1972 by Y. Bottinga I) and by H. Weymanrr') the equation of viscosity is:
Shaw"). Both used nearly the same published viscosity data 1 1
(mostly geological melts) and empirical rules to calculate the Jl = (kTI E w)2. (21tkT)2. V -2/3. P; - I. exp(EwlkT), (I)
viscosities. The heterogeneity of the accuracies of these data
m and v are mass and volume of the structural unit, E; en-
present some difficulties for application to slags.
ergy of the well, k and T Boltzmann's constant and tempera-
The proposed model originated in a collective work carried ture in K, P; is the "hole" probability connected with the
out in the sixties in Irsid's Physical Chemistry Department structural model of the liquid.
under the direction of Professor P. Kozakevitch. The probability P; is proportional to the concentration N; of
"holes" given at T by the equilibrium:
(2)
Background for estimations
!i.Hv and !i.Sv are the partial molar enthalpy and entropy as-
Viscous flow is a momentum transport process in the fluid sociated to the "hole" formation. Using equations (I) and
phase, just as thermal conductance is a transport of energy. (2) it is possible to obtain a two parameter expression for the
On a macroscopic scale, shear viscosity is the ratio between viscosity:
shear stress and the rate of flow. Units for dynamic viscosity
Jl=A· T·exp(1000·BIT). (3)
are:
The pre-exponential parameter A is a function of m, E w , v
- the S.1. unit Poiseuille = Pas = N s/rn? or and !i.Sv. Parameter B is a function of E; and !i.Hv.
- the CGS unit poise = 0.1 Pas.
In this publication the unit "poise" is used as the purpose is Relation between A and B. Parameters A and B expressed in
to give useful information needed by practicians. molar quantities are from (1):
A =Ao·exp( -!i.SJR) (4)
Expressions for the viscous flow. On a molecular scale, vis-
I
cous flow involves relative movements of the structural ele-
with Ao=R ·(2MIE)2 ·(11 V)2/3 .(11 N)I/3,
ments of the liquid. Two conditions must be realised for
such a displacement: B=(E+!i.Hv)IR. (5)
- an energetic condition: the "jumping" probability Pe, As the "hole" equilibrium is governed by the free energy
- a geometric condition: the "hole" or "free volume" prob- !i.Gv and the equation:
ability r..
!i.GvCT)=!i.Hv- T·!i.S"
Simultaneous "satisfaction" of these two conditions is nec-
it is always possible for T= T; to have !i.Gv=O, then:
essary to move some structural units when external stress is
applied. The fluidity, the reciprocal of viscosity, could be !i.Sv=!i.HvITc.
written:
Combination with equations (4) and (5) gives:
InA =lnAo+EI(RTc)-IOOOBITc

Dr-Inq. Georges Urbain, Oir. Recherche C.N.R.S., Universite de Per- as A o, E and T; are constant for a given liquid, a simple for-
pignan, France mulation is:

steel research 58 (1987) NO.3 111

Process metallurgy

- Ln A•__,---------,_----,-~.~-____,_-_,_-__, Examples: In a slag with only Si as a network former it is

clear that "T" = Si and the mole fractions are the same if
there is no change in the other mole fractions. When silicates
30 and phosphates are present in the slag calculation is
"T" (equivalent) = "Si" + "P":

