Revealing Precipitate Development During Hot Rolling and Cooling of a Ti–Nb Micro‐Alloyed High Strength Low‐Alloy Steel through X‐Ray Scattering

High‐energy synchrotron X‐ray small‐angle scattering (SAXS) is used to study the precipitate development during hot rolling and cooling of a commercial Ti–Nb micro‐alloyed, high‐strength, low‐alloy (HSLA) steel. To study precipitation during hot rolling conditions, Gleeble and dilatometer trials are made. Samples are then studied at room temperature using SAXS in conjunction with transmission electron microscopy (TEM). TEM is used to determine the morphology and composition of the precipitates, whilst both TEM and SAXS are used to study the particle sizes. One major advantage with high‐energy SAXS is the ability to make measurements after a minimum of sample preparation and in transmission geometry, as opposed to just at prepared surfaces, plus the possibility to determine volume fractions of the precipitates. The measurements show that after deformation at high temperature, particle coarsening occurs and the volume fraction of precipitates increases after holding for 20 s at 900 °C which confirms strain‐induced precipitation at finishing rolling conditions. The measurements show that holding at 600 or 650 °C for one hour gives a larger volume fraction of nanosized particles. Coiling simulations with slow cooling from 600 to 470 °C show coarsening of particles and an increase in the volume fraction of the smaller particles compared to holding at a constant temperature.

High-energy synchrotron X-ray small-angle scattering (SAXS) is used to study the precipitate development during hot rolling and cooling of a commercial Ti-Nb micro-alloyed, high-strength, low-alloy (HSLA) steel. To study precipitation during hot rolling conditions, Gleeble and dilatometer trials are made. Samples are then studied at room temperature using SAXS in conjunction with transmission electron microscopy (TEM). TEM is used to determine the morphology and composition of the precipitates, whilst both TEM and SAXS are used to study the particle sizes. One major advantage with high-energy SAXS is the ability to make measurements after a minimum of sample preparation and in transmission geometry, as opposed to just at prepared surfaces, plus the possibility to determine volume fractions of the precipitates. The measurements show that after deformation at high temperature, particle coarsening occurs and the volume fraction of precipitates increases after holding for 20 s at 900°C which confirms strain-induced precipitation at finishing rolling conditions. The measurements show that holding at 600 or 650°C for one hour gives a larger volume fraction of nanosized particles. Coiling simulations with slow cooling from 600 to 470°C show coarsening of particles and an increase in the volume fraction of the smaller particles compared to holding at a constant temperature. enable improved optimization of processing and compositions of HSLA steels.
Strain-induced precipitation and interphase precipitation during hot rolling and cooling in Nb and Ti alloyed steels has been studied for years. [1,3,5,6] TEM studies are common practice to study precipitates, but provide results from a very small sample volume and after surface polishing. Such measurements are, thus, perhaps not always representative of the studied material. In addition, volume fractions and number densities of precipitates are practically impossible to establish with TEM. High-energy synchrotron X-ray scattering techniques provide the possibility to study larger samples to yield more representative data of the material. The volume fraction of the particles of interest in micro-alloyed steels is usually below 0.2% with sizes around 5 nm. In this context, it has previously been shown that with high-energy synchrotron X-ray scattering, the increase of precipitates when annealing above 500°C was detectable in micro-alloyed Nb steels with volume fractions below 0.1%. [7,8] The primary objective of this work was to study precipitate development in a commercial Ti-Nb micro-alloyed steel in thermomechanically treated samples and to assess if high-energy synchrotron SAXS could be used to study mm-thick samples and to detect very small volume fractions of <10 nm sized particles. The high-energy synchrotron X-ray beamline P21.2 at DESY was used to track even minor changes of very small volume fractions of nanosized particles. The possibility for rapid simultaneous acquisition of SAXS and wide-angle scattering (WAXS) at P21.2 at DESY also enables future in situ measurements to study both interphase precipitation and phase transformation. A major advantage with SAXS is the ability to make measurements after a minimum of sample preparation and over large sample volumes (the beamline at DESY provides penetration of 4-5 mm steel samples), as opposed to just at prepared surfaces, plus the possibility to determine volume fractions of the precipitates. Therefore, more representative measurements could be made and in the future in situ studies could be performed during sample heating and deformation to follow the active processes of precipitation and phase transformation.

