You are on page 1of 12

AM 6632 No.

of Pages 12, Model 5+


ARTICLE IN PRESS
15 June 2007; Disk Used

Acta Materialia xxx (2007) xxx–xxx


www.elsevier.com/locate/actamat

2 Effect of partitioning of Mn and Si on the growth kinetics


3 of cementite in tempered Fe–0.6 mass% C martensite

F
a,*
4 G. Miyamoto , J.C. Oh b, K. Hono b, T. Furuhara a, T. Maki c

OO
a
5 Institute for Materials Research, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan
b
6 National Institute for Materials Science, 1-2-1 Sengen, Tsukuba 305-0047, Japan
c
7 Department of Materials Science and Engineering, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan
8 Received 29 January 2007; received in revised form 12 April 2007; accepted 15 May 2007

PR
9

10 Abstract

11 The effects of Mn and Si addition on the growth rate of cementite in Fe–0.6 mass% C martensite have been studied by means of scan-
ED
12 ning electron microscopy, transmission electron microscopy and a three-dimensional atom probe. The growth rate of the cementite dur-
13 ing tempering at 723 K decreases substantially with the addition of Si due to the redistribution of Si between the cementite and ferrite
14 matrix. Mn retards cementite coarsening more effectively than Si at 923 K. In tempering at both 723 and 923 K, the Si concentration in
15 the cementite starts to decrease from an early stage of precipitation, whereas the cementite develops initially without the redistribution of
16 Mn, before the Mn gradually enriches into the cementite during tempering. Calculations of phase boundaries for stable equilibrium (par-
CT

17 tition local equilibrium) and metastable equilibria (para and negligible-partition local equilibriums) have revealed that there is a sufficient
18 driving force for the formation of paracementite in the Mn-added alloys. On the other hand, paracementite is difficult to form in the Si-
19 containing alloy because the cementite becomes unstable due to the dissolution of Si.
20  2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
E

21 Keywords: Tempering; Partitioning; Steels; Cementite; APFIM


22
RR

23 1. Introduction and W enrich into h while Ni, Co, Si and Al enrich into 37
the a matrix [5–7]. Björklund et al. [8] derived a theoret- 38
24 Tempered martensite with a fine dispersion of alloy car- ical expression for the coarsening kinetics of h in the a 39
CO

25 bides, such as CrCx, MoCx and VCx, is a base structure for matrix of the Fe–M–C ternary systems by assuming an 40
26 high-strength steels. Recently, steels with simpler chemical equilibrium partitioning of M between a and h. The vol- 41
27 compositions that contain only Mn, Si and C have ume diffusion of M in the a matrix for the partitioning 42
28 attracted more attention because of their recyclablity and serves as the controlling process for the coarsening of h. 43
29 low material cost. In such steels, the refinement of cement- Sakuma et al. [4] reported that the coarsening rate of h 44
UN

30 ite (h), which coarsens rapidly in plain carbon steels, is observed in Fe–Cr–C alloy tempered at 973 K agreed well 45
31 required to achieve higher strength. with the rate calculated by the model proposed by Björkl- 46
32 The growth of h is known to be remarkably retarded und et al. [8], while in tempering at 773 and 873 K, the 47
33 by the addition of Mn, Cr or Si [1–4]. Such an alloying observed rates were much faster than the calculated one. 48
34 effect has been considered to occur due to the partitioning They proposed that the deviation was caused by the 49
35 of these substitutional elements between a ferrite (a) incomplete partitioning of Cr between a and h even after 50
36 matrix and h. In an equilibrium state, Cr, Mn, V, Mo prolonged tempering at these temperatures. On the other 51
hand, they concluded that the equilibrium partitioning of 52
*
Corresponding author. Tel.: +81 22 215 2049; fax: +81 22 215 2046. Cr would be established from an early stage of tempering 53
E-mail address: miyamoto@imr.tohoku.ac.jp (G. Miyamoto). at 973 K. 54

1359-6454/$30.00  2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2007.05.023

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

2 G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx

55 In the pearlitic transformation in Fe–C–M alloys, 2. Experimental 112


56 Ridley [9] clarified that the partitioning coefficient, the
57 ratio of concentration of M between h and a, approaches An Fe–0.6% C binary alloy and Mn- or Si-added Fe– 113
58 unity as the transformation temperature decreases. For 0.6% C–M ternary and Fe–0.6% C–Si–Mn quaternary 114
59 proeutectoid a and pearlite transformations, Hultgren alloys were used in this study. The nominal compositions 115
60 [10] proposed the concept of paraequilibrium (PE), in of the alloys are shown in Table 1. The alloy ingots were 116
61 which only interstitial alloying elements are considered prepared by vacuum melting and casting. The samples 117
62 to be equilibrated because the diffusivity of the interstitial cut out from hot-rolled ingots were homogenized at 118
63 element (C) is orders of magnitude higher than that for 1473 K for 86.4 ks in an Ar-filled silica tube. The homoge- 119
64 the substitutional element (M). An alternative concept nized samples were austenitized at 1273 K for 1.8 ks and 120

F
65 for no-partitioning growth is the negligible-partition local quenched into water or oil, followed by treatment at 121
66 equilibrium (NPLE), in which the local equilibrium at the 77 K for 0.3 ks to reduce the amount of retained austenite. 122

OO
67 interface and the existence of a thin spike of the The samples were tempered at various temperatures 123
68 substitutional element in front of the interface are between 523 and 923 K in a salt bath up to 86.4 ks and 124
69 assumed [11–13]. water-quenched. For longer tempering periods, these pre- 125
70 In an early stage of tempering of martensite, the for- tempered samples were tempered further in a muffle fur- 126
71 mation of h without the redistribution of alloying ele- nace after being encapsulated in Ar-filled silica tubes to 127