20 analysis gives wt. % (PzOs),

mole fraction is calculated N(PzOs),
charge equivalence T = 4/5· P
10 mole equivalence N(TO z) = 8/5· N(PzOs).
For this slag N(TOz)=N(SiO z)+ 1.6·N(PzOs), without for-
getting normalisation to unity.
o For modifier cations M + + (Mg, Ca, Fe, Ni, Co, Mn, Zn, Cd
B etc.) or M+ (Li, Na, K, Ag etc.) associated with one oxygen:
Figure 1. Correlation between A and B (equation (5)) for oxide, silicate MO or MzO the mole fractions are equivalent. Trivalent ca-
and alumino-silicate melts. Melt compositions are identified by num- tions (AI, Ga, Cr, Y, La etc.) mole fraction N(M z0 3 ) are con-
bers corresponding to the entries of table 1 in 5). F is for liquid FeO, verted into the equivalent for one oxygen: 3· N(M z0 3 ) . Te-
and A for liquid alumina
travalent (penta- etc.) cations MO z (Ti, Zr etc.) or MzOs (V,
Bild 1. Zusammenhang zwischen den Parametern A und B und der
Weymann'schen Gleichung (5)3). Die Zahlen stellen die verschiedenen Nb, Mo etc.) mole fractions are converted according to this
Zusammensetzungen (Oxide, Silikate und Aluminiumsilikate) dar, sie equation: N(equi.MO)=y·N(MxO.). Such a change in the
sind in Tafel 1 in 5) wiedergegeben. F = FeO,-Schmelze im Gleichge- number of moles alters all mole fractions and requires nor-
wicht mit einem Eisentiegel, A = Tonerde malization to unity for further calculations.

-lnA=m·B+n (6) Determination of B. To enable the estimation of viscosity of

where m and n are deduced from the experimental paramet- a ternary melt:
ers A and B. From a group of 54 liquids (oxides, silicates real ternary as SiO z-Alz0 3-CaO,
and alumino-silicates) recorded in figure 1 the mean values 11 hypothetical ternary TO z-Az0 3-MO,
of m and n are:
it is necessary to define two convenient composition param-
m = 0.29 and n = 11.57. eters. The first one is X the network former mole fraction
Such a linear equation allows the calculation of the viscosity N(TO z) and the second is a:
at temperature Twhen only enthalpy parameter B is known. X=N(TO z)
The model for viscosity estimations is now limited to an in- a = MO/(MO + AZ0 3) with
terpolation of B with the composition of the slag.
MO = ~ (mole frac. modifier oxides),
AZ03 = ~ (mole frac. amphoteric oxides).
Model for estimations Experience shows that B is well represented by polynomial
equation:
The model is coherent with the assumption of an equili-
B=Bo+BJ·X +B z·X z+B 3·X 3 , (7)
brium of polyanion entities in the melt and it needs the clas-
sification of all cations into three kinds of behaviour: glass B;=a(i)+b(i)·a+c(i)·a z , (8)
former, modifier and amphoteric. This classical classifica-
with i=O, 1,2,3.
tion is empirical in fact and must be controlled by experi-
It is possible to calculate B for two ternaries:
ments in ambiguous situations (an example for Fe 3+ cation
as seen later). B(Mg) for ternary SiO z-Alz0 3-MgO,
Glass former: Cations Si4+, Ge 4+, pH, etc., forming po- B(Ca) for ternary SiO z-Alz0 3 - C a O ,
Iyanions: Si0 4 , Siz0 7 , P0 4 , etc. are always in tetrahedral po-
and to obtain a mean B for a quaternary melt by considera-
sition. tion of the B(Mg) and B(Ca) using the mole fractions
N(MgO) and N(CaO).
Modifier: Cations Na ", K+, Mg " ", Ca " ", Fe " ", Cr " " ",
Ti 4+, etc. are modifiers associated (change balance) with po-
Determination of the viscosity. When parameter B is calcu-
Iyanions in the melt. lated, the viscosity at temperature T(K) is obtained from
equations (3) and (6). The values used for m and n in equa-
Amphoteric: Cations A1 3+ and Fe3+ may act either as glass
tion (6) depend on the nature of the liquid. For ionic melts,
formers or as modifiers following the composition of the such as slags, the experimental values are m = 0.29 and
melt (charge balance condition). n= 11.57, but for network liquids, such as SiOz, 8 Z 0 3 , GeOz
the values are m = 0.207 and n = 10.288. The borderline be-
Calculation of the equivalent mole fraction. Chemical analy- tween the two sets of values depends on the slag composi-
sis gives the weight % of each component, mole fractions are tion, but it is generally around 0.85 to 0.90 mole fraction of
calculated and normalised to unity and then the equivalent
tetrahedral TO z.
mole fraction is obtained for an hypothetic ternary:
TO z-Az0 3-MO.
Data for calculation
"T" is a cation in a tetrahedral position: T0 4 ,
"A" is an amphoteric cation: A3+, Binary melts. Polynomial equations, such as 7 or 8, allow the
"M" is a modifier cation (M + +) associated with one oxy- calculation of B with the composition of the melt in mole
gen. fraction. For silica-oxide melts a third order polynomial

