ISIJ International, Vol. 40 (2000), No. 12, pp. 1260-1268

Effect of Inclusion Size on the Nucleation of Acicular Ferrite in Welds

Tae-Kyu LEE,1) H. J. KIM, B. Y. KANG and S. K. HWANG2)

Korea Academy of Industrial Technology (KITECH), ChonAnSi, 330-825, Korea. 1) Formerly at KITECH. Now at Univ. of Calif., Berkely, CA, USA. 2) Department of Metallurgical Engineering, Inha University, Inchon, 402-751, Korea.

(Received on June 26, 2000; accepted in final form on August 17, 2000)

Low-carbon steel weld with a high density of oxide inclusions prepared using a experimental metal-cored wire has been examined to study the effect of inclusion size on the formation of acicular ferrite, and to un-derstand the role of inclusion in the nucleation of ferrite lath. Depending on the ferrite morphology associat-ed with inclusions, a total of 282 inclusions observed under TEM could be classified into two groups, i.e. the non-nucleant and the nucleant. Experimental results showed that the group of inclusions acted as nucle-ant were appreciably larger in size compared with those of non-nucleant resulting in the increased probabili-ty of nucleation with the increase of inclusion size, even though the chemical and structural natures ap-peared to be the same. The group of nucleant-inclusion was further divided into two types depending on the degree of nucleation, which was evaluated by the number of ferrite lath nucleated. Statistical analysis performed on inclusion size indicated that the larger the inclusion size is the more ferrite laths could be nu-cleated. Those laths nucleated from a large single inclusion have grown in many different radial directions and mostly had a different crystallographic orientation from those of adjacent ferrite laths. As a result of this study, it is demonstrated that larger inclusions are indeed more potent nucleation sites when compared with those of smaller size. Thus it could be concluded that the provision of the inclusion surface as for the inert surface for the heterogeneous nucleation of acicular ferrite lath would be the principal role of inclu-sions playing in the weld metal of low alloy steels. Other possible mechanisms were also considered, but they were unlikely to be operated in the present weld metal system. KEY WORDS: acicular ferrite; inclusion; heterogeneous nucleation.

1. Introduction Acicular ferrite is one of the microstructural constituents

which is most commonly observed in the weld deposit of low-alloy steel and is easily differentiated from other con-stituents by its fine looking nature. 1 , 2 ) Because of the fine inter-locking nature, this microstructure has been known to impart high toughness and strength 3 , 4 ) and a high propor-tion of acicular ferrite is generally desirable for improving the mechanical properties of weld metal. Recently, there have been efforts 5 - 7 ) to apply acicular ferrite microstructure even for developing steel plates by offering favorable con-ditions for its formation in their heat-affected zone (HAZ) and thus for improving the toughness in the region of grain-coarsened region of HAZ near fusion boundaries. For these reasons, particular importance has been given to understand the formation mechanism and the factors which affect the nucleation of acicular ferrite in steel welds.

Previously it has been well established that the chemical composition and cooling rate are most important factors controlling the amount of acicular ferrite formed. 8 1 1 ) A lot of authors also have provided convincing evidences that acicular ferrite is essentially intragranularly nucleated ban-ite and is mainly nucleated at inclusions in the we ld . 1 2 1 4 )

Accordingly, if thermal and chemical conditions in the weld

favor the ferrite transformation, acicular ferrite will nucle-ate at inclusions along with other microstructural con-stituents.

However, there are a lot of confusions in the role of the inclusions for ferrite nucleation. More specifically, it is not clear which type of inclusions are preferable and what char-acteristics of inclusions are required in order to nucleate acicular ferrite and to promote its nucleation. Various stud-ies have reported that acicular ferrite was nucleated by cer-tain type of inclusion composition and phase. Fine oxide in-clusions, 8 , 1 5 , 1 6- 1 boron nitrides precipitated on rare earth metal oxysulfides,1 7 ) aluminum-rich inclusions,1 8 - and tita-nium nitride 1 5 , 1 9 ) have all been suggested as nucleant inclu-sions. Bhatti et a/.18) has reported that the inclusions rich in aluminium were more favorable than those of manganese-rich and concluded the chemical composition would be the primary factor controlling the their efficiency as nucleants of acicular ferrite formation. More recently, however, Dowling et a/.20) failed to find any correlation between the amount of acicular ferrite and either inclusion composition or phases.

