Great!! Very good.

Thanks Dirk! The papers represent an amazing amount of work combining very expensive techniques :-). I am sure that you need to go that much into details (and I admit, I was asking for subblock and subgrains as well).

Let's going back to the more phenomenological aspects of this problem. In SEM and sometimes also light microscope you can see the packet structure, or blocks (possibly lathes). What all blocks (or lathes) have in common that they forming a packet? What all lathes have in common to form blocks? Is it the crystallographic direction (the lath-shaped habit suggests this ) or the transformation plane, or nothing of them? Maybe I am totally wrong, but to me it appears that the "macroscopic" morphology expresses some crystallographic correlation which might be very rough, and microscopically there are many questions you investigated them in a very convincing and comprehensive way. Maybe, my question is too simple that I do not find the answer in the paper (or do not spend enough time to derive it easily by myself investigating a few EBSD maps carefully).

With some reservation for varying terms, the classification is normally related to 24 variants of OR as follows. There are 3 Bain groups, each with a common axis of parent austenite and 4 cpp (close packed plane) groups, each with a common {111} plane of parent austenite. Within a common Bain group variants are relatively "weak" (approx 5 to 20 deg) dosoriented; variants of different Bain groups are considered ''highly'' (approx 50 to 62.5 deg) disoriented.

1) All variants of certain cpp group display approximately the same habit i.e. combine into the so called PACKET of roughly parallel elementary crystals -LATHS, normally visible (when etched) by optical microscopy.

2) Since each cpp group may contain martensite crystals of different (three) Bain groups "highly" misoriented to each other, a unique packet may contain "BLOCKS" of parallel laths (sometimes one lath) divided from other "Blocks" by 'high-angle'' boundaries. This important discovery by EBSD essentally developed our understanding of maretensite structure and properties (particularly fracture toughness).

3) However, even the same block (as defined above) may contain physically high-angle boundaries (10 to20 deg) distinct by optical microscopy, that sometimes evokes confusion. As to low-angle boundaries of 5 to 10 deg in the same Bain group, their visibility by TEM and even optical microscopy is related to very thin interlayers of retained austenite as well as associated carbides.

4) Laths of the same OR variant are still somewhat disoriented (less than 5 deg) and commonly are also distinct by TEM.

Dear A.A. Zisman,

only to make clear that I understood you correctly:

a) the packets define the alignment of the former {111}fcc austenite plans, i.e. they indicate the rough position of the {110}bcc

b) because of the 6 possibilities to align bcc blocks (variants) on a single {111}fcc, 6 different striations of a comparably appearing packet may occur, 3 rotated by 120° to each other and then another 3 which are about 10° rotated to the just mentioned 3, whereas this difference angle becomes 0° for N-W, and something in between for Greninger-Trojano.

c) within a single packet not only one of these 6 differently alinged blocks may exist, so that blocks with a different striation may occur.

d) lathes are parts of a single block which only show small misorientation angle. They are not the result of some orientation relationship (OR) variation but simply the consequence of mismatch and strain.

e) subgrains (or possible also subblocks) are further segmentations of lathes perpendicular to the lath axis.

f) all of them cause the observed orientation spread around the ideal poles expected from the crystallographically described and symmetry-consistent model

What puzzles me is the omnipresent explanation using Bain or Bain groups or circles (referring to the [001] pole figures). I agree that these rings are very fascinating, but I really do not see any necessary basis to refer to them since from a simple crystallographic point of view nothing is seriously discussed along [001]. The most successful models are all based on the imagination transferring closed or closely packed planes of fcc and bcc into each other...with a minimal movement of (nearest) atoms. And as far as I know, the Bain OR has been never discovered in steel. The required compression along c of 41% is far away from any realistic imagination, especially if one considers a polycrystalline material.

Is the discussion via Bain groups really beneficial, or simply present practice?

Your are mostly right, though your (d) point wants remarks. What you say is only one sort (Q

Thanks for your explanation. I already admitted that the circles are somehow fascinating. However, since the assumption of close-packed directions and planes are quite convincing I am not sure whether this model is helpful.

As example I attach an image which shows a microstructure of a plessite in an iron meteorite. Left I colored all variants belonging to a Bain group in one color (i.e. 3 colors), and right I did the same for each close packed plane {111} (i.e. 4 colors). I interpret the image in a way that the packets are a "better" indicated by an assumption of common close-packet planes than by the Bain groups. There are only a few connected fields existing which is clearly better with the assumption of the common use of close-packed planes.