112 steel research 58 (1987) No 3

 Process metallurgy

Figure 2. lso-B curves for silica-alumina-CaO melts following equa- 1.0

tions (7) and (8)'0)
Bild 2. Ternares Diagramm Si0 2- AI20 3-CaO. Iso-6-Kurven der ter-
naren Schmelzen (Gleichungen (7) and (8)10)

'"
....
. ......
' .
07 • ". ". '" ". 1D. .
, : ", ". ", " . . '38 ,'
.: \ :.. ...•... ' .)6 .
0,6 .: ~:. .... :.. .. ,. '. . 31. .'
.......
0.. 0. .32
.... ' ... '. . 30
0
: : :. °•• ••

, 28
: ": : .... 0 ••

"
. 26 .'
'.
-,
-, 2S

21.
.:
"
.'
'.
'. . \16 ':~6 ".20 ···.22 ' •. 23 ... 22,: 20.:
•.12 ".14 ". ".
1,0 0,9 0,8 0;1 D,6 o.S 0,4
N(CoOI

Table 1. Parameters a, b, C und d of the

equation B=a+b·X+c·X2+d.X3 for bi- binary melt i ref. a b c d iv
nary melts silica-oxide. Xis the mole frac-
tion of silica silica + alumina 6 5 13.2 30.5 -40.4 60.8 O<X<I
silica + boric oxide II 7 13.2 0 -4.7 55.6 O<X<I
Tafel 1. Parameter a, b, C und d der Glei- O<X<I
silica + MgO 8 5,8 11.5 0.5 12.1 40
chung B=a+b·X+c·X2+d·X3 , wobei silica + CaO 11 5,8, 10 10.7 43.7 -111.4 121.1 O<X<I
X = Molenbruch von Si0 2 fur binsre silica + MnO 4 5 7.6 0.5 35.5 20.5 O<X<I
Schmelzen ist
silica + FeO, 14 5,11 6.1 0 12 46 O<X<I
silica + PbO 10 13 6.88 7.79 1.11 48.32 O<X<I

equation is convenient. For alumina-oxide melts the small Table 2. Parameters a, band c of the equation B=a+b·a+c·a 2 for
number of experiments limits the equation to a parabolic ap- binary melts alumina-oxide. a is equivalent to the mole fraction of the
oxide
proximation. Pure oxides, such as MgO, CaO, MnO, PbO,
Tafel 2. Parameter a, b und c der Gleichung B=a+b·a+c·a 2 wobei
are not accessible to measurement. The extrapolated value a=Molenbruch des Oxid fur binare Schmelzen ist
of B obtained from binaries is in agreement with the theore-
tical value given by J. O. M. Bockris") using the "hole" mod- binary melt ref. a b c iv
el. All information for each system is given in sequence:
Alumina + MgO 3 13.2 16.9 -18.6 O<a<1
I number of measured melts, Alumina + CaO 5 14 13.2 40 -42.5 O<a<1
ii origin of data, Alumina + MnO 4 15 13.2 20 -25.6 O<a<1
III equation B - X (mole fraction),
iv validity limits,
v B values for pure oxides.

