Академический Документы
Профессиональный Документы
Культура Документы
140-147
Retsu NAGABAYASHl,1)
Mitsutaka HIN02) and ShirO BAN-YA2)
1) Graduate School, Tohoku University Nowat Research and DevelopmentLaboratories, ShunanWorks, Nisshin Steel Co., Ltd.. Tonda,
Shin-nanyo, Yamaguchi-ken, 746 Japan. 2) Department of Metallurgy, Faculty of Engineering. Tohoku University, Aza Aoba. Aramaki,
Sendai, Miyagi-ken, 980 Japan
Phosphorusdistribution between slag and liquid iron has been studied at the temperature range from 1573 to 1953 K.
The slag system of FetO-P205-M.Oy(M.Oy= CaO, MgO,Si02) ternary, and of FetO-P205-CaO-M.Oy(M.Oy MgO,
=
Si02) quaternary were studied to clarify the effect of oxides on the phosphorus distribution equiiibrium at steelmaking
process.
In view of ionic theory, the approximate validity of regular solution model
was examined to formulate the equilibrium
reaction of phosphorus distribution between slag and metal. As the results, it was confirmed that the regular solution
model was satisfied for all the experimental results including the present work and the previous stvdies by other inves-
tigators, except for extreamly high iron oxide region The phosphorus and oxygen contents in liquid iron in equilibrium
with slag can be estimated within the accuracy of ~:10 "/• by the quadratic formalism derived from an assumption of the
regular solution of slag
from a reaction chamber, and then was quenched established at the sametime.
into water bath to solidifv the liquid iron as quickly
as possible. After that, the crucible was immediate- + Q= (FeO)
Fe (1)
""" """"""
"'(3) (i* *1~g)
The MgO crucible was bought from market, and Manymodels have been proposed to express the
the Si02 crucible wasmadefrom pure silica glass tube. ionic nature of liquid slag. The regular solution
The CaOcrLrcible was fabricated from the calcined model of slag has proposed by Lumsdenl9) in 1961.
powder of' 10,/.,FeO-CaO mixture, by stamping with In present work, an attempt has been madeto quan-
the binder oflinseed o_ilkerosene (1 : l) solution, and titatively represent the equilibrium relation of phos-
by firing with C2H2-02flame at 1973 K. According phorus distribution in steelmaking process as a func-
to the recent phase diagram of Ca(_)-FeO system, the tion of temperature and slag composition by the
solid CaOmakes a solid sol_ution containing about model, in continuation of our previous studiesl8,20) on
lO o/o FeO at steelmaking temperature. Therefore, the oxygen distribution in steelmaking process. The
pure wustite powder was added in calcining process activity coefficient of a component i in a multicom-
to stabilize the solid CaO. The CaO-FeO crucible ponent regular solution is expressed by the following
madeby technique was durable for about 3.6x
this equation.
work, but was unable to be used in
l03 s in present
the measurementofvery low oxygen potential. G~=AHi=RTlnr"('.'Sf=0) ..............
(8)'
. .
Since the phosphorus dissolved in liquid iron, [Po], where Xj : the cation fraction of the componentj in
is oxidized to (P5+) in the oxidizing process of steel- slag
making, it has been considered that the reaction (Yij : the interaction energy bctween cation i-
product in the dephosphorizing of iron with liquid O-cation j.
slag
maybe phosphorus pentoxide (P205), tricalcium Whenthe of Eq. (8)' is valid to express the
relation
phosphate (3CaO.P205) or tetracalcium phosphate activity of cornponent i in the experiments,
coeflicient
(4CaO.P205) in the molecular theory of slag. On the quadratic formalism of Eq. (8)' can be applied as
the other hand, it is well-known that the liquid slag an approximate expression of activity coefficient of
is
one of the ionic melts, and that the phosphorus in componenti in slag, even if the liquid slag is not real
liquid slag exists as the species of phosphate anion regular solution (S~~~O). This is the similar idea to
(P0~-) containing one atom of phospho_rus. Even in the quaclratic formalism of metallic solution proposed
the ionic theory of slag, however, an equilibrium con- by Darken.21)
stant can not be derived without making some as-
sumption, because of the lack of the enough knowl-
edge concerning the ionic naturc of thc liquid slag.