Early workers 2 1 - 2 3 ) also have suggested that low mis-match between the inclusion and ferrite would promote nu-cleation, i.e., those inclusions which show the best 'lattice matching' with ferrite are most effective in nucleating the

© 2000 ISIJ 1260

ISIJ International, Vol. 40 (2000), No. 12

ferrite. It has even been suggested 2 4 ) that there may exist orientation relationships between inclusions and the ferrite laths that they nucleate and has been shown experimentally in the V-containing steel.5 ) Another experiments performed in various weld metals, however, failed to demonstrate the presence of a reproducible ferrite/inclusion orientation rela-tionship 2 0 ) and no evidence has been established in weld metal yet.

Another idea that has been suggested 2 5 ) was the thermal-ly induced stress, that can be developed due to the differ-ence in the thermal expansion coefficient between the austenite and inclusion. This idea is largely based on data showing a large difference in thermal expansion coefficient between austenite and several types of inclusions. If this mechanism was operative, then it would have a significant effect on the local hardenability and hence transformation behavior. However, this idea has been rejected by other re-searchers, 2 6 , 2 7 ) who calculated the resultant strain energy and found it was very trivial compared with the transforma-tion energy.

Besides, there are two more mechanisms favorably being studied as the major role of inclusions for acicular ferrite formation, i.e. the provision of inert surface and the devel-opment of chemical variations near inclusion surface. Localized variations in composition, in particular man-ganese, were initially suggested by Farrar and Watson, 2 8 )

but their idea were rejected by Ricks et a/. 8 ) and by Bhatti et a/. 1 8 ) because they could not find the direct evidence of any variation in compositions of Mn and Si in the close proxim-ity of the inclusions studied. Accordingly, it was concluded that the major effect of weld-metal inclusion was through their ability to reduce the energy barrier to nucleation by acting as an inert substrate. 8 , 2 0 ) Krauklis et a/.29) showed that a minium particle size of 0.4-0.6 /mm was required for acicular ferrite nucleation in the two different welds and agreed to the role of inclusion as an inert substrate. More recently, however, Mabuchi et a/ . 3 0 ) have succeeded to get a direct evidence of Mn depletion at the interface of MnS/fer-rite matrix by nano-probe analysis. They believed Mn-de-pletion was attributable to the formation of MnS precipitate on (Mn, Si) oxide inclusion during cooling and considered this feature to be one of the most important factors to pro-mote acicular ferrite. Another possibility for Mn-depleted region has been shown by Shim 2 7 ) and others, 3 1 ) who found such region at the diffusion-bonded interface between a layer of Ti 2O 3 powders and a low carbon steel. They be-lieved that a large number of cation vacancies in Ti 2O 3

structure facilitated the absorption of Mn from surrounding steel matrix, thereby forming a Mn-depleted region around the particle.

From this brief review of the literature, it is obvious that the role of inclusions in the formation of acicular ferrite is still far from being understood and often directly contradic-tory. Therefore, this study was intended to study the influ-ence of inclusion in terms of 'inert substrate', 'lattice matching' and 'chemical variation', and to suggest more feasible role of inclusions for the formation of acicular fer-rite in the weld metal.

Fig. 1. Geometry of welded plate and the location of specimens taken for the study.

2. Experimental Procedure All the specimens studied in this investigation were taken

from the last capping pass of the weld made with experi-mental metal-cored wire, as shown in Fig. 1. Welding was performed on 25 mm thick plate with Ar+2%O 2 gas shield-ing in the flat position. The nominal welding condition was 230A, 20V and 25 cm/min (heat input= 1.0kJ/cm) and the interpass temperature was about 150°C. The resultant weld had a chemical composition of Fe-0.042C-0.40Si-1.51Mn-0.015P-0.01 S-0.02Al-0.01 Ti-2.3Ni(wt%) with 42 ppm N and 820 ppm O, showing a relatively high oxygen content compared with that of commercial products.