Binary melts silica-oxide. For all these composrtions Table 3. Numerical values of coefficients a(i), b(i) and c(i) for the four
equations (8) with i=O, 1,2 and 3 for the three oxides MgO, CaO and
X = N(SiO z). For pure silica B = 64.1 5 ) , see table 1.
MnO
Tafel 3. Parameter a, b und c der 4 Gleichungen (8), wobei i=O, 1,2
Binary melts of type alumina-oxide. Melted pure alumina has und 3 der Schmelzen der 3 ternaren Systeme SI0 2-AI20 3-MO. Ka-
a measured B= I3.2±0.5 and a calculated B=8.6 4 ) , for tion M ~ Mg, Ca oder Mn
Tm = 2054 °C, this difference is discussed in 6). For all sys-
tems alumina-oxide the composition parameter a defined i a (i) b(i) c(i)
all Mg Ca Mn Mg Ca Mn
before a = MO/(MO + A z0 3 ) is equivalent to the mole frac-
tion MO of the oxide: a = N(MO), see table 2. 0 13.2 15.9 41.5 20.0 -18.6 -45 -25.6
1 30.5 -54.1 -117.2 26 33 130 -56
Ternary melts. Only three ternary systems are studied in the 2 -40.4 138 232.1 - 110.3 -112 -298.6 186.2
3 60.8 -99.8 - 156.4 64.3 97.6 213.6 -104.6
whole composition range: silica-alumina-MO oxide with

steel research 58 (1987) NO.3 113

Process metallurgy

Table 4. Example of B calculations for three melts M = Mg, Ca or Mn. The value of parameter B is obtained
Tafel 4. Beispiele der Berechnung von B fur drei Schmelzen from equations (7) and (8) for the composition given by X
and a. Table 3 recorded the parameters a(i), b(i) and c(i) of
sample no. 29 30 31
equation (8) for the three cations.
composition:
silica 44.9 42 33.14 Figure 2 gives iso-B curves for melts silica-alumina-CaO is-
alumina 29.9 19.8 38.36 sued from 10).
CaO 25.2 38.2 29.5
For the following numerical applications the use of a simple
X 0.5016 0.4439 0.3721 basic program is a good solution. A copy on floppy disc for
a 0.6051 0.7781 0.5830
Apple computer is available on request.
B 28.7 24.23 23.38
Numerical applications

Applications to ternary melts. Viscosities of three melts sili-

Tables 5-7. Calculated and measured viscosity (Poise) for three ca-alumina-CaO, not used to calculate the a(i), b(i) and c(i)
melts
coefficients, are obtained with the model as an example and
Tafeln 5-7. Berechnete und gemessene Viskositat (Poise) fur drei
Schmelzen a control of the estimations. These melts are samples no. 29,
30 and 31 of 5). Table 4 shows their compositions in wt. %,
Table 5. Sample no. 29
values of X and a calculated from the mole fractions and
T [DC] 1603 1653 1703 1753 1803 1853 1903 parameter B obtained with equations (7) and (8) using a(i),
b(i) and c(i) of table 1 for Ca. The comparison between
measured 19.7 14.2 10.4 7.6 5.8 4.4 3.4 measured 5) and calculated viscosities is made for each sam-
calculated 31 21 15 II 7.9 5.8 4.4
ple in tables 5-7. The differences between measured and
calculated viscosities are comparable with experimental dis-
Table 6. Sample no. 30
persions. For samples 29, 30 and 31 on a log scale deviation
T [DC] 1338 1409 1573 1652 1724 in % gives 12, 15 and 6%. This example is mostly a self con-
sistence test for the model as the three melts 29, 30 and 31
measured 69 38 12 7 4.6 are not used in the model's calculations.
calculated 35.6 14.5 3.9 2.7 2.0
Applications to complex melts. Three typical applications to
Table 7. Sample no. 31 practical problems are feldspar melts for clay ceramic sinter-
T[°C] 1305 1353 1425 1491 1545 1583 1664
ing, rocks and lava of geological interest, and metallurgical
slags.
measured 87.8 54 23.3 13.3 8.8 6.8 3.3 Feldspar melts. The model is applied to liquids: silica-anor-
calculated 67 44 28 15 11 8.6 5.3 thite-kalsilite-nepheline limited to feldspar-like melts with a
ratio alumina/modifier equal to unity. Data and results are
given in 16). But instead of using complex polynomial equa-
tions (with two sets of 27 coefficients) it is possible to com-
Table 8. Basalt composition in wt. % and mole fractions X
bine the 4 composition parameters in two sets of ternary
Tafel 8. Basaltzusammensetzung in Gew.-% und Molenbruch
melts. A mean value for B could be a first approximation
Oxide amount in mole fraction and may be sufficient.
wt.% X
Geological melts. Two examples present the opportunity of
SiO, 59.78 0.6469 discussing the influence of ratio Fe + + /Fe + + + on the vis-
Al,03 17.82 0.1137
cosity. Measurements on olivine type melts
Fe,O, 4.86 0.0193
MgO 4.29 0.0692 (SiOz-CaO-FeOx)recorded in 5), show clearly an increase
CaO 7.64 0.0886 of the viscosity with growing X. Cation Fe + + is always a
TiO, 1.63 0,0133 "modifier" but cation Fe + + + has an amphoteric behaviour
Na,O 3.45 0.0362 such as AI + + + .
K,O 1.47 0.0101
H,O 0.06 0.0022
Viscosities of a typical basalt (Ue 18q no. 2) are measured by
Piwinski and Weed 17) using a rotation cup viscometer in air.
Composition of the mineral is given in wt. % but total iron is
Table 9. Summary of calculations
taken as Fe z03' see table 8.
Tafel 9. Zusammenfassung der Berechnungen As the ratio Fe + + /Fe + + + is unknown, two simple hypo-
Hypothesis no. 2 theses are possible on the real behaviour of iron:

modifier
1) all iron (Fe + + +) is equivalent to AI + + +,
iron behaviour amphoteric
tetrahedral cations T =0.6469 T =0.6469 2) all iron is Fe + + .
amphoteric cations A =0.1335 A =0.113
modifier cations M=0.2328 M=0.2922 For a summary of the calculations see table 9.
total: 1.0133 1.0528
The calculation of X and a is obvious following their defini-
calculated: X 0.6385 0.6145 tion and paying attention to the fact that
a 0.6355 0.7198 N(TiO z)=2·N(MO) and that (Fez03)=3·N(MO). For the
B(CaO) 33.75 31.12 two ternaries silica-alumina-CaO and silica-alumina- MgO
B(MgO) 32.24 29.8 the B values are obtained with equations (7) and (8) using a
B (mean) 33.09 30.56
micro computer.

114 steel research 58 (1987) No 3

 Process metallurgy

Table 10. Measured and estimated viscosities x 10' in poise Table 12. Summary of calculations
Tafel 10. Gemessene und geschiitzte Viskositiiten x 10' in Poise Tafel 12. Zusammenfassung der Ergebnisse

TOC P& W data hypothesis 1 hypothesis 2 Hypothesis no. 2

1117 3.436 3.52 1.13 iron state Fe " + Fe " " +

1158 1.964 1.83 0.62 tetrahedral cations 0.4589 0.4589
1193 0.609 1.08 0.38 amphoteric cations 0.0485 (AI) 0.1422 (AI + Fe)
1267 0.123 0.38 0.15 modifier cations 0.5850 0.3976
1310 0.067 0.22 0.09
total 1.0923 0.9987
calculated: X 0.4201 0.4595
a 0.9235 0.7366
Measured and estimated viscosities are listed in table 10. B (CaO) 20.13 25.44
The viscosities are obtained from the calculated B values us- B (MgO) 18.02 22.76
ing equation (6) (with m=0.269 and n > 11.67) and equation B (mean) 18.86 24.09
(3). Estimated values frame the measured ones taking into
account all dispersions. Conclusion: only part of the iron is
in the melt in a tetrahedral position.
Table 13. Comparison of calculated and measured values of viscosities
Viscosity of a lunar lava. The paper of T. Murase and in poise
Mcbimey!") gives the analysis of a lunar lava sample and Tafel 13. Vergleich der berechneten und gemessenen Viskositiiten
few viscosity measurements under air and argon atmo-
sphere. Composition in wt. % and calculated mole fraction T[0C] M & B data hyp.1 hypo 2 hypo 3
are listed in table 11. In agreement with experimental condi-
Air Argon
tions two hypotheses are possible: 1395 10 7.1 7.2 41 12.9
1) argon atmosphere: all iron as Fe + + , 1450 5 5.2 27 9
1495 4.5 4.1 19 6.8
2) air atmosphere: unfortunatly no analysis of the sample
after measurement in air gives information for the ratio
Fe + + /Fe + + +. It is assumed that all iron is Fe + + ".
A summary of the calculations is given in table 12.
Table 14. Melubir 740 analysis in wt. % and mole fractions
Table 13 shows a comparison between calculated and mea- Tafel 14. Analyse von Melubir 740 in Gew.-% und Molenbruche
