fundamental metallurgy of niobium in steel

8/10/2019 Fundamental Metallurgy of Niobium in Steel

1/74

FUNDAMENTAL METALLURGY OF NIOBIUM IN STEEL

Anthony J. DeArdo

Basic Metals Processing Research InstituteDepartment of Materials Science and Engineering

University of Pittsburgh, Pittsburgh, PA 15261, U.S.A.

Abstract

It is well known that niobium is added to a wide range of steels for improving processing,microstructure, properties and performance. Over the last two decades, the use of Nb has also

permitted new steels with attractive properties to be developed. Furthermore, the addition ofNb to existing steels such as ferritic stainless steels has also led to improvement. The goal ofthis paper is to review the basic behavior of niobium in a wide range of steels, including boththe traditional steels and also some of the newer versions. Particular emphasis has been placedon the basic metallurgical principles that apply to these steels, for it is the application of these

principles that allows the composition - processing - microstructure mechanical propertyrelationships to be rationalized and exploited. The application of basic metallurgical principleshas resulted in a predictive capability that has led to alterations in composition and processingfor the purpose of producing steels with superior mechanical properties and improved overall

performance.


2/74

Introduction

As we gather here to celebrate the 200th anniversary of the discovery of niobium by CharlesHatchett, we will be reminded repeatedly of how this metal has changed the face of science andtechnology in the materials industry (1, 2). Nowhere is this more noticeable than in the case ofsteels. Although niobium is classified as a transition metal, perhaps the most importanttransition made by niobium was from being little more than a laboratory curiosity prior to 1960to being a viable, commercially produced ferroalloy suitable for addition to steel in 1965.

While we celebrate 1801 as the date of the discovery of niobium, in fact, the first successfulcommercial production of ferroniobium by the CBMM mine in Araxa, Brazil took place in1965 (3). After 1965, ferroniobium was abundantly available to the steel industry for the firsttime as a microalloying element. Prior to this date, only vanadium and titanium were availableon a commercial scale for microalloyed steel.

Although the benefits of niobium additions to low carbon steel were known since the late1930's (4), it was not until 1958 that the first heat of niobium microalloyed steel in the form ofhot strip was commercially produced by National Steel in the United States (5). Theinternational symposium, Niobium, held in San Francisco in 1981, reviewed the science andtechnology of using niobium in a broad range on materials as practiced at that time (6). By1981, the concepts of controlled rolling, austenite conditioning, and proper alloy design forferrite-pearlite microstructures with adequate properties in a broad range of products wereunderstood and practiced.

Perhaps the plate and linepipe microalloyed steels were the best understood and optimized atthat time. On the other hand, the benefits of niobium microalloying to product lines such asstrip, sheet, bars, shapes and castings were just starting to be exploited. Most of the work

published in the steel papers inNiobiumconcerned ferrite-pearlite microstructures (6).

What has changed since 1981 concerning the benefits of Nb microalloying to steels? There

have been at least two major changes in the Nb bearing steel products over the past 20 years.First, since ferrite-pearlite microstructures cannot readily exceed yield strength levels of about400 MPa in reasonable section sizes and low carbon contents, the applications-drivenrequirement for higher strength levels has led to the development of ferritic microstructures invirtually every product class that are based on low temperature transformation products such asacicular ferrite and bainite or to multiphase microstructures. Sometimes these microstructuresare achieved through the use of accelerated cooling, sometimes by using hardenabilityapproaches, or sometimes additions to improve both. Regardless of the approach, Nb continuesto play a major positive role in optimizing the manufacturability, final properties and

performance of these steels. Other papers in this conference will highlight these benefits.

The second major change since 1981 has been in the wide acceptance of Nb as a stabilizingelement, often in combination with Ti, in both ultra-low carbon sheet steel (ULC) and ferriticstainless steel. Since the Nb levels employed in the ULC steels frequently fall in the range 100-300 ppm they can be considered as a microalloying addition. However, the 1500-5000 ppm Nbsometimes used in the ferritic stainless steels would be considered an alloying addition. Theso-called dual-stabilized ULC steel and ferritic stainless steels are viable, successful, andgrowing members of their respective families.

In the Proceedings of Niobium, the forerunner of this current paper appeared (7). In that paper,the authors attempted to present the fundamental aspects of Nb and its behavior in steel, at leastas understood at that time. These fundamental aspects have remained valid over the course of


3/74

time. However, much has been learned in the intervening 20 years that will expand upon whatwas presented in the earlier paper. This current paper will be organized as follows:

A. Fundamental Considerations.B. Niobium and Austenite Conditioning.C. Niobium and Transformation.D. Niobium and Strengthening.E. Niobium and Stabilization.

The goal of this paper is to provide the reader with the background necessary to understandmodern Nb microalloyed steels. The basic principles presented here will hopefully permit thereader to design better steels and processing routes to further improve future steels, all throughthe intelligent use of Nb. The application side is becoming even more demanding: weightreduction, safety, lower cost, etc. These increasing demands can only be met through athorough understanding of the literature and of the basic principles of microalloying with Nbupon which much progress depends.

Fundamental Considerations

Electronic Structure

Iron, manganese and niobium belong to the family of metallic elements known as transitionmetals. Transition metals are characterized by the electronic structure of the atoms where theouter shell (4s energy level for period 4 and 5s for period 5 elements) contains electrons whilethe inner shell (3d energy level for period 4 and 4d for period 5 elements) is not completelyfilled. The transition metals are the only elements that have unfilled inner shells. Many of theelements that form substitutional solid solutions with iron in HSLA steels are shown in Figure 1(8).

Fe-Nb Phase Diagram

As might be expected from the crystallographic data, niobium and iron are completely miscibleat high temperatures. The equilibrium diagram for the iron-niobium system is well established:details may be found in standard texts (9). The most interesting feature of the diagram, inrelation to the use of niobium in steels, is the existence of a loop which limits the existence ofthe ? phase to alloys containing less than 0.83 percent niobium (0.50 at. %). Clearly, niobiumis a ferrite stabilizer. However, additions of up to 0.10 to 0.20 percent lower the A3temperature, i.e. the ? loop apparently displays a minimum similar to that observed in the iron-chromium and iron-vanadium systems; thus, in small quantities niobium acts as a ? stabilizer.

In the ternary iron-carbon-niobium system, the major feature of the equilibrium diagramrelevant to steels is the marked reduction of the solubility of carbon in austenite and ferrite as aresult of the ready formation of niobium carbide.

Diffusion

The early literature contains several estimates of the interdiffusion coefficients in dilutesolutions of niobium in ? iron (7). The data show considerable scatter and are consideredunreliable.


4/74

Figure 1: Abbreviated Periodic Table Showing Some of the Elements WhichForm Solid Solution with Iron (8).

Figure 2: Possible diffusion distance of Nb in ferrite and austenite based ondiffusion coefficients (11).


5/74

Kurokawa et al. (10) have studied the diffusion of niobium in high-purity iron containing nodetectable interstitial levels. Iron, Fe-0.6 Si and Fe-0.6 Si-1.5 Mn alloys were investigated attemperatures in excess of 1080C. The experiments were conducted using the radioactivetracers Nb95 and Nb96, and the concentration profiles were determined by direct sectioning.Two important observations emerge: firstly, that the interdiffusion coefficient of Nb inaustenite is not strongly influenced by the composition of the solid solution matrix, and,secondly, that the coefficient for Nb is somewhat higher than the coefficient for the self-diffusion of iron.

The diffusion coefficient in BCC a-Fe is about 100 times higher than in ?-Fe. This is thoughtto be partially caused by the looser atomic packing in the BCC crystal structure, Figure 2 (11).The diffusion coefficient is usually expressed as an Arrhenius law of the form:

D = Doexp(-Q/RT) (1)

where Do is a constant depending on the nature of the element and the bulk composition andcalled the frequency factor (12), R is the perfect gas constant = 8.314 J/molK, Q is theactivation energy constant, T is the temperature in K.

The constants for the diffusion of niobium in iron are given in Table I and can be compared tothe coefficients for the self-diffusion of iron given in Table II.

Table I Diffusion of niobium in iron.Element Bulk Q in KJ/mol Doin cm

2/s Source

Nb Fe-? 266.5 (+/- 1.8) 0.83 (+/- 0.69) (12,13)

Nb Fe-? 264 0.75 (14)

Nb Fe-? 209/334 400/530 (15)

Nb Fe-a 252 (+/- 2.5) 50.2 (+/- 3.0) (12,13)Nb Fe-a 289 100 x Do(?) (15,16)

Table II Self-diffusion coefficients for iron.

Element Bulk Q in KJ/mol Doin cm2/s Source

Fe Fe-? 284 0.49 (17)Fe Fe-a 241 2.01 (17)

Compound Forming Tendencies

Transition metals are known to form a series of simple and solid solution compounds of oxides,sulfides, carbides and nitrides. The compound forming tendencies of several of the transitionmetals in steel have been reviewed (18, 19) and have been summarized by Meyer et al. (20),Figure 3. Niobium shows a strong tendency to form carbonitrides, but relatively little tendencyto form oxides, sulfides or solid solutions of these compounds. In this regard it behavessimilarly to vanadium. This characteristic distinguishes it from titanium which does not act as acarbide former until all oxygen, nitrogen and sulfur have been consumed by initial additions oftitanium.