For light optical microscopy, welds were sectioned trans-versely and mechanically polished prior to etching in 2% nital solution. This specimen was also used for the high magnification observation using scanning electron micro-scope (SEM). To make thin-foil transmission electron mi-croscope (TEM) specimens, samples were machined out precisely from the last bead region of weld metal in the form of 3 mm diameter round bar. These bar specimens sliced into discs of 0.25 mm in thickness using a diamond cutter. The discs were mechanically ground and then thinned in a twin-jet electro-polisher using an electrolyte consisting of 10% perchloric acid and 90% glacial acetic acid at -30°C.

TEM of Philips CM200 equipped with Energy Dispersive X-ray Spectroscopy (EDS) were employed to study the details of microstructure formed around the inclu-sions and to examine the compositional variation in inclu-sions. Inclusion sizes were measured from the bright-field micrographs taken in TEM, operated at 200 kV Inclusions smaller than 0.05 /m were not counted in this investigation. The details of lath arrangement around the inclusions were also studied in TEM, and the crystallographic orientation of each ferrites nucleating from inclusions was determined by analysing their selected area diffraction (SAD) patterns ob-tained from each lath. To study the chemical characteristics of inclusions, EDS analysis was conducted both in SEM and TEM. The experimental procedure involved point analysis, line scanning and X-ray mapping to reveal the de-tails of inclusion constituent phases.

3. Results 3.1. Weld Microstructure and Inclusions

The optical microstructure of the weld is illustrated in Fig. 2, showing a microstructure developed in as-deposited

1261 © 2000 ISIJ

ISIJ International, Vol. 40 (2000), No. 12, pp. 1260-1268

Fig. 2. Optical micrographs of as-deposited weld metal taken in (a) low magnification and (b) high magnification.

Fig. 3. SEM micrographs (a) showing inclusions of various sizes and (b) illustrating ferrite nucleation from inclusion.

condition. This weld exhibits a microstructure that consists of predominantly acicular ferrite (over 80%). Prior-austen-ite grain boundaries are decorated with allotriomorphic fer-rite while the interior of the austenite grains is packed with acicular ferrite. This microstructure is characterized with its fine interlocking morpology as shown in a high-magnifica-tion optical micrograph, Fig. 2(b). SEM observation with higher magnification could reveal the imbedded inclusions as shown in Fig. 3 (a). Almost all of the inclusions observed were spherical in shape, with a size up to about 3 / m . In this figure, some of the inclusions appeared to be acted as nucleants for acicular ferrites, i.e. several ferrite laths look like to be fanned outwards from the inclusions. This feature becomes more clear in the high magnification micrograph of Fig. 3(b), which shows a ferrite lath tightly bonded on the inclusion surface. It is also noted in Fig. 3(a) that the nucleant inclusions are generally larger in size compared with the non-nucleant inclusions.

Preliminary examination with SEM-EDS has been con-ducted on the inclusion chemistry. The typical result is shown in Fig. 4, indicating that the present inclusions are rich in oxygen, silicon, manganese and aluminium with a small amount of titanium and sulfur. Both the point analysis and line scanning showed that all the inclusions, regardless of the size, were chemically homogeneous and were absent from any chemical segregation. This is consistent with the work of Bhatti et a/ . 1 8 ) The relative proportion of the major elements varied slightly from inclusion to inclusion but they were distributed quite uniformly throughout the inclu-sion. Therefore, it was assumed that the major metallic ele-ments were combined with oxygen to give single phase oxide within the MnO-SiO 2-Al 2O 3 system. More details of the chemical nature observed with ED S/TEM will be de-

Fig. 4. Typical SEM-EDS spectrum taken from inclusion.