4. Distribution Ratio of Phosphorus
Moreover, it is theoretically diflicult to determine the Manyof the experiments in the present work were
activities or activity coefficients of ions in the fused carried out over a wide range of FetO content and at
Therefore, in the present work, the hypotheti-
salts.
cal species (P02.5) containing one atom of phosphorus
3 temperatures of 1823, 1873 and 1923K. The
phosphorus distribution ratio is a function of the tem-
was assumedas the reaction product to represent the perature, oxygen potential of the system and the com-
dephosphorizing reaction of liquid iron in view of position of slag, as seen in Eqs. (1) and (5). The
ionic theory. The samcidea is also va,lid in the regu- values of the distribution ratio (o/_nP)jr~/oP] obtained
lar solution model oi' slag described in the later. The at 1873 K
were shownas a function of FetO content
equilibrium relation of phosphorus reaction in the in Figs. to 3. IFig. I
shows the results of FetO-
steelmaking process can be written as follows; P205-M.Oy(MwityhO CaO MgOSIO ) isternary sys
tems saturated oxide M*Oy, which the cruci-
~+2.5Q= (P02.5) (i* sl*g) """""""""(1)
ble material. The distribution ratio in FetO-P205-
Kpl = ap02.*/(ap 'a~'5) ........................(2)
.
CaOslag were the highest among3 ternary systems
In the equilibrium state of the system, the oxygen studied, and the maximum value of the distribution
distribution between slag and liquid iron should also ratio was observed at the range of (massol)FetO)=
be in equilibrium. The following equilibria are lO-20. The maximum phosphorus distribution ra-
141
ISIT International, Vol. 29 (1989), No. 2
4
Fet
O o + FetO-P205--CaO(Peter et al.)
3
~ ~~++
~~ ~~+
~V~~~~O
O ~e
-p2 ~e~
d'
Ca
S
~(,qa t. O e FetO-P205-CaO(Knuppel et al.)
~ ~~~~\
(,
V FetO- P2O5-CaO(Fischer et al.)
~~l I~ 2 ~ e FetO-P205-CaO
?1•*
FetO- P205 - A FetO-P205-MgO Trdmel et al.
Ngosaf I FetO-P205-Si02
1 A~-A~A_A
_
.
Fig.
---- FetO-P205
1.
4 A
massell CaO
mass'/.h4gO
F:eto FetO - CaOsat
3 " F)205
P2 -
OO * 5
ll!~'~~_1 " caoset 3 l$
5 * 10
\'~t~~o
~/z::/i0~2~~\:~t ,h~ 20
D 15 - 25
~L
~~ 2 v =,' Q
2
;~^ A 75
O
,~,,,,
_~9"~C:8~)
\ v:218
t""
,,
a, o
Al o
, /d~!Q
'\
ll
a) O
~
~ >~6
lo
BalaJjva et al CID
o 25
C1, 1
O O mass'l.CaO
V= mass'/. O2
Si FeO-PO
t 25 -SiO 2 sat FetO-P205-MgOsat.
/
o ~2 o
-1
2-3 o 10 20 30 40
A 3~
mass'/. Fet O
-2
Fig. 3. ratio between liquid iron
Phosphorus distribution
O 20 40 60 eo
masso/oFetO and FetO-P205-CaO-MgO slags saturated with
(Mg, Fe)O at 1873 K.
Fig. 2. I)hosphorus distributionratio between liquid iron
and FetO-P205-Si02-CaOslags equilibratcd with
solid 2CiaO.Si02 at 1873 K.
From the above experimental results, it will be
seen that of oxides increasing the phos-
the ability
phorus distribution ratio increases with the order of
tios in each slag system
were L~::~1 OOOin FetO- Si02 FeLO MgO
CaO. However, the sameprop-
P205-CaOslag, L~~(lO FetO-P205-MgOslag, in erty of desulfurizing of iron is the order of Ca0>>
and L~ I in FetO-P205-Si02 slag, respectively. The FetO> MgO > si02'
approximate value of Lf, in FetO-P205 binary slag It
was conflrmed that the phosphorus distribution
will be given by extrapolation of L~-values in 3 each ratio decreased with increasing temperature, but the
slag system to 100 massO/oFetO in Fig. l. The dis- results obtained were omitted from the limited page
tribution ratio in FetO-P205 binary slag of the paper.