PR
72 ments has been observed repeatedly by extraction avoid decarburization. 128
73 methods [6,7,14–16]. Recently, atom probe field ion The microstructure was observed with a field emission 129
74 microscopy (APFIM) has played an important role in scanning electron microscope (SEM; JEOL JSM-6500F) 130
75 the characterization of precipitation in tempered martens- and transmission electron microscopes (TEM; CM200 131
76 ite. Using APFIM, Barnard and Smith [17] reported that and CM-200FEG, operated at 200 kV). Specimens for 132
77 epsilon (e) carbide is formed without detectable redistribu- SEM observation were prepared by mechanical polishing 133
ED
78 tion of Si in an Fe–0.55% C–2.01% Si–0.85% Mn–0.24% followed by a selective potentiostatic etching electrolytic 134
79 Cr (hereafter composition is displayed in mass%) alloy dissolution (SPEED) method [25] at 50 to 0 mV in a solu- 135
80 tempered at 523 K for 86.4 ks. Chang and Smith [18] tion of 445 ml of methanol, 50 ml of acetylacetone and 5 g 136
81 showed that Si is rejected from h, forming an Si-enriched of tetramethylammonium chloride. Thin-foil specimens for 137
82 region beside h after tempering an Fe–0.6% C–2% Si alloy TEM observation were prepared by twin-jet electropolish- 138
CT

83 at 653 and 673 K for 3.6 ks. Babu et al. [19,20] investi- ing at 25–30 V in a solution of 500 ml CH3COOH, 20 ml 139
84 gated the partitioning behavior of Mn and Si at the h/a H2O and 100 g CrO3 at 285 K. The carbide size was 140
85 interface in Fe–2% Si–3% Mn–0.41% C martensite tem- measured from TEM or SEM photographs. Since the mor- 141
86 pered at temperatures ranging from 623 to 773 K. Unlike phology of h and e-carbide was reported to be lath- or acic- 142
E

87 the previous report [18], they found no initial partitioning ular-shaped in the initial stage of precipitation [26], the 143
88 of Si and Mn in tempering at 623 and 673 K, and minimum length of a given carbide in TEM photographs 144
RR

89 observed partitioning of Mn and Si into h and a, respec- was taken as the approximate diameter of the carbide in 145
90 tively, after tempering for an extended period of time or tempering at 723 K. While at 923 K, because h particles 146
91 at higher temperatures. Recently, using a three-dimen- nearly spheroidized, the radius of the circle having the 147
92 sional atom probe (3DAP), the present authors [21] found same area as a given h in TEM or SEM photographs was 148
93 that the redistribution of Si begins almost simultaneously taken as an approximate radius of the h particle. To deter- 149
CO

94 with the e ! h transition in Fe–2% Si–0.6% C alloy that mine the mean radius of carbide particles ðrÞ, particles of 150
95 was tempered at 723 K. The APFIM work on the parti- which total numbers were more than 30 and typically 100 151
96 tioning behavior of Cr and Mo in two Fe–2.25% Cr–1% at 723 K and more than 100 and typically 400 at 923 K 152
97 Mo steels containing 0.15 and 0.4% C reported that Cr were measured in several fields of view for each specimen. 153
98 and Mo do not redistribute in the early stage of temper- The compositions of h particles were analyzed using 154
UN

99 ing and the concentration of these elements in h gradually TEM energy-dispersive X-ray spectroscopy on carbon 155
100 increases during tempering [22–24]. Thus, the growth of h
101 is thought to occur concurrently with the gradual redistri-
102 bution of alloying elements during tempering of the mar- Table 1
103 tensite. However, so far there has been no systematic Chemical composition of the alloys
104 study on the growth kinetics of h with measurements of (mass%)
105 alloy partitioning. C Si Mn P S Fe
106 In this study, solute partitioning and the growth kinetics
Fe–0.6C 0.58 0.002 0.002 0.0008 0.0009 bal.
107 of h in a matrixes have been investigated in tempered high- Fe–0.6C–1Mn 0.61 0.016 1.02 0.0006 0.0009 bal.
108 carbon martensite. For substitutional alloying elements, we Fe–0.6C–2Mn 0.61 0.039 1.96 0.0007 0.0013 bal.
109 focus on Mn and Si, both of which are important alloying Fe–0.6C–1Si 0.59 1.01 0.001 0.0006 0.0009 bal.
110 elements in steels of simple compositions with good Fe–0.6C–2Si 0.59 2.01 0.001 0.0006 0.0009 bal.
Fe–0.6C–2Si–1Mn 0.59 2.01 1.02 0.0010 0.0009 bal.
111 recyclability.

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx 3

156 extraction replicas prepared by the SPEED method. The Mn and Si only minimally affect the hardness of the as- 177
157 composition of the alloying elements in h was also mea- quenched martensite (Fig. 1a). In tempering at 523 K 178
158 sured by 3DAP. Needle-like specimens for atom probe (Fig. 1a), the effect of the addition of Mn on the change 179
159 analysis were prepared by the standard two-step electropol- in hardness is small, whereas the Si-added alloys exhibit 180
160 ishing method. The mass spectra corresponding to C+, large resistance to softening. With an increase in the tem- 181
161 C2+, C3+, Cþ 2þ 2+
2 , C3 , Fe , Si
2+
and Mn2+ were taken into pering temperature, the resistance to softening by the Mn 182
162 account for the analysis. addition becomes larger (Fig. 1b). In tempering at 923 K, 183
163 Furthermore, the Vickers hardness (HV) was measured shown in Fig. 1c, the alloys containing Mn or Si show 184
164 using a load of 9.8 N. Seven HV measurements were taken, remarkable retardation against softening. Fig. 1d shows 185
165 of which five (excluding the maximum and the minimum) softening resistances by the addition of the alloying ele- 186

F
166 were averaged from each of the as-quenched and tempered ments at various tempering temperatures, where the soft- 187
167 specimens. The standard deviations for the five HV mea- ening resistance, DHV (=HVFe–0.6C–M  HVFe–0.6C), 188

OO
168 surements were well within 3% in most cases. represents a hardness difference between the Fe–0.6C bin- 189
169 Thermodynamical calculations were conducted with ary alloy and the Fe–0.6C–M ternary and quaternary 190
170 ThermoCalc through a programming interface, TC-API, alloys after a 3.6 ks tempering at each temperature. 191
171 using the TCFE2000 database. Details of the calculation Fig. 1d indicates that the DHV of Mn-added alloys are 192
172 method are described in Section 4.1. small at low tempering temperatures and increase with 193

PR
tempering temperature. On the other hand, the DHV of 194
173 3. Results the Si-added alloys are large at low temperatures, which 195
is attributed to the retardation of the e ! h transition by 196
174 3.1. Hardness change during tempering the Si addition [21]. The Si-added alloys also exhibit as 197
large a DHV as many of the Mn-added alloys at higher 198
175 Fig. 1 shows the effect of alloying elements on the hard- temperatures. As the amount of the alloying element 199
ED
176 ness of the tempered Fe–0.6C martensite. The additions of increases, the softening retardation effect also increases. 200
E CT
RR
CO
UN

Fig. 1. Effect of alloying elements on the hardness change of the tempered Fe–0.6%C martensite: (a) tempering at 523 K; (b) tempering at 723 K; (c)
tempering at 923 K; (d) resistance to softening by the addition of alloying elements after tempering for 3.6 ks at each temperature; DHV = HVFe–0.6C–
M  HVFe–0.6C. The standard deviations of HV are within 3% in most cases and are shown as error bars in (d) only.

