: ~ , ;!~
ELSEVIER Materials Science and Engineering A205 (1996) 133-138 A
The effect of carbon and sulphur on the character of the grain boundary population in a-iron
M. TacikowskP, M.W. GrabskP, J. Driver b, A. Kobylanski b aDepartment of Materials Science and Engineering, Warsaw University of Technology, Narbutta 85, 02-524 Warsaw, Poland
bEcole Nationale Superieure des Mines de Saint-Etienne, 158 Cours Fauriel, F-42023 Saint-Eteinne, France
Received 6 June 1994; in revised form 26 May 1995
The grain boundary misorientations and CSL frequencies of a population of about 250 boundaries in ultra high purity ~-iron doped with small quantities of carbon and sulphur have been determined by a SEM microdiffraction BKD (Backscattered Kikuchi Diffraction) technique. The aim is to examine the effect of these elements on the character of the grain boundary population (CGBP). Samples containing about 75 wt. ppm of sulphur and/or 150 (200) wt. ppm carbon were recrystallized at 725 C after cold hydrostatic extrusion to e = 1.1 It was found that the addition of sulphur and carbon modifies the CGBP of e-iron. Sulphur clearly increases the density of low angle grain boundaries (LANGB), but only in the pure Fe-S binary alloy. The fractions of the LANGB and coincidence site lattice grain boundaries (CSLGB) in the Fe-C-S alloy are rather similar to those of Fe-C and thus the CGBP could be controlled by carbon. As a consequence, the lowest fraction of high energy random grain boundaries (RANGB) appears in the Fe-S alloy at about two-thirds of the total population. The effect of carbon and sulphur on the CGBP is interpreted in terms of grain boundary selection as a consequence of an impurity-specific dragging effect or a precipitation pinning effect on the migration of boundaries during recrystallization and subsequent grain growth.
Keywords: Carbon; Sulphur; Grain boundary; Iron
It is well known that the properties of grain boundaries vary from one boundary to another. The properties of a particular grain boundary are deter- mined by its atomic structure, defined by the atom species (matrix and segregated) and their geometrical configuration. Numerous studies have been undertaken in this area to observe or simulate grain boundary structures and to correlate the structure and the chem- istry with the properties of particular types of grain boundaries [1 6]. As a consequence, it is now well established that grain boundaries can be considered as an active element of the structure of polycrystals and the properties of a polycrystal should be interpreted in terms of the particular character of the grain boundary population (CGBP) [7-11]. Grain boundaries could also be used to control the properties of polycrystalline
* Corresponding author.
0921-5093/96/$15.00 1996- Elsevier Science S.A. All rights reserved
materials in accordance with the concept of "grain boundary design and control" . However, the ques- tion still remains of how to obtain a polycrystal with a specific grain boundary population containing the de- sired fractions of a particular type of grain boundary. Theoretically the main factors that can play an impor- tant role are defined, although published studies are fairly rare. One of the reasons has been the absence of a rapid technique for characterizing grain boundaries. However, progress in SEM microdiffraction techniques such as ECP (Eelectron Channeling Patterns) and more recently the powerful BKD (Backscattered Kikuchi Diffraction, which is also referred to in the literature as EBSP: Electron Backscattered Patterns) has solved this problem. One of the pioneering systematic studies of the problem of grain boundary population was under- taken by Watanabe and his coworkers using the ECP method .
Grain boundaries are formed during solidification and in the solid state as a consequence of such pro- cesses as recrystallization, phase transformation and
134 M. Tacikowski et al. / Materiab Science and Engineering A205 (1996) I33 138
sintering. In the case of recrystallization one can expect that the character of the grain boundary population is determined by such factors as temperature, annealing time, initial texture, deformation mode, strain and also by the purity of the material. There are some general ideas of how deformation-recrystallization parameters could influence the CGBP. As far as the influence of the purity of the materials is concerned only a few cases have been examined [11-13]. One can expect that im- purities or microalloying elements in the form of segre- gated atoms or precipitates could strongly influence recrystallization and the grain boundary migration pro- cess by solute atom dragging or precipitation pinning and so affect the character of the resulting grain boundary population. In previous work on the effect of microalloying elements on the grain boundary proper- ties in relation to the mechanisms of high temperature brittleness of ultra pure iron , it appeared that carbon and sulphur, exhibiting a complex effect, could also influence the CGBP. The aim of this work is thus, using the BKD technique, to examine how sulphur and carbon affect the CGBP in ultra high purity iron. In particular the behavior of pure Fe-S and Fe-C alloys will be compared with that of a ternary Fe-C-S alloy.
