Modern HSLA steels and role of non-recrystallisation temperature

Limiting the slab reheating temperature

To make the forming process easier, relatively high soaking temperatures are traditionally employed, which cause considerable grain growth. Hence, the first step in controlled rolling is to control the grain size during reheating. It is well known that the austenite grain size is related exponentially to the soaking temperature. Therefore, the lowest soaking temperature that leads to solution of the microalloying elements should be employed. For most steel grades, this is determined by the niobium and carbon contents. Titanium forms a very stable compound TiN, which remains undissolved. This compound consequently limits austenite grain growth at relatively high soaking temperatures and also keeps the nitrogen from forming the Nb(C,N) phase, enabling the dissolution of NbC to occur more readily.

Final deformation in the low temperature austenite region

This is the temperature range within which complete static recrystallisation no longer takes place between rolling passes. Deformation in the low temperature austenite region (characterised by T<T _nr = non-recrystallisation temperature, as discussed in detail in the section on ‘The concept of the “non-recrystallisation temperature”’) leads to the retention of work hardening, which is accompanied by the formation of pancaked grains and deformation bands. Consequently, increased numbers of nucleation sites are made available for the γ to α transformation. This promotes the formation of a fine grained microstructure, fulfilling the requirements for both strength and toughness.

Accelerated cooling

By employing faster cooling rates, further grain refinement can be achieved. This lowers the transformation start temperature and thus provides more nuclei in the undercooled austenite. While the air cooling of structural steels results in a ferrite–pearlite microstructure, accelerated cooling avoids the transformation to pearlite and provides a microstructure of ferrite and bainite (or bainite and martensite). Up to a yield strength level of about 500 MPa, ferrite–pearlite microstructures are produced, while higher strength steels require bainitic constituents. Steels with about 700 MPa yield strength are typically fully bainitic (or bainitic with the presence of martensite islands).

Thermomechanically rolled steels are gaining constantly in importance because of their outstanding cost to performance ratio. Today, a variety of high strength low alloy steels are processed via hot strip mills since, in comparison to plate mills, they offer a lower cost route, especially for thinner gauges. Because the interpass times pertaining to strip mills are appreciably shorter than those associated with (reversing) plate mills, different metallurgical considerations come into play. With respect to rolling schedules, the requirements for thermomechanical processing therefore have to be optimised with regard to the conditions applicable to strip mills. Better knowledge of processing–microstructure–property relationships is the key to such optimisation.

Mechanical properties

HSLA steels are considered as ‘high strength’ steels, but only in a relative sense compared to mild steel, with yield strengths in the range 355–800 MPa. This limitation is the result of an approach towards system optimisation where the strength of the material has to be considered with respect to other properties such as fracture toughness, formability, weldability or corrosion resistance.

Gladman and Pickering developed formulas for the mechanical properties of 0·1–0·2 wt-% C–Mn microalloyed steels with predominantly ferritic microstructures, i.e. less than 20% pearlite. The total yield strength of HSLA steels is dependent on multiple strengthening mechanisms, as indicated in equation (1) (1) In this equation, σ _P–N is the intrinsic strength of the iron lattice, as specified by the Peierls–Nabarro force; this is about 40–64 MPa at 273 K.5, 6 Here σ _prec represents the contribution of precipitation strengthening, σ _SS represents that of solid solution strengthening due to both interstitial and substitutional atoms and σ _gs represents the grain size effect.

According to Pickering, the common substitutional alloying elements (Cu, Mn, Ni, P and Si) have the following strengthening effects, where the alloy concentrations are in weight per cent7 (2) Depending on the additions, typical strengthening effects of 100–150 MPa are obtained from substitutional solid solution strengthening. The interstitial elements C and N both have strengthening effects of roughly 5000 MPa wt-%⁻¹. The maximum C solubility in ferrite is 0·02 wt-%, which gives C a maximum contribution of 100 MPa to the strengthening of ferrite.

As the concentration of a solute is increased beyond its solubility limit in Fe, precipitates may form inside the grains, which may also increase the strength. The precipitates also produce strengthening by impeding the motion of dislocations through the lattice. When a dislocation moving along its glide plane encounters a precipitate, it may either loop around such a particle, if it is impenetrable (the Orowan mechanism8, 9), or cut through it,10 shearing the precipitate in the process. The increases in strength due to particle looping and particle cutting, are given, respectively, by (3) (4) where f is the volume fraction and r is the radius of the precipitates. The increase in yield strength as a function of the size and volume fraction of the precipitates is illustrated in Fig. 4.

OPEN IN VIEWER

Figure 4.

Precipitation strengthening of ferrite based on Ashby–Orowan model

Finally, the ferrite grain size also has an important effect on the strength. This contribution is described by the Hall–Petch equation11, 12 (5) where σ ₀ is the friction stress, d is the grain size in μm and K is a constant that depends on steel composition. From the equation, it is clear that avoiding grain growth during steel processing is of major importance, as it can lead to considerable loss of strength. Zener showed that, in the presence of small particles, grain growth is inhibited when the grain size reaches a maximum value d _{max, Zener}, given by (6) where r _prec is the radius and f _prec is the volume fraction of the pinning precipitates. From Fig. 5, it is clear that the grain diameter that can be reached in HSLA steels containing a fine dispersion of Nb(C,N) precipitates is below 1 μm.