6/74

Figure 3: The tendency of certain metals to form oxides, sulfides, carbides, andnitrides and their precipitation-strengthening potential (arranged similar to the

periodic table). After Meyer et al. 1977 (20).

Figure 4: Nb-NbC phase diagram. After Storms and Krikorian (22).


7/74

Precipitation of Niobium Compounds: Nb-NbC System

The Nb-NbC system has been studied extensively over the last quarter century (21, 22) themost definitive study being that of Storms and Krikorian (22). The Nb-NbC phase diagram isshown in Figure 4. There are three solid-solution, single-phase regions: a, and ?. The a-

phase is an interstitial solid solution of carbon in Nb, with a maximum solid solubility of 0.2 at.percent C near 2300C. A vanishingly small carbon solid solubility would be expected attemperatures of interest in steelmaking. The a-phase has an A2 (BCC) crystal structure with a

lattice parameter that will vary with the carbon content and is 3.294 for pure Nb (21). The -phase (Nb2C) also appears to have a very limited range of carbon solubility around thestoichiometric composition (33.3 at. % C) below 1500C. Nb2C has an HCP crystal structurewith lattice parameters a = 3.12 and c = 4.95 (21). The phase of the Nb-NbC system that is ofmost interest in steels is the ?-phase NbC, but it should be noted that Nb2C is observed insteels having high Nb/C ratios (23). The ?-phase has a range of carbon solubility which, forexample, appears to extend from near NbC.72 to NbC at 1100C. This range of possible carboncontents has led to the adoption of the symbol NbCx to describe the ?-phase, with x being equalto the mole ratio of C to Nb (at 1100C; therefore, 0.72< x < 1.0). The ?-phase NbC has a B1(NaCl) type face-centered-cubic (FCC) crystal structure (24) whose symmetry is described by

the Schoenflies point group

5

hO and Hermann-Mauguin space group (shortened) Fm3m with 8atoms per unit cell (or 4 lattice points, each consisting of one Nb and one C sites). The crystalstructure of NbCx can be represented by two interpenetrating FCC lattices with Nb atoms (or

Nb atom vacancies) residing on one set of FCC lattice points and C atoms (or C atomvacancies) residing on the other set. NbCxcan be considered, therefore, to consist of Nb atoms,C atoms and vacancies. If we let N0 equal the number of atom positions on either FCC latticecontaining Nb or C atoms, then we have the following relationship:

VCC

VNbNb

0 NNNNN +=+= (2)Where:

NbN = no. Nb atom sites per unit volume containing an Nb atom;VNbN = no. vacant Nb atom sites per unit volume;

CN = no. C atom sites per unit volume containing a C atom;V

CN = no. vacant C atom sites per unit volume.

Subtracting vNbN from both sides, we get:

( )VNbVCCNb NNNN += (3)

Since, by convention, the subscript of the Nb is taken as one, both sides of equation (2) shouldbe divided by NNb. This gives:

Nb

VNb

VC

Nb

C

N

NN

N

N1

+= (4)

wherex

x in NbCNNbN

CN = (4a)

and x1N

NN

Nb

VNb

VC =

(4b)


8/74

Figure 5: Variation of NbC lattice parameter with composition. After Storms andKrikorian (22).

Figure 6: Lattice parameter versus composition in the NbN-NbC system. After

Storms and Krikorian (22).


9/74

Equations (4) indicate that if the number of vacancies on Nb and C sites is equal, then NNb=NCand x in NbCx would be equal to one and the carbide would have the stoichiometriccomposition. If, however, there are more vacancies in the carbon lattice, then NNb > NCand x 150h) and higherniobium ( > 0.10%) and nitrogen contents ( > 0.012%) give rise to the formation of non-cubiccompounds, often of the hexagonal d or e carbonitride type. Also, as stated earlier, Nb2C-type

precipitates are observed at high Nb/C mole ratios (23, 33).

The Solubility of NbCxNyin Austenite

The formation (and dissolution) of precipitates normally exhibits sigmoidal kinetic curveswhich can be approximated by the Johnson-Mehl equation (47).

= 43p tGN

3exp1X && (5)

where:Xp = fraction of precipitates formed

N.

= nucleation rate (usually assumed constant)

G.

= growth rate (usually assumed constant)t = reaction time.

The equation shows that a precipitation reaction will attain a given level of completion only

when N.

, G.

and t have sufficiently large values. The nucleation rate is controlled by solutesupersaturation while the growth rate is controlled by both the diffusion coefficient and the

solute supersaturation (48). Since solute supersaturation controls both N.

and G.

, it is very

important to understand the factors that determine it.

The extent to which elements can be maintained in solid solution in austenite is governed by theappropriate solubility product. If, for example, we are interested in the extent of solid solubilityof Nb and C in austenite which is in equilibrium with pure, stoichiometric NbC, then we mustconsider this reaction (49):

[ ] [ ]+= CNbNbC (6)

and at equilibrium, the reaction can be written as follows:

T67.1130122G)gr(C)s(Nb)s(NbC 01 =+= (7)

where o1G represents the positive change in free energy (joules/mole) for the above reaction.It is important to realize that in this reaction, NbC, Nb(s) and C(gr) are taken as the purecomponents in a stable state of existence at some given temperature. This is referred to as theRaoultian standard state (50,51). For cases where the pure component may exist in a physicalstate which is different from that of the solution, the Henrian 1 weight percent standard statemay be more convenient.


13/74

The Henrian standard state is obtained from Henrys law, which, strictly being a limiting lawobeyed by the solute Nb (or C) at infinite dilution in austenite, is expressed as (51):

o

Nb

Nb

Nb

X

a as 0XNb (8)

where Nba is the activity of Nb in austenite with respect to the Raoultian standard state, NbX is

the mole fraction of Nb in solution and oNb is Henrys law constant. Henrys law constant isthe activity coefficient which quantifies the difference between Raoultian solution behavior of

Nb and Henrian solution behavior of Nb (51). If the solute (Nb or C) obeys Henrys law over afinite composition range, then:

NbNbNb Xa o= (9)

The Henrian standard state is obtained by extrapolating the Henrys law line deviating fromideal solution behavior to NbX = 1. This state represents pure Nb in the hypothetical, non-

physical state in which it would exist as a pure component if it obeyed Henrys law over theentire composition range (i.e., as it does for a dilute solution) (50).

Having defined the Henrian standard state, the activity of Nb in solution in austenite ( Nbh ) with

respect to the Henrian standard state having unit activity is given by:

NbNbNb Xfh = (10)

where Nbf is the Henrian activity coefficient.

It should be noted that the mole fraction of Nb in solution in austenite can be related to itsconcentration in weight percent by:

( )

FeNb

Nb

Nb

MW

Nb%wt100

MW

Nb%wt

MW

Nb%wt

X

+= (11)

where MWNb and MWFe are the respective atomic weights of Nb and Fe. This conversion isnecessary in order that a hybrid Henrian standard state can be introduced. This is referred to as

the 1 wt% standard state. This standard state differs from the previous one in that a weightpercent coordinate system is used rather than a mole fraction coordinate system. The use ofthis standard state eliminates the necessity of converting weight percentages, obtained viachemical analysis, to mole fractions for the purpose of thermodynamic calculations. Thisstandard state is particularly convenient to use in metallurgical systems containing dilutesolutes (50, 51). This standard state can formally be defined as:

1fNb as 0Nb%wt (12)

Hence, with respect to the 1 wt% standard state having unit activity, the activity of Nb is given

by:


14/74

Nb%wtfh NbNb = (13)

This expression for the activity of Nb is often simplified by two further assumptions whichreduce Nbf to unity. These assumptions are often made in the calculation of the equilibrium

constant (K), also referred to as the solubility product (52). The first of these assumes that Nbis a dilute solute in austenite. The second assumption neglects any interaction between solutes

in the system. The fact that fNb goes to unity under these assumptions can be realized given thedependence of fNb on the first order free energy interaction coefficient f or dilute, multi-component solutions (55, 56). This interaction coefficient was first introduced by Wagner (55)and later was extensively used by Chipman (56) and colleagues in the study of molten alloysteels. It is related to the activity coefficient such that:

( )=

=n

1j

jii j%wteflog (14)

where jie , (read e j on i) is the Wagner interaction parameter on a weight percent composition

coordinate and n would represent the total number of solute elements or compounds in thesystem (55). It is apparent from Equation 14 that if either a solute is dilute in a given system(wt% j ? 0) or that interactions between solutes is negligible ( jie ? 0), the activity coefficientfor the given solute will be unity.

