Incompatible effects of B and B + Nb additions and inclusions ’ characteristics on the microstructures and mechanical properties of low-carbon steels

The influence of boron as well as boron with niobium additions on the phase transformation behaviour, resultant microstructures, and mechanical properties of thermomechanically controlled hot-rolled and direct-quenched low-carbon bainitic steel plates was investigated. Also, the probable factors that could inhibit their specific merits on hardenability, phase transformation behaviour and mechanical properties, were studied. Continuous cooling transformation diagrams of both deformed and non-deformed austenite were constructed for the investigated steels. Laser scanning confocal microscopy (LSCM) and field emission scanning electron microscopy (FESEM) were employed to examine the microstructures, besides detailed analyses of the non-metallic inclusions using FESEM combined with INCA software. Moreover, the precipitates were investigated qualitatively using both FESEM as well as transmission electron microscopy (TEM). The results showed that the addition of boron or boron with niobium led to an increase in the critical transformation temperatures ( A C1 and A C3 ) covering the intercritical range. The addition of boron with niobium decreased the bainite start transformation temperature ( B s ), while the lone addition of boron had a slight or insignificant effect on B s . While the addition of boron alone had no effect on the hardness, ultimate tensile strength (UTS), yield strength (YS), and elongation to fracture, augmenting it with niobium led to a marginal increase in the UTS and YS. In general, the addition of boron with or without niobium deteriorated the impact toughness of the investigated steel. These were explained in terms of the slight changes in the chemical composition and cleanness of the investigated steels and considering various microstructural features i.e., prior austenite grain size, effective grain size, coarsest grain size and precipitates characteristics, particularly the formation of coarse (Fe,Cr) 23 (B,C) 6 .

The influence of boron as well as boron with niobium additions on the phase transformation behaviour, resultant microstructures, and mechanical properties of thermomechanically controlled hot-rolled and direct-quenched low-carbon bainitic steel plates was investigated. Also, the probable factors that could inhibit their specific merits on hardenability, phase transformation behaviour and mechanical properties, were studied. Continuous cooling transformation diagrams of both deformed and non-deformed austenite were constructed for the investigated steels. Laser scanning confocal microscopy (LSCM) and field emission scanning electron microscopy (FESEM) were employed to examine the microstructures, besides detailed analyses of the non-metallic inclusions using FESEM combined with INCA software. Moreover, the precipitates were investigated qualitatively using both FESEM as well as transmission electron microscopy (TEM). The results showed that the addition of boron or boron with niobium led to an increase in the critical transformation temperatures (A C1 and A C3 ) covering the intercritical range. The addition of boron with niobium decreased the bainite start transformation temperature (B s ), while the lone addition of boron had a slight or insignificant effect on B s . While the addition of boron alone had no effect on the hardness, ultimate tensile strength (UTS), yield strength (YS), and elongation to fracture, augmenting it with niobium led to a marginal increase in the UTS and YS. In general, the addition of boron with or without niobium deteriorated the impact toughness of the investigated steel. These were explained in terms of the slight changes in the chemical composition and cleanness of the investigated steels and considering various microstructural features i.e., prior austenite grain size, effective grain size, coarsest grain size and precipitates characteristics, particularly the formation of coarse (Fe,Cr) 23 (B,C) 6 .

Introduction
Low carbon bainitic steels are widely used in several applications such as structural steels, pressure vessels, mobile cranes, and booms thanks to their desired combinations of strength and ductility. In this context, thermomechanically controlled processing followed by direct quenching is considered the most effective production route to achieve the desired microstructures resulting in a reasonable combination of strength and toughness in the low-carbon microalloyed steels [1,2]. In structural steels, different alloying elements like Mn, Cr, Mo, Nb, and B can be used in order to facilitate phase transformation to bainite via improving the hardenability and refining the microstructure, thereby simultaneously enhancing the strength and toughness [3][4][5]. For instance, chromium is commonly used to improve strength and hardenability as well as to enhance corrosion resistance [6,7]. In the case of low carbon steels, hardenability can be improved by suppressing the formation of high-temperature transformation products like ferrite and pearlite and promoting the formation of bainite and/or martensite [8][9][10][11][12].