2. Experimental details
The ultra high purity iron, both as a reference mate- rial and alloyed with small quantities of carbon and sulphur, was prepared in the laboratory of the Ecole des Mines de Saint-Etienne  (Table 1). The alloy compositions were chosen to study first the separate influences of C and S (FC200, FS75C0) and then their combined action (FS75C150).
The CGBP obtained by recrystallization of a de- formed material was investigated in the present work. The samples were recrystallized at 725 C for 1 h after cold hydrostatic extrusion to E = 1.1 (Fig. 1). This treatment was preceded by annealing in the austenite range at 950 C for 30 min in order to reduce the texture resulting from the alloy fabrication process, i.e. cooling followed by preliminary deformation (forging)
Table 1 Chemical composit ions of the al loys (in wt. ppm) and strain in hydrostat ic extrusion
725C; 1 h
/ FORGING H. EXTRUSION
Fig. 1. The thermomechanical treatment of the alloys.
and the recrystallization annealing at 725 C for 1 h. The temperature of the basic recrystallization treatment (725 C) was chosen to ensure that the alloys contain- ing carbon were entirely ferritic. At this temperature carbon is in solid solution, but sulphur, with a solubil- ity limit about 25 wt. ppm , is partially present in the form of sulfides. All heat treatments were per- formed in vacuum-sealed quartz ampoules to avoid oxidation. The cross-sections of samples for BKD ex- amination were first polished mechanically and then electropolished in "A2 Struers reagent".
The grain boundary misorientation angle was used to characterize the grain boundary population. For this purpose the BKD technique was applied to the determi- nation of grain orientations and the misorientation angle of each boundary was calculated using a special computer program . It should be pointed out that the characterization of grain boundaries by their mis- orientation is only a preliminary approach and that the grain boundary plane is also an important element, which should be taken into account in further studies. At least 250 grain boundaries were analysed in each alloy, except for the pure iron (FP725 sample); the latter exhibited a very large grain size so that only 72 grain boundaries could be characterized. An additional, fine-grained sample of high purity iron (FP450) was also examined to check the results of the FP725 sample with better statistics. The recrystallization annealing was therefore performed at a low temperature i.e. at 450 C and the deformation was higher, e = 1.4, to obtain a fine grain size.
Al loy C S e content content strain
FP725 ~ - 1.05 FS75C0 14 71 1.11 FC200 200 < 10 1.08 FS75C150 153 71 1.08 FP450 - 1.38
" , not analysed; see Ref. .
The microstructures of the alloys resulting from the above thermomechanical treatments (Fig. 1) are shown in Fig. 2. One can notice a visible decrease of the grain size in the samples containing carbon and sulphur. This
M. Tacikowski et al./ Materials Science and Engineering A205 (1996) 133 138 135
Fig. 2. The microstructures after hydrostatic extrusion to E = 1.1 and recrystallization annealing 1 h at 725 C: (a) ultra pure iron, FP725; (b) ultra pure iron, FP450 (E= 1.4, 450 C, 1 h); (c) FC200 alloy; (d) FS75C0 alloy; (e) FS75C150 alloy.
effect is much more pronounced when sulphur and carbon are simultaneously present in iron. This differ- ence in grain size seems to be a consequence of the different grain boundary mobilities and may be related to the type of grain boundaries present in the alloy.
