Académique Documents
Professionnel Documents
Culture Documents
Mats Hillert
and Magnus Jar1
Division
of Physical
Metallurgy
Royal institute
of Technology
S-100 44 STOCKHOLM 70
Sweden
Abstract
A mathematical
representation
of the magnetic
specific
heat,
recently
suggested by Inden,
was applied to iron in an evaluation
of the difference
in Gibbs energy between the fee and bee
states.
The resulting
equations
were then used for a treatment
of alloying
effects
in ferromagThe result
was approximated
in order
netic
metals due TV the change of the Curie temperature.
term was obtained.
to conform to the subregular
solution
model. A strong asymmetric
Introduction
It was pointed
out by Zener (1) long ago that the effect
of an alloying
element on the
base metal should result
in a strong thermodynamic effect,
magnetic
state
of a ferromagnetic
This effect
has recently
attracted
considerable
attention
(2-4)
but the results
t&ke complicated
analytical
forms. An attempt will
now be made to develop a treatment
in a simpler
form.
In particular,
an attempt
will
be made to put the result
into the regular
or subregular
type of representation.
The work will
be based upon a type of description
of the thermodynamics
of a ferromagnetic
metal,recentfy
developed by lnden (3) and it wiI1 be applied
to iron and iron base
at ioys.
The regular
and subregular
solutian
models are useful
tools for
of the thermodynamic
properties
of binary alloys.
They are particularly
cerned with the coupling
between thermodynamics
and phase equilibria
sive use in that field.
Description
Provided
that
use a polynomial
state5
of
a pure
the temperature
for the description
element,
AG = A + BT 4 CTlnJ
of
Gibbs
energy
is much higher
than
of the difference
for
approximate
descriptions
valuable
in work conand they have found exten-
iron
+ DT2
to
flf
The ferromagnetic
metals constitute
an important
exception
and it has been customary to present
the result
of evaluations
of Gibbs energy for iron in Tables
(5-11).
Kaufman and Nesor (12) who
used an expression
like eq. 1 had to give different
parameter
values above and below the Curie
temperature.
lnden (3) has recently
shown that the following
types of expressions
can be used for a
rather
accurate
description
of the magnetic
contribution
to the specific
heat of a ferromagnetic
metal A
_1
ym! = KzRln -!.%?
cA
for
T <
for
T >
l-2
5
CrnB
= K;Rln
The quantity
7 -1
T is defined
as T/TC
where
TC is
the
Curie
temperature.
Ki and Kt are
two constants
228
for
= 2KaR(~3+,g/3+~15/5)
A
m8 = 2K;R(T-5+~-5/3+~-25/5)
CA
for
T < 1
(4)
for
T > 1
(5)
They approximate
eqs. 2 and 3 very well except for a temperature
range very close to T . Eqs. 2
and 3 go to infinity
at T which is in agreement wit $ many theories
pheno &na.
of ordering
Eqs.
respectively,
and may thus be an attrac4 and 5 approach finite
vglues,
46/15RKa and 46/15RK
fi.
tive alternative
from this point of viea,
also.
The erght at T, can be adjusted
by keeping
wi 11 be Test if many tel;ms are retained.
more terms but some of the advantage with eqs. 4 and
finding
KFe=0.554
and
lnden evaluated
the numerical
values of Ka and Ki? for bee iron
KFe=0.714.
According
to his procedure
eqs. 4 affd 5 yifild
the following
result.
T
B;(=)
- G;(B)
= 6
m8
c CM
$dT
+ 7 $dT
= gR(K;
(6)
+ 0.6K;).
TC
- H~~TC) = 7
$(=)
mB
cAdT = ~~CK~
(7)
TC
- H;(D)
H,m(T,)
= ILcrmdT = GRTCK;
0 A
Let f be the fraction
of the total
magnetic
ture.
Eqs. 7 and 8 then yield,
(8)
enthalpy
which
is absorbed
above
the
Curie
tempera-
(9)
When analyzing
the specific
heat data for bee iron,
inden found f=O.4.
For fee metals he found
f=O.28.