OPEN IN VIEWER

Figure 5.

Effect of precipitate volume fraction and size on maximum grain size

The strength benefits of HSLA steels can only be realised if the product remains safe as well. This means that the steel must be tough at operating temperatures. In general, HSLA steels have low DBTTs due to their fine grain sizes. Gladman et al. 13 developed the following formula for the DBTT of HSLA steels (7) where Pe is the volume fraction of pearlite, N _i is the concentration of interstitials in wt-% and d is the grain size in μm.

From equations (5) and (7), a number of important conclusions can be drawn:

The effect of Nb(C,N) precipitation on the toughness has been analyzed by several researchers in the past and various values have been proposed for this effect. The increment in impact transition temperature pertaining to the increase in yield strength associated with precipitation strengthening was determined to be 0·35°C MPa⁻¹ by Gladman et al.,14 0·55°C MPa⁻¹ by Gray15 and 0·32°C MPa⁻¹ by Vervynckt et al. 16 By contrast, Tamehiro et al. 17 have reported that there is no effect of Nb(C,N) precipitation on the impact transition temperature of a 0·03C-0·046Nb steel. Furthermore, Lin et al. 18 have also shown that precipitation strengthening does not affect the toughness of a 0·03C–0·09Nb steel.

Transition temperatures (the temperature at which the impact energy is 40 J) and yield stresses are compared for various HSLA steels with different microalloying additions in Fig. 6. The highest yield strengths and lowest transition temperatures are obtained when Nb and Ti are employed together.

OPEN IN VIEWER

Figure 6.

Overview of literature data on mechanical properties of thermomechanically hot rolled HSLA steels19

Weldability

Since welding is now the most widely used fabrication method and with the tendency towards the use of higher heat input welding processes, weldability has become an important issue in the development of HSLA steels. It is essential that chemical compositions can be employed that promote fusion of the base metal and filler (i.e. weld electrode) metal, without the formation of cracks and other imperfections. HSLA steels have excellent weldabilities due to their low carbon equivalents. The International Institute of Welding defines the carbon equivalent P _CM, an empirical measure of the weldability for steels with carbon contents in excess of 0·18 wt-%, as follows (8) Ito and Bessyo20 define the carbon equivalent (CE) of steels with carbon contents below 0·18 wt-% as follows (9) Steel grades with CE′s<0·4% are considered to have good weldability and do not require special preparation or post-treatment of the weld. Examination of the two formulae indicates that the carbon content is the dominant factor with respect to the formation of harmful phases in the microstructure and that the negative role of the other alloying elements is minimised in modern low carbon steels.

Corrosion resistance

Small additions of Cu, P, Si or Cr to low alloy structural steels can improve the corrosion resistance by the formation of a protective rust layer. This allows structures to be built without requiring painting or the use of other corrosion prevention techniques. Such steels develop a natural adherent patina in the surface layer that resists further corrosive attack. The weathering steels, structural HSLA steels with improved atmospheric corrosion properties, are commonly known as Cor-Ten steels, their original US brand name.

Cost effectiveness

The necessary dimensions of a structure are determined by calculating the stresses imposed by the service load and employing a yield strength based on the guaranteed minimum yield strength of the steel. By using a steel with a higher yield strength, the thickness and thus the weight and consequently the cost of a construction can be considerably reduced. Furthermore, by the introduction of TMCP as the main process for the production of HSLA steels, expensive heat treatments can be avoided.

Definition of the T _nr

From a theoretical point of view, the non-recrystallisation temperature T _nr is defined as the temperature below which static recrystallisation can no longer go to completion in the interpass interval or during cooling. Deformation below the T _nr leads to the accumulation of strain hardening and results in the appearance of elongated grains (pancake structure) and the formation of deformation bands. In clean steels, the most relevant nucleation sites for the austenite to ferrite transformation are the austenite grain boundaries. In the case of the transformation of deformed austenite, dislocation bands within the grains will also act as nucleation sites. After nucleation, the ferrite grains grow until impingement of the grains occurs. The ferrite grains will be finer when the transformation starts from finer austenite grain sizes and especially from an elongated austenite grain microstructure, where the ratio of grain boundary surface area to grain volume is increased. This higher nucleus density in combination with the high nucleation rate caused by accelerated cooling, will finally lead to a significant finer ferrite grain size.35, 36 The concept of the T _nr is illustrated schematically in Figs. 8 and 9.

OPEN IN VIEWER

Figure 8.

Illustration of deformation above T _nr with complete static recrystallisation of austenite between rolling passes i and i+1

OPEN IN VIEWER

Figure 9.

Illustration of deformation below T _nr, with only partial recrystallisation of austenite between rolling passes i and i+1: heavily deformed (‘pancaked’) austenite grains result in more nucleation sites for austenite to ferrite transformation

Determination of T _nr

The non-recrystallisation temperature T _nr represents the start of the inhibition of complete static recrystallisation during cooling between rolling passes. Unlike a melting point or phase transition temperature, the T _nr is not an exact quantity. Generally speaking, interpass recrystallisation begins to be incomplete some 50°C or so above the temperature at which it is arrested completely under the interpass time conditions of interest. This gradual transition in recrystallisation behaviour will be examined more closely below. The most common method for determining the T _nr consists of simulating successive rolling passes and representing the mean flow stress (MFS) versus the inverse of absolute temperature graphically for each of the simulated passes. The inhibition of recrystallisation indicated by the T _nr appears as a change in slope of the MFS curve, as demonstrated in more detail below.