In order to establish experimental evidence of rela- tions between the CGBP and the carbon and sulphur contents in the recrystallized iron, the frequency of low angle grain boundaries (LANGB) and each type of CSLGB up to 1E29 was calculated from BKD data for each of the alloys and compared in Figs. 3 to 5. The pure iron sample (FP725) was initially intended as a reference but its relatively large grain size implies a strong grain growth contribution to the final CGBP. Also several types of coincidence site lattice grain boundaries (CSLGB) are absent in the grain boundary population (Fig. 3). This is probably mainly due to the poor statistics (only 72 grain boundaries analysed), and not to a physical effect. In fact in the fine grain sample of iron (FS450; Fig. 3; about 250 grain boundaries analysed) only the Z3 and El7 are absent. However, the CGBP in this sample results from a higher deforma- tion and lower recrystallization temperature than the other samples. As a consequence neither sample of iron could be easily used as a reference for the CGBP in the
Fe-S, Fe-C and Fe-C-S alloys. However, the LANGB (El) fractions in both samples of pure iron remain almost the same. For this reason we compared the character of the CSLGB population only between the three doped iron samples.
As can be seen in Fig. 4, the CGBP is different for each composition. The most important result is the higher fraction (roughly double) of LANGB in iron
0"14 t 0.12 11 z O.08 ~ 0.06
0021 a J l . . EL..qa. 0 i , i i i i i J i , , i , ,
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 SIGMA
I I FP725 ~ FP450
Fig. 3. The frequency of low angle (El) and coincidence grain boundaries in ultra pure iron after hydrostatic extrusion (FP725, E= 1.1, recrystallized at 725 C; FP450, E= 1.4; recrystaUized at 450 C).
136 M. Tacikowski et al./ Materials Science and Engineering A205 (1996) 133-138
u_ 0 .1~
O ~ 1 3 5 7 9 11 131517 192123252729
m FC200 m FFS75C0 ~]FS75C150
Fig. 4. The frequency of low angle and coincidence grain boundaries in the binary Fe S, Fe-C and the ternary Fe -S -C alloys after hydrostatic extrusion and recrystallization annealing at 725 C.
containing sulphur (Fig. 4). This effect disappears when carbon is simultaneously present with sulphur in the ternary Fe-S-C alloy. The frequency of the LANGB in samples with carbon is almost the same as in the pure iron samples (Fig. 3). The frequency of a particu- lar type of CSLGB varies slightly from one alloy to another. In the Fe-S binary alloy the E3 and Z5 boundaries are absent. Carbon and sulphur do not seem to have a clear effect on the total fraction of CSLGB, which remain roughly the same in all alloys (Fig. 5). As a consequence, the total fraction of random grain boundaries (RANGB) decreases to about two- thirds in the Fe-S binary alloy, because of the higher proportion of LANGB.
I FP725 FP450 FC200 FS75C0 FS75C150 ]
>- 0 Z LU
0 LU r r u_
LANGB ~ CSLGB ~ RANGB
Fig. 5. The character of the grain boundary population in iron and its microalloys doped with sulphur and/or carbon (hydrostatic extrusion to E = 1.1 and recrystallization annealing at 725 C).
All of the iron-based samples examined here exhibit slight recrystallization textures, leading to higher frac- tions of LANG and CSL boundaries than expected for a random polycrystal . However, their differences are not only due to the deformation-induced modification of the CGBP; the results of the present work also provide evidence of an influence of carbon and sulphur on the CGBP. This effect concerns the "composition" of the grain boundary population (in the sense of the frequency of occurrence of a particular type of "~"), rather than the total fraction of CSL boundaries. The CSL boundary fractions of all the alloys are not very different from those in pure iron and appear rather low compared with previous results [10,11] (this is probably due to the relatively low amount of deformation of the samples). As far as the LANGB population is concerned, in pure iron, the larger strains and lower recrystallization tem- peratures (compare samples FP725 and FP450) do not seem to produce significant differences in the fraction of the LANG boundaries. The increase of LANGB frac- tion (sample FS75C0) is associated with the sulphur addition. Our results do not confirm the tendency to higher random boundary populations observed in the case of phosphorus-doped iron by Elm'Rabat .