He suggested
that the f value depends upon the structure
and proposed that 0.4 could
be used for all
bee metals and 0.28 for all
fee metals.
The total
magnetic entropy
is often expressed
by the following
expression
where 8,
is the
mean atomic moment expressed
in Bohr magnetons (9),
q-4
- St(O)
Combination
with
= Rln(BA
eq.
+ 1)
(0)
6 yields
(11
By combining
Kfi
=
A
K; =
this
5ii8
with
eq.
9 and using
the value
f=O.4
we obtain
for
bee metals,
n(OBA + 1)
= 0.64t71n(BA
(12)
+ 1)
518.
1125
o.g80~(~A
+ 1)
of 0BA=2.22
(3)
yields
KE=1.073
and Ki=O.7504.
For
fee metals
the
value
Ki = 0.426Bn(BA
+ 1)
(14)
K; = .o46gn(BA
+ 1)
(15)
zero
When evaluating
the magnetic contribution
at Td.
For temperatures
above the Curie
IF
A
= *K;RTC[/10r4
By continuing
tic
to) state.
om
GA
the
+ [35T
integration
8
= -KART&79/140
define
this quantity
as
expression
is obtained
(17)
+ /YOT241
TC one abtains
below
- 58~/125J
-K;BTC[.4/6
the
folIowing
expression
for
the
ferromagne-
~~D/l~~~r6/600+71./120-51~~/675]
(19)
in
The difference
the
Below
Curie
OGY
- G$
Fe
temperature
*
3883.4
we obtain
+ 36.07
- 6,3TnT
+ D.0D24T2
+$309[T4f6T~+TD/35~~+T16/600T~6)
J&no
(21
rkese exprassions
a value of TC=lOQK
must be used,
The function
given by eqs. 2D and 23 was evasrated n~~~~~~~~
and in Figs,
t and 2 it is
compared with the evaluations
present&
by Orr and ChIpman f]
and by Kaufman and Besot (2).
Above the trans~~~rn~t~on
point at il85K the new evaluatian
agrees wet1 w3th Grr and Chipman
and below the transformation
it agrees well with Kaufman and Nesor. This is partcuiary
satisfactory
because they based their
evaluation
on the previous
evaluation
by Kaufman, Clougherty
and Weiss (14) who used a realistic
astIrnate
of the Debye temperature
for fee iron.
Eqs. 20 and
21 thus seem to give an adequate
representation
of the Gibbs; energy for iron,
Wwever,
they
shauld not be usc?d much below the Debyo temperature.
fn
ho&
Previous
treatments
of
the magnetic
aloying
effect
QS;edTC/dx
(221
230
--
-.-
Koutman
awl
--
0~
-.-
Kc&non
and
Now
Fig.
1 Difference
in Gibbs energy
between FCC and BCC iron in the
temperature
range 1000-1800
K
according
to different
works.
Fig.
2 Difference
in Gibbs energy
between FCC and BCC iron in the
temperature
range 300-1lOD
K
according
to different
works.
Chipman
and
Nuor
500
lOOa
231
contribution
to the Gibbs energy of an iron-base
alloy
in the bee state
the magnetic entropy
in pure bee iron.
Z&ers
suggestion
was somewhat arbitrary
and cannot be strictly
justified
on thermodynamic grounds.
A formally
more satisfactory
treatment
could be based on any mathematicl
model of
Such treatments
were recently
discussed
by Miodownik (4).
HOWthe magnetic
properties
of iron.
ever,
apart
frcnn the work of HWWno treatment
has yet been put in analytical
form.
The change in the magnetic
contribution
to the Gibbs energy due to the addition
of an
alloying
element can be written
in the following
general
form
a& OSrnis
AG; = G; - xAoG; -
xBG;
(23)
The superscript
m is here used to denote magnetic contributions
whereas the subscript
m denotes
lnden (16) has suggested
that one should use the same type of demolar integral
quantities.
inserting
the
acription
for the magnetic contribution
in an alloy,
Gm, as in a pure elementoby
8,. An al ternamean value of the atomic moments for the mixture
of elements
in the alloy
as
It is based upon a separation
of the effects
from
tive will
be examined in the present
paper.
different
elements.