Material
The studied material was a commercial Ti-Nb-alloyed hot rolled strip steel with the chemical composition given in Table 1. Results from Thermo-Calc (Thermo-Calc Software TCFe Steels/Fe-alloys version 11) calculations of precipitate evolution are shown in Figure 1 in terms of the stable phases and compositions of carbides as a function of temperature. To study strain-induced and interphase precipitation during hot rolling conditions, a Gleeble 3500 and a Bähr dilatometer 805A/D were used to produce various precipitation states. Samples were prepared to study strain-induced precipitation after finishing rolling conditions and possible interphase precipitation during cooling and coiling conditions at high temperature. A high-temperature dissolution treatment was first carried out to maximize the Ti and Nb in a solid solution. According to Thermo-Calc calculations, the carbides should be dissolved at 1245°C and heat treatment was accordingly done at 1250°C for 30 min. One heattreated sample was used as a reference.
To establish a temperature when recrystallization is completely retarded, presumably by precipitation pinning, Gleeble stress relaxation tests were performed. [9,10] Preheating at 1050°C was done before the deformations to get a grain size comparable with finishing rolling conditions. The trials were then carried out with two 20% reduction deformations at a strain rate of 20 s À1 . The first deformation was made at 1000°C followed by 10 s hold to allow a recrystallized structure to form. The second deformation was made at 900°C where recrystallization was completely retarded. All investigated samples were thermo-mechanically prepared in the same manner before various cooling treatments. A schematic presentation of the heat treatment, deformation, and cooling is shown in Figure 2.
To study phase transformation at typical coiling temperatures, dilatometer trials were performed and the samples were held at 650 and 600°C for 1 h subsequent to deformation at 900°C. Before cooling to 650 and 600°C, the samples were held for 2 s at 900°C to allow for some nucleation of precipitates in the austenite at high temperature, similar to hot rolling conditions, before cooling. Fraction transformed from dilatometer trials was evaluated using the manual determination of fraction transformed in the DIL805 software. It was established that %60% ferrite is already transformed when reaching 600°C, as seen in the dilatometer results for cooling down to room temperature (RT) in Figure 3. For both holding temperatures, phase transformation was rather rapid and complete within two minutes after reaching the temperature. To evaluate the effect of the cooling rate, cooling was performed at 50 and 100°C s À1 .
To study the evolution of precipitation in the austenite, two samples were deformed at 900°C followed by holding at 900 for 1 and 20 s, respectively. The samples were then rapidly cooled to room temperature with a maximum cooling rate of 200°C s À1 . During cooling and, hence, austenite to ferrite phase transformation, interphase precipitation could occur, but there should be no time for coarsening of those precipitates. To study interphase precipitation and precipitation in the ferrite, samples were cooled for 2 s after deformation at 900°C at a cooling rate of 50 and 100°C s À1 . To study interphase precipitation, samples were directly cooled to room temperature. To study precipitation in the ferrite, samples were cooled to 650 or 600°C, holding for 1 h, followed by rapid cooling to room temperature. The effect of slow cooling, as is the case for the coiling of a strip, was also investigated during a coil cooling simulation. Slow cooling was done with a cooling rate of 0.5°C min À1 , from 600 to 470°C.

TEM
Carbides formed during processing were studied on carbon extraction replicas. [11] For the preparation of carbon extraction replicas, sample surfaces were ground and polished down to 0.25 μm with diamond paste followed by a rapid etching for 5 s in 2% Nital etchant (HNO 3 in ethanol). A carbon film was deposited on the etched surface using a precision etching and coating system (PECS) (Gatan, Pleasanton, CA, USA) to a thickness of about 20 nm. The film was removed by immersing the samples in 10% Nital etchant and the replicas were placed on TEM copper grids. Electropolishing in 10% Nital was required to release the replica from the sample surface in some cases. The voltage used for electropolishing was 15 V. [12] Scanning transmission electron microscopy (STEM) together with energy-dispersive X-ray spectroscopy (EDX) was performed at Swerim (Kista, Sweden) using a Jeol 2100 F (Jeol, Akishima, Tokyo, Japan) microscope with an X-MAXN 80 TLE detector (Oxford Instruments, Abingdon, Oxfordshire, England). The incident beam energy was 200 keV.