The non-recrystallisation temperature is usually determined by multideformation testing. After a reheating step, a sample is subjected to a series of consecutive deformations separated by specific time intervals, while the specimen is cooled at a selected cooling rate. The stress σ and strain ϵ are registered during each of the deformation passes. An example of the stress–strain curves determined in such a multideformation test is presented in Fig. 10a .33

OPEN IN VIEWER

Figure 10.

a stress–strain curves obtained from multideformation test on low alloy steel (0·02 wt-%C, 1·50 wt-%Mn, 0·26 wt-%Si, 0·043 wt-%Al, 20 ppm Ti and 18 ppm N) and b mean flow stress versus inverse of absolute temperature33

The simulation of a 23 plate rolling pass schedule for a low alloy steel is illustrated in Fig. 10, where fixed interpass times of 20 s have been employed in conjunction with a cooling rate of 1°C s⁻¹. The flow stress increases as the temperature is decreased (Fig. 10a ), and the curves can be seen to change their shapes on entering into region II. Later (and thus at lower temperatures), the flow stress drops, region III, before increasing again, region IV. These zones can be better interpreted with the aid of Fig. 10b , in which the MFS is plotted against the inverse absolute temperature. The MFS is calculated for each deformation step by dividing the area under the stress–strain curve by the applied strain. It should be noted that the simulations employed are always based on constant pass strains, strain rates, cooling rates and interpass times. By contrast, all these quantities vary continuously in industrial mills. However, it is the trends in the MFS versus 1/T relationship that are employed to define the transitions between the zones. These could suggest false transitions if the pass strain, strain rate or temperature interval was changed from pass to pass instead of being held constant.

Four different zones can be distinguished in Fig. 10b :

The intersection of the straight regression lines fitted to regions I and II defines the T _nr and the intersection of the regression lines fitted to regions II and III determines the value of Ar ₃, i.e. the ‘upper critical’ or austenite to ferrite transformation start temperature. This temperature is only applicable to deformed, i.e. unrecrystallised austenite and depends on the chemistry, cooling rate, interpass time and applied strain (reduction).37 Similar remarks apply to the Ar ₁, which is defined by the break between regions III and IV. It should be noted that the ‘true’ MFS versus 1/T relationship in Fig. 10b is actually curved in the vicinity of the vertical T _nr line; that is, the actual behaviour is not correctly described by the two straight lines that intersect at this point. Instead, the transition in the recrystallisation behaviour is more gradual than shown.

Because the T _nr depends on interpass time, it is important to note that a simulation involving 10 or 20 s interpass intervals does not apply to schedules in which the interpass time is in the range of 1 s. The point of importance is that plate or reversing mills involve interpass times that permit carbonitride precipitation to take place, as discussed in more detail below. Such precipitation arrests both dynamic and static recrystallisation. By contrast, the much shorter interpass times applicable to strip mills mean that carbonitride precipitation is unable to play a significant role and the inhibition of recrystallisation largely takes place by means of solute drag instead.38 This phenomenon is also examined in some detail below. For this reason, simulations employed to determine the T _nr should be carried out using the length of interpass interval appropriate to the process of interest.

Strip mill simulations are more difficult to perform than plate mill simulations, because much less cooling (temperature decrease) can be accomplished during the rather short interpass intervals. Accordingly, several sets of tests must be carried out to cover a given rolling temperature range and average interpass condition. As a result, this type of test is still under development and presents a challenge to the experimentalist.

Recrystallisation kinetics

Ideally, process improvements and rolling schedules for new products should be tested under actual industrial conditions. Such experiments, however, are often impossible to carry out or at least expensive to perform. Consequently, preliminary tests on a laboratory scale can provide useful information regarding the hot rolling behaviour of selected materials before industrial trials are carried out. Of importance is knowledge regarding the high temperature recrystallisation behaviour of the material of interest. Two types of experimental approaches have been proposed in the literature for the quantification of the recrystallisation kinetics: (1) direct observation methods such as optical microscopy39 and electron backscatter diffraction;40 and (2) indirect mechanical methods, such as multideformation tests,37, 41 double deformation tests42 ^– 49 and stress relaxation tests,50 ^– 53 which are based on measuring the progress of softening.

The direct measurement of the recrystallised fraction in microalloyed steels is difficult because the material transforms during cooling and special etching techniques must be used to reveal the former austenite grain boundaries. One of these is the Béchet–Beaujard54 method, the use of which was illustrated in Fig. 7. This procedure is tedious and time consuming; furthermore, in many cases, it cannot be used because of low hardenability as well as the practical difficulties associated with observing quenched austenite. Moreover, it is often difficult to distinguish between the recrystallised and deformed prior austenite grains, so that the analysis often involves a measure of subjectivity.

For this reason, mechanical testing techniques are generally preferred, which can be divided into two main groups. The first consists of multideformation tests carried out under continuous cooling conditions. These have been described above and are mainly focused on determination of the non-recrystallisation temperature.55 Such multideformation tests are mainly performed in torsion due to the possibility of applying large total strains. For example, Maccagno et al. 55 showed that there is excellent agreement between torsion testing simulations of this type and rolling mill data. This technique, however, does not provide any basic information regarding the static recrystallisation behaviour of microalloyed steels in the interval between two rolling passes. Although laboratory rolling and deformation dilatometry can also be used to provide multiple deformations, the use of these methods has not been reported in the literature.

