Académique Documents
Professionnel Documents
Culture Documents
net/publication/309166723
Internal Stress and Elastic Strain Energy in Pearlite and Their Accommodation
by Misfit Dislocations
CITATIONS READS
6 83
2 authors:
All content following this page was uploaded by Nobuo Nakada on 26 October 2016.
1) Department of Materials Science and Engineering, School of Materials and Chemical Technology, Tokyo Institute of Tech-
nology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, Kanagawa, 226-8503 Japan.
2) Formerly Department of Materials Science and Engineering, Tokyo Institute of Technology. Now at Nippon Steel & Sumitomo
Metal Corporation, 20-1 Shintomi, Futtsu, Chiba, 293-8511 Japan.
(Received on April 27, 2016; accepted on June 28, 2016)
Internal stress and elastic strain energy in pearlite caused by misfit between ferrite and lamellar cement-
ite were theoretically analyzed based on micromechanics while taking into account an accommodation
mechanism of the misfit strain between ferrite and cementite on pearlitic transformation. Two-dimensional
large deformation analysis reveals that the Pitsch-Petch orientation relationship is most appropriate among
already reported crystal orientation relationships under the condition that the interface contains a lattice
invariant direction. However, the micromechanics analysis using a periodic function proves that the misfit
strain generates very large elastic strain energy in pearlite even when the Pitsch-Petch orientation relation-
ship is satisfied, which is almost comparable to the chemical driving force for pearlitic transformation.
Assuming that an interval of misfit dislocations dynamically introduced at ferrite/cementite interface upon
pearlitic transformation depends on the growth rate of pearlite, the total elastic strain energy reduces
more effectively as the growth rate becomes lower. As a result, the elastic strain energy in pearlite
changes widely depending on interlamellar spacing.
KEY WORDS: pearlite; elastic theory; misfit strain; internal stress; misfit dislocation.
chanics. In addition, the interlamellar spacing dependency phases have good lattice correspondence on these planes.
of elastic accommodation will be discussed by taking into Indeed, each set of them corresponds to the following crys-
account the arrays of misfit dislocations which are dynami- tal orientation relationships (OR).
cally introduced on pearlitic transformation. a–c; Bagaryatsky OR (001)C//(11 2 )F, [100]C//[1 10]F,
[010]C//[111]F28)
b–c; Isaichev OR (101)C//(11 2)F, [10 1]C//[1 10]F,
2. Calculation of Misfit Strain by Two-dimensional
[010]C//[111]F29)
Large Deformation Theory
a–d; Pitsch-Petch OR (001)C//( 2 15)F, [100]C2–3°[ 31 1]F,
Let us first evaluate the elastic state of pearlite by [010]C2–3°[131]F30,31)
assuming that coherency is maintained between ferrite and b–d; Unknown OR (101)C//(2 15)F, [10-1]C2–3°[311]F,
cementite in pearlite. For this purpose, we must know the [010]C2–3°[131]F
interface orientation composed of crystal planes of the two Except for the set of b–d, they had been actually identified
phases. in the previous researches.
Figure 1 displays unit cells of (a) cementite (orthorhom- As an example and for reasons mentioned below, we
bic crystal,) and (b) ferrite (body-centered cubic crystal) will consider the Pitsch-Petch OR and misfit strains at the
graphically-represented by a software Crystal Maker. To interface will be evaluated using two-dimensional large
evaluate misfit strains at pearlitic transformation tem- deformation analysis.
perature, the lattice constants of cementite (C) and ferrite To change the iron rectangular lattice of cementite (Fig.