Returning to the dissolution of NbC precipitates, the change in free energy (joules/mole) as Cand Nb are dissolved in austenite is given by (52):

C(gr) = [C] T22.3335062G 2 = o (15)

Nb(s) = [Nb] T54.2927949G 3 = o

(16)

where o2G ando

3G represent the free energies associated with the change between theRaoultian and Henrian standard states. The overall reaction is given as:

NbC = [Nb] + [C] (17)

The corresponding change in free energy is given by the addition of Equations 7, 15, and 16:

NbC

CNb

a

hhlogTR303.2T43.64137235G == o (18)

where NbCa is taken as unity for the pure component NbC in its standard state. Furthermore,

assuming that Nb and C are dilute in austenite, Nbh and Ch may be represented by their weight

percents. Hence, the solubility product under this theoretical treatment is found to be:

[ ] [ ]T

716736.3CNblog = (19)


15/74

The importance of Nb as a microalloying element in steels is apparent from the numerousstudies over the past 30 years. Much of this work is concerned with the solubility of niobiummonocarbides (7, 52, 57-64), mononitrides (7, 52, 60, 64, 65) and carbonitrides (7, 52, 66-70)in austenite. The results of these studies are shown in Table III in the form of solubility

products. The differences among these products are considerable and may be attributed to anumber of reasons. Foremost among these reasons would be the methods used in obtaining thegiven solubility product as each technique has its own assumptions and limitations. Thesetechniques used in obtaining the solubility products of Table III are classified as a-e and are

briefly described below:

a) Thermodynamic calculations;b) Chemical separation and isolation of precipitate;c) Equilibrating a series of steels with different Nb contents with a H2-CH4

atmosphere at various temperatures, after which the carbon contents are analyzed;d) Hardness measurements;e) Statistical treatment of previous solubility products.

Although techniques such as chemical separation and methane equilibration indirectly accountfor chemical interactions, both have their own limitations. One problem arising from theseparation technique Is that very fine precipitates may not be included in the analysis (63) andthat discrepancies may exist as to the exact composition of the precipitate (52). This problemalso plagues the equilibration methods, as carbon contents are often analyzed assuming that auniform stoichiometric or non-stoichiometric compound is present (52, 63). Additionally, bothof these methods, including the thermodynamic methods, neglect the effect of precipitate sizeon solubility. Thermodynamics indicate that small particles are more soluble than large

particles (71). Hence, solubility products obtained via these methods may predict a more stable

precipitate than would be expected. A possible exception to this may be found in the work bySimoneau et al. (72) who used electrical resistivity measurements to determine the solubility ofNb (CN) in austenite. This technique is based upon the measured reduction in bulk resistivityas Nb, C and N in solid solution precipitate to form Nb (CN). The reduction in resistivity iscaused by the loss of solid solution atoms which act as scattering centers. The results ofSimoneau et al. (72) show that the thermodynamic stability of Nb (CN) is less than that

predicted using equilibrium methods.

Finally, hardness techniques are questionable since they are based upon the assumption that anincrease in hardness is proportional to the amount of Nb dissolved into austenite, andsubsequently precipitated in ferrite as NbC (52). Although this does occur, all of the carbon

and nitrogen in the alloy is not necessarily associated with the precipitate. Also, difficulties

There are three things that must be taken into account when applying a solubility product of theform shown above that has been derived from thermodynamic considerations:

i) It applies strictly to equilibrium conditions that rarely exist in practice,

ii) It might be strongly altered by the presence of other solutes through their effecton the interaction coefficients presented above, and

iii) It will surely be altered by the Gibbs-Thompson effect that accounts for particle

curvature. The solubility of a given particle varies inversely with its curvature.

Hence, a particle of diameter 5 nm would be theoretically much more solublethan one of 5 m, by a factor of 103.

Nb(CN) Solubility Products in Austenite


16/74

arise in separating this hardness increment from those due to other mechanisms such as grainsize strengthening, solid solution strengthening and dislocation strengthening.

Table III Solubility products for Nb-C, Nb-N and Nb-C-N systems in austeniteSystem Product Method Reference

Log[Nb][C] = 2.9 - 7500/T d 52Log[Nb][C] = 3.04 - 7290/T b 52Log[Nb][C] = 3.7 - 9100/T c 57

Log[Nb][C] = 3.42 - 7900/T b 58Log[Nb][C] = 4.37 - 9290/T c 59Log[Nb][C]0.87= 3.18 - 7700/T b 60Log[Nb][C]0.87= 3.11 - 7520/T e 52Log[Nb][C] = 2.96 - 7510/T e 52Log[Nb][C]0.87= 3.4 - 7200/T a 52Log[Nb][C] = 3.31 - 7970/T + f* b 61Log[Nb][C]0.87= 2.81 -7019.5/T a 62Log[Nb][C] = 1.18 - 4880/T 64Log[Nb][C] = 3.89 8030/T 64Log[Nb][C] = 4.04 - 10230/T c 65

Nb-C

Log[Nb][C] = 3.79 - 10150/T b 60Log[Nb][N] = 2.8 - 8500/T b 58Log[Nb][N] = 3.7 - 10800/T b 67Log[Nb][N]0.87= 2.86 - 7927/T a 62

Nb-N

Log[Nb][N] = 4.2 - 10000/T 64Log[Nb][C]0.24[N]0.65= 4.09 - 10500/T b 52Log[Nb}[C + 12/14N] = 3.97 - 8800/T c 68Log[Nb][C + N] = 1.54 - 5860/T b 52Log[Nb][C]0.83[N]0.14= 4.46 - 9800/T b 52

Nb-C-N

Log[Nb][C + 12/14N] = 2.26 - 6770/T c 66* f = [Mn](1371/T - 0.9) - [Mn]2(75/T - 0.0504)

A comparison between some of the solubility relations listed in Table III is shown in Figure 11(7). This diagram depicts the solubility product in austenite, K, versus temperature. The resultsshown by Figure 11 indicate two important points. First, the solubility product is substantiallylowered as the compound becomes enriched in nitrogen. Second, the solubility of a

precipitating compound in austenite is decreased as the vacancy content becomes smaller.Solubility products play a vital role in understanding the physical metallurgy of microalloyedsteel, especially those aspects which are concerned with precipitation-related phenomena.Solubility products can be plotted in either of two ways. Consider the case of the solubility

product of NbC in austenite at some given temperature. If the abscissa (C) and ordinate (Nb)axes have a linear scale, then any given solubility isotherm will have the shape of a hyperbole.If, however, the axes have a logarithmic scale, then the solubility isotherm will be a straightline. Both approaches have been widely used in the literature.

A hypothetical solubility isotherm for NbC in equilibrium with austenite at 1000C is shown inFigure 12(a) (7). The solubility isotherm gives the locus of Nb and C products whichrepresents the limit of solid solubility of NbC in austenite at 1000C, i.e. any combination of

products located above the line will be in the ? + NbC phase field at 1000C. The straight linewith a positive slope on the diagram represents the stoichiometric ratio. The precipitation of

NbC can be followed by considering the slightly curved line which is nearly equidistant fromthe stoichiometric line and passes through the point with the Nb and C coordinates which


17/74

Figure 11: Solubility products for various Nb precipitates in austenite. AfterNordberg and Aronson, 1968 (7).

Figure 12: Hypothetical solubility diagrams describing equilibrium between NbCand austenite: (a) relationship at 1000 C and (b) interrelationships amongisotherms, steel composition, processing and potential amount of precipitation.After DeArdo, Gray and Meyer, 1981 (7).


18/74

describe the composition of the steel. A schematic illustration is presented in Figure 12(b).Consider a steel of composition given by point B which is reheated to 1300C and is veryslowly cooled to 900C. Since 1300C is the solution temperature for the NbC underconsideration here, the NbC can be assumed to be completely dissolved after reheating. If wefurther assume that precipitation occurs during the cooling, then the composition of theaustenite in equilibrium with NbC would move along the curve passing through point B andwhich is equidistant from the stoichiometric line. The distance moved along the curved linethrough point B is proportional to the volume fraction of precipitate formed as a result of the

cooling, assuming equilibrium is established at all temperatures. In other words, the distancemoved along the curved line through B during cooling is proportional to the supersaturation andwould also be proportional to the volume fraction precipitate formed if equilibrium prevailed.Therefore, if steel B is reheated to 1300C and then cooled to and rolled at 900C, the distanceBC in Figure 12(b) will be proportional to the volume fraction of precipitation of NbC formedin austenite. This precipitation would be of two types: that formed during cooling and thatformed during or after rolling. The first type is not likely to be significant since precipitation inrecrystallized austenite is normally very sluggish, as will be discussed later. The second type isthe strain-induced precipitation. A second interesting case would occur if steel of compositionA in Figure 12(b) is considered. If this steel is reheated at 1300C and hot rolled at 900C, thentwo arrays of NbC particles would be expected. The first would be the precipitates thatsurvived the reheating treatment (volume fraction proportional to AB) and the second strain-induced precipitate (volume fraction proportional to BC).

Several investigators have used solubility diagrams to help explain the physical metallurgy ofmicroalloyed steels, and, in particular, the precipitation phenomena, (73,74). Ourunderstanding of precipitation in these steels has been substantially increased by the work ofWadsworth et al. (75), Roberts et al. (45), and Keown and Wilson (76). The study byWadsworth, et al. (75) concluded that the supersaturation that could be developed between anytwo temperatures is a strong function of the position of the steel relative to precipitatestoichiometry on the solubility diagram. This effect, illustrated in Figure 13, shows that the

largest possible supersaturation occurs when the microalloying element and the interstitial arepresent in the steel in the stoichiometric ratio, and deviations from this ratio will lead todecreasing supersaturations.

