Isn't that how high quality steel swords are produced?

material surface will burn 

Dear Shitong,

In addition to the previous comments, just to answer your question, I think there would not be a difference after the first and the second quenching in terms of the microstructure if the two quenching processes are performed at the same conditions (same austenitizing temperature, same cooling rate etc..).Because you have to heat your quenched sample for the first time up to the austenitizing temperature one more. Then you normally transform the martensite structure into the fcc austenite.   

But in practice, the steel can be decarburized and can lose some carbon after the first quenching  depending on the austenitizing and quenching environment. Then when you quench it for a second time, it is possible to not to obtain the same microstructure and mechanical properties compared to the first one.   

Dear Professor Lawrence Margulies, Professor Tarik Ömer Oğurtani, Professor Akhilesh Singh, Professor Atilla Atli, thank you very much for your kind replies.

I am sorry I have made a mistake in the description of my question.It should be 'For high strength steels, fully martensite microstructure can ...'

Dear Shitong:  In general, the second austenitize results in a finer austenite grain size.  This happens because when a martensitic microstructure (the result after the first austenitize and quench) is heated above the Ac1 temperature, the new austenite grains nucleate at the many carbides and grain boundaries that are characteristic of as-quenched (and heated) high strength steels.  So after the second austenitize and quench, the resulting microsturucture--in generall--has finer lath martensite than after the first austenitize and quench.  The hardness is not affected but the impact properties are--in general--improved by the second austenitize.  Note that this effect of "double-austenitizing" is effective for many steels but not for all steels.

You need to define what type of steel you are talking about. For low-alloy medium C high-strength steels, quenched to produced martensite and then tempered to reduce quench stress, the resulting microstructure is usually Martensite with a certain amount of retained austenite (Mf is lower than ambiant temperature for these steels). Both phases undergo transformation during tempering. Carbides are produced. Thus, even though martensite is the major phase, there are other phases such as carbides (usually FexC) that do not readily dissolve during further austenizing treatment. That's why Martensite will get less hard for each new austenizing treatment (each time it loses a little amount of C). We have done tensile tests after several austenizing-quenching-tempering treatment and the tensile strength decreased each time (3 to 4 ksi) until the steel stops meeting the specification. This scenario will be different if you deal with different types of steel.

Shitong,

You need to define the exact composition of the steel, the particular section size involved, the times and temperatures involved of all the proposed thermal cycles, and the atmospheres employed, i.e. original austenitising treatment, type of quenching, re-austenitisation details, protective atmosphere, carbon potential of atms. oxygen potential of atms. etc. etc.

What actually happens will depend strongly on all these parameters, so here you can receive several different, but still correct answers !

But you actually asked a sort of philosophical question, i.e. on repeated austenitisation and quenching what will happen, so what can be said re such a general question ?

Fred already mentioned refinement of the austenite grain size, but of course that depends upon the original austenite grain size, and on the details of the second austenitisation period, i.e. how long you hold the steel, and at what temperature. It's perfectly conceiveable that by a certain choice of re-austenitising temperature and hold time, you get back to square one, i.e. the original austenite grain size and distribution of the elements. Then the second quench will give the same properties as the first. Of course in practice you might degrade the alloy by losing carbon through oxidation, then the resulting martensite would be softer, as its basic hardness is fixed by the carbon retained in the quenched BCT structure. That's likely to be a surface effect, but again the exact effect would depend upon the depth of carbon loss, time for  homgenisation, distribution of other elements etc.

If on the other hand you rapidly cycle, then you will affect the nature of the austenite that re-forms prior to the quench. So as you speed up from the rates at which you allow the steel to move more towards equilibrium, (NB it can take weeks to disperse segregation of substitutional elements !), it is possible to decrease the austenite grain size by this process, increase its inhomogeneity, increase the number of nucleation sites, leave carbides out of solution and so on. This then affects the details of the subsequent transformation on cooling, martensite packet sizes etc. On refinement of the austenite grain size, you can decrease the effective hardenability of the steel, i.e. you increase the possibility of formation of the other possible transformation products, i.e. bainites and pearlites.

Dissolution of carbides on re-austenitisation is a function of their type and chemistry, with cementite being the least resistant to dissolution (quickest).

Kinetics of the movement of the carbide/austenite interface in this situatuion has been treated by several authors, and with competing models, see the work of the Bhadeshia group on one hand, and that of Hillert on the other. I guess Prof. Agren is the current active researcher of the latter Swedish group.

You didn't mention tempering or ask about it, but for some reason other posters did, now if during the processing of your alloys, you allowed the time for alloy carbides to form, then in effect you have increased the segregation of those elements. What happens when you re-austenite such tempered alloys ? Whilst carbon is highly mobile at the austenitisation temperature, these larger elements are not that mobile (and you have to take into consideration the differences in their diffusion rates in the ferrite and austenite lattices). Where will the carbon 'like' to be ? At equilibrium programs like MTDATA and Thermocalc will give you the answer, (i.e. normally in the austenite), but under kinetically controlled conditions, the instantenous answer can be different, that's why you get reluctance to dissolve of such carbides (temporary stabilisation).

Even when they have dissolved, local chemical analysis might indicate that their distribution is none uniform and 'peaks' at the site of the former carbides.

NB A non-uniform distribution of carbides in materials often follows patterns of segregation, e.g. manganese, here although carbon could have been uniformaly distributed in the thermal cycle, thermodynamically it 'likes' to be with the managnese.