The addition of boron is known to improve the hardenability of steels through its non-equilibrium segregation to austenite grain boundaries. Boron may be considered either the largest interstitial or the smallest substitutional alloy element in steel, as the atomic size ratio of boron/ iron is ≤ 0.6 for interstitial and ≥0.85 for substitutional alloying role in steels. This contributes to its poor solubility in austenite and thus describes its ability to segregate at the austenite grain boundaries [13]. This decreases the grain boundary energy and the self-diffusion of iron at the grain boundaries, which lead to a reduction in the favorable nucleation sites for ferrite thus hindering its formation [4,[14][15][16][17]. Given an appropriate cooling path, boron facilitates the transformation of austenite to much desired lower bainite with a concomitant increase in toughness alongsided strength. Besides, in small fractions of few ppm, boron may not affect the weldability adversely, while imparting high hardenability and maintaining alloying cost at low levels. For instance, addition of 10-30 wt ppm of protected boron to steel is equivalent to the addition (in wt.%) of 0.6 Mn or 0.7 Cr or 1.5 Ni or 0.5 Mo [14,18]. However, in order to reap the benefits of boron alloying, it should be available as soluble atoms at the grain boundaries, i.e. formation of boron nitrides (BN) or carbides ((Fe,Cr) 23 (C,B) 6 ) or oxides should be prevented [14]. In order to ensure that boron will be available to increase the hardenability, the melting should be carried out under a protective atmosphere and the molten metal should be protected using Ti to lower the free oxygen and nitrogen levels [19].
Although boron has been used over decades in different types of steel families, there is still indeterminacy on the effect of boron on toughness. Some studies [18,[20][21][22][23] showed that boron resulted in deteriorating the toughness, while others [24][25][26][27][28][29][30][31] believed that it leads to an increase in the toughness. These controversies may be due to several factors such as: i) difference in measuring methods for the toughness properties i.e. Charpy tests with standard or sub-size samples with V or U shape, ii) difficulties in measuring the small amounts of the added B, which can remain protected either as a free element or get combined in a precipitate, iii) the main elements in combination with boron like Cr, Ti, Al, N, and P are not being taken to the account properly, iv) the degree of protection of the molten steel before boron addition and v) type of the boron-additives such as boric acid, boric oxide, borax, ferro-boron ferroalloys and calcium boride (CaB 6 ) [19].
Alloying with niobium is widely used in different steel families due to its ability to refine the prior austenite grain size during hot rolling. At high temperatures, the relatively larger atomic size and electronic structure of niobium compared to iron develop a strong solute drag effect on the grain boundaries, thus exerting a strong retarding effect on recrystallization [32]. Moreover, Nb can hinder recrystallization owing to the strain-induced precipitation of Nb(C,N) at low temperatures. Also, remaining unprecipitated Nb retards phase transformations to lower temperatures, thereby increasing transformation hardening and toughening [14,[33][34][35][36][37]. Also, there is a synergistic effect of Nb with B, as the presence of soluble Nb prevents the precipitation of M 23 (B,C) 6 (where, M = Fe, Cr) on the prior austenite grain boundaries, thus combining with C to form NbC or Nb (C,N) [38,39].
Many controversies do still exist about the effect of B addition with or without Nb on the microstructure and mechanical properties. Hence, the current work aims at studying the effects of B as well as B + Nb additions and the prior controlled deformation on subsequent phase transformation behaviour during continuous cooling of a newly developed Cr containing low-carbon bainitic steel. Also, one of the objectives was to study the probable factors that could inhibit the explicit merits of B as well as B aided with Nb on hardenability, phase transformation behaviour and mechanical properties of the investigated steels including evolving microstructural features i.e., phase fractions, bainite effective grain size (with high-angle grain boundaries), and coarsest grain size (determined as 80 percentile effective high-angle grain sizes (D80%)), precipitate characteristics and non-metallic inclusion characteristics.