For a binary
alloy
we shall
write
(24)
Zeners
original
suggestion
one should
This equation
can be used in different
ways. Following
neglect
the direct
effect
of the alloying
element and describe
the magnetic contribution
from
the base metal by displacing
the Gibbs energy function
for the pure base metal along the temperature
axis by the same aa-ount AT as the Curie temperature
has been changed.
AG; = xAtoG;(T-AT)
- G;(T)
(25)
HW approximated
this expression
by the first
term in a series
expansion
and assumed that the
This method actually
yields
the
Curie temperature
varies
linearly
with the alloying
content.
following
expression
which has sometimes been used (17)
AGE = -ATxAdoG;/dxB
= xAxBoS;(T)dTCA/dxB
(26)
The magnetic
alloying
effect
thus takes the form of the regular
solution
model although
the
temperature
dependence is quite
unique.
Nishizawa
et al.
(18) have emphasized that it may be essential
to retain
another
term.
This is particularly
evident
if one wants to calculate
a miscibility
gap. They derived
an equation for the spinodal
in the following
way by putting
the second derivative
of the Gibbs energy
equal to zero.
From eq. 25 one obtains
the following
expression
for the case where the Curie
temperature
varies
linearly
with the alloying
content,
d2AGm
dT
dx;
The following
d2Gm
-=
dx2,
d2'Gm(T-AT)
dG;(T-AT),,,
-=2* m
equation
dXg + B
is
thus
obtained
A
dT2
for
(ddT)2
dxB
the
(27)
spinodal.
dS;(T-AT)
-2L
+ RT/xAxB
- 2'S;(T-AT)$$ - xA
B
dT
(daT)2=G
dx
(28)
where L is a regular
solgtion
parameter
whic,hmdescribes
the nonmagnetic
deviation
from ideal
solution.
The quantity
d S /dT is equal to c IT and it is thus evident
that the last term in
eq. 28 has a very strong e e feet
in the neigh 6 orhood of the Curie line where the specific
heat
goes to large values.
In fact,
Nishizawa
et al. were able to predict
that a miscibility
gap,
which is primarily
due to a chemcial
effect
expressable
by a regular
solution
parameter
L, may
develop a horn along the Curie line
if intersected
by such a line.
New approach
tive
As already
emphasized,
would be to inserteqs.
to the maqnetic
alloying
effect
Zeners
suggestion
was somewhat arbitrary
and an attrative
alterna17 and 19 in eq. 24 which could be done by using the individual
232
B = XABA+ XBB8
When other kinds of infor~tion
is lacking
an arbitrary
assumption must be made. According
to
Bates (19) it is often
reason?bie
to put G,=O for nonmagnetic
alloying
elements
in iron and to
They were now used for a comparitreat
GA as a constant.
These were also Zeners
assumptions.
son between the HW version
of Zeners
model and the neir model at low alloy
contents.
As a
basis for the comparison
the magnetic
Gibbs energy was defined
as zero at Tm. The results
are
presented
in Fig. 3. They are quite
similar
but the new model predicts
a slightly
smaller
effect.
Miodownik (4) discussed
the short-comings
of the Zener treatment
and pointed out that it
might be important
to include
the effect
of changes in the saturation
magnetisation.
For the
alloys
Fe-It% Cr and Fe-lo% Co he made a quantitative
comparison with values obtained
from his
own treatment.
In order to compare the treatment
developed
in the present
report with Miodowniks
treatment
the new treatment
was applied
to the same two alloys.
The comparison
is made
in Figs. 4 and 5. For the Fe-11% Cr alloy
the new treatment
was applied
in two different
ways.