The second group of experiments employs the isothermal deformation technique. The best known ones in this group are the double hit test56 (DHT) and the stress relaxation test57 (SRT). These techniques permit determination of the progress of softening as a function of temperature in the time interval after or in between deformation passes. Isothermal deformation tests can be performed with a variety of deformation modes on different types of equipment, such as torsion on a hot torsion machine,58, 59 uniaxial compression on a deformation dilatometer45, 46 or a high speed press,43, 60 and plane strain compression on a cam plastometer,61 a mechanical testing system/Instron machine62 or Gleeble.63, 64

Although there is a considerable amount of literature data available regarding the recrystallisation kinetics of microalloyed steels, it is difficult to compare these data since the deformation schedule, equipment used and analysis method all influence the resulting derived kinetics. So far, only a few comparisons have been made between the different techniques and methods. Jonas et al. 55 compared the T _nr values obtained from hot torsion simulations with those obtained from industrial mills and Gomez et al. 39 and Liu and Akben65 compared the results of isothermal double deformation tests to continuous cooling multipass hot rolling simulations. Iparraguirre et al. 66 studied the differences in recrystallisation behaviour deduced from different double deformation mechanical testing modes, i.e. uniaxial compression on a dilatometer and torsion tests. They observed some important differences between the recrystallised fractions determined by these methods. Fernandez et al.,58 Andrade et al. 60 and Yanagida and Yanagimoto67 described the differences in recrystallisation behaviour deduced using different methods to analyse a double deformation test. Other studies63, 64, 68, 69 have compared the recrystallised fractions obtained after a double deformation test and after a stress relaxation test. Significant discrepancies can be found, since some authors63, 64 find good agreement between the fractions obtained from a DHT and a SRT, while others69 have reported that recrystallisation is accelerated in a SRT. However, since these authors used different equipment and analysis methods to obtain their results, it is difficult to determine the reasons for the observed discrepancies. An additional problem stems from the non-uniform ways employed to evaluate the softening ratio, i.e. to separate the softening fraction due to recovery from the softening fraction due to recrystallisation.

The above survey indicates that, although comparisons between the different methods have been reported in the literature, some discrepancies remain to be accounted for. Recently, some work was reported in which a systematic comparison was presented for a single alloy in order to determine the relationships between the recrystallisation kinetics in the static recrystallisation region,70 different test procedures (double deformation and stress relaxation), different types of equipment (hot torsion, deformation dilatometry and mechanical testing system compression testing) and different analysis techniques. This comparison was then extended to different test procedures (multideformation and double deformation) combining different types of equipment (laboratory rolling, hot torsion and deformation dilatometry) and different analysis techniques.71

The former publication70 led to the conclusion that it is preferable to determine the fractional softening in double deformation tests using either the 5% true strain or the 2% offset method to define the yield stress. This is because they tend to limit the effect of recovery on the fractional softening, as was also demonstrated by Li et al. 43 and which can be assumed to be a reasonable assumption for austenite as the amount of recovery is relatively low. The latter definition was proposed as the more accurate and reliable one. Details and practical aspects of the different methods have also been discussed by Fernandez et al. 58

Although the method used for analysing data influences the derived kinetics of recrystallisation, double deformation tests have been identified as the most useful when comparisons are to be made between the results of different authors. The type of testing equipment does not seem to have an important influence on the calculated softening fractions. On the other hand, stress relaxation tests appear to display several drawbacks when evaluating the recrystallisation behaviour of microalloyed steels. First of all, the precise set-up of the equipment and control of the temperature are both difficult, so that apparent differences in recrystallisation kinetics are determined when the same test is performed with different types of equipment. Furthermore, there is no standard method for analysing relaxation curves applicable to all equipment types and testing temperatures. Finally, comparison of the recrystallisation rates determined in this way with those from double deformation tests is not straightforward.

The conclusion was drawn in the latter publication71 that laboratory rolling, torsion testing and deformation dilatometry can all be used to evaluate the influence of chemical composition and of various process parameters on the T _nr. The results showed that, for all equipment types, the T _nr was increased by approximately 65°C when 0·16 wt-%Nb was added to a C–Mn reference steel. However, the application of T _nr values derived in laboratory tests to industrial conditions needs to be done with care. Moreover, it also appeared to be possible to link the recrystallisation behaviour deduced from isothermal double deformation tests to the T _nr values determined from multideformation tests under continuous cooling conditions.