Materials
Three steels with the compositions listed in Table 1 were vacuum induction melted (VIM) and cast into 70-kg ingots at Outokumpu, Tornio, Finland. Various pieces with the dimensions 200 x 80 × 55 mm were cut from the ingots and soaked for 2 h at 1250 • C. The reheated slabs were then hot-rolled into 11-mm thick plates in accord with a thermomechanically controlled process (TMCP) using a 1000-kN laboratory rolling mill and direct-quenched in water. The TMCP rolling comprised     six hot rolling passes with total thickness reduction of ~81%: i) five hotrolling passes in the recrystallized controlled regime above 1000 • C (with thickness reduction ~77%) and ii) the final controlled rolling pass performed in the non-recrystallization regime at ~880 • C (with thickness reduction ~20% of the previous thickness in the rolling scheme). A thermocouple was placed in the middle of each sample in the hole drilled in the edge of the sample at mid-length to the mid-width in order to monitor the temperature of the pieces during the TMCP rolling and subsequent quenching. The final thickness of the plate was 11 mm at the finish rolling temperature (FRT) at 880 • C prior to direct quenching in water in order to achieve a relatively high cooling rate (CR) of about 40-50 • C/s. The non-recrystallization temperatures (T NR ) were predicted for different experimental steels (Table 1) using equation (1) [40] and given in Table 1.

Continous cooling transformation diagrams
To determine the CCT diagrams, cylindrical specimens of dimensions Ø5x7.5 mm were machined from hot rolled plates of experimental castings for dilatation tests on a Gleeble 3800 thermomechanical simulator. The samples were cut with the axis of the cylinders longitudinal to the rolling direction in the rolling plane. Some dilatation curves were used to determine the critical transformation temperatures (A c1 and A c3 ). Fig. 1 shows the experiments conducted on the Gleeble to obtain the dilatation curves for the construction of continuous cooling transformation (CCT) and deformation continuous cooling transformation (DCCT) diagrams. Two types of dilatation tests were made: with or without prior strain in the non-recrystallization regime. The critical transformation temperatures (A c1 and A c3 ) were determined from the dilatation curve obtained by heating the sample at 10 • C/s to 1100 • C (see Fig. 1a). For obtaining the dilatation curves for CCT diagrams, the samples were heated at 10 • C/s to 1100 • C and held for 180 s before cooling at various linear rates ranging from 80 to 2 • C/s (see Fig. 1b). For obtaining the dilatation curves for the DCCT diagrams, the samples were heated at 10 • C/s to 1100 • C for 180 s and then cooled to 880 • C at 2 • C/s and held for 15 s. The samples were then compressed with three hits each having a strain of ~0.2 (total strain ~0.6) at a strain rate of 1 s − 1 . The time between hits was 20 s. The specimens were then held 20 s before continuous cooling at various linear rates in the range of 80 to 2 • C/s (see Fig. 1c).

Microstructural characterization
The final microstructures were examined using a Keyence VK-X200 LSCM and a Zeiss Sigma FESEM. were used for the prior austenite grains reconstruction, while the small size areas (145 × 145 μm) were used for the determination of other microstructural features. Effective grain sizes were determined from the EBSD data as equivalent circle diameter (ECD) values considering high angle grain boundary misorientations (>15 • ). The coarsest grains were determined as 80 percentile effective high-angle grain sizes (D80%).
A reconstruction method was employed on the EBSD data using MATLAB software supplemented with the MTEX toolbox [41] in order to reveal the prior austenite grains. The reconstruction technique was performed on the basis of the previous works [42][43][44] through two main steps. First, the orientation relationship between the parent austenite and product ferritic phase, such as martensite or bainite (here, mainly bainite), was determined using Kurdjumov-Sachs (K-S) relationship [45] Then, in the second step, the grains' map was divided into separate clusters and parent austenite orientation was calculated for each cluster discretely to reconstruct the austenite orientation map and grain structure. The mean prior austenite grain size (PAGS) was calculated using the linear intercept method.