No offense, one can try to find or use different models or descriptions, but why a special feature in a pole figure should be the reason to explain orientation relationships (which are anyway exclusively linked to atoms and actually not to crystallographic directions and planes) I really don't get. However, this might be related to my relatively limited understanding of this material science topic.

Dear Gert:

A ref figure is shown below. If we want to shown the packet structure of lath martensite, we will choose close packet plane {111}fcc to plot since the habit plane in the same packet is close to {111}fcc~//{011}bcc. That is clear as shown in your figure.

Why we concern other plots, because there are substructures in the packets, such as variant pairing (variant selection) etc, and these substructures vary with alloy composition, transformation temperature, external field etc. The combination of several plots will help us to visual the substructure directly.

BTW, using Bain circle doesnot mean there are Bain OR, but instead there are Bain correspondence (i.e. Lattice correspondence which is valid for a range of OR). By combination of actual OR and lattice correspondence, one can calculate the transformation matrix for fcc-->bcc transformation.

Best regards,

Xinfu

Dear Xinfu,

Thanks for your valuable explanation. However, for a better visualization of differently orientated blocks in a packet you don't need the Bain groups. You can simply use IPF plots which have the additional advantage that you see as well the crystallographic alignment (variant pairing) of the respective reference direction, cf. the attached images, although the used IPF encoding is based on the improper Laue-group coloring and a coloring of the rotational group would be better since it would really fit to the orientation description.

Bain circles don't mean Bain OR. This is definitely clear (the pole distribution looks much easier). I am only concerned about the continuous reference to the so-called Bain groups although the concerned variants do not have anything in common (except of a high-symmetrical appearing alignment in {001} pole figures, i.e. the same misorientation to the formerly existing parent phase). Please don't misunderstand me. We use {111} and as major criteria for the description of the transformations, and finally we group them using {001} - since accidentally there a ring like distribution appears...with the same misorientation angle? Atoms moving parallel to {111} keeping one of the directions unchanged. They don't know anything about [001]. They even do not know anything about symmetry which forces us to use centered lattices so that the [001] becomes a [001] , since it is the LATTICE direction parallel to the 4-fold symmetry.

Dear Gert,

Sorry for seemingly trivial words, I cannot agree with your unexpected statement that OR appears due to "atoms, not crystallography". Indeed, the OR, by definition, is nothing else but (approx) parallelism of certain CRYSTALLOGRAPHIC directions or/and planes! Similarly, I disagree that atoms "do not know" where to move in transformation. The latter happens because at certain conditions (temperature and cooling rate) the free energy of WHOLE LATTICE (appearing BCC/BCT/) become lower than that of parent FCC lattice. Thus, the whole array of atoms (not each individual one) has to collectively rearrange by small individual shifts. The Bain transformation (lattice deformation) is a unique indispensible way to do so; "further" deviations from the virtual Bain OR merely result from interaction of nascent martensite origins with surrounding austenite matrix. Anyway, in terms of MECHANICAL /stress conrolled/ interaction, the classical theory of martencite(PTMC) complies IN AVERAGE with enormous data (habit planes, OR, shears by individual laths or plates), when reducing interfacial misfits of Bain deformation with outer matrix by plastic(accomodative) shears within freshly appearing martensite. Of course, PMTC is oversimplified and cannot explain observed variety of (predicted set of possible) ORs and, more so, their dependence on chem. composition, cooling rate etc. Besides, this ignores that mechanical interaction of each plate (lath) with austenite essentially changes while the martensite fraction increses from zero to unity. With the latter remark, ORs are most probably predefined (decline from Bain) just at the nucleation stage by interaction in terms of reducing energy of inter-phase boundary , e.g. by allignment of close packed planes and directions of both phases (both sides from the boundary). As to specific questions:

1) Any Bain group (one of three compression axes in Bain transformation) only partly overlaps with (any) cpp group. That is why each packet of your meteorite is made of parallel laths (certain cpp group) belonging to various Bain groups.

2) After the Bain transformation happens, the mentioned allignments may go only by fitting one of 4 appearing {110}a to the closest {111}g, whereas within the fitted {110}a one of 2 a should approach the closest g. All these 4x2 options are SYMMETRIC (!!) relative to the Bain (compession) axis, whereas 3 such axes make 3x(4x2) = 24 variants of OR, whichever degree of fitting. Just the indicated symmetry is the reason for the circle-lke appearance of specific clusters in pole figures.