(F) at 873 K (600°C) (aC = 0.45235 nm, bC = 0.50895 nm, 2(a)) into that of ferrite (Fig. 2(d)) under the Pitsch-Petch
cC = 0.67433 nm and aF = 0.28665 nm, respectively) will OR, the lattice transformation matrix T is expressed as the
be used.27) Figure 2 shows iron atom configurations of product of a simple tensile and compressive deformation
(a) (001) and (b) (101) in cementite and (c) (11 2) and (d) matrix A and a simple shear deformation matrix B, as shown
( 2 1 5) in ferrite.18) This figure reveals intuitively that both in Fig. 3. When mutually perpendicular x1 and x2 axes are
fixed along [010]C and [100]C, respectively together with
the x3 axis being normal to the ferrite/cementite interface,
T is written as
Fig. 1. Unit cells of (a) cementite and (b) ferrite. from the plane correspondence in Fig. 2, where θ = 5.2°
Fig. 2. Fe atomic configurations of (a) (001) and (b) (101) planes of cementite and (c) (11 2) and (d) (21 5) planes of fer-
rite.18)
from Fig. 2(d). Table 1. Principal strains of lattice transformation and M value
Our next task is to find the principal distortions of the for respective orientation relationships.
transformation T. Since T is non-symmetric, it does not principal strain
Orientation
have mutually perpendicular principal axes of distortion. Combination M value
relationship ε11* ε22*
However, a regular matrix T is known to be decomposed
into a rotation (orthogonal) matrix R and a distortion (sym- Bagaryatsky a–c 0.02508 0.11585 0.01599
metric) matrix D as32) Isaichev b–c 0.00151 0.02508 0.00066
Pitsch-Petch a–d −0.06114 0.08968 0.00813
T = RD. ................................... (2)
Unknown b–d −0.15126 0.08188 0.02133
The symmetric distortion D can now be diagonalized to
find its principal axes and principal strains. From Eq. (2),
we have
explained above is defined by setting cementite as a start-
T tr T = (RD)tr (RD) = D tr R tr RD = D tr D .......... (3)
ing reference state (see again in Fig. 3). This is due to
where Ttr is the transposed matrix of T. Equation (3) indi- the simplification of deformation matrices, since the iron
cates that the principal strains of D, and, thus, those of T, rectangular lattice in cementite can be described by the
are obtained as square roots of the diagonal components of orthogonal coordinate system. On the other hand, when fer-
the matrix (TtrT)dia, the diagonalized matrix of TtrT. There- rite is regarded as a reference phase, the principal distortions
fore, we write η1, η2 from ferrite to cementite is obtained as the inverse
of those from cementite to ferrite (η1 = 1/η1C, η2 = 1/η2C).
T11 0 T11 T12 T112 T11T12 Principal misfit strains at the interface are now obtained as
S = T tr T = = ... (4) ε11* = η1 − 1 and ε22* = η2 − 1.
T12 T22 0 T22 T11T12 T122 + T222
By considering the similar analysis for other ORs, the
Then, the rotation of the matrix Sdia with respect to the x3 principal strains were obtained as listed in Table 1. Here,
axes results in the parameter M is nearly proportional to the elastic strain
energy and is defined as
cos α sin α T112 T11T12 cos α − sin α
Sdia =
− sin α cos α T11T12 T122 + T222 sin α cos α M = (ε11
* 2
) + (ε 22
* 2
) + 2 ε11ε 22 . ................. (8)
* *
3
T112 cos2 α + 2T11T12 sin α cos α + (T122 + T222 )sin 2 α
= 2 From the energy point of view, ORs with smaller M values
(T12 + T22 − T11 )sin α cos α + T11T12 (cos α − sin α )
2 2 2 2
are more favorable than those with larger M values.33,34)
(T122 + T222 − T112 )sin α cos α + T11T12 (cos2 α − sin 2 α ) Then, it appears from Table 1 that the Isaichev OR is most
. favorable followed in order by the Pitsch-Petch and Baga-
T112 sin 2 α − 2T11T12 sin α cos α + (T122 + T222 ) cos2 α
ryatsky ORs. Another factor often discussed for interface
.......................................... (5)
crystallography and energetics is that whether the interface
where α is the angle of rotation. In order for Sdia to be a contains the invariant-line direction.33–35) It is known that the
diagonal matrix, the following equation must be satisfied. necessary and sufficient condition for the existence of the
invariant line is ε11* < 0 < ε22*.32) As can be seen in Table
(T122 + T222 − T112 )sin α cos α + T11T12 (cos2 α − sin 2 α ) = 0. ... (6) 1, only the Pitsch-Petch OR satisfies this condition among
Solving Eq. (6) by substituting T11, T22 and T12 obtained in reported ORs.