In Model 1 is was assumed that the G value of the Fe atoms is independent
of the composition
a fair
representation
of the
and that the B value for the Cr atoms is always zero. This yields
satusation
magnetisation
of the Fe-Cr sy5tem. A more accurate
description
is obtained
with
-0.8x
which makes the b value for Fe atoms decrease
from 2.2 for pure Fe to 1.4 at a
!!$
t!Eh diifltion
in Cr. This value was used in Mode1 2. Both models were used with a value
of dT /dx
=-700 K/moi which is a very rough approximation
and does not take into.account
the
initr .E 1 I?i crease of T when Cr is added to Fe. The curve given, for the HWW treatment
in Fig. 4
For the Fe-lo% Co alloy
Model 1 may not be very realistic
was also calculated
w$th this value.
It was found that the saturation
magnesince the G value for the Co atoms is not negligible.
tisation
curve for the Fe-Co system could be represented
rather
accurately
with B =OB + 1.9X,
The curve denoted by Model 3 in Fig.
5 was obtar *i 8d wfeh
and a constant
value of B =1.7.
of 1200K for pure bee Co and with dT Idx
=1050 K/m01
these values,
with a Curl &temperature
for Fe rich alloys.
The latter
value was also used in the calculation
of th 5 cu F
ves from Hw
treatment
and from Model 1. Ail the curves are very simi iar and it may thus be concluded that
all
the models are rather
equivalent.
Figs. 4 and 5 indicate
that the change in the saturation
magnetisation
has not a drastic
effect
but in the case of Fe-lo% Co the effect
is appreciable
at low temperatures.
Sub-regular
solution
representation
A MDEL
-101
0
500
Subrrgutaf
--
~il~rt,~da,Wada
3 Comparison
contents
between
according
233
This model
--
1000
Temperature
Fig.
METALS
1500
apprax
2000
(K f
EE
x
T~m~mture
Fig.
-lrnmW
4K 1
4 Comparison of predictions
of the magnetic
alloying
effect
in an Fe-11% Cr
alloy
according
to various models.
Temperature ( K)
Fig.
5 Comparison of predictions
of
alloying
effect
in an Fe-lo%
according
to various
models.
the magnetic
Co alloy
234
vary
M. Hillert and
linearly
TC = TCA
with
Jar1
composition,
(l+kxB)
(31)
+ ax
(32)
GA = OGA
GB = OGB
+ bx
(33)
are
where OB and G
factor
ii? eq. 30*gives
ln(BA+l)
M.
the values at
the following
= In(ORA+axB+l)
z ln(OGA+l)
and a similar
approximation
is obtained
by inserting
eqs. 30 and 34 in eq.
AG; = xAln(BA+l)
+ xBg(fC)
(g(TC)
- g(TCA))
~n(*~~+l)
have
the
.2x2
axB
B
+ 2(oGA+l)2
OBA+l
value
GR+b.
The second
(34)
for ln(B,+l).
24 one oltaine
+ xAg(TC)
- x~g(TCB)ln(~~+b+l)
a2x2
B
2(Of3,+1)2 >
+ xag(TC)r
bXB
Be+
cm8
=-KBRT5,10T4
A C
A
GF
When the
(36)
- KoR[T4/6T3+T
/2-2T/3]
A
c c
-K;R[TC/2-2T/51
K values
are
derived
for
this
approximation
(37)
one finds
by the
In(OBA+l)
= -#n(bA+l)
K; =
= 0.71431n(BA+1)
(38)
{+?j($l)
Ko = ZKG = 1 07141n(B
A
2A
+l)
(39)
1 + 5kx B * 10(kxg?
= (l+kx b ) 5 =
(T,/T,,~~
(TC/TC~) 4 = (l+kxg15
(TC/TC,.jW3
= f l+kxg)-3
= 1 -
(TCJTC&-~
= (l+kx,)-3
= f + Na
=r+kx
f3
(40a)
+ LB(kxL@
(4fJb)
I + fjkxg + MRkxg
(T,IT,,)-~= (l+kxg)-3
TC/TCB
235
METALS
I - 3kxg + 6(kx,12
(40dl
La(kxB)*
(40e)
3kxN * MkxN
f40fI
cf-
=1+0
(4Og)
where
t = TIT
f4Tf
CA
It
at
is interesting
to note that all
T=T
.