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

4 G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx

201 The Mn and Si added quaternary alloy shows the highest Table 2
202 DHV at all tempering temperatures. Comparison between the calculated and experimentally obtained inter-
planar spacing of r or s in Fig. 2

203 3.2. Growth of h and partitioning of alloying elements Calculated interplanar spacing Measured interplanar spacing
(Å) (Å)
204 between h and a matrix
e h Fig. 2a Fig. 2b
205 3.2.1. Transition of carbide during tempering r d000 1e = 4.35 d01 1h = 4.06 4.36 4.10
s d 110 0e ¼ 2:38 d20 0h = 2.26 2.40 2.29
206 To clarify the effect of Mn and Si additions on the pre-
207 cipitation of carbides, lath-martensite microstructures of The d-spacing was calculated using the parameters a = 2.752 Å,
208 the Fe–0.6C, Fe–0.6C–2Mn and Fe–0.6C–2Si alloys tem- c = 4.353 Å for e-carbide [27] and a = 4.525 Å, b = 5.900 Å, c = 6.744 Å
for h [28].

F
209 pered at 523, 723 and 923 K were observed by TEM. The
210 carbides were identified by selected area diffraction

OO
211 (SAD) analysis; here only the e-carbide and h were taken Table 3
212 into consideration. SAD patterns and the corresponding Transition of carbides in the tempering of Fe–0.6C–M martensite
213 key diagram are shown in Fig. 2 as examples of SAD anal- Alloy 523 K 723 K
214 ysis of carbides, in which (a) and (b) represent the results of 30 s 1.2 ks 30 s 120 s 300 s 1.2 ks
215 the Fe–0.6C–2Si specimens tempered at 723 K for 30 and
Fe–0.6C e h h – – h

PR
216 300 s, respectively. Table 2 compares calculated and exper- Fe–0.6C–2Mn e h h – – h
217 imentally obtained interplanar spacing for r and s indi- Fe–0.6C–2Si e e e e/h h h
218 cated in Fig. 2. Although the measured interplanar spacing
219 deviates slightly from the calculated one, Table 2 indicates
220 that the e-carbide and h can be clearly distinguished. Table sections, the growth behavior of h after completion of the 228
221 3 summarizes the carbides observed in the tempered speci- e ! h transition is investigated. 229
ED
222 mens including those in the previous study [21]. e-Carbide
223 is formed after tempering at 523 K for 30 s regardless of 3.2.2. Cementite precipitation at 723 K 230
224 Mn and Si addition, while the transition from e-carbide Fig. 3 shows TEM images of the Fe–0.6 C, Fe–0.6C– 231
225 to h is retarded by the Si addition after tempering at 2Mn and Fe–0.6C–2Si samples tempered at 723 K for 232
226 523 K for 1.2 ks. Similarly, the e ! h transition is delayed 1.2 ks. The mean radii of h ðrÞ in the Fe–0.6C, Fe–0.6– 233
CT

227 by the Si addition in tempering at 723 K. In the following 2Mn and Fe–0.6C–2Si samples in this tempering condition 234
E
RR
CO
UN

Fig. 2. SAD patterns and corresponding key diagrams taken from the Fe–0.6C–2Si specimens tempered at 723 K: (a) for 30 s; (b) for 300 s.

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx 5

for several fields of view, as stated in Section 2. This shows 239


that both Mn and Si (especially Si) additions are effective in 240
refining the size of h. Fig. 4 shows the r as functions of tem- 241
pering time at 723 K. Si is shown to retard the h growth 242
strongly even after the completion of the e ! h transition 243
shown in Table 3. Although the initial size of h in the 244
Fe–0.6C–2Mn alloy is almost the same as that in the Fe– 245
0.6C alloy after 30 s, a large difference appears after a pro- 246
longed tempering (e.g. 3.6 ks). 247
Partitioning of the alloying elements between h and the a 248

F
matrix is expected to play an important role in such a retar- 249
dation effect of the h growth. Fig. 5 shows an elemental 250

OO
map and a composition–depth profile measured by 3DAP 251
in the Fe–0.6C–2Si alloy tempered at 723 K for 1.2 ks. In 252
the elemental map of Fig. 5a, in which C and Si atoms 253
are displayed as blue and green dots, respectively, h 254
appears as a C-rich region. In the composition–depth pro- 255

PR
file (Fig. 5b) and the ladder plot (Fig. 5c), h is positioned on 256
the left-hand side. The error bars in Fig. 5b represent ±2 257
standard deviations. A ladder plot displays the integrated 258
number of a detected element as a function of the total 259
number of detected atoms, in which the slope of the plot 260
indicates a local concentration of the element. In Fig. 5c, 261
ED
the slope of Si atoms in h is much lower than that in the 262
a matrix, indicating that Si atoms are already rejected from 263
h. Note that the Si concentration is slightly higher near the 264
h/a interface, as indicated by the arrows in Fig. 5c. The Si 265
depletion in h was observed even after a 120 s tempering in 266
CT

the same alloy at the same tempering temperature at which 267


e ! h transition takes place [21]. 268
Fig. 6 shows a composition–depth profile and the corre- 269
sponding ladder plot across an a/h interface in the Fe– 270
E

0.6C–2Mn alloy tempered at 723 K for 30 s. The concen- 271


tration of Mn in h increases slightly near the a/h interface 272
RR

although macroscopic redistribution of Mn between h/a 273


does not occur. After tempering for 3.6 ks at the same tem- 274
perature, TEM-EDX measurements of carbon replica spec- 275
imens indicate that the fraction of Mn atoms in the 276
substitutional lattice in h is 0.026 ± 0.004 which is nearly 277
CO

the same as the alloy composition for Mn (0.020). Thus, 278


UN

Fig. 3. TEM images of the martensite tempered at 723 K for 1.2 ks: (a)
Fe–0.6C; (b) Fe–0.6C–2Mn; (c) Fe–0.6C–2Si alloys.