scribed later. A typical microstructure of acicular ferrite observed

under TEM is shown in Fig. 5. As shown in this figure, the acicular ferrite constituent consists of fine ferrite laths or plates and the individual lath contains a relatively high den-sity of dislocations, which was noted to be significantly higher than that of grain boundary ferrite. In this micro-graph, one can see three inclusions different in size signifi-cantly. Through the close examination, it can be realized that the one located in the center, which is the largest, is surrounded by a number of ferrite laths emanating from the inclusion while the other inclusions are just located either within the ferrite lath or on the lath boundary. These two small inclusions seem to be not relevant with ferrite nucle-

© 2000 ISIJ 1262

ISIJ International, Vol. 40 (2000), No. 12

ation. Another TEM micrograph in Fig. 6 shows the growth morphology of nucleating ferrite; the ferrite lath nucleating from inclusion 'A' grows in a long distance and its width increases as it grow. During its growth, it can sweep over other inclusions resulting in other inclusions to be non-nu-cleant.

Being aware of such a morphological difference of fer-rite, authors were able to classify all the inclusion particles revealed in TEM micrographs into two groups, i.e. the nu-cleant and the non-nucleant. Each group was further divid-ed into two different types as shown and schematically drawn in Fig. 7. Firstly, type 1 and type 2 are the inclusions located within the ferrite laths and on the lath boundaries, respectively. Both types of inclusions can be reasonably re-garded as the non-nucleant of ferrite because of the absence of ferrite emanation. Secondly, inclusions classified as type 3 are the ones acting as nucleation site but the number of lath nucleated is few, either one or two, so that not more

Fig. 5. TEM micrograph showing ferrite lath structure associat-ed with inclusions.

than a half of the inclusion surface is covered by nucleated ferrite. The last type of inclusion classified as type 4 is ac-companied with a large number of ferrite laths emanating in all directions and thus covering whole inclusion surface. TEM/SAD analysis on each of these laths confirmed that these were separated largely by high-angle boundaries, as will be described later. 3.2. Distribution of Inclusion Size

A total of 282 inclusions observed under TEM micro-graphs were classified into four different types following the guidelines described above. Among these inclusions, 108 inclusions (38% of the total) were classified to type 1. Type 2, 3 and 4 were recorded as 73 (26%), 73 (26%) and 28 (10%) respectively. It implies that over 60% of inclu-sions (type 1 and type 2) belong to a group of non-nucleant.

Figure 8 shows the size distribution of each type, mea-sured in two-dimensional TEM micrographs. The average size of the total inclusions was calculated to be 0.51 /mm. The data in Fig. 8 show that each type of inclusion has a wide variation in size, overlapping with other types. However, it is clear that the average size increases with in-

Fig. 6. TEM micrograph showing the growth morphology of nu-cleated ferrite.

Fig. 7. Typical TEM micrographs and schematic drawings of 4 different types of inclusions classified by accompanying ferrite structure.

1263 © 2000 ISIJ

ISIJ International, Vol. 40 (2000), No. 12, pp. 1260-1268

Fig. 8. Size distribution of each type of inclusion measured under TEM micrographs.

creasing the number of type, i.e. 0.33, 0.51, 0.65 and 0.83 /m for type 1, 2, 3 and 4 respectively. It is also inter-esting to note in Fig. 8 that a minimum inclusion size of about 0.3 and 0.5 /m seems to be required for being the type 3 and type 4 nucleation respectively, but this require-ment is not sufficient for being these types because a lot of inclusions over these sizes are being classified as either type 1 or type 2. For this reason, the probability of inclu-sions for nucleating acicular ferrite lath at a given size, di, was estimated by dividing the sum of type 3 and 4 inclu-sions by the total number of inclusions, as shown in follow-ing equation.