was esti-
mated to be L~c~ 1. The distribution ratio in FetO-
P205-Si02-CaOsystem saturated with 2CaO•Si02 5. Equilibrium of OxygenDistribution
is shown in Fig. 2. This series of the measurement
was conducted usin" solid CaOcrucible but the in- In our previous papers, 18,20) it was confirmed that
terface between th~ solid CaOand the liquid the regular solution model of slag was satisfactory for
was
always covered with 2CaO.Si02, as was expected predicting the oxygen content in metal in the equi-
from the phase diagram of CaO-FetO-Si02 system. librium state of steelrnaking process over the wide
The distribution ratio in this slag increased with in- range of silicate slag composition. To establish the
creasing basicity defined with (masso/oCaO)/(massO/o relation for the dephosphorization in steelmak-
same
Si02), and an optimum FetO content, at which the ing process, the validity of the model for phosphate
value of Li, is maximum,was observed at the com- slag should be conflrmed, and the interaction energies
position of (massO/oFetO)~:20. As shown in Fig. 3, between cations and phosphorus ion such as (x(Ca2+_
thc distribution ratio in FetO_-P205-CaO-MgO satu- p5+), (r(Mg2+_p5+) and (t(Si4+_p5+) should be cleter-
rated with (Mg,Fe)O increased from I.j,c::: 10 to L' ::: mined if the model was valid for the data obtained.
l OOOwith increasing CaOcontent, and it was clear The experimental results of oxygen distribution equi-
that CaOand MgO
were not equivalent for the de- libriurn was applied for this purpose.
phosphorizing of iron in steelmaking process. The equilibrium relation oi' FeO-Fe203-P205-CaO
142
ISIT International, Vol. 29 (1989), No. 2
(~!45(Xc*o/Xp02'5) Fe3+_Na+
- 74 890 25)
Fe3+_Mg2+ -2 930 20)
(1
/Ypo, ,) {R T In (XF.o/ao) ~ 18660X~•.o,.5 Fe3+_Ca2+ -95 810 20) I02 510 Fujita et al.
31 380X~02'5~31380X{2.~o -
Fe3+_Mn2+
- 56 480 26) -12 550 Suito et al.
E YA(J) .......
.........(ll) Fe3+_p5+ + 14 640 22)
linear function of the term of (Xc'o!Xpo,.,), Passing Mg2+_Ca2+ Ioo 420 20)
through the origin.
-
Mg2+_Mn2++61 920 27) -23 850 Suito et al.
The oxygen distribution ratio, L0=JVF.to![o/oO], ob- +38 910 Suito et al.
Kntippel al.5) and Trdmel and Fritre,6) and Ca2+_Ti4+ 360 Sommerville et al.
al.,4) et
133 890
- 138 490 Fujita
obtained by Fischer Ca2+Si4+ 20) ei al.
the oxygen distribution ratios - Prcsent
-
and Ende3) were a little lower than those of others. Ca2+_p5+ -2~)1 040
work
Onthe other hand, the phosphorus distribution ratio Sommerville
Mn2+_A13+
- 20 920
et al.
in the samesystem obtained at present work agreed
well with those by Fischer and Ende,3) Kntlppel et Mn2+_Ti4+ -66 409. Sommcrville et al.
Present
through the origin. The slope of a line is the inter- Si4+_p5+ +83 6ao work
action energy between Ca2+ and P5+. The following
interaction energies between cations were given by RTIn rF*o
the sameprocedure.
= - 18 660X~.oi 5~31 380X~o,.5
a (Ca2+_p5+) = _251 040 -41 840X~2io,~3 1380X~*o
a(Mg2+_p5+)=
_ 37 660
(J) .........(12) +33 470X~Igo~64 680XF.ol " Xp02'5
a(Si4+_p5+) + 83 680
= +45 770XF.o, 5 Xo'0+ 17740XF.o, , Xltgo
' '
From the above results, the activity coeflicient of -93 140XF.ol 5'Xsio,+ 188 280Xp02"Xc'o
FeO in the FeO-Fe203-P205-Si02-CaO-MgO slag, +39 750Xpo, , 'XMgo~ 156 900Xro, , 'Xsio,
which are major componentsin steelmaking slag, can + 102 5lOXc'~o'XMg0+60670Xc'~o'Xsio,
be written as follows; +58 570X~lgo'Xsio, '
(J) .(13)
143
ISIJ International, Vol. 29 (1989), No. 2
8 1.0
o Present work /' : oo ,
6
D KnOppel et
e Tr6rnel
Fischer
et ai
et ai
al
eCI
o
/D AA
l
0.5
v
xyJ)~)
~A
DC~;~0 ,d' A o
+ Peter et a[
~~
J04
'D:n
D
8 o
d,
C!D
o
r)
ol
o
o
ldi~;. /
+0 ~
2 CD
-
o.