235 are 18.5, 12.0 and 5.0 nm, respectively. In the estimation of
236 r, the minimum length of a given carbide in a TEM photo- Fig. 4. Mean radius of the carbides ðrÞ as functions of tempering time at
237 graph was taken as an approximate diameter of the carbide 723 K. The error bars indicate ±1 standard deviation of r for each field of
238 in tempering at this temperature and those were averaged view.

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

6 G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx

F
OO
PR
ED
Fig. 6. (a) Composition–depth profile across the a/h interface in the Fe–
0.6C–2Mn alloy tempered at 723 K for 30 s and (b) the corresponding
ladder plot. The error bars represent ±2 standard deviations.
CT

4 shows the values of experimentally determined n and k 295


that were measured by fitting the data in Fig. 8 to a rate 296
equation, r ¼ k  tn . The n values of the Fe–0.6C and Fe– 297
0.6C–2Si alloys were determined to be approximately 1/3, 298
E

indicating that the coarsening rate of h is volume diffusion 299


controlled [29–31], while that of the Mn-added alloys were 300
RR

Fig. 5. (a) 3DAP elemental maps of C and Si in the Fe–0.6C–2Si alloy found to be smaller than 1/3. 301
tempered at 723 K for 1.2 ks; (b) composition–depth profile and (c) and Fig. 9 shows a change in the concentration of M in h 302
ladder plot across the h/a interface. The black box in (a) indicates the during tempering at 923 K. The concentration is expressed 303
region for the analysis and the error bars in (b) represent ±2 standard
deviations.
as a site fraction in the substitutional lattice 304
Y hM : Y hM ¼ X hM =ðX hM þ X hFe Þ where XM and XFe represent 305
CO

atomic fractions of M and Fe, respectively. In the Mn- 306


279 alloy partitioning between h and a is clearly slower for Mn added alloys, the concentrations of Mn in h after a short 307
280 than that for Si at 723 K. tempering time are found to be close to the alloy composi- 308
tions for Mn. The concentration of Mn then increases 309
281 3.2.3. Cementite precipitation at 923 K gradually during tempering, eventually reaching the equi- 310
UN

282 Fig. 7 shows SEM images of h in the martensite tem- librium composition that was estimated from ThermoCalc 311
283 pered at 923 K for 18 ks. The h particles in the Fe–0.6C– calculations. On the other hand, most of the Si atoms are 312
284 2Si (Fig. 7b), Fe–0.6C–1Mn (Fig. 7c) and Fe–0.6C–2Mn rejected from h even after a short tempering for 30 s. 313
285 alloys (Fig. 7d) are much finer than that in the Fe–0.6C
286 base alloy (Fig. 7a). Fig. 8 shows the mean radii of hðrh Þ 4. Discussion 314
287 in the different alloys as functions of tempering time at
288 923 K. The radius of the circle having the same area as a 4.1. Phase equilibria in the Fe–Mn–C and Fe–Si–C ternary 315
289 given h in TEM or SEM photographs was taken as an systems 316
290 approximate radius of the h particle at this temperature
291 and those were averaged for several fields of view, as stated In this study, we have investigated the partitioning 317
292 in Section 2. Fig. 8 indicates that, unlike the case of temper- behavior of Mn and Si between the h and a matrix and 318
293 ing at 723 K, the Mn addition retards the h growth more the resultant retardation of h growth during tempering at 319
294 effectively than the Si addition at this temperature. Table 723 and 923 K. We found that Si is rejected from h in an 320

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx 7

F
OO
PR
ED
Fig. 7. SEM images of h in the martensite tempered at 923 K for 18 ks: (a) Fe–0.6C, (b) Fe–0.6C–2Si, (c) Fe–0.6C–1Mn, (d) Fe–0.6C–2Mn alloys.
E CT
RR

Fig. 9. Change in the concentration of Mn and Si in h in tempering at


Fig. 8. Variation in the mean radius of hðrh Þ in tempering at 923 K. The 923 K. Here concentration of Mn or Si is expressed as a site fraction,
CO

error bars represent ±1 standard deviation of rh for each field of view. YM = XM/(XM + XFe), where XM and XFe are the atomic fractions of an
alloying element and iron, respectively.
Table 4
Experimentally determined values for constants (n and k) in the rate dent at a later stage (Fig. 4). After all, the growth retarda- 327
equation ðr ¼ k  tn Þ of h in tempering at 923 K tion by Mn is more effective than by Si at 923 K (Fig. 8). 328
UN

n k These results indicate that the growth kinetics of h is closely 329


related to the redistribution of the alloying elements. Thus, 330
Fe–0.6C 0.31 ± 0.02 17 ± 2.9
Fe–0.6C–2Si 0.33 ± 0.01 7.5 ± 0.8 the partitioning behavior of Mn and Si is discussed first 331
Fe–0.6C–1Mn 0.30 ± 0.02 5.9 ± 0.9 from a viewpoint of thermodynamics of h precipitation in 332
Fe–0.6C–2Mn 0.23 ± 0.02 7.8 ± 2.2 Fe–C–M ternary systems. 333
A number of concepts have been proposed [10–13] 334
regarding the growth kinetics of a with and without the 335
321 early stage of its precipitation, whereas h forms without the partitioning of an alloying element in the decomposition 336
322 partitioning of Mn in the early stage of tempering and the of austenite (c) in Fe–C–M ternary systems. They are (i) 337
323 partitioning of Mn into h gradually proceeds during tem- PLE: a local equilibrium at the a/c interface with partition- 338
324 pering. While the growth of h in the Si-added alloy is ing of M between the two phases; (ii) NPLE: a local equi- 339
325 strongly retarded from the initial stage of precipitation at librium at the a/c interface with a pile up of M in front of 340
326 723 K, the retarding effect by the Mn addition becomes evi- the growing interface but no partitioning in the phases; and 341