To evaluate the average radius of formed carbides from STEM micrographs, image processing was performed using the ImageJ software. [13] The image processing included a band-pass filter to level out the background and to enhance the contrast noise of the particles for better thresholding. After thresholding of the images to separate the particles from the background, each particle was labeled and the radius was calculated from the area of each particle, assuming a spherical shape.
Scanning precession electron diffraction (SPED) was used to gain crystallographic information on the formed carbides. An ASTAR (NanoMEGAS, Brussels, Belgium) setup was used, which includes a DigiSTAR precession device and a CCD camera to record patterns. A digital STEM (TOPSPIN, NanoMEGAS, Brussels, Belgium) was used to scan the beam over an area of interest. Measurements were performed with a step size of 5 nm. The total frame size was 200 Â 200 nm 2 . SPED is a modification of the TEM mode. A convergent electron beam is focused to give a beam size of %1 nm on the sample. A virtual STEM is used to scan the beam. A diffraction pattern was recorded at each scan position. Dictionary indexing was used, where collected  www.advancedsciencenews.com www.aem-journal.com diffraction patterns are matched against a database of known patterns (simulated patterns of possible phases) generated from crystallographic information files (CIFs). The DiffGEN software was used to generate sets of simulated electron diffraction patterns for possible crystallographic structures within the samples, e.g., cementite (Fe3C), titanium carbide (TiC), and niobium carbide NbC.

SAXS
SAXS measurements were performed at the Swedish materials science beamline P21.2 at the PETRA III storage ring at the Deutsches Elektronen Synchrotron (DESY) in Hamburg. The energy used was 82.5 keV and a Pilatus CdTe 2 M detector was placed 14.7 m behind the sample to acquire the 2D SAXS patterns, giving a q-range of 0.05-3 nm À1 . The size of the beam was 0.2 Â 0.2 mm 2 at the sample position and the exposure time was 60 s. A NIST reference glassy carbon sample was used to scale for absolute intensity. Image corrections for background and transmission effects were made using ImageJ and data reduction, including radial integration of the 2D patterns to give 1D SAXS curves, were performed in Fit2d. [14] The analysis of the processed 1D-scattering curves was made with SANSFit. [15] For the calculation of the contrast of the (Ti,Nb)C precipitates, a ratio of 66% Ti and 34% Nb was assumed.
The contrast of 17.4 · 10 10 cm À2 between iron and the carbide particles was then determined from tabulated values. [16] In addition, a spherical form factor and a log-normal size distribution were assumed for the data analysis with SANSFit.

Optical and Electron Microscopy
Optical (OM) micrograph reveals that the steel has a ferritic microstructure with varying amounts of pearlite after processing, see   www.advancedsciencenews.com www.aem-journal.com shows a representative dark-field image of carbides on carbon extraction replicas. Carbon-enriched features such as grain boundaries and carbides are transferred to the extracted carbon film and large grain boundary precipitates can be seen in the dark-field image, as well as small carbides within grains. To further study precipitates formed just before, during, or after the austenite to ferrite phase transformation STEM and EDX was performed at higher magnification on carbides located within grains, see dark field image in Figure 4c. The carbides can be seen to have primarily spherical shapes. However, some larger carbides are ellipsoidal, which were most likely formed at elevated temperatures in austenite. Representative EDX spectra from carbides found in the steel are shown in Figure 4d. No significant difference in chemical composition could be detected between large and small carbides. The crystal structure was determined by SPED to be TiC, see phase map and corresponding diffraction pattern in Figure 4e. Crystallographic details of the TiC phase are given in Table 2.