3) Note that each round cluster (above) for a=g is commonly associated with two rectagular ones corresponding to related a & a. Shapes of these rectangles are not symmetric since rotations of the LATTER two directions due to the above considered allignments are not symmetric relative to the "active" Bain axis a=g!

Dear A. A. Zisman, crystallography is the science of describing crystals. You can change the description, you can change the rules for the description...but not the behavior of the atoms. Therefore I said, atoms don't know anything about crystallography. We USE or DEVELOP crystallography to describe the atoms behavior in a proper/meaningful/reasonable way.

The definition of non-centered unit cells is also the reason, that we create artificially absent reflections which are not existent. If we would describe the same lattice by basis vectors defining a primitive unit cell, and there are infinite ways to do this, we would not have any absent reflection for ferrite and austenite anymore. But we would pay a high price since many descriptions would not be that convenient anymore. The first version of the International Tables still considered a H lattice type, and since then some of the descriptions of space group types have been adapted as well. Sometimes we are changing the perspective, and this might cause some changes in definitions. But I am sure that this is all very clear to you...

What I want point out is, that atoms and lattice directions/planes are for me two different things.

Anyway, you are the expert, and if you think the Bain concept is useful (what I do not doubt) or even necessary (where I have more doubts) then you are certainly correct. With my limited insight I do not see it that clearly at the moment. For me it is only a pleasant definition, which is also absolutely OK. Therefore I was asking :-).

Thanks a lot for your critical comments. I guess I have to think a bit more about your objections. Especially about the difference or identity of Bain axis and parent symmetry along {001}g.

to 1) it is clear that a Bain group cannot contain a ccp group entirely (or vice versa)

to 2) the symmetry comes from the parent phase which defines a 4mm (you can call this Bain axis). Thus, a ring-like arrangement is a very special one since the OR has to be exactly between the two mirror planes. Consequences are, that Pitsch and NW only show 4 poles (sitting exactly on one of these m's), K-S has really 8 but there are other relationships possible with smaller diameter; and any other pole position in between also generates 8 pole positions around [001]g, but they are asymmetrically distributed (like Greninger-Trojano). I assume that this (possibly more phenomenological) interpretation is similar to yours?

3) yes, each round cluster is associated with two rectangular-shaped distributions. This can be clearly seen in the first pole figure, e.g. by the blue drawn poles.

Well, you seem to be a pure mathematician (do not be offended pls!) considering reality in terms of formulas. Meanwhile the crystallography is not merely the convenient DESCRIPTION of collective atoms' behavior, but

reflects standalone PHYSICAL REGULARITIES peculiar to the macroscale behaviour of regularly arranged atoms. Otherwise, to be logical, you should presume that human's behaviour (love, jelosy, fear, braveness, ...) AT THIS PARTICULAR TIME AND PLACE can be accurately simulated on the atom (OK, macromolecules) scale. I beleave this is not possible, and the principal reason is not in complex computations, but in SCALE-RELATED rules irreducible to microscopic terms.

As to your last remarks 1),2)3), I agree to what you say.

Alexander

Dear A.A. Zisman, we all have our weak points (sorry for mine). I am happy that I've got the answers on my questions (what was the reason of everything) and because of you and other colleagues I've also learned some additional lessons. Many thanks to all of you!

Neat discussion, I want to add some more input at this point:

The Bain strain is a continuum concept and has only recently been proven (mathematically, intuitively it was long clear) to be optimal in certain energy norms - see:

K Koumatos, A Muehlemann Proc. R. Soc. A 472 (2188), 20150865 "Optimality of general lattice transformations with applications to the Bain strain in steel"

--- and ---

Determination of the stretch tensor for structural transformations - X. Chen, Y. Song, R. D.James

Since the Bain strain is optimal - if a martensitic domain bigger than, lets say 100nm transforms approximately under a homogeneous strain than this strain should naturally incorporate the Bain deformation (hence the Bain correspondence). However, since other constraints have to be fullfilled during the transformation as well (e.g. Orientation relation), this is additionally achieved by plastic deformation.

Well, the main insight is that despite any crystallographic description using planes and directions (OR models) it starts with an fcc and ends with an bcc...