Eq. (1) with lattice parameters of cementite and ferrite and As a result, it is concluded that the Pitsch-Petch orienta-
θ = 5.2 degree, α is either 71.26 or − 18.74 degrees, both tion relationship is the most appropriate one to optimize the
being crystallographically equivalent. Furthermore, substi- plane matching at the interface between ferrite and cement-
tuting these values into (5), the detail Sdia is ite among already reported crystal orientation relationships.
In fact, the Pitsch-Petch OR has been observed most fre-
1.1345 0
Sdia = . ..................... (7) quently in pearlitic steels.15,18,23,24) In the following session,
0 0.8422 the principal strains for the Pitsch-Petch OR is used as the
The square root of the diagonal components in (7) indicates misfit strains and the eigenstrains in the lamellar cementite.
principal distortions η1C = 1.06512 and η2C = 0.91769.
Here, it should be noted that the lattice deformation
∞
2 nπ 1 − ν
ε ij* ( x3 ) = ε ij* f + ∑ sin n π f cos x3 .... (9) .................. (12)
π
n =1 L 2µ *
σ 22 (F) = f (νε 11 + ε *
2 2 ) in ferrite
1 −ν
Here, εij* are given as follows.
σ 33 (F) = 0
ε11
*
0 0 ....................... (10) where μ is the shear modulus and ν is the Poisson ratio.
ε ij = 0 ε 22
* *
0
0 0 *
ε 33 3.2. Elastic Strain Energy Caused by Misfit Strain in
Pearlite
where ε11* and ε22* are the same misfit strains as mentioned Now that eigenstrains and internal stresses in the two
above and given in Table 1. ε33* is unknown at this moment phases are found, elastic strain energy per unit volume of
but will soon be found immaterial for the present analysis. pearlite E0 can be calculated according to the following
equation.37)
L
1
2 L ∫0
E0 = − σ ij ε ij* dx3 ....................... (13)
2a
E0 = −
2L
{
σ 11 (C)ε11*
+ σ 22 (C)ε 22
*
} ... (14)
µ * 2
= f (1 − f )
1 −ν
{
(ε11 ) + (ε 22 ) + 2νε11ε 22 .
* 2 * *
}
Cementite fraction in pearlite is 0.12 in a conventional
eutectoid Fe-0.8mass%C alloy. Therefore, substituting f =
0.12, μ = 80 GPa, ν = 0.3, ε11* = − 0.0611 and ε22* =
0.0897 into Eq. (14), E0 is calculated to be 1.04 × 108 J/m3
(737 J/mol). This value is as high as the free energy change
from austenite to pearlite at 873 K (600°C) in this alloy
(approximately 1 150 J/mol) calculated by a thermody-
namics calculation software Thermo-calc (Ver. 3.0 SSOL
database). This proves that the ideal coherency strain energy
of pearlite is almost comparable to the chemical driving
force of pearlitic transformation, even when the most stable
Pitsch-Petch OR is kept between ferrite and cementite. In
addition to periodic function, another calculation method for
internal stress analysis using Eshelby’s inclusion problems37)
with Mori-Tanaka method38) will be shown in Appendix.
Fig. 6. Schematic illustration explaining the two difference conditions in the interval of misfit dislocations introduced
upon pearlitic transformation.
4.1. Constant Interval of Misfit Dislocations Array; Burgers vector of ferrite (0.25 nm), while d is set to be 20
Type 1 nm by reference to the observation results by Shiflet et al.18)
Assuming that the interval of misfit dislocations array is As can be seen, E1 is smaller than E0, which demonstrates
constant at d, Δε11* and Δε22* are written as b/d and –b/d that the introduction of misfit dislocations accommodates
with Burgers vector b, respectively. Here, we approximate misfit strains. However, most of the elastic strain energy
that b is the same for all the misfit dislocations. Substitut- still remains in pearlite. This agrees well with the results in
ing these values into Eq. (15), elastic strain energy per unit previous researches23–25) and suggests that the misfit disloca-
volume E1mis is modified as tion interval of 20 nm is insufficient to effectively reduce the
coherency strain energy. 2 On the other hands, it is understood
µ * 2 b that E * increases as
binterlamellar spacing becomes finer due
E1 = f (1 − f ) (ε11 ) + (ε 22 ) + 2νε11ε 22 + 2(1 −ν ) (ε11 − ε 22 ) + 2(1 −ν ) dis
mis * 2 * * * 1
.