Quantitative and qualitative analyses of the non-metallic inclusions (NMIs) in hot-rolled samples were performed using an energy-dispersive X-ray spectroscopy (EDS) equipped with a Jeol JSM-7000F FESEM, combined with an automated particle explorer and analyzer (INCA software). The operating parameters used were 15 kV accelerating voltage, 10 mm working distance and 3.5 nA current. Each inclusion was measured for 1 s live time and the minimum inclusion size included in the results was 1 μm. The results included the number densities, location, shape, area fractions, and chemical composition of NMIs. The size of each NMI was determined using its maximum length. The inclusions were measured from samples in the rolling directionnormal direction (RD-ND).
A JEOL JEM-2200FS EFTEM/STEM was employed for TEM study to investigate the precipitates' characteristics of the investigated steels on various carbon extraction replicas and to determine their chemical compositions using energy-dispersive X-ray spectroscopy. Different samples were cut from the hot-rolled steel plates, ground, polished, and left for 2 days prior to etching in fresh 2 vol % nital reagent and then coated with 10-15 nm thick carbon films using physical vapour deposition. The samples were scored into small squares with dimensions about 3 mm × 3 mm. In order to extract the carbon films along with the precipitates, the surfaces of the samples were then exposed to 10% HNO 3 acid at a potential of 10 V. The extracted replicas were subsequently washed using ethanol (C 2 H 5 OH), hydrochloric acid (HCl) and distilled water in succession [46].
The fracture surfaces of Charpy V-notched samples at − 40 and − 80 • C were examined using FESEM with an accelerating voltage of 5 kV and a working distance of 8.5 mm. All the samples were cleaned in acetone for 10 min using an ultrasonic cleaner prior to the fractographic investigations.

Image quality (IQ) analysis
The EBSD-IQ analysis method proposed by DeArdo et al. [47] has been employed for detailed microstructural characterization and phase type and fraction discretization. Based on the varying intrinsic dislocation densities of different phase morphologies, it is obvious that different microstructural constituents generate different IQ values in the EBSD measurements [48]. Hence, the normalized IQ data can be analyzed to quantify the microstructures. In this technique, the normalized IQ histogram is deconvoluted into multiple peaks with a normal distribution shape and the ratio of area under each peak to the area of the IQ curve was considered as a fraction of a particular phase morphology or constituent. In the calculations, the existing possibility of the following phases or phase constituents was checked based on their relative peak locations in the normalized IQ axis (from 0 to 100): (Polygonal) ferrite ≥85, coarse bainite 65-75, upper bainite 40-50, lower bainite 25-35, and martensite ≤ 20.

Characterization of mechanical properties
The tensile properties were measured at room temperature using a 100-kN Zwick/Roell tensile testing machine on round tensile specimens with a diameter of 4 ± 0.1 mm and gauge length of 16 ± 0.1 mm, machined in the longitudinal (rolling) direction according to ASTM standard E8. The results are reported as an average of three samples per steel composition.
Charpy V tests were performed using a 300-J Zwick Roell PSW750 Charpy testing machine according to EN ISO 148 in the temperature range from − 120 • C to +20 • C with two standard Charpy V-notched samples (10 × 10 × 55 mm) per test temperature machined in the longitudinal direction. The 35 J/cm 2 Charpy V transition temperature (T35J) of the rolled plates was determined by plotting suitable ductile to brittle transition curves.
The hardness of the CCT/DCCT samples was measured using a Duramin-A300 (Struers) macro-hardness tester with 1 kgf indenter load (HV1) at 5 random positions, while the hardness of the hot-rolled samples was measured with 10 kgf indenter load (HV10), taken from the average of 5 sets of seven suitably spaced hardness impressions through the thickness.

CCT and DCCT diagrams
The CCT and DCCT diagrams were constructed based on the dilatation data, final microstructures, and macrohardness values of the dilatation specimens tested in the Gleeble simulator. Example of typical dilatation curves with and without prior strain are shown in Fig. 3 for 2.5CrB steel. The decomposition of austenite to bainite with or without ferrite can be observed clearly from the dilatation curves. Moreover, the effect of cooling rate and prior strain on the phase transformation characteristics is clearly observed from these curves. The results of CCT and DCCT diagrams are presented in Fig. 4. Compared to 2.5Cr steel, the addition of 25 wt ppm of B in the case of 2.5CrB steel led to a slight increase in the critical transformation temperatures A C1 and A C3 , which were further increased by a combined addition of 25 wt ppm of B along with 0.06 wt% of Nb (2.5CrBNb steel), as both elements stabilize and enlarge the ferrite phase field [49].