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

8 G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx

342 (iii) PE: a paraequilibrium at the a/c interface without par- obtained by extrapolating the formation enthalpies 382
343 titioning of M even at the interface. Generally, PLE occurs reported by Chipman [34] to 0 K, and only the data below 383
344 at low supersaturation, and its kinetics is governed by the the Curie temperature of Fe3C were used here for the 384
345 volume diffusion of M, while NPLE and PE need high extrapolation. The calculated formation enthalpy of Fe3C 385
346 supersaturation, and the growth kinetics in these models agrees well with the experimental data shown in Table 5. 386
347 are controlled by carbon diffusion. Although there are deviations between the ab initio calcu- 387
348 In contrast to the precipitation of a in the c matrix, ther- lations and the experimental data for the other M3C, the 388
349 modynamics of the h precipitation in the a matrix in the calculated formation enthalpies match the experimental 389
350 Fe–C–M ternary systems is not understood well yet. In data very well. Thus, we conclude that reliable formation 390
351 the following sections, we will attempt to apply these mod- enthalpies could be estimated by the present ab initio calcu- 391

F
352 els to the precipitation of h in a. The retardation of the h lations. The formation enthalpy of Si3C, which cannot be 392
353 growth by the addition of Mn and Si will be discussed in determined experimentally, is positive and much larger 393

OO
354 Section 4.2 than the others, indicating that Si is highly unstable in h. 394

355 4.1.1. Formation enthalpy of M3C 4.1.2. h/a Phase boundaries in the Fe–Mn–C and Fe–Si–C 395
356 Thermodynamical parameters of a and M3C in the sys- systems 396
357 tems are necessary to calculate the equilibrium partition- The result of ab initio calculations is used for the calcu- 397

PR
358 ing. Among these parameters, the formation enthalpy of lation of Si3C by neglecting the temperature dependence of 398
359 Si3C is not available in the TCFE2000 database in Thermo- the formation enthalpy, and the values extracted from the 399
360 Calc because the equilibrium concentration of Si in h is too TCFE2000 database in ThermoCalc are used for the calcu- 400
361 low to obtain a precise measurement. Accordingly, the for- lations of Mn3C and a. 401
362 mation enthalpy of Si3C was deduced by ab initio calcula- The phase boundaries at 723 K for equilibrium (PLE) 402
363 tions in this study. and metastable equilibriums (NPLE and PE) between a 403
ED
364 The spin-polarized calculations with lattice relaxation and h in the Fe–C–Mn and Fe–C–Si ternary systems are 404
365 were carried out using the Vienna Ab Initio Simulation shown in Fig. 10a and b, respectively. The concentration 405
366 Package (VASP) [32,33]. The interaction between the ions
367 and valence electrons was described by a projector aug-
368 mented-wave method. A plane-wave basis set with a cutoff
CT

369 of 500 eV was used. The exchange correlation function was


370 described by the generalized gradient approximation.
371 Under these conditions, energies of graphite-C, M (=Cr,
372 Mn Fe, Ni, Co, Al and Si) of a reference state and M3C
E

373 were calculated. Then the formation enthalpy of M3C at


374 0 K was obtained as follows:
RR

graphite
376 DH 0MK3 C ¼ H M3 C  3H ref
M  HC ð1Þ
377 Table 5 compares the formation enthalpies estimated
378 from the ab initio calculations (column 2) with the values
379 determined based upon experiments (column 3). The values
CO

380 in column 3 except for Fe3C were calculated using Thermo-


381 Calc with the TCFE2000 database. That of Fe3C was

Table 5
UN

Comparison of formation enthalpies of M3C obtained by ab initio


calculations with the values determined based upon experiments at 0 K
refð0 KÞ graphiteð0 KÞ
DH ¼ H 0MK3 C  3H M  HC ðkJ=formula unitÞ
Present study (VASP) Experimental data (0 K)
Cr3C 33.0 47.9
Mn3C 26.7 40.4
Fe3C 18.8 18.3*
Ni3C 39.4 40.4
Co3C 51.9 24.0 Fig. 10. Phase diagrams of an isothermal section at 723 K: (a) Fe–C–Mn
Al3C 58.0 No data ternary system; (b) Fe–C–Si ternary system. The lines of a/a + h (PLE)
Si3C 260 and a/a + h (PE) indicate boundaries between the a and a + h region in
Values indicated with an asterisk were estimated by extrapolating the partitioning local equilibrium (PLE) and paraequilibrium (PE), respec-
formation enthalpies reported by Chipman [34] to 0 K; the others were tively. In the region below the PLE/NPLE line, h would grow under
calculated using ThermoCalc with the TCFE2000 database. negligible-partitioning local equilibrium (NPLE).

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx 9

406 of M is expressed in site fraction, YM. The tie lines in PLE lizing effect of h by the addition of these alloying elements 463
407 and PE are shown as broken and dotted lines, respectively. on the driving force for h precipitation under PE is much 464
408 In Fig. 10a, the a/ah) (PLE) boundary is shown to shift smaller than the decrease by the addition of Si. Therefore, 465
409 slightly towards the low-C side with increasing Mn content, more detailed studies on the h precipitation during temper- 466
410 and Mn enriches in h in PLE as indicated by its tie line in ing of Fe–C–Si–M quaternary alloys are necessary to clar- 467
411 the Fe–Mn–C system. Shifts of a/ah (PE) and PLE/NPLE ify the presence of the para-h in the Si-added steels. 468
412 boundaries with Mn are also small. In the Fe–Si–C system The present ab initio calculations clarified that the addi- 469
413 (Fig. 10b), the a/a + h (PE) boundary shifts considerably tion of Si increases the formation enthalpy of the e-carbide 470
414 towards the high-C side with increasing Si content. Fur- greatly, as well as that of h. This means that Si unstabilizes 471
415 thermore, the Si content in h in PLE is always nearly zero the e-carbide as well as h. However, in the previous study 472

F
416 regardless of the composition of Si, as assumed in the [21], para-e carbide was reported in the Fe–0.6C–2Si alloy 473
417 TCFE database of ThermoCalc. The negligible solubility tempered at 723 K for 30 s. Similarly, Barnard and Smith 474

OO
418 of Si in h under PLE means that the PLE/NPLE boundary [17] reported that the redistribution of Si between the e-car- 475
419 in the Fe–Si–C system almost coincides with the horizontal bide and the a matrix was not detectable by APFIM in an 476
420 axis at YSi = 0, and the NPLE region below the PLE/ Fe–0.55C–2.01Si–0.85Mn–0.24Cr alloy tempered at 523 K 477
421 NPLE boundary is virtually non-existent (Fig. 10b). for 86.4 ks. Therefore, further studies on the precipitation 478
422 The nominal compositions of the alloys used in this of e-carbide are required to clarify the effect of alloying ele- 479
ments on the e ! h transition from the viewpoint of