Probability of nucleation at di

(number of type 3) d + (number of type 4) d

(total number of inclusions)d

Figure 9 documents the calculated result as a function of inclusion size and shows the increased probability as the in-clusion size increases. This result is quite consistent with that of previous workers, 2 9 ) who performed TEM studies in the hot-rolled weld metals subsequently heat-cycled for simulating heat affected zone. Although there is some scat-ter in data at about 1 / m , it is quite evident that the inclu-sion size affects ferrite nucleation in a systematic way, showing a transition from non-nucleant to nucleant with in-creasing the inclusion size. When the inclusion size is less than 0.2 / m , the probability for ferrite nucleation is nearly zero. With increasing the size over 0.2 /mm, the probability enhances markedly and reaches 1.0 at about 1.1 / m . This result reflects the fact that the larger the inclusion size is the more potential the inclusion has for ferrite nucleation. It is also important to note that only a portion of inclusions is acting as nucleation sites of ferrite. In the present weld, only 36% of total inclusions, classified as type 3 and 4, are playing an actual role for ferrite nucleation but able to gen-erate a weld microstructure containing acicular ferrite of

Fig. 9. Effect of inclusion size on the probability of ferrite nu-cleation.

over 80%. 3.3. Crystallographic Orientation of Ferrite Lath

Due to the absence of grain boundary-like features in acicular ferrite, it is difficult to trace the boundaries be-tween areas of different crystallographic orientation in the bright-field images. However, this difficulty can be solved by taking selected area diffraction (SAD) patterns and dark-field (DF) images using TEM.

The TEM micrographs of acicular ferrite accompanying with both type 1 and 2 inclusions are shown in Fig. 10. The bright-field micrograph contains two parallel laths, lath 'A' and lath 'B', that are different in width. Note that type 2 in-clusion is located on the lath boundary of these two laths. SAD patterns taken from each lath show that both have a identical orientation, i.e. [315]. It follows that they are sep-arated by low angle boundaries and thus act as a single crystal against the orientation-sensitive fracture such as brittle cleavage. Such an arrangement of laths as a single unit is commonly observed within the packet of lath martensitic structure, which takes no account of inclusions

© 2000 ISIJ 1264

ISIJ International, Vol. 40 (2000), No. 12

Fig. 10. TEM micrographs containing inclusions of type 1 and type 2: (a) SAD pattern from lath 'A', (B) bright-field image containing two inclusions and (c) SAD pattern from lath 'B'.

Fig. 11. Results of TEM analysis performed on type 3 inclusion: (a) bright-field micrograph and (b) its selected area dif-fraction pattern showing two different orientations, each of which is operating in the area illustrated in the insert-ed figure.

during its transformation. Therefore, it is expected that those two inclusions were swept over by the ferrite laths nu-cleated from other locations, as was illustrated in Fig. 6.

Figure 11 shows the typical TEM image of type 3 inclu-sion having a single nucleation of ferrite and the SAD pat-tern taken from the surrounding area. The SAD pattern was indexed into two different orientations of [012] and [113]. The areas operating each orientation were identified by dark field technique and the resultant area map is drawn in the inserted figure. As shown in this map, each orientation is bounded by lath boundaries that appear in the bright-field image. The ferrite lath nucleating from the inclusion has [012] orientation while the rest of area is in [113] orienta-tion. Accordingly, the lath boundaries revealed in Fig. 11(a) can be characterized as high-angle boundaries.

Crystallographic analysis on type 4 inclusion was also conducted with the same procedure. The bright-field image in Fig. 12 (a) shows typical micro structure of ferrite laths nucleating from a single inclusion of type 4. A total of 8 laths can be figured out by tracing the lath boundaries. Because of the large number of ferrite laths, SAD patterns taken from all the laths were very complicated. However, it could be safely indexed by combining 8 individual patterns taken from each lath. Finally, the SAD pattern was turned out to be comprised of 4 different orientations, i.e. [001]1, [001]2, [111] and [112], as shown in Fig. 12(b). There is an rotational difference of about 13 degrees between [001]1

and [001]2. Each lath was assigned its orientation and the

resultant orientation map is presented in the inserted figure. This figure shows that several laths have same orientation with its neighbours but most of them have not. Accordingly, the lath boundaries can be regarded as either low- or high-angle boundaries. These results indicate that each lath nu-cleates individually with a different orientation to some ex-tent from its neighbors and grows in radial direction, result-ing in a inter-locking nature of acicular ferrite. 3.4. Inclusion Composition