From both EDX and SPED it was determined that the carbides are TiC, enriched in Nb with small amounts of V, Cr, and Fe. Particle size distributions were evaluated using image processing of micrographs taken with STEM, such as the example STEM dark-field micrograph in Figure 4c. The size distributions agree with a log-normal distribution, which was fitted to the sample data; see representative fitting results in Figure 4f.
Representative STEM micrographs of carbon replicas from samples that have undergone various heat treatments simulated in Gleeble and dilatometry are given in Figure 5a-d. Micrographs of samples held for 1 s and 20 s at 900°C after deformation, which was subsequently quenched to room temperature, are given in Figure 5a,b respectively.
The average carbide composition depending on heat treatment is presented in the bar plot in Figure 5e. No significant differences in carbide composition depending on holding time at 900°C could be detected. The fitted size distributions with varying heat treatments are plotted in Figure 5f. In Figure 5a, some larger particles are seen, probably precipitated before deformation, at higher temperature. There are also small particles, precipitated just before quenching. In Figure 5b, more large and small particles are seen due to nucleation after deformation (strain-induced precipitation). Although the size of the larger particles in Figure 5a appears larger than in Figure 5b, the average radius of precipitates is larger after holding at 20 s, 9 nm compared to 5.5 nm after 1 s, which shows the rapid coarsening of carbides in austenite.
Carbide precipitation during interrupted cooling from 900°C and an isothermal hold at 600°C for 1 h, was compared to direct Table 2. TiC crystallography. [17,18] Carbide   Figure 5c), the average carbide size is about 5 nm. The size distribution is shifted toward smaller sizes after 1 h hold at 600°C (see micrograph in Figure 5d), and the average carbide size is close to 4 nm. This indicates interphase precipitation during the phase transformation or random precipitation in ferrite during the isothermal hold. As mentioned earlier, the phase transformation is complete within 2 min at 600°C. Due to the fast transformation and the relatively low transformation temperature at 600°C, the interphase precipitation is believed to be restricted and general, random precipitation in ferrite is assumed to be the dominant precipitation mechanism. A similar trend is seen for samples cooled at 100°C s À1 . Figure 6 shows examples of the 2D and reduced 1D SAXS data. The particles are assumed to be formed during or after phase transformation. The sample heat treated at 1250°C for 30 min was assumed to have no particles left, due to dissolution. It cannot be excluded that some very large particles remain and very small particles <1 nm are formed during cooling. For this reason, only changes in comparison to this sample were considered for the evaluation. It is also noted that the different sample shapes and orientations could induce some uncertainty in the comparison between the samples, but the effect of this is expected to be minor with respect to understanding the overall variations in precipitate volumes. Figure 6d shows results from the fitting of the SAXS measurements for samples to be compared with TEM results, see Figure 5f. The SAXS and TEM results show very good agreement, only the two samples held at 900°C show more coarsening in the SAXS data compared to the TEM results. Mean particle radius and volume fraction for all SAXS results are shown in Table 3. Calculated equilibrium volume fractions from Thermo-Calc, as seen in Figure 1a, are also given. All samples in the table were first heated to 1050°C and then 20% deformed at 1000 and 900°C.