1 −ν d to the increase in Ed1 . But, the magnitude value of E1dis
is much lower than that of E1mis, thus the dependency of
b
2
b * interlamellar spacing on E1 is very small.
ε11
* 2
) + (ε 22
* 2
) + 2νε11 ε 22 + 2(1 −ν ) (ε11
* *
− ε 22
*
) + 2(1 −ν ) .
d d
4.2. Variable Interval of Misfit Dislocations Array;
........................................ (16)
Type 2
If d is changeable depending on the growth rate of pearl-
ε11
*
≤ − b d or ε 22 *
≤b d ite υ, it is reasonable to consider from the kinetics point
In addition to E1mis, elastic strain energy of misfit disloca- of view that d is an increasing function of υ. Here, let us
tions themselves E1dis must be considered. In pearlite with assume for simplicity that d is proportional to υ, as written
interlamellar spacing L, total area of ferrite/cementite inter- by Eq. (19).
face per unit volume is counted as 2/L. Taking into account
dmin
that misfit dislocations are arranged along both directions of d= υ ................................ (19)
x1 and x2 (see Fig. 5), the density of misfit dislocations per υmin
unit volume is 4/(L × d). Thus, E1dis is given as dmin and υmin are the minimum interval of misfit dislocations
and the minimum growth rate to fully accommodate misfit
µb 2
R ........................ (17)
E1dis = ln . strains, respectively. From geometry, dmin can be estimated
Ldπ r0 as either –b/ε11* or b/ε22*. Some researchers including
Here, r0 and R indicate the radius of dislocation core and Ridly investigated the relation between υ and L,3,4,8,11) and
stress field, respectively. Given that r0 = 5b and R = d/2, reported that υL2 is constant (k), which emphasizes that
the total elastic strain energy per unit volume in Type 1 E1 pearlitic transformation is controlled by volume diffusion of
is described as carbon in austenite near the growing interface. According to
this empirical equation, d is 2expressed as a function of L, as
µ * 2 b * in Eq.
shown * (20). b µ b
2
d
E1 = E1 + E1 = f (1 − f ) (ε11 ) + (ε 22 ) + 2νε11ε 22 + 2(1 −ν ) (ε11 − ε 22 ) + 2(1 −ν ) +
mis dis * 2 * *
ln .
1 −ν d d Ldπ 10b
* 2 * * b *
ε 22 ) + 2νε11ε 22 + 2(1 −ν ) (ε11 − ε 22 ) + 2(1 −ν ) +
* b µ b
2 2
d
ln .
d=
k
υmin L
d =
2 min
kb
υmin L2ε *
(
ε * = ε11 *
or ε 22
*
) ... (20)
d d Ldπ 10b
Then, substituting Eq. (20) into (18) with ε22* leads to the
........................................ (18)
total elastic strain energy per unit volume in Type 2 E2.
Figure 7 shows the change in E1 as a function of interla-
E2 = E2mis + E2dis =
mellar spacing, with f = 0.12, μ = 80 GPa, ν = 0.3, ε11* =
− 0.0611 and ε22* = 0.0897. In addition, b is treated as the µ * 2 υmin L2
f (1 − f ) (ε11 ) + (ε 22 ) + 2νε11ε 22 + 2
* 2 * *
(1 −ν ) ε11 ε 22 − (ε 22
* * * 2
)
1 −ν k
E2 = E2 + E 2 =
mis dis
µ υmin L2 υmin L2ε 22 *
µ bυmin Lε 22
2 *
k
f (1 − f ) (ε
11
* 2
) + (ε * 2
) + 2νε ε
* *
+ 2 (1 − ν ) ε ε
* *
− (ε * 2
) + 2( 1 − ν ) + ln
1 −ν
22 11 22
k 11 22 22
k kπ 10υ
min L
µ bυmin Lε 22
2
υmin L2 υmin L2ε 22
* *
k
)2 + (ε 22
* 2
) + 2νε11ε 22 + 2
* *
(1 −ν ) ε11ε 22 − (ε 22
* * * 2
) + 2(1 −ν ) + ln 2 *
k k k π 10 υ min L ε 22
........................................ (21)
Figure 8 displays the change in E2 as a function of interla-
mellar spacing using the same numerical values as above.