5
(:
o
O 0.1
O2 03 0.4 O5 O6 10
NFetO
-1.0 -o 5 o o5 1.0
J,
C]
I Present work
C!
Fig. 6. Comparison between measured and calculated log
> _ 750
~0' CZlee o rFeto'
Xcao
O ' YA= 251040 (J)
' Xp025
-1000 O
o C]
6. Equilibrium of PhosphorusDistribution
O
O
- 1250
Fromthe above results, the equilibrium relation of
O 2 4 6 8 10 12 dephosphorization with FeO-Fe203-P205-Si02-CaO-
XcaO/Xp02.5 MgO slag can be written as follows;
Fig. 5. Relation between YA and XcaO/Xp02. in FeO-
FcO1.5-P02.5-CaOslags.
~+2.5Q= (P02.5) (i* sl*gl """"""""""""""""'(1)
RT In Kpl = RTln {Xro,.*/(ar' a(2j.5)}
+RT In rp02.5
Since the above equation quadratic formalism, the
.
(17)
minor componentswhich are less than 5mass~/o can
is
RT In rp02.5
taken to the hypothetical stoichiometric FeOin which -25 1040XF.o ~ 102 51 XMgo
Xc'*o OX},.o
' '
is the
pure iron oxide in equilibrium with metallic -20 090XF.ol" XMg0+ 65 680Xl'o, Xsio,
'
5
'
FetO(~+(1 -t) Fe (s or l)
= FeO
(1, R.S.) According to the all ofexperimental results, the
range
.(14) of slag composition covered the wide range of JV:F*t0
AG'14 = -8540+7.142T (J) ....... ........(15) 0.5 and JVI"o, in whicll the relation of Eqs. (17)
RTIn . ~ o RTln
al aF~0~8 540+7. 142 T (J) and (18) is valid to evaluate the equilibrium constant.
.
144
ISIT International, Vol. 29 (1989), No. 2
by the sameprocedure was shownin F'rg. 8, in which model by Eq. (21 ), because the oxygen content in iron
the data reported by other 5 groups of Winkler and in the equilibrium was not determined in their mea-
Chipman,i) Kntppel et al.,5) Trdmel et al.,6,8) Suito surement. As shown in Fig. 9, the points of data
et al.,9) and Kor28) were also shownby the recalcula- recalculated from the results by Balajiva et al. at
tion. The results obtained by different kinds of slags l 858 Iie on a horizontal line of log K/'2 (1 858 K)=
K
-0.76 within experimental error. Tllerefore, the va-
and by many different investigators were in fairly
good agreement each other, and the following equa- lidity of the model and the accuracy of Eq. (21) were
tions were given. also verified from the data by Balajiva et al. The
effect of tcmperature on the equilibrium constant of
p+2 50 =(PO2.5 )(i* *1*g) """"""" Eq. (5) were shownin Fig. lO, in which the results by
log Kpl_ 17060/T-8.510(a = 0.4) ..... .......(19)
_
'
Where, the standard phosphorus oxide in state for Fig. 10 were a little scattered in comparison with
slag is pure liquid hypothetical P02.5 under an as- those in Fig. 8, but be seen that they fairly
it will
sumption of' regular solution, and those for oxygen agreed with each other within the experimental error.
In Fig, I l, the phosphorus distribution ratio L~ cal-
and phosphorus being I o/o by massin iron.
The equilibrium relation of phosphorus in iron culated from the model was plotted against the mea-
with iron oxide in slag can be derived as follows by sured I., f'or the data by present authors and other in-
the combination ot' Eqs. (4) and (20). vestigators.1,2,5,6,8-10) The agreement between both
calculated and measuredvalues was also good within
f n I +2.5Fe(~
E+2.5(FeO)(i**1*g)
= \Pv2.5' (i* *1*g) the experimental error.