The problems appear later when we realise that these kinds of transformations need space...and despite of the situation that start and end volume look identical, they are rotated against each other (which is unfortunately not shown in the file below :-(

Thank you, Manuel.

Presentation Orientation relationships between fcc-bcc

The majority of the rotation you refer to comes from plasticity. Actually, it is a special kind of plastic deformation that appears very locally at the interface and enables the lath transformation at its characteristic (comparably) high transformation temperature in the first place. I am currently writing a program that deals with the question of why blocks are forming

(see. https://github.com/ManuelPetersmann/Martensite_Calculator - if you have Matlab you can start the GUI by typing "Martensite_Calculator". I started my work based on the paper "The microstructure of dislocated martensitic steel: Theory - by L Qi AG Khachaturyan and JW Morris - Acta Materialia, 2014 -, but was not fully satisfied, since therein the "lath scale" is basically skipped and they directly go to the "block scale". I want to build blocks from laths. Along the way, I tried to fulfill all experimentally observed "constraints" like characteristic misorientations, orientation relations etc...

Indeed, it seems hard to incorporate the Bain strain into the total deformation of an individual lath without deviating from at least one of the experimentally justified constraints. But, I am confident that there are many ways all of them could be fullfilled, I just have not figured out the right optimizer for the plastic shears yet...

Dear colleagues!

Please consider one more evidence that Bain's deformation (and its axis) does matter. Under consideration (Fig.1) is axially symmetric plastic strain localized within an infinite thin band. Degree of interfacial plastic misfit (and related internal stress) depends on the band plane orientation relative to the deformation axis- Fig.2- with the minimum at angle of about 35 deg.

However, the Bain (inelastic) strain in M-transformation is of the same kind, and the lath habit plane plays the same role as above. The related dependence of phase stess maginitude (planar incompatibility of phase strains) is shown in Fig.3. Interesting, the most experimentally observed habits are concetrated around the minimum!

REF: [MICROMECHANICAL REASON FOR 35° ORIENTATION OF LOCALIZED SHEAR UNDER TENSION OR COMPRESSION (2004) 25th Riso Int.symp. on mater. sci.]

Conclusions:

1) The habit of martensite nucleus is most often predefined by the Bain deformation i.e. BEFORE plastic (accoomodative) shears develop in the transformed crystal (and eventually reduce the indicated measure to zero).

2) The only exclusion for h={111}g indicates a special role of twinning as a fast accommodation mechanism.

3) Since the result is particularly relevant to the NUCLEATION stage, the least interfacial INCOMPATIBILITY of Bain strain seems to be a condition to minimize the enegry of inter-phase boundary.

////As far as the cassical PTCM is also based on inter-phase compatibility (concept of "invariant strain plane") while predicting crystallographic combinations (parallelism of CPPs and CPDs), there is a feeling that just the common geometrical (symmetry) roots of mechanical (stress in volume) and interfacial (specific energy) aspects of the phase interaction lead to the same results,

Great discussion. Martensite is a microstructure with many components, but we must remember that it is the formation of individual crystals by diffusionless, shear mechanisms that is the basis of those microstructuresf. Each crystal has a crystallographic relationship to the parrent austenite and a crystallographic habit plane, irrational in lath martensite as determined by Marder and myself in the beginning stages of the characterization of lath martensite.The way the individual crystals come together in packets, blocks, separated by retained austenite in parent austenite grains then contribute to the morphology which we attempt to characterize by better analytical techniques as discussed.

To continue, another component of individual martensite crystals, and which then becomes an important component of lath martensitic microstructures, is a very high dislocation density. Not often considered is the fact that as carbon content of austenite increases there may be subtle changes in dislocation density and/or morphology as required by the lattice invariant deformation of the martensitic phase transformation. Those changes may affect block/packet structures in lath martensite, and may certainly affect the strain hardening which produces the very high strength and hardness of lath martensitic microstructures. Since as-quenched martensite is invariably tempered, I have questions about the changes and stability of the dislocation density and the very fine individual martensite crystal size of tempered martensite, as discussed in my 2017 review "Tempering of Lath Martensite in Low and Medium Carbon Steels: Assessment and Challenges" in Steel Research International. Best wishes to all for the excellent discussion. George Krauss

Dear George, the main reason why I was asking this question was, that I became accidentally involved a project where during mechanical loading at 620°C a softening of a martensitic microstructure is observed. The dislocation density decreases dramatically. It changes that much that they practically disappear accommodated by a coarsening of blocks and packets. Of course, there are many questions like where are the former lathes, how pinned dislocations become again "mobile" and diminish. I guess it is all related to the temperature. Well, I am no specialist in this topic so that I can only try to deliver reliably experimental data (EBSD in high precision).