Physical metallurgical principles behind T _nr: interaction between deformation, recovery, recrystallisation, solute drag and precipitation

The T _nr temperature is a complex concept since it is influenced by the interaction between four different mechanisms: recovery, recrystallisation, solute drag and precipitation. These mechanisms influence each other and all depend on the parameters of the preceding deformation, characterised by the deformation temperature T _def, the strain ϵ and the strain rate , not to mention the composition. The effects of strain ϵ, strain rate , soaking temperature T _reh and deformation temperature T _def on the recrystallisation behaviour are illustrated in Fig. 11. These results were obtained on a C–Mn alloy (0·02wt-%C, 1·50 wt-%Mn, 0·26 wt-%Si, 0·043 wt-%Al, 20ppm Ti and 18 ppm N). It is clear from this figure that the recrystallisation rate increases with increasing prestrain (Fig. 11a ), strain rate (Fig. 11b ) and deformation temperature (by comparing Fig. 11c and d ) and with decreasing soaking temperature (visible in both Fig. 11c and d ). The effect of soak temperature is related to the way the grain size increases with increasing reheat temperature, as illustrated in Fig. 12. These increased from 100 μm at a reheat temperature of 1000°C to 200 μm at 1100°C and 343 μm at 1250°C. It has to be added, however, that changing the reheat temperature can also lead to the increased solution of precipitates before double deformation testing. This makes it difficult to investigate the isolated effect of reduced grain size in Nb microalloyed steels.

OPEN IN VIEWER

Figure 11.

Influence of a strain ϵ, b strain rate ·ϵ, deformation temperature T _def (by comparing c and d), and soaking temperature T _reh (both visible in c and d) on recrystallisation behaviour of C–Mn steel:33 standard test parameters were T _reh = 1250°C for 300 s, ϵ = 0·2, ·ϵ = 0·5 s⁻¹

OPEN IN VIEWER

Figure 12.

Influence of reheat temperature on austenite grain size in C–Mn steel of Fig. 11 (etched with Béchet–Beaujard reagent)33

A schematic overview of the different interacting parameters is presented in Fig. 13 and the influence of these basic parameters is discussed in detail in this section. The roles of the main alloying elements in determining the T _nr will be outlined mainly in the section on ‘Recrystallisation behaviour of HSLA steels’.

OPEN IN VIEWER

Figure 13.

Schematic illustration of interactions between deformation, recovery, recrystallisation and precipitation

Interaction between recrystallisation and precipitation

In microalloyed steels, recrystallisation and precipitation can interact in at least three distinct ways: (1) a decrease in dislocation density through recrystallisation reduces the number of precipitate nucleation sites available and can thus retard the onset and rate of precipitation; (2) a precipitate dispersion can provide a pinning (Zener) force72 that can retard or halt the progress of recrystallisation; (3) the mobility of grain boundaries is strongly affected by the solute content of the matrix (solute drag). The progress of precipitation, which reduces the matrix solute content, also affects the mobility of grain boundaries. Fine precipitates can pin boundaries and coarse precipitates can enhance the progress of recrystallisation by removing Nb from solution.

When precipitation starts before recrystallisation is finished, this results in a shift of the curve that indicates the end of recrystallisation R _F to longer times, as shown in Fig. 14a . Thus, the recrystallisation is retarded. If precipitation starts before the onset of recrystallisation (at high deformations due to strain-induced precipitation or at low temperatures), then both the recrystallisation start R _S and finish R _F curves are shifted to longer times. Frequently, recrystallisation is halted completely. Alternatively, as was already mentioned above, solute drag also delays the R _S and R _F curves as illustrated in Fig. 14b . The solute drag effect of Nb can even shift the R _S line past the R _F line for plain carbon steels.

OPEN IN VIEWER

Figure 14.

Interaction between a recrystallisation and precipitation (P = precipitation, R _s = recrystallisation start, R _f = recrystallisation finish) and b between recrystallisation and solute drag (C = plain carbon steel, SD = solute drag)

Influence of C

Although carbon has received wide attention, there is still disagreement with respect to the effect of carbon on the hot deformation behaviour of austenite as well as on the recrystallisation kinetics. To date, research has mainly focused on the effect of carbon on the hot deformation of plain carbon steels and only limited results have been reported for microalloyed steels. In the latter, carbon interacts with the microalloying elements (e.g. Nb), affecting the recrystallisation behaviour. Speer and Hansen86 showed that the carbon content clearly has a substantial influence on the recrystallisation kinetics of austenite containing 0·05 wt-%Nb. As the carbon level was increased from 0·008 to 0·10 wt-%, recrystallisation was retarded by almost two orders of magnitude. Recently, Beladi and Hodgson87 reported that the T _nr temperature of a 0·03wt-%Nb microalloyed steel was increased by 40–60°C when the carbon content was raised from 0·04 to 0·11 and 0·16 wt-%, respectively. Furthermore, they found that at temperatures above the T _nr, recrystallisation was only slightly influenced by the C concentration. By contrast, recrystallisation was significantly retarded by carbon at deformation temperatures below the T _nr due to the formation of Nb(C,N). If the precipitates were formed during deformation, the recrystallisation rate was markedly reduced, even at very high strains.

Influence of ‘standard’ microalloying elements: Nb, Ti and V

The addition of microalloying elements such as Nb, Ti and V can increase the T _nr, with Nb being the most effective element. Figure 15 indicates that a steel with just 0·02wt-%Nb will not recrystallise completely below 950°C.

OPEN IN VIEWER

Figure 15.

Retardation of recrystallisation due to presence of Nb, Ti and V for 0·07C–1·4Mn–0·25Si steel88

The increase in the T _nr by microalloying additions can be explained in terms of two complementary mechanisms:

OPEN IN VIEWER

Figure 16.