Although k and υmin are varied depending on composition of
steels, they could be considered as constant values in a steel.
Thus, k and υmin were determined to be 1.0 × 10 − 19 m3/s and
50 nm/s, respectively, by adopting the experimental data
Fig. 7. Relation between elastic strain energy and interlamellar reported by Bolling & Richmann using a low alloy eutectoid
spacing. Elastic strain energies E1 is given by Eq. (18). steel.3) E2 is lower than E0 within all region of interlamellar
REFERENCES
1) F. C. Hull and R. F. Mehl: Trans ASM, 30 (1942), 381.
2) J. W. Cahn and W. C. Hagel: Acta Metall., 11 (1963), 561.
3) G. F. Bolling and R. H. Richmann: Metall. Trans., 1 (1970), 2095.
4) M. P. Puls and J. S. Kirkaldy: Metall. Trans., 3 (1972), 2777.
5) S. D. Kulkarni: Acta Metall., 21 (1973), 1539.
6) C. Zener: Trans. AIME, 47 (1946), 550.
7) G. Zhang and M. Enomoto: ISIJ Int., 49 (2009), 921.
8) G. E. Pellissier, M. F. Hawkes, W. A. Johnson and R. F. Mehl: Trans
ASM, 30 (1942), 1049.
9) J. W. Cahn and W. C. Hagel: Decomposition of Austenite by Dif-
fusional Processes, eds. by V. F. Zackay and H. I. Aaronson, John
Wiley, New Nork, (1962), 131.
10) M. Hillert: Decomposition of Austenite by Diffusional Processes, eds.
Fig. 8. Relation between elastic strain energy and interlamellar by V. F. Zackay and H. I. Aaronson, John Wiley, New Nork, (1962),
spacing. Elastic strain energies E2 is given by Eq. (21). 197.
11) N. Ridley: Metall. Trans. A, 15A (1984), 1019.
12) A. R. Marder and B. L. Bramfitt: Metall. Trans. A, 24 (1975), 365.
13) T. Takahashi, M. Nagumo and Y. Asano: Trans. Jpn. Inst. Met., 42
spacing, however, the lowering behavior is completely dif- (1978), 708.
14) T. Takahashi, M. Nagumo and Y. Asano: Trans. Jpn. Inst. Met., 42
ferent from E1 in Fig. 7. It is obvious from Eq. (21) that the (1978), 716.
first term E2mis depends more heavily on L than the second 15) M. A. Mangan and G. J. Shiflet: Metall. Trans. A, 30A (1999), 2267.
term E2dis since the former contains higher power exponent 16) S. A. Hackney and G. J. Shiflet: Acta Metall., 35 (1987), 1007.
17) S. A. Hackney and G. J. Shiflet: Acta Metall., 35 (1987), 1019.
of L. This means that the accommodation of misfit strains by 18) D. Z. Zhou and G. J. Shiflet: Metall. Trans. A, 23A (1992), 1259.
misfit dislocations is more effective when the interlamellar 19) D. V. Shtansky, K. Nakai and Y. Ohmori: Acta Mater., 47 (1999),
1105.
spacing becomes coarser for slower pearlite growth rate. In 20) S. Zaefferer, S. I. Wright and D. Raabe: Metal. Mater. Tans. A, 39A
this calculation, E2 significantly decreases when interlamel- (2008), 374.
lar spacing is larger than 200 nm. As a result, E2 varies more 21) Y. T. Wang, Y. Adachi, K. Nakajima and Y. Sugimoto: Acta Mater.,
58 (2010), 4849.
significantly than E1 depending on interlamellar spacing. 22) Y. Tomota, O. Watanabe, A. Kanie, A. Moriai, N. Minakawa and Y.