Regarding "diffusionless": In iron meteorites we have a lot of diffusion, quite strong chemical gradients of Ni and Fe, no carbon, and anyway martensite. How do you interpret such martensitic microstructures? Or do you mean with diffusionless, that at the moment of transformation there was no considerable diffusion?

If I hear or read "shear" mechanism, it sound to me like an external force, however, I don't think that there is something like that. Only a movement of atoms which looks like a shear movement. The funny thing form my point of view is that this movement actually generate a stress and strain so that the driving force to transform must be bigger than the stress generated by this collective shift of atoms. Is this correct, or are these interpretations already proven as incorrect?

Dear Gert: What you describe in your first paragraph is exactly what I try to evaluate from the state-of-the- art in the review paper I mentioned in my previous comment. In that review I try to systematically characterize what happens to martensitic microstructures as tempering temperature increases. Please read that paper if you are able to obtain a copy. You will see that I too have many questions despite the considerable knowledge about martensite and its tempering.

The shear that drives the martensitic transformation is internal, generated by the thermodynamic driving forces between austenite and martensite. It is real, demonstrated by surface tilting, changes in dimensions, plastic deformation accomplished by either dislocations or twins. The tilting looks like the result of plane strain deformation on an undistorted habit plane of the austenite and is accomplished by the change in crystal lattice (the lattice deformation)and deformation of the bcc/bct lattice (the lattice invariant deformation). It is the dislocation content of as-quenched lath martensite, its reduction during tempering, and the associated effects on mechanical properties that interests me from a practical standpoint. Keep doing your best to characterize that microstructure.

The martensitic deformation is assumed to be diffusionless. Therefore the chemical composition of the parent austenite is the same as the product martensite. In Fe- 30 Ni alloys, the austenite structure is so shear resistant that martensite cannot form until temperature is well below room temperature. At those temperatures dislocation motion is difficult, and the lattice invariant deformation is accomplished primarily by twins. The exceedingly low cooling rates and temperatures in meterorites in outer space and variable chemistries are interesting to unravel, and I have seen some interesting papers on that subject.

Best wishes, George Krauss

I want ask a question:

Do the laths in a block form simultaneously or in succession? what determine the block width?

1) Diffusion. // A conventional implication termed as a diffusionless mechanism concerns CARBON (other alloys do not martter as notably less mobile) ; specifically C atoms has no time to leave just formed martensite that hence becomes C-supersaturated. In the meteorite case (no carbon claimed) we may forget the above issue. Here, the strong overcooling (reason to form the new phase in a marnensite form) is completely due to high amounts of Ni, which IN ANY CASE is much less mobile than carbon while within a solid, i.e. the transformation is diffusionless "by default". On the other hand, the mentioned Ni non-uniformuty can result from surfacial remelting at high meteorite velocities.

2) Shear. // The lattice transformation (Bain) is driven by a difference of alpha and gamma energies as G.Krauss mentioned; such a latice deformation displays itself as shear-like essentially because a specific shape of appearing laths is the most efficient to accommodate a notable SIMPLE SHEAR COMPONENT of this deformation at the least phase stress (energy cost for the interaction with austenite surroundings). Other components, since the lath appeared, are accommodated by the following plastic accommodation of the lath.

[Formally, in the classical PTMC theory all events, reducing the phase stresses (phase strain mismatch), happen simultaneously..]

3) Blocks/laths (to Hou Wang)//// Second generation laths appear (later) at sides of previously nucleated ones as far as similar lattice deformations in the latter (same Bain group and, obviously, same cpp group) suggest least (ideally, no) phase stresses in their interaction. Such a lateral extension of any block is limited by a neighboring one. In other words, the block width is predefined by density of nucleation sites for the first generation laths.

Really interesting questions and comments. Many thanks...