The work of Roberts et al. (45) presents an interesting way of considering the precipitation ofcarbonitride precipitates in microalloyed steels. They suggest a thermodynamic analysis whichattempts to predict the change of x and y in VC xNy with changes of temperature and/or extentof precipitation. Their model predicts that nitrogen-rich precipitates are the first to form andthat nitrogen plays a central role in controlling the precipitation until it is completely consumed.As stated above, solubility products can be influenced by the presence of elements that do notdirectly participate in the precipitation reaction. This effect occurs through the influence of

third and higher order elements in solution on the interaction coefficients, hence activities, forNb and C in austenite. Third elemental solutes that raise the activity of Nb or C through apositive interaction coefficient decrease their solubility while those that decrease the activitythrough negative coefficients increase it. Interaction coefficients for C, N, and Nb in liquidsteel at 600C are shown in Table IV (77). Unfortunately, similar data for these species inaustenite at 1000C, for example, do not exist at this time. If we assumed that the data of TableIV did apply to austenite at 1000C, then we would find that certain solutes would increase thesolubility of C and Nb in austenite, leading to less than expected NbC, while others woulddecrease it, leading to more.


19/74

Figure 13: Amount of niobium carbide available for precipitation at 923 K (aftersolution-treatment at 1373 K) as function of degree of deviation fromstoichiometry, r. Positive values of r indicate C-rich compositions, negative values

Nb-rich. After Wadsworth et al, 1976 (75).

Figure 14: Solubility for each pure compound, NbC, TiC and TiN in austeniteand ferrite. After S. Akamatsu et al., 1991.


20/74

Examples of this effect have been studied by Koyama, et al. (78) who have examined theinfluence of several elements on the solubility of NbC in austenite. By way of an example, thesolubility product for NbC in otherwise unalloyed austenite is about 5 x 10-3 at 1150C. Anaddition of Mn increased log K by about 5% per % Mn added, while an addition of Si reducedlog K by about 45% per % Si added. The solubility product at 1150C for a steel that contained1.5% Mn and 0.4% Si was 4 x 10-3; the addition of Mn and Si acted to reduce the solubility

product by about 20%.

Table IV Influence of several selected elements on the interaction coefficients forC, N, and Nb in liquid iron at 1600C (77)

Solute C N Nb

Al 5.3 5.2C 6.9 5.86 -23.7Cr -5.1 -10H 3.8 -1.5Mn -2.7 -8.1Mo -4 -4.9

N 5.86 0.8

Nb -23.7 -26 -26Ni 2.9 1.5O -22 4.0 -54P 7 6.2S 6.5 1.4 -5.8Si 9.7 5.9

The latest and perhaps the most accurate solubility product for nominally stoichiometric NbC inaustenite in a steel containing 0.08C-1.5Mn-0.008N-0.02Nb was published by Palmiere et al. in1994 (79). The following expression was derived from solubility data obtained from reheatedand quenched samples as measured using the atom probe facility of an APFIM instrument:

Log [Nb][C] = 2.06 6700/T (20)

It is important to note that a comparison of this product with several earlier ones indicates thatthe earlier products substantially overestimated the amount of Nb in solution in austenite andunderestimated the dissolution temperature (79). This product has been used successfully insubsequent research (80).

Solubility of NbCN in Ferrite

It is well known that the solubility of various elements is higher in austenite than in ferrite (81,82). An example, published in 1994, for the solubility of various microalloyed carbides andnitrides in austenite and ferrite in an ultra-low carbon steel is shown in Figure 14 (83, 84). Atabout the same time, solubility products for various microalloyed precipitates in ferrite were

published by Taylor (85).

It is interesting to compare the expected behavior for MC in austenite and ferrite for the variousmicroalloying elements. One way that this can be accomplished is by comparing the solubilitylimits for the various carbides in austenite and ferrite at a given temperature, for example, neara typical Ar3. The result of using this approach is shown in Table V below, where thesolubility limits in austenite and ferrite were calculated at 800C using the products suggested


21/74

by Taylor (85). This table shows that the limiting product for NbC is reduced by a factor ofnear 20 when going from austenite to ferrite at 800C.

Table V. Comparative Solubility Limits for MC in Austenite and Ferrite at 800CSolubility Product [M][C] X 10 4

NbC in 8.9 X 10-5NbC in 4.5 X 10-6

TiC in 1.7 X 10-4

TiC in 3.0 X 10 5VC in 7.9 X 10-3VC in 1.1 X 10-3

Crystallography of Precipitation: Orientation Relationships and Lattice Matching

The crystallography of precipitation in steel has been reviewed by Jack et al. (86), Davenport etal. (87), and Honeycombe (88). There are two aspects of the crystallography of precipitationthat are of interest in this discussion. These are (a) the orientation relationship that exists

between the crystal structure of the precipitate and that of the matrix, and (b) the degree oflattice registry between the precipitate and the matrix.

Carbonitrides of niobium, vanadium and titanium can precipitate in both austenite and ferrite.Several studies (87, 89-91) have shown that when these carbonitrides precipitate in austeniticstainless steel, they do so such that the lattice of the precipitate with the NaCl crystal structureis parallel to the FCC lattice of the parent austenite, i.e.

[100]M(CN)|| [100]?[010]M(CN)|| [010]?

Davenport et al., have provided direct evidence that this same relationship holds for the strain-induced precipitation of NbC in austenite in a microalloyed steel (87). When NbCxNyprecipitates in ferrite (87, 92) or martensite, (93) it does so with the Baker-Nutting orientationrelationship (94):

[100]NbC || [100]a[011]NbC || [010]a

The parallel and Baker-Nutting orientation relationships can be simply illustrated through theuse of the appropriate metal-atom octahedra (86). The metal-atom octahedra for austenite,

NbCxNy and ferrite are shown in Figure 15 (86). The structures of the austenite and NbCxNyare positioned to represent the parallel orientation relationship and the structures of the ferriteand NbCxNy are positioned to represent the Baker-Nutting orientation relationship. Note thatthe three octahedra shown in Figure 15 all have a cube plane at the base.

The octahedra shown in Figure 15 also permit the calculation of lattice misfit strain or thelattice strain required for coherency. This is done by calculating the linear strain in the matrixlattice parameter that would be required to bring the two lattices into coincidence at the matrix-

precipitate interface, i.e.


22/74

Figure 15: Metal-atom octahedral for (a) austenite, (b) NbC and (c) ferrite.Adapted from Jack & Jack, 1973 (86).

Figure 16: Strain induced precipitation of NbCN in austenite in a steel containing0.09%C - 0.07%Nb. Specimen reheated at 1250, rolled 25% and held at950, and air cooled to RT. Centered dark field electron micrograph using a(111) NbC reflection. After Santella, 1981.


23/74

100L

LL%

m

mp

=StrainLinear (21)

where Lp= appropriate length of octahedron of the precipitate, andLm= appropriate length of octahedron of the matrix.

Examples of the required strain or strain of the matrix for several types of precipitates areshown in Table VI. The magnitudes of the elastic matrix strains (e = 0.255 for NbC in austenite

and e = 0.105 and 0.563 in ferrite) required for lattice registry would appear to rule out anylarge degree of coherency between the precipitate and the matrix. However, the elastic strainsrequired of the matrix to achieve partial lattice registry could be easily accommodated by the

presence of a few interfacial dislocations for a precipitate of dimension 100 .

Table VI Lattice mismatch for NbCxNyprecipitates in austenite and ferriteRequired Linear Strain in Matrix, %

?

MatrixOrientation

Relationship NbC NbC.8 NbN.8

[100]ppt|| [100] 25.5 26.6 23.0[010]ppt|| [101] 25.5 26.6 23.0[001]ppt|| [001] 25.5 26.6 23.0[100]ppt|| [100] 56.3 57.7 53.1[011]ppt|| [010] 10.5 11.5 8.4[011]ppt|| [001] 10.5 11.5 8.4

? All matrix strains are tensile (+).

A first-order approximation indicates that about seven and three dislocations would be requiredto cancel the lattice mismatch for a precipitate of NbC of size 100 in austenite and ferrite,respectively.

The orientation relationship that is observed between the NbC and ferrite can be used todistinguish the NbC which had nucleated in austenite from the NbC which had nucleated inferrite. As was noted above, NbC forms in austenite with a parallel orientation relationship andin ferrite with the Baker-Nutting orientation relationship. Therefore, all NbC precipitates thatshow the Baker-Nutting relationship must have formed in the ferrite. The NbC that forms inaustenite will not have the Baker-Nutting relationship with the ferrite.

When austenite transforms to ferrite or martensite, it does so with the Kurdjumov-Sachsorientation relationship (95).