PR
423 study are indicated by solid circles in Fig. 10a and b. The 480
424 composition of the Fe–0.6C–2Mn alloy is well within all thermodynamics. 481
425 of the PLE, PE and NPLE regions (Fig. 10a). Hence, h Nakamura and Nagakura [36] reported that h 0 -carbide 482
426 without the partitioning of Mn can precipitate initially in nucleates on the a/g-carbide interface and g-carbide acts 483
427 this alloy under both PE and NPLE, and the redistribution as a carbon source for h 0 -carbide in Fe–1.5C martensite 484
428 of Mn occurs gradually during prolonged tempering under tempered at 470 K. The crystal structures of h 0 -carbide 485
ED
429 PLE. On the other hand, the composition of the Fe–0.6C– and g-carbide are very similar to those of h and e-carbide, 486
430 2Si alloy is within the PLE and PE regions, but outside of respectively. Thereby h might form with inheriting the par- 487
431 the NPLE region. Since the driving force for the h nucle- ticle distribution from e-carbide, as in the case of the g ! h 0 488
432 ation under PE (29.6 kJ mol1) was estimated to be only transition. This is supported from that fact that the carbide 489
433 57% of that under PLE (51.8 kJ mol1) in the Fe–0.6C– size in the Fe–0.6C–2Si alloy changes continuously regard- 490
CT

434 2Si alloy, the nucleation of h would be difficult under PE. less of the e ! h transition, as shown in Fig. 4. The inher- 491
435 Furthermore, the growth of para-h is inhibited after a local itance of particles distribution from e-carbide to h is also 492
436 equilibrium is established because the composition of the expected in the Fe–0.6C and Fe–0.6C–2Mn alloys. 493
437 Fe–0.6C–2Si alloy is outside of the NPLE region. Accord- Although the h sizes in the Fe–0.6C and Fe–0.6C–2Mn 494
E

438 ingly, para-h does not appear in the tempering of the Fe– alloys are much larger than that of the Fe–0.6C–2Si alloy 495
439 0.6C–2Si alloy. after tempering for 30 s, this might be due to completion 496
RR

440 For the retardation of h precipitation by the Si addition, of the e ! h transition at the very early stage of tempering 497
441 Owen [35] considered that Si is rejected from the growing h and following rapid growth of h controlled by carbon dif- 498
442 particles, and the rejected Si increases the activity of C fusion in these alloys. After initiation of the partitioning of 499
443 around the growing particles, resulting in the reduction Mn between a and h, h growth in the Fe–0.6 C–2Mn alloy 500
444 of the h growth as the flux of C to the particle, is sup- is controlled by the diffusion of Mn in a matrix. Since coef- 501
CO

445 pressed. The quite narrow NPLE region in the Fe–Si–C ficient for impurity diffusion of Mn in a is about half that 502
446 system calculated in this study supports Owen’s [35] theory for Si at this temperature [37], it is expected that growth of 503
447 in most of the compositional regions in this system. h is more sluggish in the Fe–0.6C–2Mn alloy than in the 504
448 However, para-h has sometimes been observed in the Fe–0.6C–2Si alloy, as seen in Fig. 4. 505
449 tempering of martensite containing Si with the other substi-
UN

450 tutional alloying elements. Sato et al. [14] reported that car- 4.2. Coarsening kinetics of h in Fe–C and Fe–C–M alloys 506
451 bides, although not identifying them as either h or e,
452 contain almost the same concentration of Mn and Si as As listed in Table 4, the n value for the Fe–0.6C alloy at 507
453 the alloy composition of them after tempering an Fe– 923 K is close to 1/3, indicating that the coarsening kinetics 508
454 0.58C–1.25Si–0.87Mn alloy at 873 K for 120 s. Similarly, of h are volume diffusion-controlled according to the Lif- 509
455 Babu et al. [19,20] and Thomson and Miller [22–24] shitz–Slyozov–Wagner (LSW) [29–31] theory. Several 510
456 observed para-h in an Fe–0.15C–2Si–3Mn alloy tempered researchers also reported that the coarsening kinetics of h 511
457 at 673 K for 1.8 ks and Fe–(0.15,0.4)C–0.3Si–2.2Cr– in the a matrix satisfies the cubic law [1,2,4,38–40]. How- 512
458 1Mo–0.5Mn alloys tempered at 623 K for 144 ks, respec- ever, Lindsley and Marder [41] and Nam and Bae [42] 513
459 tively, using APFIM. The discrepancy between the present recently reported that the n values in the tempering of mar- 514
460 and these previous studies might arise from the difference tensite fall between 0.20 and 0.23, and concluded that the 515
461 in the concentration of other alloying elements, such as coarsening of h is predominantly controlled by the diffu- 516
462 Mn, Cr or Mo, all of which stabilize h. However, the stabi- sion of C along dislocations and grain boundaries. The fol- 517

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

10 G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx

518 lowing two factors could be considered as possible reasons


519 for the difference in the n values compared with the previ-
520 ous results. The first one is a magnification of microscopic
521 measurements; Lindsley and Marder [41] measured the
522 particle size at magnifications of up to ·1000, while magni-
523 fications of up to ·20,000 is used in this study. For exam-
524 ple, Lindsley and Marder [41] measured the areas of
525 particles in the photographs, the mean diameter of which
526 was 400 nm in the shortest tempering period. However,
527 the particle seems to be too small to measure precisely at

F
528 the magnification of ·1000, and they may have overesti-
529 mated the mean particle size by missing the smaller parti-

OO
530 cles. This error would decrease with an increase in the
531 mean particle size, resulting in an underestimation of the
532 n value. The second factor is an effect of impurity elements.
533 As shown in Table 4, Mn-added alloys exhibit lower n val-
534 ues than that of the Fe–0.6C alloy. Since Nam and Bae [42]