T64 five terms in eq. 35 will
noted by the superscript
B because
the first
term we obtain
fromeq.
AG; = x~o~~~(T~~-T~~~~~T4
The superscript
these
correction
parameters
take
the
value
zero
at
?=l,
now be evaluated.
The values at high temperature
wilt
they wilt
depend upon the properties
of the B state.
36 by applying
the approximation
given by eq. 40a,
= -~~xgkRTGAOL~Ikl10+C3~kxgli10+4
o indicate3
that the K value shoutd be calculated
characteristic
of pure A,
8,. The value of the first
term in eq.
luated from eqs. 37 and 40d,
i.e.
be deFor
$431
from eq, 38 with the
35 at low temperature
3 value
is eva-
236
+ xAK$T4/T;A
= -xAxBkRTCAIoK;/2
116
(44)
+ oK~[3-3_r4+(6cLa)kxB?4]/6~
(451
4 - oK~LakxB~4/6}
=. x (Gm8 - OGy)
A
A
= xARTCA~oK~t5-4~-l/~4]/10
(46)
+ K;[3-4r+r4],6}
Formally,
this correction
affects
the standard
state
for pure A rather
than the regular
or
regular.solution
parameters.
On the other hand, all
the correction
terms should be included
in eq. 44 if the Curie
paerature
is increased
by the alloying
addition,
i.e.
k is positive.
AG; = -xAxBkRT~A{oK~[5-L~kxB/~4~/~0
In
this
+ oK~[3-3~4+(6~La)k~B~4]/6~
the correction
should be used below the Curie temperature
of the alloy,
TC,
by putting
Curie temperature
of pure A, TCA. Below TCA they can be omitted
~~LB~xp@=G
for T < T
Forthe
second terhAin
eq. 35 we obtain
the following
expressions
at high temperature
arranging
the correction
terms as just discussed,
AxBaRTCA
{$[+
(~A~~)ln~o~A+l)
+ kxB(y
+ $,(:;;:~4)]
(48)
where
MB=Ha=Oa=NB=Na=O at
AG; - -
T < TCA,
AxBaRTCA
{$[
For
low temperature
s-T+kxB(s
- $
we obtain,
$h~;;;;~)]
(OBA+l) ln(OBA+l)
A
4 _ 3-4~+30~+
(No+1 ) r4
(49)
2 (OBA+t 1 k/a
MB= Ma=Oa=NB=No=O at T < f
For the third
term in eq. 3SAwe obtain
arranging
the correction
terms as before,
where
at
high
temperature
by multiplying
tem-
(47)
- Ai+&
case,
AG; = -
sub-
with
xA and
and
by
4-oKykxBT4/6]
AC; = -xAxBRTCAtoK~[1+(5+MB)kxg]/10-r
where
h=Mo=O
T > T CA. At
for
temperatures
low
237
METALS
(50)
we obtain
AG; = - xAx~RTCA{oK~[5-4~+kx~(5-~)/~4]/,0
+ K;[3-4r+r4+kxB(3+(Mo-3)T4)
(51)
l/6]
oB
KR and OKi are evaluated
from eqs.
where MR=Mo=O for T C T
, The quantities
the R value at x =O, i%.
BBOB *
at high temperature
by multiplying
For the four Bh term in eq. 55 we obtain
12 and 13 with
with
xA,
xAx~oK~RT~Bln(o~B+b+l)
(52)
A&;" =
~OT41(08B+~)
Below
the
AGaB
4
Curie
point
of pure
B, TCR, we find
x x RT In(bB+b+l)
AB
CR
{K;t5-4T/TCBI/10
+ oK~[T4/T~s+3-4T/TCB]/6)
In(BB+l)
An alternative
method to treat
this
term is to include
it in the term x OG which appears in
The result
wi?l Ee x (G - G 1 and
the complete expression
for the Gibbs energy of thealloy.
is equivalent
to choosing non-magnetic
B as the state of reference
for B, In f!ct,Bthi!
is how
Kirchner
et al.
treated
Mn in the Fe-W system (20).