(111)? || (110)a

[110]? || [111]a

The consequence of this relationship is that when the matrix transforms from austenite toferrite, the orientation of the original austenite and the precipitates that formed in that austenitewould be related to the ferrite by the Kurdjumov-Sachs relationship. Therefore, the precipitatesthat formed in the austenite can be identified because they will have the Kurdjumov-Sachsorientation relationship with the ferrite when observed at room temperature. If the orientationof the austenite grains would change after precipitation had occurred, e.g. by grain rotationsaccompanying deformation or recrystallization, then there would be no rational crystallographicrelationship between the NbC that had formed in the austenite and the final ferrite matrix. In

principle, this change in austenite grain orientations after precipitation would not appear to be a


24/74

very frequent occurrence since most of the precipitation that forms in austenite is strain-induced, and these precipitates act to suppress subsequent recrystallization of the deformedaustenite in which they formed.*

Precipitation of NbCN: Typical Morphologies and Distributions

The precipitation of NbCN in austenite and ferrite is heterogeneous in nature, i.e. it alwaysoccurs in conjunction with crystalline defects such as grain boundaries, incoherent twin

boundaries, stacking fault boundaries, subgrain boundaries, or dislocations. One reason whythe precipitation forms in this fashion is the rather large mismatch between the lattice of NbCNand the matrix, austenite (18, 23, 31, 33-35) or ferrite (10, 23, 29, 34, 36-42) Table VI. Thesecrystalline defects are sources of dislocations which can act to cancel some of the elastic strainwhich may develop during the formation of the precipitates.

In one of the few studies that have been conducted on the precipitation of NbCN inrecrystallized austenite, Santella has shown that the NbCN forms almost exclusively on grain

boundaries (29) as is the case for precipitation in casting (42). Much more research has beendone on NbCN precipitation which has formed in deformed austenite and in ferrite. Goodillustrations of strain-induced precipitates of NbCN in deformed austenite are available in theliterature (87, 96-100) and an example is presented in Figure 16 (29). In every case, the

precipitates appear to decorate what was once a grain or subgrain boundary in the prior-austenite.

When NbCN precipitates in ferrite, the nature of the precipitate distribution is related to thenature of the austenite-to-ferrite transformation. If the ferrite that forms has a polygonalmorphology, the NbCN precipitation will have the interphase distribution (40, 96). Duringthe interphase precipitation, the NbCN forms along the advancing austenite-ferrite phase

boundary. When the boundary moves to a new location, the precipitates are left behind in asheet-like array. The final microstructure consists of numerous sheets of precipitates, where

each sheet denotes the location of the interphase boundary during the course of thetransformation. This form of precipitation in Nb steels has been observed by Gray and Yeo(101) and others (87, 102, 103), and an example is given in Figure 17 (29). It should be notedthat this form of precipitation only occurs at very high temperatures in ferrite. When

precipitation occurs substantially after transformation is complete, the precipitate has a moreuniform or general distribution (101). These precipitates are responsible for the precipitationhardening effect of NbCN.

Similar general distributions occur in ferrite which has an acicular or bainitic character (87).Furthermore, there is great similarity in distributions and morphologies between the NbCN thatforms in acicular ferrite and the NbCN that forms during secondary hardening in the tempering

of a quenched steel (93). An example of this general type of precipitation of NbCN in acicularferrite is given in Figure 18 (29).

Niobium, Thermomechanical Processingand Austenite Conditioning

Background

*Hereinafter, NbCN will be used to denote NbCxNy


25/74

Although Nb can play many important roles in steel, none is more important than austeniteconditioning. Here, austenite conditioning means having the terminal hot rolled austenite withthe proper microstructure and composition to allow the desired final ferrite microstructure to be

Figure 17: Interphase precipitation of NBCN in ferrite in steel containing 0.09%C- 0.07%Nb. Specimen was reheated to 1250, hot rolled to 1000 and aircooled to RT. Bright field electron micrograph.

Figure 18: General precipitation of NbCN in austenite in a steel containing0.09%C - 0.07%Nb. Specimen reheated at 1250, hot rolled to 1000 and aircooled to RT. Centered dark field electron micrograph using a (111) NbCreflection. After Santella, 1981.


26/74

achieved after suitable cooling. For simple ferrite-pearlite steels, this means controlling(increasing) the crystalline defect content of the austenite that can act as nucleation sites forferrite upon transformation. This was the original view of austenite conditioning and waswidely discussed in the early 1980s. These defects, as we shall see, include grain boundaryarea, deformation band area, and incoherent twin boundary area per unit volume. This catalyticeffect has been discussed earlier (104-106). In the past 20 years, we have expanded this viewof austenite conditioning and its benefits to include acicular ferrites, bainites and martensites(107). Furthermore, the benefits of austenite conditioning have been extended to high strength

strip and sheet with improved stretch forming, e.g., the dual-phase and TRIP steels. Here, thedistribution of low temperature transformation products and their contribution to workhardening is controlled by the defect structure of the austenite (108-115). This work shows thatthe finer the austenite, the finer the distribution of low temperature transformation products,and the higher the work hardening rate.

Role of Processing

Before the role of Nb in austenite conditioning can be discussed, it is important to brieflyreview the various commercial rolling practices being used. Since Nb can act as either a soluteor precipitate in austenite, each with a different effect, it is important to recognize where the Nbwould be during rolling, and this will vary with the rolling practice. For precipitation to occurin austenite, there must exist adequate levels of both supersaturation and interpass time. Hence,slow processing in austenite, typical of reversing plate and heavy section rolling, where therolling involves many light passes with long interpass times and long total rolling times, israther ideal for extensive static precipitation of NbCN in austenite. However, fast processing ofaustenite, such as found with strip or bar rolling, where both the interpass times and the totalrolling times are short, is not conducive for extensive precipitation. It is, therefore, notunreasonable to assume that Nb will influence the behavior of austenite first as a solute andthen as a precipitate in plate and structural rolling, whereas it is expected that Nb would largelyremain in solid solution during strip and bar rolling (116, 117).

Behavior of Niobium in Austenite

When a Nb-bearing low-alloy steel is in the austenite phase field, the Nb will be in both thesolid solution matrix and in the precipitated NbCN. At equilibrium, this partitioning of Nb

between the matrix and precipitate will be controlled by the solubility relations discussedearlier. Since one of the primary requisites of a successfully conditioned austenite is the

presence of a large number of crystalline defects that can act as sites for ferrite nucleationduring cooling, i.e., a high Sv, the motion of subgrain and grain boundaries associated withstatic recrystallization and grain growth after hot deformation must be retarded. Evidence forthis retardation by Nb in both solid solution (118-120) and in precipitate (121-123) can be

found in the literature.

The different effects of NbCN precipitation during hot rolling have stimulated a large numberof studies concerned with the kinetics of precipitation in austenite as influenced bycomposition, strain, strain-rate, temperature and overall heat treatment. This group of studiesof precipitation kinetics has utilized a wide variety of techniques including: chemical analysis(20, 26, 28, 43, 124, 67), electrical resistivity (124, 125), X-ray diffraction (98), quantitativeelectron microscopy (100), flow curves (120), and hardness testing (124, 126, 127). The

precipitation which has been studied in these experiments is, with one exception, the strain-induced precipitation which occurs during and/or after deformation. The exception is thedynamic precipitation which accompanies deformation and which has been studied by Jonas et

al. (120, 128).


27/74

Studies of the precipitation in recrystallized austenite have shown that the kinetics are verysluggish (24, 124, 125). The results of the study by Simoneau et al. (124) on precipitation ratesat temperatures above 900C are shown in Figure 19. The kinetics of NbCN precipitation at900C is shown in Figure 20; these data are from the work of LeBon et al. (118). Finally, the

precipitation kinetics at low temperatures, below 950C, has also been determined by Watanabeet al. (25); these results are shown in Figure 21. These studies illustrate the slow rate of

precipitation in recrystallized austenite; it takes several thousands of seconds at 900C for 50percent of the potential precipitation to form. Also, by combining the results of Simoneau et al.

(124) and Watanabe et al. (25) the overall precipitation behavior does appear to conform to C-curve kinetics.

The precipitation rate is remarkably sensitive to the level of strain imparted prior to aging, as isillustrated by the results of LeBon et al. (118) Figure 20, and Hoogendorn and Spanraft (67),Figure 22. The strain-induced precipitation of NbCN in austenite appears to follow C-curvekinetics. Three C-curves, each based on a different technique, are shown in Figures 22, 23, and24. These results are from the work of Watanabe et al. (25) Figure 21; Ouchi et al. (126) Figure23; and Hansen et al. (100) Figure 24. In all three cases, the nose of the C-curve appears to bein the temperature range 900 to 950C.

The overall composition of the steel appears to have a strong effect on the kinetics ofprecipitation (25, 120, 127). For example, the presence of Mo appears to shift the C-curve tolower temperatures and shorter times, (25) whereas an increase in Mn level acts to shift the C-curve to longer times (128). These effects are probably caused by the influence of theseelements on the activities of solute Nb and C, as discussed above.