Solubility products of main precipitate species in HSLA steels87, 90 ^– 94

Figure 16 clearly shows that the solubility product of TiN in austenite is very low. As a consequence, in HSLA steels, all the Ti will be bound to N in the austenite hot rolling temperature range. This means that, although the solute drag potential of Ti is very high (Table 2), the actual effect of Ti as a solute drag element in retarding recrystallisation in HSLA steels will be negligible due to the very low (or even nonexistent) solute Ti content. On the other hand, the solubility of VC in austenite is rather high. As a consequence, the role of VC precipitates in retarding the austenite recrystallisation and in increasing the T_nr will be of minor importance as can already be concluded from Fig. 15.

Maruyama et al. 98 have proposed a further reason why Nb is the most effective element for delaying austenite recrystallisation. Nb atoms will trap or combine with the interstitials and/or defects more quickly than the other elements. Along with the strong interstitial–Nb interaction and dislocation–Nb interaction, the greater diffusivity of Nb is an important factor contributing to its significant retarding effect on recovery and recrystallisation. It can be seen in Fig. 17 that the diffusion coefficients of the interstitials (C,N) are more than 1000 times higher than the diffusion coefficients of the substitutional elements Ti, Nb and V. Therefore, the diffusion distance (Dt)^1/2 (in which D is the diffusion coefficient and t is the time) is completely determined by the diffusivity of Nb, Ti and V and not by the diffusivity of C and N.

OPEN IN VIEWER

Figure 17.

Diffusion coefficients of main precipitate species in HSLA steels

The exact mechanism of recrystallisation inhibition (solute drag or precipitation) is of considerable importance in the design of microalloyed steels because the microalloying elements contribute substantially to the alloying cost. As a result, optimisation of the use of these elements in combination with the production schedule permits the total production cost to be decreased. Nevertheless, it is not possible at present to distinguish clearly between the influence of these two mechanisms on the recrystallisation behaviour. If the recrystallisation kinetics is controlled by solute drag, it should only be necessary to consider the concentration of the microalloying elements dissolved in the austenite matrix. On the other hand, if carbonitride precipitation controls the recrystallisation kinetics, the levels of carbon and nitrogen in solution should also be important. Because of the important influence of interpass time on this issue (short in strip mills and long in plate mills, as discussed above), generally speaking, solute effects are more important in strip mills and precipitation effects in plate mills.

Speer and Hansen86 investigated austenite recrystallisation in Nb microalloyed steels and showed that the solute drag effects are relatively small compared to those of carbonitride precipitation. Nevertheless, other researchers have argued strongly for the importance of the solute drag effect (e.g. Coladas et al. 99) or for a combination of the two effects.38, 60, 100 Some recent results101, 102 that throw light on this topic are discussed below.

So far, several authors38, 60, 86, 98, 99 have discussed the influence of the Nb content on the non-recrystallisation temperature and on the recrystallisation kinetics. Less attention has been paid to the influence of Ti and V additions on this phenomenon.103 ^– 106 Still less data are available on the simultaneous influence of Ti–V, Nb–V or Ti–Nb although these elements are frequently added together in modern microalloyed steels.107 With regard to the Nb–Ti steels in particular, a full understanding of the influence of Ti is still lacking. Nevertheless, it is known that the T _nr increases with increasing Ti content and that this effect saturates at Ti levels of about 0·025 wt-%.⁸⁰ Furthermore, it has also been noted that the T _nr is always lower in Nb–Ti steels than in steels containing Nb alone (for equal Nb amounts). The explanation for these observations lies in the effect of increasing the Ti/N ratio (by increasing the Ti content) as a result of which there is a larger volume fraction of undissolved (Ti,Nb)(C,N) particles present in the austenite. This in turn reduces the supersaturation level, thereby leading to a lower driving force for precipitation during deformation. Additionally, these undissolved particles can favour the precipitation of carbonitrides on the particles, with the result that precipitation on dislocations, which is responsible for the retardation of recrystallisation, can be seriously impaired. These interactions can significantly reduce the effectiveness of Ti addition in increasing the T _nr.

As mentioned above, Nb is the element that most effectively increases the T _nr and its role in retarding austenite recrystallisation in HSLA steels has been the subject of considerable interest and discussion over the past 30 years.86, 108 Previous research109, 110 indicates, as already mentioned above, that the retardation of austenite recrystallisation in Nb containing steels results from the pinning of austenite grain boundaries and sub-boundaries by niobium carbonitride precipitation as well as from the solute drag effect of niobium atoms in solid solution in the austenite.38, 60, 99 Here, some results are summarised from a recent study101 in which attempts were made to quantify and separate the effects of solute drag and of precipitate pinning by combining various experimental techniques and using the following well designed alloys: (1) a C–Mn reference alloy (0·02wt-%C, 1·50 wt-%Mn, 0·26 wt-%Si, 0·043 wt-%Al, 20 ppm Ti and 18 ppm N); (2) a low C alloy (11 ppm C, 1·47 wt-%Mn, 0·26 wt-%Si, 0·022 wt-%Al, 0·16wt-%Nb, 40 ppm Ti and 12 ppm N) designed to study the effect of Nb in solid solution; and (3) a 0·16Nb alloy (0·02 wt-%C, 1·50 wt-%Mn, 0·26 wt-%Si, 0·067wt-%Al, 0·16 wt-%Nb, 80 ppm Ti and 20 ppm N) designed to produce a high fraction of NbC precipitates. More detailed results and further information regarding the experimental conditions are presented elsewhere.101, 102

Torsion testing was employed to determine the T _nr for the three steels, using the methodology described above and in the literature.37, 111, 112 Interpass times of 20 s, a cooling rate of 1°C s⁻¹, a strain rate of 1 s⁻¹ and a pass strain of 0·3 were used. The reference alloy had a T _nr of 880°C, while the T _nr for the low C alloy was 920°C and the one for the 0·16Nb alloy was 950°C. The isothermal recrystallisation kinetics were determined from double deformation compression tests with increasing interpass times. The results for the reference alloy were already presented above (cf. Fig. 11), while those for the two other alloys are illustrated in Fig. 18.