.(5)
As similar to the case of the activities of FetO and
FeO in slag, the conversion factor between the ac-
= 328/T-0.936((r =
log Kp2 0.4) ..................(21)
D Treme[ et aL -e
-6
e KnUPPet et a[.
-8 2.
-8 o 25 30 3.5 4O
1 8 9 (XcaotXb,1gO~XFeO)'( Xsi02'X p02 )
(XcaO+XFeO)! Xp025 5
Fig. 9. Relation betwcen logKp2and slag basicity in FeO-
Fig. 7. Relation between log Kpl and (Xca0+Xl'*o)/ FeOl.5-1)02.5-Si02-CaO-MgO slags equilibrated
Xpo,.* in FeOFeO1.5P02.5-CaO
slags equilibrat-
with liquid iron and solid 2(Ca, Mg)O•Si02 at
ed with liquid iron and solid CaOat 1873 K. l 858 K froln the data of Balajiva et al.
Temp,K Temp K .
2000 1900 1800 1700 1600 aooo l900 1800 wOO 1600
8 8
present work present work
6 ~ Feo- Feoi 5- PO2.s ~ cao sat.
~Feo - Feol 5- p02s ~ SiOi - Cao 6
al ol o ~~,~~- It,
~:
J90
~) _{
o o -2 o
-2 ~ Tr5mel et a[. log Kpl = 17060/T -8.510
~ Trdmel al
~ Suito et al.
_jil~~~~~
~ Kor
~HnUppel et al. ((1=0 ~ ) -4
et
-6
~ Winkler et aL
~Ror -e
~ Hn~ppel et aL,
~T
Ba[aiiva et al.
-8
~ Suito et al.
-8
~ Winkler et al.
~ Shirota et al.
O50 o s5 o 60 o 65 o 50 o.55 o 60 o 65
T~t/ 10~aK~t T~1
/ I 0~3 K~1
Fig. 8. Temperature dependenceof log Kpl" Fig. 1O. Temperature dependenceof log Kp2'
145
ISIT International, Vol 29 (1989), No. 2
k 9
3 -, 1
IS I
I
I
-CL ;~]D o
J 2
x~/~~]
Ix ~
Olo
Lr,
-13
~(V'
,i O
QJ
I' ~L
aJ 1 t~
rU
Al oooo O1 . ~~
l/V'( O _1 5 ov
l~i O I' O
U ~L
-1
-1 7 ~~
IPv
-2
1 O 1 2 3 ~
-1 9
-19 -17
-
Measured [ogL~ -
15 -13 -1 1 - 9
O Kndppel et al.
A Present work + Iog (Xl~02.5/JVp*0=) + 2754/ T- 12.05
O Trt)mel et al. FetO-P205-Si02-CaO FetO-P205-CaOsat. FetO-P205-CaO-MgOsat.
e Prescnt work (2CaO.Si02 sat.)
FetO-P205-CaO
O Kntippel et al. A Prcsent work
V Present work o Tr6nlel et al. FetO-P205-Si02-CaO
d) Shirota et al. FetO-P205-Si02-Ca0=MgO Present work (2CaO.Si02 sat.)
FetO-P205-MgO
s~t.
e
sat . ~ Winkler el al.
FetO-P205-CaO V Present work
x Present work V Balajiva al.et al.
FetO-I]205-Si02-CaO-MgOsat.
Fet O-P205-SiO2 sat. O Shirota et al.
A Suito et FetO-P205-MgO
sat. (~ Winkler et al.
[] Tr6mel et al.
x Present work V Balajiva et al.
l Present work Fet O-P205-SiO2 sat. A Suito et al.
Acknowledgements
In Fig. 12, the activity coefncients of P205 directly
derived by applying Eq. (24) to the present data and The authors wish to thank former undergraduate
previous data by other investigatorsl,2,5,6,8-10) students at Tohoku University-Messrs. M. Tobita
were
comparedwith the values estimated from the model and H. Yamaguchi, now Spring Co., Ltd.) Mr. HNK
by using Eq. (25). As shown in Fig. 12, the most T. Tanemo, now Nisshin Steel Co., Ltd., Mr. T.
points of the data showedgood agreement within the Megumi,now Nihon Unisys, Ltd., and M. Mukaida,
experimental error. now Graduate school, Tohoku University-for their
effective assistance in making their experiment.