Studies of the precipitation of NbCN in deformed austenite are subject to certain problemswhen the temperature of deformation (and subsequent holding) are raised. The problem arises

because the density of crystalline defects in the austenite does not vary continuously withtemperature over a large temperature range, say, 800 # T # 1200C. For a given set of

conditions of composition, temperature, strain and strain-rate, the austenite deformed at hightemperatures may experience static recrystallization between passes if the interpass time issufficiently long. This would be expected at large strains, high temperatures and high strainrates. The austenite deformed at low temperatures will be highly elongated and the austenitedeformed at intermediate temperatures will have a mixed equiaxed plus elongated grainstructure. Behavior of this kind would be expected to give a discontinuous relationship

between the crystalline defect density that can act as nucleating sites for NbCN precipitationand the deformation temperature. This type of austenite deformation behavior has beenobserved in microalloyed steels (46, 98) and has been shown to influence the results of

precipitation kinetic studies (100). An example of this effect is shown in Figure 25, which hasbeen taken from the work of Davenport et al. (98). Figure 26, where the relative integrated

intensity is assumed to be roughly proportional to the volume fraction of precipitate, indicatesthat there are high rates of precipitation in the temperature range 950 to 1100C and lower ratesat temperatures which are either above or below this range. Davenport et al. (98) have notedthat the lowering of precipitation kinetics which results from higher temperature deformationand holding may be due to two effects. The first is the reduction in solute supersaturation withincreasing temperature; this alone would lead to a decrease in precipitation rate. The second isthe change in deformation structure of the austenite near 1100C. Austenite rolled below thistemperature leaves the rolls with an elongated, deformed microstructure that contains numeroussites for strain-induced precipitation. Austenite rolled above this temperature quickly achievesthe statically recrystallized state; this structure contains few sites for strain-induced

precipitation; hence, low precipitation kinetics can be expected.


28/74

Figure 19: Isothermal precipitation of Nb in undeformed austenite in steelcontaining 07%C-.04%Nb-.010%N. After Simoneau (124).

Figure 20: Influence of strain level on the kinetics of Nb precipitation at 900C.After Lebon and Saint-Martin (118).


29/74

Figure 21: Time-temperature-precipitation diagram showing effect ofdeformation of Nb(C,N) in austenite. After Watanabe et al. (25).

Figure 22: Influence of deformation on precipitation in a steel containing .06%C-

.041%Nb and .06%N. After Hoogendorn and Spanraft (67).


30/74

Figure 23: Precipitation of Nb(CN) and recrystallization kinetics duringisothermal holding in Nb-bearing steel: (a) hardness changes after tempering at1112F (600C); (b) precipitation kinetics; (c) recrystallization progress. AfterOuchi et al., 1976 (126).

Figure 24: Recrystallization-precipitation-temperature-time (RPTT) diagram forsteel 3 after solutionizing at 1250C and hot rolling 50% at 950C. After Hansenet al. (100).


31/74

Figure 25: Relative integrated intensity versus aging time for samples deformed60% at the indicated temperatures. After Davenport et al (98).

Figure 26: PTT curves for dynamic precipitation of NbCN in austenite. AfterAkben et al. (128).


32/74

The studies of precipitation kinetics described to this point have involved static precipitation.That is, the austenite has been deformed in one operation and then aged in a second operation.A different approach to studying strain-induced precipitation has been developed by Jonas et al.(120, 128). This approach enables the kinetics of strain-induced precipitates to be determinedin a dynamic context, i.e. where the austenite is being strained and aged at the same time.This approach yields C-curves describing the dynamic precipitation and is based on the analysisof hot flow curves. C-curves describing this dynamic precipitation (128) are shown in Figure26.

Principles of Austenite Conditioning

There are two different approaches available for austenite conditioning during hot rolling.They are known as conventional controlled rolling (CCR) and recrystallization controlledrolling (RCR). The purpose of this paper is to describe these practices, to show how they differfrom conventional hot rolling (CHR), and to indicate the microalloying additions required fortheir respective implementation (116, 117, 129). In addition, the requirements of the millequipment suitable for these practices will be discussed.

Conditioning of Austenite

Obtaining fine ferrite grain sizes from the transformation of austenite requires high rates offerrite nucleation, and low rates of growth and subsequent coarsening. Conditioning ofaustenite means that the microstructure of the austenite has achieved, through controlled hotdeformation, the proper predetermined metallurgical state or condition prior to transformationthat will embody these requirements. It was shown earlier that high ferrite nucleation ratesresult from having a large number of potential nucleation sites and a high nucleation rate persite (105, 130, 131). The sites for ferrite nucleation are austenite grain and incoherent twin

boundaries and deformation bands (105, 132-137). The density of these sites per unit volume isexpressed as the total interfacial area of the near-planar boundaries per unit volume and has the

units mm2

/mm3

or mm-1

. This stereological concept was first discussed by Underwood (138),and was later adopted to describe austenite by Kozasu et al. in Microalloying' 75 (132). Kozasuet al. called this measure of the total interfacial area per unit volume, the parameter Sv. The

parameter Sv can be considered as an approximate austenite grain size, and its magnitude isan indication of the degree of austenite conditioning. The principal goal of TMP and austeniteconditioning is to maximize Sv. The influence of Sv on ferrite grain size is shown in Figure 27(139).

There are two entirely different approaches to increasing Sv. In the first, the initiallyrecrystallized reheated austenite grains undergo repeated recrystallization during subsequenthot deformation leading to grain refinement. This deformation would take place at

temperatures above T95 in Figure 28. Here, one set of equiaxed grains is replaced by a new setof finer, equiaxed grains. Since these fine grains would have a strong tendency to coarsenduring the interpass time, this fine grain size can be retained only if a suitable mechanism isavailable to suppress grain coarsening. The process that involves repeated recrystallization andinhibition of grain coarsening is called Recrystallization Controlled Rolling (RCR), and wasoriginally proposed by Sekine et al. (140, 141). Clearly, the lower the T95, the larger will be the

processing window between the reheat temperature and the minimum finishing temperature ofT95. Hence, steels designed for RCR conditioning must have a low recrystallization-stoptemperature and a pre-existing grain coarsening inhibition system. This practice is well-suitedfor production conditions that require high finishing temperatures, e.g. underpowered rollingmills, heavy section and heavy plate rolling, and forging.


33/74

Figure 27: Ferrite grain sizes produced from recrystallized and unrecrystallizedaustenite at various Sv values [139].

Figure 28: Schematic illustration of austenite microstructures resulting fromvarious deformation conditions [142].

The second approach involves substantial deformation below T5,

where, for example, the thickness at T5 is three or four times the final


34/74

gauge. In this method, the grain size at T5 is deformed and stays unrecrystallized through theinterpass time for all subsequent passes. Hence, there is a change in grain shape and theoccurrence of transgranular twins and deformation bands. The process that involves repeatedflattening or "cold working" of the grains by repeated deformation below T5 is calledConventional Controlled Rolling (CCR). Clearly, the higher the T5 the more passes can be usedand the more effective is the practice. Hence, steels designed for CCR conditioning must havea high recrystallization-stop temperature.

Although there are two different approaches to austenite conditioning, i.e. RCR and CCR, theyboth have the same objective of resulting in structural refinement in the final plate, coil, beam,or forging. The differences between RCR and CCR can be easily understood with the aid ofFigure 29 which shows the different paths used by each to achieve high values of Sv(142).

The increase in Sv for the RCR practice comes exclusively from the increase in grain boundaryarea per unit volume which results from a decrease in average grain volume, while the increasein Sv for the CCR practice results from the increase in grain boundary area per unit volumeresulting from a change in grain shape and through the addition of the transgranular twins anddeformation bands.

Physical Metallurgy of Thermomechanical Processing: Driving Forces and Retarding Forces

It is evident from the above that temperatures such as T95 and T5are significant in considerationof steels for TMP. Steels that are to be used for RCR processing must achieve, throughmicroalloying, two essential features. First, they must have a relatively low T95 to enable the15-21 passes needed, for example, for wide plate rolling to take place considering the usualdrop in temperature of about 300 - 400C from the reheat to the finishing temperature. Second,they must have a pre-existing grain boundary motion inhibition system that is small enough to

permit static recrystallization but large enough to suppress post-recrystallization graincoarsening. The difference in high temperature rolling behavior of austenite during CHR and

RCR is shown schematically in Figure 30. The principal difference between the two is that theRCR steel has a grain coarsening inhibitor whereas the CHR steel does not.

Grain Coarsening

Driving Force for Coarsening

Fine grain sizes result from high rates of "nucleation" and low rates of "growth" duringrecrystallization and from the inhibition of subsequent grain coarsening. Grain coarseningoccurs in polycrystalline metals and alloys because of the need to achieve an equilibrium

condition. This condition requires that the grain boundaries move to reduce curvature and tomeet at certain angles (143). The non-equilibrium forces on grain boundaries cause them tomove toward their respective centers of curvature. This boundary motion causes the smallergrains to disappear and the larger grains to grow with a net increase in grain size with, forexample, holding time at a suitable temperature. The pulling force per unit area of grain

boundary is:

F = ?G/VM (22)


35/74

Figure 29: Grain size evolution through succeeding passes depending on TMP.

Figure 30: Grain size evolution through succeeding passes depending on TMP.

where ?G is the lowering of the free energy caused by the boundary migration and VM is the

unit volume of material. Equation 22 applies to any boundary whose motion leads to a


36/74

lowering of the free energy. For grain coarsening, the ?G arises from the boundary curvature,and, in the case of fine grained austenite, leads to a driving force of approximately 100 kN/m2(144). For recrystallization, the ?G is a result of the dislocation density difference between theunrecrystallized and recrystallized grains. A typical driving force for the recrystallization ofaustenite as a result of a single low temperature rolling pass is approximately 20 MN/m2 (49,80, 145, 146), or about 200 times larger than that for coarsening.