OPEN IN VIEWER

Figure 18.

Softening fractions for low C and 0·16Nb steels determined from double deformation tests (ϵ = 0·2, dϵ/dt = 0·5 s⁻¹): temperatures and times at which state of precipitation was investigated by TEM and ICP-MS measurements are indicated by circles on figure

For several of the curves shown in Fig. 18, a temporary recrystallisation stop was observed. This feature has already been observed in the literature.96 Note that the various arrests of recrystallisation in Fig. 18 take place in the interval 30–40 s. This is the time that it took for precipitation to become effective under the conditions of these experiments (such times are closer to 10 s under industrial conditions of strain induced precipitation). Thus, precipitation can only be effective in arresting recrystallisation under rolling conditions where the interpass time is of this order, or at least when the total rolling sequence takes more than 10 s. This signifies that the role of strain induced precipitation is particularly important in reversing mills, such as plate mills. There is clearly insufficient time for strain induced precipitation under strip mill rolling conditions.

However, in the absence of Nb addition, recrystallisation takes place fairly rapidly, as shown in Fig. 11. Here it is evident that recrystallisation has gone to completion in 10 s at 950°C after deformation at 10 s⁻¹ (Fig. 11b ). It is also well under way at this time at lower temperatures (again, these rates are more rapid under industrial, higher strain rate, conditions). It is clear from the softening curves of Fig. 18 that, when Nb is added, recrystallisation only begins at about 10 s. Thus, the presence of Nb in solution is what retards recrystallisation sufficiently in microalloyed steels to prevent its occurrence during rolling sequences that takes a total of 10 s or less, i.e. under strip mill conditions. Nevertheless, it should be mentioned that a strain of 0·2 was employed in the present experiments, while in hot strip mills, strains of more than 1·0 are applied. As demonstrated by Dutta and Sellars113 in their model, the precipitation start time is proportional to the inverse of the strain and so shorter precipitation start times are likely to apply to industrial conditions than suggested by Fig. 11.

In order to evaluate the interaction between recrystallisation and precipitation, the driving force for recrystallisation was calculated114, 115 and compared with the Zener pinning force72 of the precipitates on the grain boundaries. The latter was deduced from a detailed analysis by transmission electron microscopy (TEM) and inductive coupled plasma-mass spectroscopy (ICP-MS), which were performed on the samples identified by the encircled points in Figure 18b .

The precipitate diameter distributions at two interpass times are illustrated in Fig. 19, as determined by TEM on C extraction replicas. The precipitates can be divided into two groups, i.e. precipitates that did not dissolve during reheating and the SIPs. From Fig. 19, it is clear that the distribution of the undissolved precipitates is unimodal, with a lognormal average diameter that remains constant at about 60 nm, as shown by the right hand curve in the various graphs. The distribution of the deformation induced precipitates is also unimodal, with a log normal average diameter that increases with increasing interpass time, as shown by the left hand curve in the different graphs.

OPEN IN VIEWER

Figure 19.

Precipitation state studied by TEM at interpass times t _ip of a 10 s and b 1000 s after deformation at 1000°C for 0·16Nb alloy (0·02 wt-%C, 1·50 wt-%Mn, 0·26 wt-%Si, 0·067 wt-%Al, 0·16 wt-%Nb, 80 ppm Ti and 20 ppm N)102

After determination of the weight fraction of Nb precipitates by ICP-MS, the evolution of the Zener pinning force with time was calculated. This was compared with the recrystallisation driving force. Good agreement was observed between the present experimental data102 and a number of theoretical models developed to describe the interaction of precipitation and recrystallisation116, 117 in a quantitative way. It was concluded that, at first (up to 30 s), the driving force for recrystallisation is higher than the Zener pinning force. The average SIP diameter is small, because these precipitates have just nucleated, but since the volume fraction of the precipitates is also small, the resulting Zener pinning force is insufficient to halt recrystallisation. Later (at about 30 s), the Zener pinning force exceeds the recrystallisation driving force because of the growth of the precipitates; consequently, the recrystallised fraction remains constant for a specific time interval. Finally (after 300 s), the Zener pinning force once again drops below the value of the recrystallisation driving force because of the further growth and coarsening of the precipitates and recrystallisation recommences.

The next step was to quantify the solute drag effect and to compare it with the impact of the precipitation effect. As described elsewhere101 in detail, quantification of the solute drag effect was based on literature findings;108, 118 ^– 120 the results are reproduced here in Fig. 20. The dashed line in this figure represents the retardation of recrystallisation due to the solute drag of Nb in solution. This ‘Nb solute drag line’, which is nearly parallel to the t _0·5 line of the C–Mn reference steel, permits quantification of the magnitudes of the solute drag and precipitation effects directly from the softening time–temperature diagram. Figure 20 shows that, at 50% recrystallisation, i.e. during the growth stage of the recrystallising grains, the dominant mechanism in retarding austenite recrystallisation changes from solute drag, at relatively high temperatures, to precipitation, at temperatures just below the solution temperature, to solute drag once again, at precipitate coarsening temperatures.