PR
535 used an alloy containing 0.62% Mn, the n value that they
536 obtained might be smaller than that in an Fe–C binary
537 alloy. The decrease in the n value caused by the addition
538 of Mn does not indicate a change in the controlling process
539 but could be attributed to the partitioning of Mn during
540 tempering, as explained below. Accordingly, we consider
ED
541 that the coarsening kinetics of h in the Fe–0.6C alloy is vol-
542 ume diffusion-controlled.
543 The coarsening rate of the particles controlled by the
544 volume diffusion in the Fe–C binary system was predicted
545 by the LSW theory [29–31] as follows:
CT

546
8rV hm xC D
548
r3  r30 ¼  ðt  t0 Þ ð2Þ
9RT
549 where r is the mean particle radius of h at time t and r0 is
E

550 that at time t0 when steady-state growth begins. r is the a/h


551 interfacial energy, V hm is the molar volume of h, xC is the
solubility of carbon, and D is the diffusion coefficient of a
RR

552
553 solute controlling coarsening process. If the coarsening of
554 h is controlled by the volume diffusion of carbon, D is equal
555 to the diffusion coefficient of carbon in the a matrix, DaC .
556 On the other hand, the coarsening rate of h calculated by Fig. 11. Comparison of the coarsening kinetics of h between experiments
CO

557 Eq. (2) using DaC has been reported to be much faster than and calculations in tempering at 923 K: (a) Fe–0.6C; (b) Fe–0.6C–2Si; (c)
Fe–0.6C–Mn alloys. The error bars represent ±1 standard deviation of rh
558 the experimentally observed one [1,24,39]. A possible
for each field of view.
559 explanation for this discrepancy is that the volumetric dif-
560 ference between h and the a matrix is relaxed by the diffu-
561 sion of Fe. As a result, the coarsening of h would be coarsening rates in the Fe–0.6C alloy and the ones calcu- 575
UN

562 controlled by the diffusions of both C and Fe in a. This sit- lated by Eq. (2) using the diffusion coefficient of carbon 576
563 uation is called coupled diffusion. The effective diffusion (DC) and the effective diffusion coefficient (Deff). Here DC 577
564 coefficient when the volumetric mismatch is completely re- was taken to be 4.29 · 1011 m2 s1 [45] and the value of 578
565 lieved by the diffusion of Fe was proposed as follows r was taken as 0.7 J m2 [46]. xC was calculated by Ther- 579
566
567
[43,44]: moCalc to be 3.94 · 104. r0 and t0 were assumed to be 580
  negligibly small. In the calculation of Deff, DFe were taken 581
nFe Dc DFe V Fe nc
Deff ¼ 2 2
ðV Fe þ V cÞ ð3Þ to be 4.65 · 1018 m2 s1 [37], VFe and VC were taken as 582
569 nFe DFe ðV Fe Þ þ nc Dc ðV c Þ n Fe
7.09 · 106 and 2.12 · 106 m3 mol1, and nC/nFe was as- 583
570 where ni and Di are the number of moles and the diffusion sumed to be approximately equal to xC. As shown in 584
571 coefficient of i (i = Fe and C) in the a matrix, respectively. Fig. 11a, the observed coarsening rate is slower than the 585
572 VFe is the atomic volume of Fe in a and VC is the difference one calculated by Eq. (2) using DC but faster than the 586
573 between the volume of one molecule of Fe3C and the vol- one using Deff, implying that the volumetric mismatch 587
574 ume of three atoms of Fe in a. Fig. 11a compares observed accompanying the h precipitation is not fully compensated 588

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx 11

589 for by the diffusion of Fe but partly relieved by the intro- retarded by Si addition from an early stage of pre- 641
590 duction of dislocations, as proposed by Hillert [47]. cipitation. At 923 K, the coarsening rate of cement- 642
591 A coarsening rate of h particles in a Fe–C–M ternary ite in the Fe–0.6C binary alloy is intermediate 643
592 system was derived by Björklund et al. [8] as follows: between the carbon diffusion-controlled and cou- 644
h pled-diffusion-controlled rates. Unlike the tempering 645
8rV m DM
r3  r30 ¼ 2
 ðt  t0 Þ ð4Þ at 723 K, the retardation effect of coarsening of 646
594 27RT ð1  KÞ Y aM cementite by Mn addition is stronger than that by 647
595 where DM is the diffusion coefficient of an alloying element Si addition. 648
596 in the a matrix and K is a partitioning coefficient expressed (3) In tempering both at 723 and 923 K, Si is rejected 649
597 as K ¼ Y hM =Y aM . Fig. 11b and c compares the experimen- from cementite at an early stage of precipitation. 650

F
598 tally measured coarsening rates and calculated values Cementite without partitioning of Mn is initially 651
599 based on Eq. (3). Here, K and Y aM were calculated by Ther- formed, and then Mn gradually enriches cementite 652

OO
600 moCalc by assuming the equilibrium partitioning of M, during tempering. 653
601 and K2Si, K1Mn and K2Mn were obtained as 0, 16.7 and (4) Calculations of phase boundaries for para-, partition 654
602 15.0, respectively. DMn and DSi were taken to be local and negligible-partition local equilibriums have 655
603 1.49 · 1017 and 2.67 · 1017 m2 s1, respectively [37]. As revealed that there is a sufficient driving force for the 656
604 shown in Fig. 11b, the experimental data agree quite well formation of paracementite in the Mn-added alloys. 657

PR
605 with the calculated rate in the Fe–0.6C–2Si alloy. On the On the other hand, paracementite is difficult to form 658
606 other hand, the observed coarsening rate is much faster in the Si-added alloy due to the instability of cement- 659
607 than the calculated one in the Fe–0.6C–(1,2)Mn alloys, ite containing Si. 660
608 although the difference between them gradually decreases
609 with increasing tempering time (Fig. 11c). The coarsening Acknowledgement 661
610 rate of h in the Mn-added alloy is slower than that in the
ED
611 Fe–0.6C alloy even after tempering for 30 s, as shown in G.M. gratefully acknowledges the financial support of 662
612 Fig. 11c. According to Eq. (3), the coarsening rate de- the Research Fellowships of the Japan Society for the Pro- 663
613 creases as the coefficient K increases. The experimentally motion of Science for Young Scientists. Appreciation is ex- 664
614 observed coarsening rate is faster than the calculated one pressed to the Nippon Steel Corporation for providing the 665
in the initial stage of tempering because the partitioning materials. The authors are grateful to Prof. Shigeto Nishi- 666
CT