The coarsening of austenite grains has been treated by several investigators (147-149). The

approach adopted here is that of Gladman (150, 152). Gladman suggested that grain coarseningis accompanied by two types of boundary motion: the inward motion of boundaries of shrinkinggrains and the outward motion of boundaries of growing grains (150-152). An energy balanceshowed that only grains with a size larger than 4/3 that of the average could grow. This clearlyshows the importance of homogeneity in grain size distribution. According to this view, verynarrow grain size distributions would be expected to show little, if any, grain coarsening,independent of average grain size.

Coarsening Inhibition Using Particles

While the tendency for grain coarsening will always be present in real steels, it is important torealize that this coarsening tendency can be slowed or eliminated by the presence of suitableforces that can inhibit boundary motion. Two of the more important mechanisms are particle

pinning and solute drag. When a boundary intersects a particle, a portion of the boundary iseliminated. To pull the boundary away from the particle requires the re-creation of boundaryarea, and, hence, the application of a force. This force is the force for grain coarsening.Gladman's treatment of grain coarsening also included the effects of particles (150-152).Gladman's analysis is based on the earlier of work of Zener (153). His energy balanceconsidering the shrinking of small grains, the expanding of large grains, and the pulling of

boundaries away from particles led to what is considered the Gladman equation:

( )[ ]1

Oc 2/Z-3/2pf6Rr

= (23)

where rc is the critical or maximum particle size required to suppress coarsening, R0 is theradius of the grain of average size, f is the volume fraction of particles, and Z is the ratio ofradii R/Ro of growing to average grains. Hence, for any average grain size, there exists acombination of particle size and particle volume fraction that can suppress coarsening. TheGladman equation is shown in Figure 31, assuming Z = 1.5 (152).

The negatively sloping lines represent combinations of rc and f that can pin all grain sizeswhich fall above the line. The finest grain sizes that can be stabilized by particles in real steelsare shown in the lower hatched area. The upper hatched area shows the finest grains that can

be anchored by large inclusions. The nearly two orders of difference in grain sizes that can beanchored by these two very different particle distributions is striking.

The particles best suited for inhibition of grain coarsening are those that have the smallest sizeand the largest volume fraction. This requires that the particles have a low solubility inaustenite and themselves coarsen slowly. The volume fraction will be governed by the bulkcomposition, the appropriate solubility product, and the temperature while the particle size will

be controlled by rate of coarsening of the particles. The particle coarsening rate due toLiftschiftz-Wagner (154, 155) is given as:

r3- ro3 = 8 v2? C D t [9kT]-l (24)


37/74

where r is the average particle size at time t, ro is the initial average particle size, V is theatomic volume of the precipitate, ? is the particle-matrix surface energy, C is normally taken to

be the concentration of the metal atom in solution at the particle interface, D is the diffusioncoefficient of the metal atom in the matrix, t is the coarsening time, k is Boltzmann's constantand T is the absolute temperature. Equation 24 shows that particle coarsening rates will bereduced when ?, C and D are as low as possible. The choice of particle will determine thesethree parameters. The structure and local composition of the particle-matrix interface

determine ? and is controlled by the lattice matching across the interface. Once the particle ischosen, the matching is essentially set. The more coherent the interface, the lower the ?. Themagnitude of C is determined by the composition of the particle, the nature of the particle, andthe solubility product. Systems that have low concentrations of metal atoms in the matrix inequilibrium with the particles obviously have a low C. Two requisites for low C are a lowsolubility product for the particle in austenite and a rather high interstitial level. This meansthat systems with low solubility products will have not only large volume fractions of particles

but also low rates of coarsening (152, 156). Alternatively, low particle coarsening rates wouldbe favored by steels that are hypostoichiometric in the interstitial element, i.e. on the right sideof Figure 13.

Coarsening Inhibition Using Solute Drag

When solute atoms are located on and attached to a grain boundary, the migration rate of theboundary resulting from a given driving force will be influenced by the solute atoms. Thetheory of solute drag has been the subject of numerous studies and reviews (157-159). Most ofthis work was conducted in recrystallization experiments where, as discussed above, the drivingforces are rather high. Grain coarsening, on the other hand, takes place at very low drivingforces.

According to Cahn's theory, the rate of grain growth or coarsening G can be given as:

G2= (2 s VMn / t) / (? + aC) (25)

where s is the specific grain boundary energy per unit volume, VM is the molar volume ofaustenite, n is the isothermal grain coarsening law exponent, t is the coarsening time, ? is thereciprocal boundary mobility of "pure" austenite, a is the reciprocal boundary mobility at unit,concentration solute, and C is the bulk solute concentration. The linear relationship predicted

by this theory between the G2 vs- C has, in fact, been observed (159). The coarsening ratewill fall with increasing solute levels. The choice of solute will influence the values of s, VM,nand a; hence, each solute would be expected to exert its own particular influence on G.

Recrystallization: Driving Force for Recrystallization

The driving force for static recrystallization can be taken to be the difference in dislocationdensity between the deformed or recovered matrix and the recrystallized matrix. This can beestimated by using the approach of Keh and Weissmann (160) who related the increase in flowstress caused by work hardening to the increase in dislocation density. Their expression is:

?s = s 0+ a b (??)1/2 (26)


38/74

Figure 31: Effect of particles on inhibition of grain growth [152]

Figure 32: The effect of deformation temperature on the softening behavior in0.002C and 0.002C-0.097Nb steels. From Yamamoto et al. (165).


39/74

where ?s is the observed increase in flow stress caused by work hardening, s 0 is the initialyield strength, a is a constant, is the shear modulus, b is the Burgers vector and ?? is thechange in dislocation density. The driving force for recrystallization is then given by:

FRXN= b2?? / 2 (27)

As mentioned above, driving forces resulting from individual hot roll pass simulations havebeen found to be on the order of 20 MN/M2using this technique (49, 80, 145).

Recrystallization: Pinning Forces Resisting Recrystallization

When reheated austenite is deformed at a lower temperature where the austenite issupersaturated with respect to the precipitating species, a temperature will be reached wherestrain-induced precipitates will form in the deformed or recovered austenite thereby preventingthe nucleation of recrystallization that would otherwise occur during the interpass time in theabsence of precipitation (49, 80, 145, 162). The temperature where this initially occurs isshown as T95 in Figure 28. At a somewhat lower temperature, the suppression of nucleation iscomplete resulting in T5 in Figure 28. The most homogeneous austenite microstructures resultfrom rolling either above T

5or below T

5.

To retard recrystallization, the pinning forces developed by the strain-induced precipitate mustbe larger than the 20 MN/M2 driving force available to cause recrystallization. The pinningforce can be given as:

SPIN NrF 4= (28)

where r is the particle radius, s is the interfacial energy of the austenite grain boundary, and NSis the number of particles per unit area. Three models have been proposed for NSin calculatingthe pinning forces exerted by arrays of microalloy precipitates (100, 161). These are the rigid

boundary, the flexible boundary and the subgrain boundary models. Recent work has shownthat the subgrain model of Hansen et al. (100) is probably the most realistic (49, 80, 145). Thismodel predicts the pinning force to be:

( ) 1223 = rlfF VS

PIN (29)

where s is the austenite grain boundary energy, fv is the volume fraction particles, l is thesubgrain size, and r is the particle radius. From equation 29 it appears to be rather simple tomeasure an average r and to measure or calculate the fv of the particles. An example of the

precipitates causing this pinning effect in a Nb steel is shown in Figure 16 (29).

It is immediately apparent from Figure 16 that the particles are not randomly distributed, butappear to have formed on the subgrain boundaries of deformed and recovered austenite.Clearly, an average value for Ns would be inappropriate since the distribution of particles is solocalized. It has been found that true localized pinning forces at the sites for preferential

precipitation are far higher than that expected from a uniform distribution of particles (49, 80,145). Equation 29 shows that high pinning forces will be promoted by high values of fv andsmall particle sizes. Those locations with very high localized values of fv, i.e. grain andsubgrain boundaries as shown in Figure 16, will have especially high magnitudes of FPIN.

In order to have the largest number of passes possible below T5 in Figure 28, it is clearly

necessary to have T5 in the correct temperature range. This means that the strain-induced


40/74

precipitation should begin at a temperature which is well up in the range of temperatures wherethe rolling passes actually occur. If the precipitate forms at too high of a temperature it will not

be able to generate a sufficiently high pinning force to suppress static recrystallization duringrolling at lower temperatures. If it forms at too low of a temperature T5will be too close to thefinishing temperature to allow enough passes to occur. The details of RCR processing will not

be discussed here, but they have been presented elsewhere (116, 117, 129).

Role of Niobium in Thermomechanical Processing

The value of an alloying or microalloying addition to a steel from an austenite conditioningstandpoint for controlled rolling derives from its ability to generate sufficiently large pinningforces to retard the motion of crystalline defects, i.e., recovery and recrystallization, that wouldotherwise occur to lower the free energy of the system. Hence we are interested in aquantitative assessment of how a given addition can retard recovery and recrystallization. Thisis accomplished through the use of the well-known, so-called double hit test (80, 145, 162,163), where a series of samples is reheated to a given temperature then cooled to thetemperature of interest. One specimen is deformed continuously to a large strain, the resultingflow curve acts as a reference for the interrupted tests to follow. Other specimens are deformedto a constant pre-strain, often approximating a rolling pass strain, after which they areunloaded, held at temperature for various delay times, then reloaded past yielding. Somemeasure of softening is recorded for each holding time and plots of fractional softening versusholding time at a given temperature are constructed (80, 145, 162, 163). Earlier work onaustenite has shown that the initial 25% softening is caused by static recovery while theremaining 75% is due to recrystallization (80, 145).