This means that Type 2 accommodation is more reason- Moriai: Mater. Sci. Technol., 19 (2003), 1715.
23) N. Nakada, N. Koga, T. Tsuchiyama and S. Takaki: Scr. Mater., 61
able to explain the fact that ferrite lattice strain in pearlite (2009), 133.
clearly decreases with increasing interlamellar spacing.25) 24) N. Koga, N. Nakada, T. Tscuhiyama, S. Takaki, M. Ojima and Y.
E2 is concave down function of L and becomes almost con- Adachi: Scr. Mater., 67 (2012), 400.
25) N. Nakada, N. Koga, Y. Tanaka, T. Tsuchiyama, S. Takaki and M.
stant when interlamellar spacing is finer than 200 nm. This Ueda: ISIJ Int., 55 (2015), 2036.
is because we have assumed for simplicity that the interval 26) A. Benedetti, H. Bender, A. Lauwers, C. Torregiani and K. Maex: J.
Microsc., 223 (2006), 249.
of misfit dislocations is simply proportional to the pearlite 27) M. Umemoto and K. Tsuchiya: Tetsu-to-Hagané, 88 (2002), 117.
growth rate, as expressed by Eq. (19). In practice, the rela- 28) Y. A. Bagaryatsky: Dokl. Akad. Nauk. SSSR., 73 (1950), 1161.
29) I. V. Isaichev: Zh. Tekh. Fiz., 17 (1947), 835.
tion between them should be more complicated. 30) W. Pitsch: Acta Metall., 10 (1962), 79.
31) N. J. Petch: T Acta Crystallogr., 6 (1953), 96.
32) M. S. Wechsler, D. S. Lieberman and T. A. Read: Trans. AIME, 197
5. Conclusions (1953), 1503.
33) M. Kato, M. Wada, A. Sato and T. Mori: Acta Metall., 37 (1989),
For theoretical analysis of internal stress and elastic 749.
34) M. Kato: Mater. Sci. Eng. A, A146 (1991), 205.
strain energy in pearlite caused by misfit between ferrite 35) U. Dahmen: Acta Metall., 30 (1982), 63.
and lamellar cementite, micromechanics was applied. The 36) T. Mura: Micromechanics of Defects in Solids, 2nd ed., Martinus
obtained results are as follows. Nijhoff Publ., The Hague, (1987), 7.
37) J. D. Eshelby: Proc. R. Soc., Lond., A241 (1957), 376.
(1) Two-dimensional large deformation analysis reveals 38) T. Mori and K. Tanaka: Acta Metall., 21 (1973), 571.
that the Pitsch-Petch orientation relationship is the most 39) M. Kato and T. Fujii: Acta Metal. Mater., 42 (1994), 2929.
appropriate one to optimize the plane matching at the inter-
face between ferrite and cementite among already reported Appendix
crystal orientation relationships in pearlite.
(2) Micromechanics analysis using a periodic function Analysis of internal stress by Eshelby’s ellipsoidal inclu-
demonstrates that the misfit strain generates very large elas- sion problems37) with the Mori-Tanaka method.38)
tic strain energy in pearlite even when the Pitsch-Petch ori- When a penny-shaped ellipsoidal single inclusion Ω
entation relationship is satisfied, which is almost comparable perpendicular to the x3 axis with uniform eigenstrains εij*
to the chemical driving force for pearlitic transformation. of Eq. (10) is within an isotropic infinite body (Fig. 9(a)),
Therefore, the coherency strain energy should be reduced internal stresses in the inclusion σijΩ can be expressed by the
through the strain accommodation mechanism, such as the following equation, as reported by Eshelby.
introduction of misfit dislocations.
σ ijΩ = Cijkl (Sklmnε mn
*
− ε kl* ) .................... (22)
(3) When misfit dislocations are dynamically intro-
duced at ferrite/cementite lamellar interface upon pearlitic Here, Cijkl and Sklmn are the elastic constants and Eshelby’s
transformation, the elastic strain energy is surely reduced, tensors, respectively. For a penny-shaped inclusion, Sklmn
although most of that still remains. Especially, given that the are given as follows.
interval of misfit dislocations depends on the growth rate of