sured viscosities. The third hypothesis is based on a realistic
oxide amount mole fraction
proportion between ferrous and ferric iron in such a melt in in wt.% X
air: 2/3 ferrous and 1/3 ferritic. The B (mean) interpolated
between 18.86 and 24.09 for such a ratio is B=20.60 and the SiO, 34.4 0.4424
viscosities obtained are in the last column (see hypothesis 3). TiO, 0.3 0.0029
P,Os 0.1 0.0005
AI,O, 10.5 0.0796
Metallurgical slag. Different protecting slags for continuous MgO 1.3 0.0249
casting are prepared from mixed powders under the trade CaO 24.6 0.3390
name "Melubir" by the Societe Denain-Anzin Mineraux CaF, 4.9 0.0485
FeO 1.4 0.0151
(France). From chemical analysis and viscosity data given MnO 0.05 0.0005
by the manufacturer we have the opportunity to discuss the Na,O 3.0 0.0374
applicability of the model for slags containing fluorides. As K,O 1.1 0.009
an example the product "Melubir 740" for the raw material Fire loss 17.3
the analysis is given in table 14. The elevated humidity trap-
ped in the powders may introduce some water dissolved in
the melt. We have no analysis of the slag after viscosity
measurements to either check this possible water dissolution
or the evolution of fluoride content during experiments. Table 15. Summary of the calculations
Tafel 15. Zusammenfassung der Berechnungen

tetrahedral cations 0.4433 (Si + P)

Table 11. Composition of lunar lava sample amphoteric cation 0.0796 (AI)
Tafel 11. Zusammensetzung der Mondlavastein-Probe modifier cations 0.5773 (Ti gives two MO and CaF, in ag-
reement with its dissociation
oxide amount in mole fraction schema gives the equivalent of
wt.% X three MO)

SiO, 43.0 0.4589 total 1.1002

AI,O, 7.7 0.0485 calculated X = 0.4029 and a = 0.8788
Cr,O, 0.41 0.0017
TiO, 11.0 0.0883
B (CaO) =21.05
ZrO, 0.14 0.0007
B (MgO) = 18.47
FeO 21.0 0.1874
B (mean) = 20.84
MnO 0.26 0.0024 Viscosities in poises
NiO 0.04 0.0003 T[°C] manufacturer estimated
MgO 6.5 0.1034
CaO 9.0 0.1029 1100 190 169
Na,O 0.4 0.0041 1300 24 28
K,O 0.21 0.0014 1500 5.1 7.1

steel research 58 (1987) No.3 115

Process metallurgy

Again, the calculations are summarized, see table 15. Data under well defined conditions. The different possibilities of
published by the manufacturer are interpolated from viscos- this model are:
ity measurements made in Irsid laboratories. The good
agreement obtained here is a control of the hypothesis made - estimations used as a first approximation,
for the calcium fluoride behaviour in this slag. Some other - an aid for planning experiments and avoidance of numer-
examples can be found in 12). ous preliminary trials,
- a test for structural hypothesis, and
- viscosity-composition-temperature equations compatible
Conclusions with micro computers.