Figs. 5 and 6 show LSCM micrographs of the representative transformed microstructures of the non-deformed and deformed austenite for the three steels cooled at the linear rates of 80 and 2 • C/s. Over a wide range of cooling rates (80-2 • C/s), the final microstructures comprised of one or more of the following phase constituents: i) bainitic ferrite (BF), which is a mixture of lower bainite and upper bainite, ii) granular bainite (GB), iii) polygonal ferrite (PF), and iv) pearlite (P). As reported earlier by Jun et al. [4], increasing the cooling rate promoted the formation of bainitic ferrite in low carbon steels and suppressed the formation of high-temperature transformation products e.g. polygonal ferrite, granular bainite, and pearlite, thereby increasing the hardness and hence, strength of the final transformation products.
All the CCT diagrams are summarized in Fig. 7. The positive effect of B on hardenability is well known. However, in the current study, this effect is somewhat counteracted by the marginally lower contents of C, Mn, and Al in the case of 2.5CrB steel compared to those of 2.5Cr steel. In addition, the formation of coarse M 23 (B,C) 6 precipitates consume a significant fraction of B at the grain boundaries, as explained below. This explains the small variation in the hardenability and the bainite start transformation temperature (B s ) between 2.5CrB steel and 2.5Cr. However, at high cooling rates, the available small fractions of free B atoms at the grain boundaries suppressed the formation of hightemperature transformation products and promoted the formation of low-temperature transformation products i.e., BF (see Fig. 4 a, b, c, and  d). These effects were enhanced by the addition of Nb along with B, as the presence of Nb led to provide a further increase in the hardenability and better suppression of the high-temperature transformation products even extended to somewhat lower cooling rates. This resulted in a lower phase transformation start temperature and promoted the formation of bainite [50]. The formation of high-temperature transformation products i.e. GB, PF, and P as a consequence of prior deformation (see Fig. 4) are attributed to the high number of nucleation sites without adequate B atoms protecting the grain boundaries and enhanced diffusional paths provided by the dislocation substructure that formed in the austenite due to prior straining in the non-recrystallization region [51,52]. Fig. 8 shows the variation of hardness as a function of the cooling rate measured on the transformed microstructures of the studied steels in both the non-deformed and deformed austenite state. The variation in  hardness is directly related to the combination of the phase constituents in the transformation microstructures as discussed above. In the nondeformed and deformed cases, the slight difference in the hardness values between steels 2.5Cr and 2.5CrB is attributed to the slight compositional difference in respect of C, Mn, and Al contents (Table 1), though the formation of M 23 (B,C) 6 at the grain boundaries can largely influence the nucleation of transformtation products, which counteracted the positive effect of B in respect of enhancing the hardenability and hence, the hardness at a given cooling rate. However, the addition of B and Nb led to a marginal increase in the hardenability and hence, the hardness of 2.5CrBNb steel compared to the other investigated steels, especially at relatively slow cooling rates, both in the case of nondeformed as well as prior deformed samples, essentially due to the higher solid solution strengthening combined with the microstructural strengthening [50].
In the case of 2.5Cr steel, prior deformation in the nonrecrystallization regime resulted in accelerated nucleation and transformation of high temperature products, as a consequence of reduced hardenability, and consequently the hardness values were lower in the strained samples. However, with the addition of B or B + Nb, the  hardness values were at par with or close to those of unstrained samples, essentially due to enhanced hardenability and significant suppression of the high-temperature transformation products, though relatively finer grain size in Nb-bearing steels can accelerate phase transformation at least at low cooling rates, Fig. 9.