615
616 coefficient of Mn is much smaller than that in the equilib- tani (formally at Kyoto University, now at Kwansei Ga- 667
617 rium state (Fig. 9). Since Mn gradually enriches into h dur- kuin University, Japan) and Dr. Atsuto Seko (Kyoto 668
618 ing tempering, the experimental rate gradually approaches University, Japan) for allowing them to use the VASP 669
619 the calculated one. On the other hand, most of the Si is re- software. 670
E

620 jected from h and an equilibrium partitioning is nearly


621 accomplished even after tempering for a short period of References 671
RR

622 30 s, as shown in Fig. 9. Consequently, the observed coars-


623 ening rate coincides well with the calculated value in the [1] Mukherjee T, Stumpf WE, Sellars CM, Tegart WJM. JISI 672
1969;207:621. 673
624 Fe–0.6C–2Si alloy even at an early stage of tempering.
[2] Airey GP, Hughes TA, Mehl RF. Trans Metall Soc AIME 674
625 Thus, we conclude that the partitioning kinetics of substi- 1968;242:1853. 675
626 tutional elements should be taken into account to predict [3] Hobbs RM, Lorimer GW, Ridley N. JISI 1972:757. 676
CO

627 the growth of h in multicomponent systems. [4] Sakuma T, Watanabe N, Nishizawa T. Trans JIM 1980;21:159. 677
[5] Hultgren A, Kuo K. Rev Metall 1953;50:847. 678
[6] Sato T, Nishizawa T. J Jpn Inst Met 1955;19:385. 679
628 5. Conclusions
[7] Sato T, Nishizawa T, Honda H. Tetsu-to-Hagané 1955;41:1188. 680
[8] Björklund S, Donaghey LF, Hillert M. Acta Metall 1972;20:867. 681
629 The effect of Mn and Si additions on the growth rate of [9] Ridley N. In: Proceedings of the phase transformations in ferrous 682
UN

630 cementite during the tempering of Fe–0.6amss%C martens- alloys. AIME; 1984. p. 201. 683
631 ite was investigated in the temperature range between 723 [10] Hultgren A. Trans ASM 1947;39:915. 684
[11] Kirkaldy JS. Can J Phys 1958;36:907. 685
632 and 923 K. The following results were obtained.
[12] Hillert M. Paraequilibrium Internal report. Stockholm: Swedish 686
Institute of Metals; 1947. 687
633 (1) The Si-added alloys exhibit large resistance to soften- [13] Coates DE. Metall Trans 1972;3:1203. 688
634 ing at all the tempering temperatures investigated, [14] Sato T, Nishizawa T, Ohashi M. Tetsu-to-Hagané 1957;43:485. 689
635 whereas the softening retardation by Mn addition is [15] Gurry RW, Christakos J, Darken LS. Trans ASM 1961;53:187. 690
[16] Ghosh G, Olson GB. Acta Mater 2002;50:2099. 691
636 small at low tempering temperatures and increases
[17] Barnard SJ, Smith GDW. In: Proceedings of the solid solid phase 692
637 with temperature. transformation; 1981. p. 881. 693
638 (2) At 723 K, the retardation of cementite growth by [18] Chang L, Smith GDW. J De Phy 1984;C9-397. 694
639 Mn addition is observed only after tempering for [19] Babu SS, Hono K, Sakurai T. Metal Mater Trans A 695
640 1.2 ks, while the growth of cementite is strongly 1994;25A:499. 696
[20] Babu SS, Hono K, Sakurai T. Appl Surf Sci 1993;67:321. 697

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
AM 6632 No. of Pages 12, Model 5+
ARTICLE IN PRESS
15 June 2007; Disk Used

12 G. Miyamoto et al. / Acta Materialia xxx (2007) xxx–xxx

698 [21] Miyamoto G, Furuhara T, Maki T. In: Proceedings of the [33] Kresse G, Furthmüller J. Phys Rev B 1996;54:11169. 713
699 solid ! solid phase transformations in inorganic materials; 2005. p. [34] Chipman J. Metall Trans 1972;3:55. 714
700 363. [35] Owen WS. Trans ASM 1954;46:812. 715
701 [22] Thomson RC, Miller MK. Appl Surf Sci 1995;87/88:185. [36] Nakamura Y, Nagakura S. Trans JIM 1986;27:842. 716
702 [23] Thomson RC, Miller MK. Appl Surf Sci 1996;94/95:313. [37] Oikawa H. Tech Rep Tohoku Univ 1982;47:215. 717
703 [24] Thomson RC, Miller MK. Acta Mater 1998;46:2203. [38] Bannyh O, Modin H, Modin S. Jernkont Ann 1962;146:774. 718
704 [25] Kurosawa F, Taguchi I, Matsumoto R. J Jpn Inst Met 1979;43:1068. [39] Vedula KM, Heckel RW. Metall Trans 1970;1:9. 719
705 [26] Ohmori Y, Davenport AT, Honeycombe RWK. Trans ISIJ [40] Das SK, Biswas A, Ghosh RN. Acta Metall Mater 1993;41:777. 720
706 1972;12:112. [41] Lindsley BA, Marder AR. Acta Mater 1998;46:341. 721
707 [27] Duggin MJ. Trans Met Soc AIME 1968;242:1091. [42] Nam WJ, Bae CM. Scr Mater 1999;41:313. 722
708 [28] Fasiska EJ, Jeffrey GA. Acta Cryst 1965;19:463. [43] Oriani RA. Acta Metall 1964;12:1399. 723
709 [29] Lifshitz IM, Slyozov VV. J Exp Theo Phys USSR 1958;35:479. [44] Li CY, Blakely JM, Feingold AH. Acta Metall 1966;14:1397. 724

F
710 [30] Lifshitz IM, Slyozov VV. J Phys Chem Solids 1961;19:35. [45] Ågren J. Acta Metall 1982;30:841. 725
711 [31] Wagner C. Z Elektrochem 1961;65:581. [46] Kramer JJ, Pound GM, Mehl RF. Acta Met 1958;6:763. 726
712 [32] Kresse G, Hafner J. Phys Rev B 1993;47:R558. [47] Hillert M. Jernkont Ann 1957;141:67. 727

OO
728

PR
ED
E CT
RR
CO
UN

Please cite this article in press as: Miyamoto G et al., Effect of partitioning of Mn and Si on the growth kinetics ..., Acta Mater (2007),
doi:10.1016/j.actamat.2007.05.023
View publication stats

You might also like