The scope of such an experimental model has to be as sim- (A 00065; received: 8. August 1986
ple as possible in order to give estimations of slag viscosities in revised form: 17. November 1986)

References

') Bottinqa, Y.; Weill. D. F.: Am. J. Sci. 272 (1972), p. 438/75. '0) Urbain. G.; Cambier, F.; Deletter, M.; Anseau, M. R.: Trans. J. B. Cer-
') Shaw. H.: Am. J. Sci. 272 (1972), p. 870/93. am. Soc. 80 (1981), p. 139/41.
J) Weymann, H. D.: Kolloid Z. Polymer 181 (1962), p. 131/37. ") Ikeda. K.; Tamura. A.; Shiraishi, Y.; Saito. T.: Bul. Res. Inst. Min.
4) Bockris, J. O. M.; Reddy. A. K. N.: Modern Electrochemistry, v. 1, Tohoku 29 (1973), p. 24/36.
chapter 6. Plenum Press, New York, 1970. 12) Urbain. G.: J. Mat. Education 1985 Vol. 7 No.6, p. 1007/78.
5) Urbain. G.; Bottinqa, Y.; Richet, P.: Geochim. et Cosmo. Acta 46 13) Urbain. G.: Rev. Int. H. Temp. Refrat. 21 (1984), p. 107/11.
(1982), p. 1061/72. 14) Urbain. G.: Rev. Int. H. Temp. Refract. 20 (1983), p. 135/39.
6) Urbain. G.: Rev. Int. H. Temp. Refract. 19 (1982), p. 55/57. IS) Bernard. G.; Riboud, P. v.; Urbain. G.: Rev. Metallurgic 78 (1981)
7) Urbain, G.; Millon. F.; Cariset, S.: C. R. Acad. Sc. Paris 290 (1980), p. No.5, p. 421134.
137/40. 16) oa« I. N.; Cambier, F.; Deletter, M.; Urbain. G.: B. Ceram. Trans. J.
') Bockris, J. O. M.; Mackensie, J. D.; Kitchener, J. A.: Trans. Farad. 83 (1984), p. 105/12.
Soc. 51 (1955), p. 1734/48. ") Piwinski, A. J.; Weed. H. c.: Thermochimica Acta 37 (1980), p. 189/95.
9) Panek. z.. Silicaty 22 (1979), p. 97/102. IS) Murase. T.; Mcliirney, A. R.: Science 167 (1970), p. 1491/93.

Letters to the Editor

Letter to the Editor from Professor Dr.-Ing. H. Berns, Ruhr- tool steel is of a hot working type and in service carefully
Universitiit Bochum, FRG preheated to > 150°C. This enlarges the plastic zone size
several times, thus enhancing void formation and dimple
Concerning: Pacyna, J.: Effect of non-metallic inclusions fracture in CT specimens. Under these conditions the frac-
(NMI) on fracture toughness of tool steels. steel res. 57 ture toughness, in transverse specimens at 150°C, is for in-
(1986) No. 11, p. 586/92. stance raised by about 50% if the sulphur content is low-
ered from 0.0070 to 0.0027% via a Ca-treatment*). As the
The author's last conclusion is that "application of argon- author tested longitudinal specimens at room temperature
ing, electroslag remelting or vacuum which radically de- only, his results are misleading as far as the effect of refining
crease the NMI fraction is useless as regards fracture tough- processes on the fracture toughness is concerned. For ESR
ness; there is only an increase in the cost of steel produc- steel with a hardness of - 460 HY a 40% increase was mea-
tion". sured at 20°C in longitudinal and transverse specimens*).
(ALOOOl)
In my opinion this is a strong statement based on insuffi-
cient evidence. The three melting procedures are compared
*) Wendl, F.: Influence of production procedures on the microstructure
at a hardness (500 to 575 HY) and a testing temperature and toughness of hot working too) steels with 5% Cr. Dr.-lng. thesis,
(presumably 20°C) where quasicleavage prevails. Here, of Ruhr-Universitat Bochum, 1985; see also: Fortschr. Ber. VOl Reihe 5,
course, little influence of NMI is to be expected. But this Nr. 91, VOl-Verlag Diisseldorf 1985.

116 steel research 58 (198?) NO.3