However, we are still talking about the current believe of martensite formation, isn't it? I mean, who really knows whether or how much models really describe (quite limited) observations (using the nowadays available equipment)? No offence against the currently existing and under continuous development existing models and theories. However, obviously they are not entirely convincing since there are thousands of scientists around the world trying to use it and find limitations which needs a readjustment of some imaginations, meeting at several international conferences.... It is not problematic at all to have a improvable model, however, I am not sure whether the killer argument of energetic reasons is very helpful for a scientific investigation. Everything has energetic reasons, even if we do not understand the problem at all. This is simple believe...and from my point of view we all believe from a certain level. (I am not talking about a higher power and religion. I am talking about the insight that nobody can know everything and, therefore, everybody starts from some fundamental knowledge which he/she believes). I for myself have no idea why some materials form fcc, other bcc, hcp or even quite strange structures like Se (although I believe it has certainly energetic reasons), and why stable crystals transform into other structures (certainly again energetic reasons). If this would be all that obvious (except for energetic reasons which can be used in a model considering all experiences we have), we would not have to do sometimes endless experiments in order to find out which phases are available in a material...and how much. The production of alloys is nowadays no science (yes, some experience is available). There are companies which are mixing thousands of compositions until they find a useful one and can sell it for amazing costs. Once the atom was claimed to be indivisible. Nowadays we are starting to assume or already know that our elementary particles (still learned in school or university) are no more indivisible as well. My opinion is that a model is very useful, but it is never a rigid dogma. The world can change very fast...

Another opportunity: The theories are actually able to describe any material, however, our experimental work is so bad that we cannot generate the (ideal) phases designed in theory? Therefore, the always observed differences to the theories? (This is also an acceptable result...and would show where the real problem is. )

@ A.A. Zisman: The Ni-variation (and martensitic microstructure) in iron meteorites is in plessites (see the EBSD maps above). There is no remelting or reheating (as far as we can prove it on many meteorites). Only a continuously changing Ni-concentration which is OK for me.

But I have another question which you might answer (you could help me a lot): I rad very often, that the Ni diffusion in fcc is horribly small. This is the explanation for the so-called M-profiles observed first (?) by Goldstein in remaining taenite (austenite) in meteorites. On the other hand it is said that the Ni-diffusion is ferrite is very fast (at least compared to austenite). Why is this so, especially when the density of ferrite seams to be bigger than from austenite? Doesn't it mean that Fe changes the size during transformation? Don't we need than several radii for atoms, or what do they denote? What changes Fe during transformation? Or where is my mistake?

Thank you in advance.

Best greetings, Gert

This is a comment relative to the question of Hou Wang. Hou you asked a good question that has several parts. First of all each martensite crystal forms independently by the shear mechanisms we have discussed. Those shear mechanisms produce the surface tilting that are a major characteristic of the martensitic transformation. The surface tilting allows the observation of the formation of individual crystals without the necessity of polishing and etching, and because of that Arnold Marder by hot stage microscopy in his PhD thesis 1969/1970 showed that individual crystals form by nucleation and growth parallel and adjacent to other crystals to form packets and blocks. I have preserved some of those observations by joint publications (Marder and Krauss) and in my book on the physical metallurgy of steel (Steels: Processing, Structure, and Performance, Second Edition, ASM, 2015). Figures 5.4, and 5.6 show how surface tilting relates to the formation of martensitic mirostructures. There are still many questions about how composition, especially carbon, affects block and packet sizes, but there is good work going on in this area. Best wishes, George

Great discussion on Lath martensite crystallography and transformation. The common myth about martensite is, it is Hard and Brittle"". But recent automotive lath martensite based AHSS steels have UTS of 1700 MPa with minimum uniform elongation of 4-8%, yet still formable to automotive parts. Could some one comment about the lath martensite formability? Does retianed austenite flims present between lath boundaries responsible for the formability or is the low carbon content ()

Dear Vignesh,

It is hard to say for sure what is a reason of formability while the composition is hidden. In particular, whether the TRIP effect or just austenite deformation does matter. Anyway, 0.4%C and the indicated strength level allow one to expect 6 to 10% of retained austenite in the QUENCHED state. Thus, if TEMPERING is not applied or is implemented at low temperature (say

Dear Vignesh: Yes unfortunately there is still the myth that martensite is hard and brittle. Extewnsive research shows this is not the case in steels containing up to 0.5 pct C and quenched and tempered to ultra high strengths, although there are embrittlement phenomena that do cause brittleness. All this is discussed in my book STEELS PROCESSING STRUCTURE AND PERFORMANCE SECOND EDITION 2015, ASM. Martensite in low carbon sheet steels, although not brittle , is still not very formable. What makes it useable in low caarbon steels is forming and hot stamping in the austenitic condition and then quenching to martensite (STEELS pages 261-263). I have also discussed martensite formability and strenghening in low carbon steels in more detail in my article Tempering of Lath Martensite in Low and Medium Carbon Steels: Assesment and Challenges, publishe din Steel Research Interrnational this year.d