This technique has been used in numerous previous studies (80, 145, 162, 163), and Nb hasrepeatedly been shown to strongly retard the progression of both recovery and recrystallizationof austenite (49, 80, 145, 164). An example of the retarding effect of solute Nb is shown inFigure 32 taken from the work of Yamamoto et al. (165).

In the experiment leading to Figure 32, the Nb is assumed to be in solution in the austenite.Hence, the solute Nb was responsible for delaying the onset of recrystallization at 850C from20 sec in plain carbon steel to 400 sec in the Nb steel for a true pre-strain of 0.69 applied at astrain rate of 10 per sec. A comparison of the solute retarding effects of the three microalloyingelements Nb, Ti and V is shown in Figure 33 (165). It is obvious that V has the weakestretarding effect, with Ti intermediate and Nb the strongest. It is also clear that the behaviorexhibited in Figure 34 represents the same trend as found in the Cuddy diagram, Figure 34(161).

A comparison of the relative retarding effects of Nb as a solute and as a precipitate is shown in

Figure 35 (165). In Figure 35, the Nb in the steel with 0.002C is in solution whereas the Nb inthe steel with 0.019C is present as precipitate. The precipitation has caused an order ofmagnitude longer delay for the onset of recrystallization and an even much larger retardation inthe progress or rate of recrystallization.

A somewhat different example of this softening behavior is shown in Figure 36 (145) where thebehavior of three steels is compared. The difference in the nature and magnitude of theinfluence of solute Nb and precipitated Nb is clear in Figure 36. The arrows shown in Figure36 denote the earliest appearance of strain-induced precipitation in austenite. Solute Nb iscapable of suppressing recrystallization for short times while precipitated Nb can for muchlonger times. Hence, either solute or precipitated Nb would be expected to suppress

recrystallization in strip mill rolling which has short interpass times, while solute Nb alone


41/74

could not in a reversing plate mill with much longer interpass times. In this case precipitatedNb would be needed.

When a deformed, supersaturated solid solution of Nb in austenite is held at differenttemperatures, there is a competition concerning kinetics between precipitation andrecrystallization and their interaction. Several studies have attempted to treat this rathercomplex situation (120). One example of such work is shown in Figure 37 (145), where the

behavior of plain carbon steel is compared to that of a Nb steel.

In an attempt to explain the behavior shown in Figure 34, Cuddy tried to calculate the pinningforce required by NbC particles to suppress recrystallization (161). This was done by equatingthe driving force for recrystallization with pinning force generated by the particles. In Cuddyswork, the volume fractions required for the pinning force estimate were calculated using

published solubility equations. He found that the assumed array of particles could suppressrecrystallization. In later work, Palmiere et al., conducted a similar experiment but where thecharacteristics of the precipitate array were taken from experimental observations (80). Thiswork differed from Cuddys in one important way. In the study by Palmiere et al., real andlocal volume fractions and particle sizes were measured on the austenite grain boundaries,subgrain boundaries and grain interiors. The pinning forces that resulted from theseobservations showed that the forces calculated at the grain boundaries were well in excess ofthe driving force and could easily be responsible for the observed suppression ofrecrystallization. These results are summarized in Figure 38 (80).

While it is now possible to explain why the Nb in the Cuddy diagram behaves the way it does,Figure 34 (161) i.e., that the recrystallization stop temperature increases strongly withincreasing bulk Nb content, what remains to be explained is why the microalloying elementsdiffer in effectiveness from one another. One possible explanation is given in Figure 39 (164),where the temperature dependence of the driving force for precipitation, i.e., thesupersaturation, is plotted versus temperature for various microalloying precipitating systems.

Of all the possible precipitating systems, only NbC can have high supersaturations over a largeportion of the typical hot rolling temperature range (80).

Niobium and Transformation of Austenite

What is presented here are the manifestations of Nb beyond the ferrite grain refining effectsattributed to the austenite conditioning described above. The high Sv that results fromcontrolled rolling by itself would lower the hardenability of the austenite. In order to obtainlower transformation temperatures and low temperature transformation products, this effectmust be overcome by a combination of higher alloying and/or accelerated cooling. When

relatively large amounts of Nb are in solution in austenite at the transformation temperature,i.e., high Nb contents, low C contents, and high reheating temperatures, it often has animportant effect on the CCT diagram and subsequent transformation during continuous cooling.This is manifested as both lower transformation start temperatures and a higher probability ofachieving non-polygonal ferrite microstructures, especially at higher cooling rates. The firsteffect was reviewed earlier (7, 166, 20, 67, 167), and a good example is shown in Figure 40(168).

More recent work has shown that more than just the temperature is altered by solute Nb. Theincrease in hardenability also means that for the same general conditions, the Nb steels willexhibit larger amounts of low temperature transformation products such as acicular ferrite,


42/74

Widmanstatten ferrite and bainitic ferrite, particularly at higher cooling rates (168-169). Anexample of this effect is shown in Figure 41(168).

Figure 33: The comparison of Nb, V, and Ti effect on the softening behavior in0.002C steels. From Yamamoto et al. (165).

Figure 34: The increase in recrystallization temperature with increase in the levelof microalloy solutes in a 0.07C, 1.40Mn, 0.25Si steel (161).


43/74

Figure 35: The effect of C content on the softening behavior in 0.10 Nb steels at900C. From Yamamoto et al. (165).

Figure 36: Static restoration behavior of three steels at 900C and 1000C.Arrows denote first observable precipitation. From Kwon et al., 1990, (145).


44/74

Figure 37: RPTT diagram showing the interaction between precipitation andrecrystallization. R. and R, refer to the start and finish of recrystallization,respectively, in HSLA steels; and R, and R refer to the start and finish ofrecrystallization, respectively, in plain carbon steels. P, and P", refer to thehypothetical precipitation-start times in deformed and undeformed austenite,respectively. P, is the actual precipitation-start time (145).

Figure 38: Comparison between FPIN and FRXN versus deformationtemperature. Data to the right of FRXN will result in the complete suppression of

austenite recrystallization. Data to the left of FRXN will result in a partial or fully

recrystallized austenite microstructure (80).


45/74

Figure 39: Precipitation potential of various microalloying systems (164).

Figure 40: Corrected Ar3 temperatures of microalloyed steels with standardaustenite grain size of 100 pm (168).

What is particularly striking about the effects shown in Figures 40 and 41 is the influence onproperties, Figure 42 (168). While it is true that the Nb addition had a strong effect on ferrite


46/74

grain size, Figure 41, the principal effect was on the nature of the ferrite. At 0.04 Nb, about80% of the ferrite was bainitic, Figure 41. It seems obvious from Figure 42 that the strengthincreased directly with the amount of bainite. Several similar examples have been also shownelsewhere (169-174). The question then arises as to how the solute Nb is responsible for theimprovement in FATT. Is it the grain refinement of 20% of the structure that is polygonalferrite? Or, is it the presence of low carbon bainite? In simple terms, this effect on toughnessis likely due to the Nb acting as both a hardenability agent and as a grain refiner.

The transformation products resulting from the decomposition of austenite containing Nb as asolute have a surprisingly high dislocation density. This was to be expected for acicular ferriteand bainite, which, by their nature, exhibit very high dislocation densities (171, 173). Whatwas not expected was the rather high dislocation densities found in what is usually called

polygonal ferrite in Nb steels (170, 171, 175-177). This often overlooked characteristic of Nbsteels has very important ramifications regarding mechanical properties as will be shown

below.

Niobium and Strengthening

There is no doubt that the addition of Nb to most low alloy steels results in higher yieldstrengths. This is true for both plate and strip products. What remains elusive is the preciseway that Nb causes this effect. The analysis of strength usually starts with the expanded Hall-Petch equation, where a linear additivity of strength components is assumed:

2/1ypptndislTextureSSNPobs DkYS]YSYSYSYS[YS

+++++= (30)

where:

YSobs = observed yield strength;

YSP-N = lattice friction stress;?YS SS = stress increment caused by solid solution;?YS texture = stress increment caused by texture;?YS disl = stress increment caused by dislocation;?YSpptn = stress increment caused by precipitation;kyD 2/1 = contribution by the ferrite grain size.

The universal use of this equation has been questioned lately from several perspectives. First,the general accuracy of this approach has been questioned and new kinds of summation have

been suggested (178-182). It appears that the linear additivity approach sometimes

overestimates the observed yield strength, and other means of summing such as a root-mean-square summation might be more appropriate (178-182). However, a recent study of several350 grade strip steels containing Nb showed that the linear approach was reasonably accurate(175, 176,183). Second, if the equation is inaccurate or invalid, then its use may lead toerroneous interpretations as to how Nb actually strengthens ferrite. For example, the

precipitation hardening increment has been determined in the literature by subtracting all othercomponents from the measured yield strength. Clearly, this would lead to incorrect values of

fundamental metallurgy of niobium in steel

Documents