TEM-EDS analysis was carried out on the thin-foil spec-imens to determine the inclusion composition and to figure out the constituent phases. Inclusions of varying size and type were examined but all the inclusions observed were chemically homogeneous with some difference in the frac-tion of the major elements as was found previously in SEM. The homogeneous nature of inclusions classified type 1 and type 4 is shown in Fig. 13. Regardless of the inclusion type and size, all the major elements even including sulfur are uniformly distributed within a inclusion, reflecting a very complex deoxidation product. Considering that the iron is not a characteristic of the present inclusion but the result of beam-matrix interaction, the present inclusion would be a mixture of SiO 2-MnO-Al 2O 3 containing a small amount of titanium and sulfur.

Many authors have reported the heterogeneous nature of inclusions in weld metal, i.e. the existence of several phases within a single inclusion generally in the form of oxide par-

1265 © 2000 ISIJ

ISIJ International, Vol. 40 (2000), No. 12, pp. 1260-1268

Fig. 12. Results of TEM analysis performed on type 4 inclusion: (a) bright-field micrograph and (b) its selected area dif-fraction pattern showing four different orientations, each of which is operating in the area illustrated in the in-serted figure.

Fig. 13. TEM-EDS mapping images of various elements taken from type 1 and type 4 inclusions.

ticles partly coated by manganese sulfide 3 0 , 3 2 ) or by copper sulfide. 3 3 , 3 4 ) However, in the present study it was failed to find such inclusions formed with sulfide coating. Although the exact reason of this has not been identified, the present inclusion may be the type which has low precipitation ratio of MnS. 3 5 , 3 6 )

4. Discussion 4.1. The role of Inclusion

The results of present study have confirmed that the acic-ular ferrite constituent of weld metal microstructure is com-prised of intragranularly nucleated ferrite and that the pri-mary nucleation sites were oxide inclusions. Besides, the present study demonstrated that the inclusion size indeed is

an important factor for ferrite nucleation and the larger the inclusion size is the higher potential the inclusion has for ferrite nucleation. As for the nucleants, the number of fer-rite nucleation tends to increase with inclusion size.

However, the questions still arise as to why and how in-clusion size affects the ferrite nucleation. One of the possi-ble explanations is a 'inert substrate model' suggested by Ricks et a/. 8 ) This model is based on the classical theory of heterogeneous nucleation, considering the inclusion to be inert, incoherent and non-deformable. The major findings obtained from this model can be summarized that nucle-ation at inclusion is always more favorable than homoge-neous nucleation but less favorable than nucleation at austenite grain boundaries. The more important thing to note is that the energy barrier to nucleation at inclusion

© 2000 ISIJ 1266

ISIJ International, Vol. 40 (2000), No. 12

lowers rapidly with increasing the size of inclusion up to about 1 jjm. It indicates that the larger inclusion is energeti-cally more favorable site for ferrite nucleation. This is quite consistent with the present experimental results, Fig. 9, which confirmed that larger inclusions are indeed more po-tent heterogeneous nucleation sites when compared with smaller size. As a result, it can be concluded that the princi-pal role of the particles in nucleation may be simply to pro-vide an inert surface, in agreement with proposals by Dowling et al . 2 0 ) and by Krauklis et al.29)

Nucleation assisted by inert surface is a heterogeneous process in which interfacial energy at the nucleating inter-face is one of the major controlling factors. Accordingly, an inclusion will be more favorable nucleation site if it has a higher interfacial energy with austenite matrix. Using the general assumption that surface energy increases with melt-ing temperature, then high melting phases such as TiO, Al 2 O 3 and SiO2 would be expected to have high surface energies. 2 0 ) In the weld studied, the inclusion presented a single phase but showed some variation in chemistry. Presumably, even with the same size, each inclusion will have a different chemistry and thus have a different surface energy. A local variation in chemistry to a less extent also can be expected along the inclusion surface. If this is so, then nucleation potential will be different from inclusion to inclusion at a given size and from location to location even at a single inclusion. 4.2. Other Mechanisms