Dear Professor Fred Fletcher, Nihad Ben Salah, John Butler, thank you very much for your replies. I learn a lot from your replies.

Although broadly it has been pointed out that the microstructure after the second austenitising treatment is finer, it is restricted to the reducing the prior austenite grain size. Packet and lath sizes are not affected by the second austentising treatment carried out at the same temperature.

heating, quenching, and heating, quenching, then again and agian? I think  the metals will get thick oxide layer and the mechanical property will reduce.

If one anneals the austenitic steel at very high temperature still staying in the same gamma phase,   it is in avoidable to have grain growth (which not desirable of course), where the surface free energy reduction phenomenon becomes to dominate all other factors, which play  positive role t the lower homogenizing temperatures on the desired grain refinements! 

If you quench a carbon steel from its austenite regime, you will have phase transformation and at high cooling rates to room temperature you will have martensite and possibly retained austenite, if the carbon content is high. If the cooling is stopped at a higher temperature, retained austenite fraction is higher. If you later continue cooling from that elevated temperature, depending on the holding time before the continued cooling, some part of the retained austenite will transform to martensite with a higher carbon content than the firstly formed martensite (the Q&P process). Continued cooling from RT to subzero temperatures will lead to the transformation of retained austenite, even though some stabilization is possible.

If you re-austenitize the martensitic sample, you will get a refined austenite grain size  (if excluding the memory-effect of austenite) and then after subsequent quenching, refined martensite laths. Martensite "grain size", lath width, block and packet sizes are somewhat dependent on the grain size of the prior austenite, even though the effect is not very drastic (though depending on the austenite grain size). 

Steel can be quenched to number of times but there is a limit to it as constant quenching/anealing/normalising will lead to the disability of material reaching optimum material properties. 

There should not be any difference provided you have solutionized it well to ensure single austenitic phase-field with all the precipitates dissolved into it. If however, the solutionizing was incomplete, you will get inferior mechanical properties.

I think the repeated quenching in martensitic (high carbon) high strength steel can reduce prior austenite grain size, increase the retained austenite content and refines the martensitic structure. 

Please see the following papers:

1) "Effect of repeated induction heating on fatigue crack propagation in SAE 52100 bearing steel:, Hitonobu KOIKE et al., Advanced Materials Research Vols. 217-218 (2011) pp 1266-1271.

2) "Effect of Repeated Quenching on the Rotating Bending Strength of SAE52100 Bearing Steel", Koshiro Mizobe et al., Advanced Materials Research (Volumes 457 - 458), 2012, pp. 1025-1031.

(also please see "Martensite in steel: strength and structure", by George Krauss, Materials Science and Engineering A273–275, 1999, pp. 40–57).

As I said every thing depends on the isothermal annealing temperature and time in austenite domain and the subsequent quenching rate. At very high temperature grain growth of austenite  is unavoidable, and upon quenching fast down to the Ms and lower temperatures up to Mf by passing the nose of TTT diagram generates rather very fine  size of martensite  needles, and  almost no retained austenite left in the matrix. Martensite is supersatured with carbon and its has  tetragonal structure c NE b= a.

Ogurtani,

Retention of austenite is a function of the composition of the austenite. In high alloy steels it is not uncommon for the Mf temperature to be sub-zero, and therefore for austenite to be retained in the structure on quenching to room temperature. In highly alloyed steels their hardenability is often such that particularly high quench rates are not required (this is section size dependent of course) to 'beat' the nose of the curves in the TTT/CCT diagrams for the diffusional and baintic transformations. In the case of tool steels repeated tempering is often employed, this eliminates retained austenite and then ensures that any fresh martensite is tempered.

 Gupta,

You may be correct in your asertions about about the relationship between austenite grain size and the martensite lath size, (as in the study of M.J.Roberts: Metall. Trans., 1970, 1(12), 3287.), however, on packet sizes, Gareth Thomas, one of the early transmission electron microcopists studying such aspects, asserted that they were related, and he repeated this view in 2000 in a paper with Greg Kusinski at the ASM George Krauss symposium:- "If the γ grain size is reduced by controlled rolling, then the resultant α′ packet size is also reduced."

Also if you plot the data of Wang et al. in

"Effect of Microstructure Refinement on the Strength and Toughness of Low Alloy Martensitic Steel", J. Mater. Sci. Technol., Vol.23 No.5, 2007, as he did in his Fig.4, there is some sort of relationship between PA GS and the martensite packet size..

There may be other papers with different and conflicting observations to these authors, but these observations stand,and show that, at least sometimes, they are related.

Dear Shitong Wei,

It is thermal stability of austenite which determines whether or not austenite transforms into martensite on quenching. Thermal stability of austenite, in turn, is determined by its Ms temperature. The Ms temperature itself is most seriously affected by the carbon content of austenite. Unlike diffusional transformations, no change in the carbon content of austenite is expected during its transformation to martensite. Therefore, the carbon concentration in austenite is the same before and after the martensitic transformation (or in the primary austenite and in the retained austenite). So, no further transformation of austenite into mrtensite is likely during quenching for more than one time.

Practice of double quenching and tempering has been tried in AISI403 forgings by doing first quenching from 1030 deg.C cooling to ambient temperature and then reheating and quenching from 965 deg.C followed by tempering.This modified process resulted in grain refining of ASTM 5 to ASTM 7-7.5 and impact strength substantially.Process was first tried in Canada, nuclear power plant endfittings.