The mechanism leading to the selection of a particu- lar grain boundary population during recrystallization appears to be a rather complicated process since at least two different stages could control the final CGBP: (1) the nucleation of grains and (2) grain growth. The CGBP is directly related to the deformation texture, which results from the initial texture, mode and amount of deformation, and also to the recrystallization anneal- ing temperature and time. However, in our work all samples were prepared in the same manner, i.e. by hydrostatic extrusion with the same amount of defor- mation followed by recrystallization annealing. The deformation and heat treatment conditions are thus a constant factor so that the role of the impurities can be isolated. The effect of impurities on the CGBP could be explained in terms of grain boundary mobility control, resulting in the selection of certain types of grain boundaries by selective elimination during grain growth [10,12]. Both elements segregate to the c~-iron grain boundaries  and it is probably segregation produced during primary recrystallization  that is at the origin of a dragging effect  on migrating boundaries. The grain size in the sulphur-doped iron is lower than in the iron-carbon alloy (Fig. 2) and this is despite a carbon content almost 3-times higher than that of sulphur. This seems to suggest that the effect of sulphur on the grain boundary mobility is stronger than that of car- bon. One possible explanation is the easier diffusivity of carbon, an interstitial element, that enables the carbon atoms to follow migrating grain boundaries in iron, in
M. Tacikowski et al./ Materials Science and Engineering A205 (1996) 133 138 137
contrast to the sulphur atoms, which can hardly be dragged after the migrating grain boundary. Another possibility is the pinning effect of small grain boundary sulfide precipitates. The stronger effect of sulphur on grain boundary mobility could explain the higher pro- portion of LANGB, which, by their nature are rather "slow" [8,11]. However, the nature of the grain boundaries after extensive grain growth is still the subject of discussion [8,12,21]. For example, the grain growth computer simulations of Shibayangi et al.  show that low mobility grain boundaries should be retained during grain growth. The case of the iron car- bon-sulphur alloy seems to be much more complicated. The behavior of this alloy, which is characterized by the smallest grain size, but which has very similar fractions of CSLGB and LANGB to those in the iron-carbon alloy, cannot be interpreted in terms of a higher fre- quency of low mobility grain boundaries. The strongest grain refinement effect seems to be a consequence of a synergetic action of carbon and sulphur, the nature of which is not clear at present. The interpretation requires more data concerning segregation on grain boundaries. The few available experimental results [16,22,23] are insufficient to give a reasonable estimate of the levels of carbon and sulphur segregation at the grain boundaries in the present alloys. According to the work of Suzuki et al. , on materials of similar composition and purity to those examined here, one can assume that there is a low level of sulphur segregation accompanied by cosegregation of carbon. However, there are no experimental data about the kinetics of the observed phenomena of segregation competition between carbon and sulphur and in what conditions the equilibrium of segregated species on grain boundaries is achieved.
It is difficult to explain why the grain size diminishes so markedly for the Fe C-S alloy. However, taking into account the fact that the CGBP (Fig. 4) of this alloy is closer to that of the iron-carbon alloy than that of the iron-sulphur alloy suggests that it is carbon which controls the behavior of this alloy. Our unpub- lished studies on a similar iron-carbon-sulphur alloy composition  indicate that carbon, having higher diffusivity than sulphur, segregates first to the grain boundaries and then is displaced to the matrix by the competition with sulphur segregation. It may signify that recrystallization, at least in the early stages, is controlled by carbon segregation.
Low contents of carbon (150 and 200 wt. ppm) and sulphur (75 wt. ppm) influence the CGBP in recrys- tallized ultra high purity c~-iron. The fractions of particular coincidence grain boundaries are differ- ent for each alloy after the same heat treatment.
2. The Fe-S alloy exhibits a much higher proportion of LANGB than the other alloys. In this alloy, containing 75 ppm of sulphur, the fraction of RANGB decreases to about two-thirds of the total boundary population. This effect of sulphur only occurs in the absence of carbon. It should be also noted that the 2;3 and I25 boundaries are absent in the Fe-S alloy.