SAXS
Comparing samples held at 600 or 650°C shows an increase in volume fraction at higher temperature. Slower cooling rate before holding at 600 or 650°C for one hour provides www.advancedsciencenews.com www.aem-journal.com slightly larger volume fractions. The sample quenched (cooling rate >200°C s À1 ) 1 s subsequent to deformation at 900°C have clearly less volume fraction than the equilibrium value. The sample quenched 20 s after deformation has a volume fraction close to the equilibrium value. For the sample quenched after 1 s, the measured volume fraction was 0.0010 and the size was 15 nm. Holding for 20 s provided a significant increase in volume fraction and size, 0.0014 and 28 nm, respectively. The samples cooled to room temperature at cooling rates 50 and 100°C s À1 show a similar volume fraction of precipitates as the sample held for 20 s at 900°C, although only a time of 2 s is provided for nucleation and growth at high temperature, implying that interphase precipitation occurs. Despite less time for nucleation and growth, the sample cooled with 100°C s À1 has a slightly larger volume fraction of precipitates compared to the sample cooled with 50°C s À1 , which means that more nucleation sites are present in the material. The difference is not substantial and is most likely due to the fact that small variations in strain and temperature occur during dilatometer testing. Hence, cooling rates of 50 or 100°C s À1 do not seem to affect the volume fraction to a large extent. The mean size of the precipitates is somewhat smaller due to a shorter time for growth compared to lower cooling rates. Highest volume fraction of particles is measured in the sample slowly cooled from 600 down to 470°C at a cooling rate of 0.5°C min À1 . The measured volume fraction (0.00236) is actually higher than the equilibrium value from Thermo-Calc (0.00180). The equilibrium value is the calculated volume fraction of (Ti,Nb)(C,N) particles and there could be other particles present. 600 and 650°C are typical coiling temperatures after hot rolling and a comparison of the measured particle size and volume fraction for those samples is shown in Figure 7a. The increase in strength due to precipitates is usually described by the Orowan-Ashby model [19] where D is the particle diameter, in μm, and f v is the volume fraction of the precipitates. The strengthening contribution from the precipitates for the different cases is shown in Figure 7b, where it is seen that the strength contribution is highest for the slowly cooled sample. Larger volume fraction, provides a higher strength contribution. Hardness measurements were done with Vickers microhardness testing, using a load of 50 g. A minimum of seven hardness indentations were made on each specimen and the average values are shown in Figure 7b. The calculated hardness peak in the sample slowly cooled from 600 to 470°C is not seen in the hardness measurements. This implies that the measured volume fraction does in fact

Conclusion
High-energy SAXS has successfully been used, in conjunction with more standard measurement approaches, to study the precipitate development during hot rolling and cooling simulations of a commercial Ti-Nb micro-alloyed, HSLA steel. The investigated samples were prepared by simulating hot rolling conditions using Gleeble and dilatometer. In addition to the SAXS analysis, supporting measurements were made with TEM, EDX, and SPED to determine the morphology and composition of the precipitates, as well as to provide standard measurements of particle sizes to compare with the SAXS results. For comparison with theoretical values, calculations were made with Thermo-Calc. Variation in size and fraction of small volume fraction of nanosized precipitates could clearly be distinguished using the high-energy synchrotron SAXS. The preparatory TEM study provided supporting details on size, morphology, and composition and the size measurements from TEM and SAXS were found to be in good agreement. It was established that at temperatures near the finishing rolling temperature, 900°C, precipitation occurs and is enhanced by strain. After deformation, longer holding time (20 s compared to 1 s) at 900°C, clearly showed an increase in the volume fraction of precipitates and also coarsening of the particles. Different cooling rates and/or holding temperatures did not show a clear difference in mean particle size. A slight increase in volume fraction was seen for samples cooled at a lower cooling rate, before holding for one hour at 600 or 650°C, when comparing samples cooled at 100°C s À1 with samples cooled at 50°C s À1 . Higher holding temperature (650 compared to 600°C) also provided a slight increase in volume fraction. Slow cooling from 600 to 470°C provided the largest volume fraction of particles.
In this study, the evolution of precipitates has been followed by studying quenched samples at different stages in the thermomechanical treatment based on the assumption that the sample state is "frozen" upon rapid cooling and that no precipitation occurs during rapid phase transformation. Measurements made during in situ thermomechanical treatment tests would provide the ability to study the active precipitation processes without quenching. This study has shown that such measurements are possible and can yield the desired information on the particle size and volume fraction evolutions. Additional measurements have also indicated that SAXS measurements of sufficient quality for analysis can be performed on samples of 4 mm thickness in 1 s at the P21.2 beamline, which is required to capture the fast processes during such in situ measurements. Furthermore, SAXS measurements can be combined with wide-angle scattering (WAXS) to provide the opportunity for fast measurements of phase transformation and precipitation simultaneously and in situ.