Best wishes, George (sorry for the typos)

Dear George,

Thanks for sharing your book and recent Lath Martensite paper. Yes, hot stamped low carbon steels gains formability since it is stamped at austenization temperature. I am curious about the formability of recently introduced fully martensitic M1700 cold rolled low carbon lath martensite grades that are emerging into auto industry for (parts such as: bumper reinforcement beams, door intrusion beams, rocker panel inners and reinforcements, side sill reinforcements) which are cold stamped at ultra high strengths with 4-8% minimum levels of total elongation values. I am surprised how they forms to actual complex shapes

I disagree with the hypothesis that the said carbon steel will transform fully to martensite. If you quench hard and trap all carbon atom into bct structure; you will still have austeniteausterity. Therefore you will have deformable structure. Especially if you low temper below 200C

Vignesh: Relative to your recent comment, it is the degree of formabilty that is required for the manufacture of a given part with a martensitic sheet steel that establishes wheter formabilty is accomplished with a martensitcic microstructure or with austenite and press hardening. Martensite has long been used in parts with simple geometries such as bumper reinforcement beams by roll forming but for more complicted shapes where stretching, bending, drawing, and or straightening are required the ductility of as-quenched martensite is not sufficient. Therefore the mechanical design of the parts you mention must have been carefull evaluated and many other parts can only be made by hot stamping if a martensitc microstructue is required. Nevertheless it is exciting to see martensite used wheere its high strength is valuable for parts resistant to side impact and roof collapse in vehicle accidents.

Gabi, Modessir: I do not understand your comments. In low carbon steel with martensitic microstructures there is only a very small amount of austenite present. See Fig. 18.4 in STEELS, Second Edition. Many other factors determinte the strength, ductility, and fracture resistance of martensitic microstructures.

Thanks George, your comments are very valuable, especially for me as crystallographer (never heard anything about steel during study :-)). I am one of those guys who connect martensite with brittle and hard. It is a good opportunity to learn that everything can be different.

I was trying to emphasize that the capture rate depend on the cooling rate. Although there is small amount of austentite but there is other properties that make deformibility possible specially if you temper to less than 200 C.

Gert: Happy to learn that you are a crystallographer. The materials community needs you.

Crystallography is the basis for the characterization of ferrous microstructures. I gave a lecture once that was entitled "Steel consists of crystals, within crystals, within crystals", meaning that transition carbide crystal form in crystals of martensite and that the martensite forms in crystals of austenite. The early metallurgists left us the legacy of calling crystals grains, and so we talk about grain size to this day, but what we should be talking about is crystal size. That is why in some of the previous comments in this series I have emphasized the formation of individual crystals of martensite. The martensite crystals are the finest that can be produced in any heat treated condition of a steel producing an extremely fine crystal (grain) size. What happens to these crystals in blocks and packets and the boundaries between them on tempering needs further characterization as I discuss in my 2017 review in steel research international.

Crystal structures in steels, i.e. alloys of iron, are extremely important because deformation and cleavage fracture occur on well defined crystal planes. The deformation occurs by the mechanisms of dislocation motion in response to shear stresses acting on well defined slip planes. Dislocations are crystal defects and so crystal defect structure controls mechanical behavior. In steels not only strength (resistance to dislocation motion) but also ductility (the ability to move without fracture) makes for the great versatility of steel as a structural material.

The above comments are merely overviews of the importance of crystal structure in steel, but important to keep in mind before we get deeply immersed in details of various facets of steel processing, structure, and performance. Those details are important and there is a huge technical and scientific knowledge base that exists and is expanding because of new instrumentation and the continued use of steel as a structural, load- carrying material.

Gert: keep asking questions. Without questions we don't look for answers.