OPEN IN VIEWER

Figure 20.

Softening time–temperature diagrams for three experimental steels: plotted is time for 50% softening (SD: solute drag; P = precipitation)101

Influence of ‘new’ alloying elements: Mo, Ni and Cu

In order to compensate for the loss in strength of HSLA steels due to their decreased carbon contents, alloying elements such as Mo, Ni and Cu are often added. The impact of these elements on the mechanical properties (e.g. strength and toughness) and corrosion resistance is fairly well known and has already been discussed above. For example, Bacroix et al. 121 described the increase in strength and toughness produced in low carbon steels by the addition of Mo. This increase was attributed to formation of the low temperature transformation products associated with the addition of Mo. However, the effect of these elements on the austenite recrystallisation behaviour during hot rolling remains to be clarified. Better knowledge of the influence of these elements could help to optimise the addition levels of these elements as well as the rolling schedules for new thick plate products.

The influence of nickel and copper on the recrystallisation behaviour is still far from understood. To date, little literature data could be retrieved on their influence on the T _nr and on the recrystallisation kinetics. Cho et al. 122 observed that Ni levels of 0·2 and 0·5 wt-% had no significant effect on the activation energy for recrystallisation in the solute drag region, although some retardation of recrystallisation was identified. Their results confirmed the previous work of Yamamoto et al.,123 who reported a minor increase in recrystallisation rate in decarburised steels containing 1 wt-% Ni. Maehara et al. 124 also evaluated the effect of Ni and Cu and the effect of Cu on the recrystallisation behaviour was reported to be similar to that of Nb, while the effect of Ni was far less.

Some work has already been presented in the literature regarding the role played by Mo. For example, Bouet et al. 125 have described the strong solute drag effect of Mo in Nb bearing transformation induced plasticity steels. Mo addition improved the mechanical properties and also contributed to delaying recrystallisation and precipitation. Furthermore, Pereda et al. 126 reported that Mo in solid solution had a strong retardation effect on dynamic recrystallisation.

In 1985, Wada and Pehlke127 reported that Mo increases the solubility of C and N in austenite. This seems to be linked to the effect of Mo addition in reducing the activities of C and N. Furthermore, Mo also increases the solid solubility of Nb in Nb bearing steels for the same reason. The increased solubilities of C, N and Nb decrease the driving force for Nb(C,N) precipitation.128 Junhua et al. 129 proposed another reason for the retardation of Nb(C,N) precipitation in Mo containing steels: they suggested that Mo increases the carbon diffusion activation energy and decreases the carbon diffusivity in this way. Moreover, Mo may also decrease the diffusivity of Nb. The decreased diffusivities of both C and Nb will delay both the nucleation as well as the growth of Nb(C,N) precipitates. Therefore, the resulting precipitates will be finer and thus more effective in pinning the grain boundaries. As a consequence, recrystallisation in Mo containing Nb steels will be retarded in comparison with Mo free Nb steels, i.e. the T _nr will be higher. In conclusion, the presence of Mo in steel appears to influence both solute drag as well as the pinning effect associated with the strain induced precipitates.

A further demonstration of the effect of Mo addition on the recrystallisation kinetics was recently quantified by Pereda et al. 130 In their study, the effect of Mo addition on the T _nr of Nb microalloyed steels was investigated by means of multipass torsion tests (ϵ = 0·4, ·ϵ = 1 s⁻¹ and t _ip = 30 s). Their results obtained on four different steels are presented in Fig. 21. The 3Nb steel contained about 0·03 wt-%Nb in addition to 0·05wt-%C, 1·6 wt-%Mn and 0·007 wt-%Mo. The 6Nb steel had the same composition but the Nb level was increased to 0·06 wt-%. The Mo containing steels were modified by the addition of 0·31 wt-%Mo.

OPEN IN VIEWER

Figure 21.

Dependence of mean flow stress on inverse absolute temperature for steels 3Nb, 3Nb–Mo31, 6Nb and 6Nb–Mo31: results were obtained from multipass torsion tests with ϵ = 0·4, ·ϵ = 1 s−1 and t _ip = 30 s130

As illustrated in Fig. 21, Pereda et al. 130 showed that, when 0·31 wt-%Mo was added to a 0·03 wt-%Nb steel, the T _nr was increased by 40°C, from 970 to 1010°C. However, when the Nb content was increased to levels as high as 0·06 wt-%, the contribution of Mo became less relevant, since the T _nr only increased from 1030 to 1045°C. For the experimental conditions and chemical compositions employed in their work, Pereda et al. 130 concluded that the T _nr in Mo–Nb steels is mainly determined by the solute drag effect. In the case of the 0·03 wt-%Nb steel, Mo addition produced an additional solute drag effect, which accounts for the significant increase in the T _nr in these steels. In the 0·06 wt-%Nb steel, the accelerated kinetics of strain induced precipitation rendered the increase in the T _nr due to the solute drag contribution of Mo less relevant.