For the fifth
term in eq. 35 we obtain
at high temperature
by multiplying
with xA and
arranging
the correction
terms as before,
where
N~=O~=~a=O for
AG; = -
T > T CA.
2
AxsbRTCA
At
tow
temperatures
we find
{K;[5+50a-4r-NB/T4]/10
+ oK;[3+30a-4T+(1+No)~4]/61
(551
("BB+l)ln(oBB+l)
Nb=Ob=N=O for T < T
.
mentioned
% may often be a good approximation
to put @=O and bB=O. The subresolution
representation
then reduces to the two terms AGa and AG
gc&
&
between TCQ and .TC. In addition
to the curves
discussed
efore,
AG1
curve for the su regular
sotution
approximation,
calculated
with
these
@ values
at low attoying
contents.
Lt is similar
to the other curves.
When compared
to the HWW treatment,
it should
be
where
As already
noticed
that
both versions
of the new treatment
predict
asymmetric
properties,
i.e.
non-regular solution
behavior.
This
fact
becomes
important
at larger
alloying
contents,
in particular
when one wants
to calculate
a phase equilibrium.
The new treatment
might
give
more realistic
predictions
for
such cases.
Of course,
the exact
form,
which
is obtained
by inserting
eqs.
17
gap because
the subregular
be preferred
expression.
a discontinuity
approximation.
to
the subregular
This
is
particularly
in
the
derivative
approximation
true
of
the
when
whenever
one
wants
Gibbs energy
it
to
has
is possible
calculate
been intro-
Conclusions
Indens
description
of the magnetic
specific
heat can be conveniently
using a series
expansion.
Analytical
expressions
can thus be derived
for
Gibbs energy between the fee and bee states
of iron, which yield
results
tabulation
sor.
published
by Orr
and
Chipman
and
analytical
expressions
applied
to
the difference
iron,
in
very
close
to the
presented
by Kaufman and
Ne-
238
This
Valuable
comments from
work has been financially
1.
C. Zener,
Trans
2.
Drs.
Peter Miodownik
supported
by the
alloying
efbase metal.
solution
mo-
and Gerhard
lnden are gratefully acknowledged.
Swedish Board for Technical
Development.
References
AIME,
1955,
vol.
203,
p.
619.
2.
G.
3.
G. Inden,
Project
Meeting
CALPHAD V, 21-25 June 1976, Max-Planck-inst.
G.m.b.H.
D~sseldorf/W-Ge~any,
1976, p. 111. 4-l.
4.
A.P;
Miodownik,
5.
C.H.
Johansson,
Arch.
EisenhDttenwes.,
AIME,
Inden,
6.
C. Zener,
7.
J.C.
Trans
Fisher,
Trans
8.
L.S.
Darken
9.
R.J.
Weiss
10,
R. Kohlhaas
11.
R.L.
12.
L.
Orr
Metallkunde,
and R.P.
and K.J.
and J.
13.
M. Braun
L.
Kaufman,
15.
M.
Hillert,
16.
G.
Inden,
1949,
Smith,
Chipman,
Z.
and R. Kohlhaas,
17.
H. Harvig,
18.
T.
19.
L.F.
20.
G. Kirchner,
E.V.
Nishizawa
Bates:
AIME,
stat.
and R.J.
Metallkunde,
G. Kirchner
1977,
and M. Hillert,
and M. Hasebe
Modern
T.
4th
ed.,
43,
p.
1815.
Met.,
Trans.,
Cambridge
and B. Uhrenius,
1972,
vol.
3,
p.
329.
Communications.
and M. Ko (Private
Magnetism,
Nishizawa
JISI,
vol.
VO!.
1951,
Metallkunde,
Phys.
Clougherty
Chem.,
Review,
Eisenh~ttenwes.,
Eng.ng.
Physical
Arch.
513.
185, p. 688.
vol.
Trans.
and H. Nesor,
Industr.
Tauer,
and M. Braun,
Kaufman
14.
AlME,
Eisenforschung,
Met.
1963, p.
Trans.,
301.