Dear George, for a crystallographer (like me) everything in a crystal is perfect (in the beginning). This makes a discussion often not easy between material scientists and engineers. Personally, I like to use the crystal lattices, points, directions, planes, vectors etc. On the other hand I have to stop myself since unfortunately (like in steel science) everything is more complicated than it seems (what you certainly know). It is sometimes a burden that in case of ferrite and austenite atoms have the same positions like lattice points which irritates many people so that they mix everything. Quite dangerous since some terms are only related to the structure , other only to the lattice. Therefore, I often have to ask first, about what they are talking about. Typical example: dislocations which represent a nice model for the movement of atoms, although finally a Burgers vector (related to the lattice) is used. Luckily, it is not a big deal to distinguish between both because the atomic layers are parallel to lattice planes etc. However, thais is not necessarily true (for more complicated structures, which crystallographers are trained for :-) ). Crystallography uses geometrically tools from algebra to simplify a complex atomic arrangement called crystal. There are no lattice planes which generate reflections in a diffractometer etc... We often forget that these are convenient models only which might be not optimal and can be changed every time. One example is the quite hue number of crystal structure descriptions which do not follow the definitions given in International Tables for Crystallography, Vol A. I only want to remember from time to time, that we are discussing about the current level of understanding. Many things are well investigated, some other are more an educated guess. Sometimes we are so happy about a fundamental insight that we forget it could be the effect of a very local investigation or preparation. As you said in one of your comments: it needs further investigations, and I want add, that there are possibly a few surprises hidden which turns around many of our interpretations. We are far away from knowing everything about martensite. Sometimes we even cannot find an agreement about the name of the phenomenon investigated. Anyway, I apologize for some naive or trivial questions, however, as you mentioned, they are not useless...

Gert mentioned that during mechanical loading at 620C, softening of martensite was observed, (reduction in dislocation density). Although the full details of the deformation schedule were not supplied, that's not too surprising, . Iron based martensitic structure are generally not stable and will start to revert to the more stable ferritic forms, when tempered, (i.e. heated above ~150C). 620C is a relatively high tempering temperature, so the tempering reaction(s) will be fairly rapid. I understand his comments about crystallographic constructs having what might be seen as having an artificial/ 'arbitrarily'/ imposed nature, however like all human constructs/theories their usefulness is judged by their utility, and specifically here in their ability to classify a given material's crystal structure, (some systems with atom locations, site occupancies etc. etc.), in relatively unambiguous manner, once one gets your head around the system's rules.

It is hard to distinguish because of the atomic complexity even when you temper. Lath are better observed than others

Gentlemen: Please read my review article "Tempering of Lath martensite in Low and Medium Carbon Steels: Assessment and Challenges" Steel Research International, 2017,where the effects of high temperature tempering on microstructure and properties on martensite after tempering are described and discussed in detail.

Best wishes, George Krauss

i think you mean by saying packed, block, lath,subgrain. During the thermal process cristalline materials such as metals they have a transformation temperature which allows the grain growth which makes it become more ductile. There are several combination of cristalline structures such as centered face cubic, simple cubic and body centered cubic for most of metals. But there are several complex cristalline structures such as close packed hexagonal structure.

The "trick" is the different mechanical loading applied which obviously affects the finally observed microstructure considerably.

@ J.M.S. Filho: a closed-packed hexagonal phase is not really complex. In fact, it has the same coordination number 12, identical element planes like fcc, but only the stacking is a bit different so that this very similar structure has to be described differently (hexagonal). BTW: There are - as far as I know - only two metallic elements described by a simple cubic arrangement. Most are body-centered, cubic- or hexagonal closed-packed.

I know different loading conditions make it more complex on the hexagonal close packed if compared to fcc structure

for sure... You have less symmetry-equivalent slip systems...

It does make sense.

Thanks

Salutes to Gert Nolze sir for the question and discussions with really really knowlegable researchers.It will be of great help for me in developing low alloy ultra high strength steel with highest possible hardness of about 850 VPN and good ductility for ballistic resistance properties, if some one can tell how to get martensite having more than 2500 MPa tensile strength in static test condition with good dynamic mechanical properties.

thank you prof. Dierk Raabe . you have shared some useful links.

The martensite lath is a single crystal of martensite with a thickness of less than 1 mm in general. The block is an aggregation of laths having a similar crystallographic orientation. In low-carbon steels, each block consists of laths of two particular K-S variants with small misorientation, which is referred to as a sub-block or bivariant block structure. For high-carbon steels, blocks consist of laths with a single variant. The packet is an aggregation of blocks with almost the same habit plane orientation by sharing the same parallel relationship for close packed planes.

I think all depend on the carbon content, I low carbon we have the formation of single crystals whereas for high carbon we can form blocks and packet.