As discussed in the introduction, several other mecha-nisms have been proposed for ferrite nucleation at inclu-sions. One of them is the nucleation by low mismatch inter-faces between nucleant and ferrite, which is also based on the heterogeneous nucleation. In fact, inclusion nucleation will be promoted further if the ferrite can adopt a rational orientation relationship with both the austenite and inclu-sion. When nucleation proceeds by creation of low mis-match interfaces, the inclusion will have a certain orienta-tion relationship with the inclusion but not with the austen-ite matrix. Thus all the ferrite laths nucleated from a single inclusion, if it is a homogeneous single phase which is most likely the case of the present study, should show a same ori-entation which is determined by the inclusion orientation. However, the present result shown in Fig. 12 demonstrates that four different orientations can be produced from a sin-gle inclusion. It implies that ferrite nucleates without hav-ing any crystallographic relation with inclusion and thus above proposal would not seem to be applicable.

In reality, the inclusions formed in the liquid steel are randomly oriented with respect to surrounding austenite and that the acicular ferrite has a limited number of orienta-tion relationship with the parent austenite due to its charac-teristic of banitic transformation. It necessarily implies that the inclusion/ferrite orientation relation also has to be ran-dom. In view of the nature of the inclusion, again above proposal based on low mismatch interface would appear to be highly unlikely.

Another possible mechanism is the nucleation favored by the Mn-depleted layer possibly formed around the inclu-sions during cooling either by the secondary precipitation of MnS 3 0 ) or by Mn diffusion into inclusions of Ti 2 O 3 . 2 7 , 3 1 )

Both of them has been shown experimentally but only in the heat-treated samples. No convincing evidence has been reported in the real weld metal. Considering the limited time available for Mn diffusion during cooling in weld ther-mal cycle, it seems to be very difficult to form Mn-depleted region. Yet, it will be more difficult to be developed without having either MnS precipitates or Ti 2O 3 particles. Therefore, considering the absence of such phases in the present weld the authors have concluded that nucleation by Mn-depleted interfacial layer does not occur in the weld metal presently studied.

5. Conclusion The effect of inclusion size on ferrite nucleation was

confirmed quantitatively through the extensive TEM study. The experimental results shows that the larger inclusions are more preferable sites for the nucleation of acicular fer-rite, resulting in transition from non-nucleant to nucleant with increasing inclusion size. As for the nucleant inclu-sion, the increase in size promotes multiple nucleation of ferrite laths having different crystallographic orientations. Such an effect of inclusion size is in good agreement with the expectation made by the classical theory of heteroge-neous nucleation. Because the present inclusions are homo-geneous and single phase in chemical nature, the increased probability of ferrite nucleation and the increased number of ferrite nucleation with inclusion size strongly suggest that the major role of inclusion is to provide the inert sub-strate for ferrite nucleation in weld metal. Acknowledgments

Authors would like to acknowledge Pohang Steel Company (POSCO), Korea for the financial support and also wish to thank Dr. W. Y. Joo of POSCO and Dr. K. Nagai of NRIM in Japan for their technical discussions.

REFERENCES

1) D. J. Abson and R. E. Dolby: IIW Doc. IXJ-29-80, (1980). 2) O. Grong and D. K. Matlock: Int. Met. Rev., 31 (1986), 27. 3) D. J. Abson and R. J. Pargeter: Int. Met. Rev., 31 (1986), 141. 4) R. A, Farrar and P. L. Harrison: J. Mater. Sci, 22 (1987), 3812. 5) F. Ishikawa, T. Takahashi and T. Ochi: Metall. Mater. Trans. A, 25A

(1994), 929. 6) I. Madariaga, J. L. Romero and I. Gutierrez: Metall. Mater. Trans. A,

29A (1998), 1003. 7) J.-H. Shim, Y. W. Choo, S. H. Chung, J.-D. Shim and D. N. Lee: J

Kor. Inst. Met. Mater., 36 (1998), 1993. 8) R. A Ricks, P. R. Howell and G. S. Barritte: J. Mater. Sci., 17 (1982),

732. 9) R. A. Farrar and P. L. Harrison: J. Mater. Sci., 22 (1987), 3812.