3. In the iron-carbon-sulphur alloy, carbon seems to control the CGBP, as the fraction of LANG re- mains similar to that of the iron-carbon alloy. The grain size in this alloy is smaller than those of the iron carbon and iron-sulphur alloys, probably as a consequence of a synergetic effect of these ele- ments.
4. The influence of carbon and sulphur on the GBCD is probably related to the selection of particular types of grain boundaries via specific dragging effects on the migration of boundaries during re- crystallization and subsequent grain growth. This may be due to the sulphur atoms in solid solution or small sulfides precipitates.
The authors would like to acknowledge financial support through a grant of the State Committee for Scientific Research (Poland) No. 3 0173 91 01, and also C. Rinaldi, P. Jouffrey and J. Mizera of the Ecole Nationale Superieure des Mines de Saint-Etienne for their precious help in carrying out the BKD investiga- tions. Interesting discussions with Prof. Tadao Watan- abe of Tohoku University in Japan are also acknowledged.
 Proc. Int. Cong. on lntergranular and lnterphase Boundaries in Materials, Paris, Sept. 4 8, 1989, Colloque CI, suppl, to No, 1, 51, 1990, pp. 1 378.
 Proc. 6th Int. Cong. on Intergranular and Interphase Boundaries in Materials, Thessaloniki, Greece, June 21 26, 1992, Mater. Sci. Forum, 126-128 (1993) 5-292.
 A. Kobylanski and C. Goux, C.R. Acad. Sci. Paris, Series C, 272 (1971) 1937.
 H. Kurishita, A. Oishi, H.Kubo and H. Yoshinaga, Trans. JIM. 26 (1985) 341.
 L.C. Lim and R. Raj, Acta Met., 34 (1984) 1183.  T..Watanabe, S.Shima and S. Karashima in M.H. Kamdar (ed.),
Embrittlement by Liquid and Solid Metals, AIME, 1984, p. 161. [7l T. Watanabe, Res. Mechanica, I1 (1984)47.  M.W. Grabski, J. Phys., 46 (1985) C4-567.  W.A. Swigttnicki and M.W. Grabski. Mater. Sci. Eng., 100
(1988) 85-100.  K.T. Aust and G. Palumbo, Proc. Int. Syrup. on Advanced
Structural Materials, Montreal, Canada, August 22 26, 1988, Vol. 9, CIM Proc., Pergamon, Toronto, 1989, p. 215.
138 M. Tacikowski et al. / Materials Science and Engineering A205 (1996) 133-138
 T. Watanabe, Mater. Sci. Forum, 46 (1989) 25.  B. Elm'Rabat, Ph.D. thesis, Universit6 Paris XI, Paris, 1986.  D. Farkas, M.O. Lewus and V. Rangarajan, Scripta Met., 22
1988 1195.  M. Tacikowski, A. Kobylanski and M.W. Grabski, Colloque CI,
suppl, to No. 1, 51 (1990) 653.  M. Tacikowski, Ph.D. thesis, EMSE-INPG, Saint-Etienne, 1986.  S. Suzuki, S. Tanii, K. Abiko and H. Kimura, Metall. Trans.,
18A (1987) 1109.  A. Akef, Ph.D. thesis, EMSE-INPG, Saint-Etienne, 1993.
 A. Garbacz and M.W. Grabski, Arch. Nauki Mater., 11 (1990) 199.
 M.B. Kasen, Acta Metall., 31 (1983) 489.  J.W. Cahn, Acta. Metall., 10 (1962) 789.  T. Shibayanagi, H.Takatani and S. Hori, Mater. Trans. JIM, 32
(1991) 25.  C.J. Briant, Acta Metall., 33 (1985) 1241.  G. Tauber and H.J. Grabke, Ber. Bunsernges. Phys. Chem., 82
(1978) 298.  M.Tacikowski, unpublished work.