In the past, the slag models, proposed by Flood
and Grjotheim,12) and Turkdogan and Pearson,13)
REFERENCES
were enough accurate to predict the phosphorus con-
tent in metal and the phosphorus distribution ratio 1) T. B. Winkler and.1. Chiprnan: Trans. Metall. Soc. AIME,
between slag and metal for basic slag, but the models 167 (1946), 1.
11
of' the
2) K. A. G. Quarrel and P. Vajragupta:
Balajiva, J. Iron
same idea were no_t successful to predict the 153 (1946), I 15.
Steel Inst.,
oxygen content in metal and the oxygen distribution K. Balajiva and P. Vajragupta: J. Iron Steel Inst,, 155
between slag and metal. However, in the approxi- (1947), 563.
mate application of the regular solution model, the P. Vajragupta: J. Iron Steel Inst., 158 (1948), 494.
both oxygen and phosphorus contents in metal and 3) W. A. Fischer and H. Ende: Stahl Eisen, 72 (1952), 1398.
the distribution ratios of oxygen and phosphorus can 4) O. Peter, W. Esche and E. Oelsen: Arch. Eisenhtittenwes.,
be estimated within the error of ~ 10 r/o over the wide 27 (1956), 219.
146
ISIJ International, Vol. 29 (1989), No. 2
G. Tr6mel and E. Fix: Areh. Eissenhattenwes., 33 (1962), 22) S. Ban-ya and T. Watanabe: Tetsu-to-Hagard, 63 (1977),
745. l 809 .
Suito, R. Inoue and M. Takada: Tetsu-to-Hagani, 67 24) S. Ban-ya and R, Nagabayashi: Tetsu-te-Hagani, 63 (19S2),
9) H.
(1981), 2645; Trans. Iron Steel Inst. Jpn., 21 (1981), 250 & 261.
22 (1982), 705. Ban-ya, T. Watanabe and R. Nagabayashi: Japan-
S.
H. Suito and R. Inoue: Tetsu-to-Hagand, 68 (1982), United States Seminar on Advancesin the Science of lron-
1541. and Steelmaking, Japan Soc. Promotion of Sci.. Tokyo,
10) Y. Shirota, K. Katohgi, K. Klein, H.-J. Engell and D. (1983), 27.
Janke : Trans. Iron Steel Inst. Jpn., 25 (1985), I 132. 25) S.Ban-ya, M. Hino and H. Takezoe: Proc. Int. Symp.on
ll) P. Herasymenkoand G. E. Speight: J. Iron Steel Inst., 166 Metall. Slags and Fluxes, ed. by H.A. Finc and D. R.
(1950), 169. Gaskell, TMSIAIMF., Warrendale, (1984), 395; Tetsu-to-
12) H. Flood and K. Grjotheim: J. Iron Steel Inst., 171 (1952), Hagand, 71 (1985), 1765 & 1903; Trans. Iron Steel Inst. Jpn.,
64. 25 (1985:), 1122.
J. Iron Steel Inst. , 175 26) S. Ban-ya, M. Hino and N. Yuge: Tetsu-to-Hagani, 71
13) E. T. Turkdogan and J. Pearson:
(1953), 398. (1985), 853.
G. W. Healy: J. Iron Steel Inst., 208 (1970), 664. 27) S. Ban-ya,M. Hino, N. Yuge and M. Ejiri: The 19th
14) (Steelmaking), thc Japan Soc. Promotion of
15) T. Kai, K. Okohira, M. Hirai, S. Murakamiand N. Satoh: Committee
Tetsu-to-Hagand, 68 (1982), 1946. Sci., Rep. No. 10748, (Oct., 1986).
16) D. R. Gaskell: Trans. Iron Steel Inst. Jpn., 22 (19a2), 997. 28) G.J.W. Kor: Metall. Trans. B, 8B (1977), 107.
17) D. Guo: Arch. Eisenhtittenwes., 55 (1984), 183.
18) J.-D. Shim and S. Ban-ya: Tetsu-to-Hagani, 67 (1981),
1735 & 1745; Can. Metall. Qjuart., 23 (1983), 319.
(Originally published in Tetsu-to-Haganl, 74 (1988), 1577, 1585,
: Physical Chemistry of Process Metallurgy,
19) .T. Lumsden I,
147