10) S. Liu and D. L. Olson: Weld. J, 65 (1986), 139-s. 11) R. A Farrar, Z. Zhang, S. R. Bannister and G. S. Barritte: J. Mater.

Sci., 28 (1993), 1385. 12) A. A. B. Sugden and H. K. D. H. Bhadeshia: Metall. Trans. A, 20A

(1989), 1811. 13) H. K. D. H. Bhadeshia and J. W. Christian: Metall. Trans. A, 21A

(1990), 767. 14) S. S. Babu and H. K. D. H. Bhadeshia: Mater. Trans., JIM, 32

(1991), 679. 15) C. I. Choi and D. C. Hill: Weld. J., 57 (1978), 232-s. 16) N. Mori, H. Homma, S. Okita and M Wakabayashi: IIW Doc. IX-

1196-81, (1981). 17) T. Funakoshi, T. Tanaka, S. Ueda, M. Ishikawa, N. Koshizuka and K.

Kobayashi: Trans. Iron Steel Inst. Jpn., 17 (1977), 419-427.

1267 © 2000 ISIJ

ISIJ International, Vol. 40 (2000), No. 12, pp. 1260-1268

18) A. R. Bhatti, M. E. Saggese, D. N. Hawkins, J. A. Whiteman and M. S. Golding: Weld. J, 63 (1984), 224-s.

19) Y. Ito and M. Nakanishi: IIW Doc. XII-A-113-75, (1975). 20) J. M. Dowling, J. M. Corbett and H. W. Kerr: Metall. Trans. A, 17A

(1986), 1611. 21) T. H. North, H. B. Bell, A. Koukabi and I. Craig: Weld. J., 58 (1979),

343-s. 22) N. Nori, H. Homma, S. Okita and M. Wakabayashi: IIW Doc. IX-

1196-81, (1981). 23) Steels, Microstructure and Properties, 2nd ed., ed. by R.

Honeycombe and H. K. D. H. Bhadeshia, Edward Arnold, London, (1981), 148.

24) A. R. Mills, G. Thewlis and J. A. Whiteman: Mater. Sci. Technol., 3 (1987), 1051.

25) L. Devillers, D. Keplan, B. Marandet, A. Ribes and P. V Ribond: Proc. of Int. Conf. on The effect of residual, impurity and microal-loying elements on weldability and weld properties, The Welding Institute, Cambridge (1984), Paper 1.

26) C. B. Dallam and D. L. Olson: Weld. J., 68 (1989), 198-s.

27) Jae-Hyeok Shim: Ph.D. thesis, Seoul National University, Feb. (2000).

28) R. A. Farrar and M. N. Watson: Metal Construction, 17 (1985), 374R.

29) P. Krauklis, F. J. Barbaro and K. E. Easterling: Proc. of Int. Conf. on Martensitic Transformations, Monterey Institute for Advanced Studies, Monterey, (1992), 439.

30) H. Mabuchi, R. Uemori and M. Fusioka: ISIJ Int., 36 (1996), 1406. 31) J. M. Gregg and H. K. D. H. Bhadeshia: Metall. Mater. Trans. A,

25A (1994), 1603. 32) A. O. Kluken and O. Grog: Metall. Trans. A, 20A (1989), 1335. 33) S. G. Court and G. Pollard: J. Mater. Sci. Lett., 4 (1985), 427. 34) E. S. Kayali, J. M. Corbett and H. W. Kerr: J. Mater. Sci. Lett., 2

(1983), 123. 35) S. Ogibayashi: Advances in technology of oxide metallurgy, Nippon

Steel Technical Report, No. 61, (1994), 70. 36) M. Wakoh, T. Sawai and S. Mizoguchi: Tetsu-to-Hagané, 78 (1992),

1697.

© 2000 ISIJ 1268