General microstructure
Microstructures recorded at the centreline of the hot rolled plate materials are displayed in Fig. 10. As a result of the accelerated cooling, the transformed microstructures of the investigated hot rolled and direct-quenched steel plates comprised mainly of a mixture of lower bainite (LB), upper bainite (UB), and coalesced bainite or coarse bainite (CB) as can be seen from Fig. 10 (a, d and g). It can be noticed from Fig. 10 (a, d and g) that in comparison to 2.5Cr steel, the fraction of the CB decreased with the addition of B in the case of 2.5CrB steel and further decreased with the addition of B + Nb, which was confirmed also by detailed EBSD-IQ data analysis, see Fig. 10. EBSD-IQ data were deconvoluted and analyzed to determine and compare the fractions of each microstructural constituent in the studied samples. The estimated results based on the areas under the curves are shown in Fig. 11. As it can be seen, 2.5CrB steel and 2.5CrBNb steel showed a similar trend with an almost equal amount of upper (43%) and lower bainite (45-49%) morphologies as the major component. While 2.5Cr steel exhibited a bit different trend such that fraction of coarse bainite (24%) was significantly higher than in the other steels (8-12%). However, the fractions of upper and lower bainite were also identical (38% each) in 2.5Cr steel, marginally lower than in the other steels.
Grain boundary maps including high-angle boundaries (15 • -65 • ) and low-angle boundaries (2 • -15 • ) are shown in Fig. 10 (b, e, and h). The reconstructed prior austenite grain structures obtained by analyzing the EBSD data with MTEX and MATLAB software along with the mean grain sizes are shown in Fig. 10 (c, f, and i).
The mean PAGS for steels 2.5Cr, 2.5CrB, and 2.5CrBNb were calculated as 23 μm, 30 μm, and 20 μm, respectively. The slightly higher PAGS in the case of 2.5CrB steel is not necessarily related directly to B addition, but it may be related to the slight compositional difference between 2.5Cr steel and 2.5CrB steel (see Table 1). However, PAGS decreased marginally by the addition of B + Nb due to the retarding effect of Nb on recrystallization via solute drag on the grain boundaries at high temperatures, and the formation of large volume fraction of Nb (C,N) precipitates, as discussed below, which increased the pinning force against the grain boundary migration and growth of the prior austenite grains [53][54][55]. Moreover, pancaked austenite grain structure was obtained in the case of 2.5CrBNb steel (see Fig. 10i) due to the retardation of austenite recrystallization kinetics as a result of Nb addition [56]. The predicted T nr temperatures for steels 2.5Cr, 2.5CrB and 2.5CrBNb were 837, 818 and 1091 • C, respectively (Table 1). This indicates that the last rolling pass at 880 • C (FRT) was in the recrystallization regime for 2.5Cr and 2.5CrB steels, whereas it was in the non-recrystallization regime in the case of 2.5CrBNb steel. Effective and coarsest (D80%) grain sizes are given in Fig. 12 a. Grain boundary misorientation distributions based on the grain boundary maps ( Fig. 10 (b, e, and h)) are shown in Fig. 12b. There is a slight or no effect of B addition on the effective and the coarsest (D80%) grain sizes. This is attributed to some interrelated parameters (as discussed below), the coarsening of the PAGS of 2.5CrB steel compared to 2.5Cr steel (as illustrated above and also shown in Fig. 10 (c and f)), and the variation in the morphologies of the bainitic structure in 2.5CrB steel compared to 2.5Cr steel. As exhibited by 2.5Cr steel, there is a relatively higher volume fraction of CB (24%) and lower volume fractions of UB (38%) and LB (38%) (see Fig. 11) compared to those seen in 2.5CrB steel. On the other hand, the addition of B + Nb led to a further decrease in the effective and the coarsest (D80%) grain sizes, which is attributed to the refinement of the PAGS. Based on the IQ-analysis results given in Fig. 11, it can be concluded that an increased volume fraction of coarse bainite led to an increase in the value of D80%, while an increased volume fraction of lower bainite as well as upper bainite led to a decrease in the value of D80% (see Fig. 13). It is known that as a result of the different variants of the Kurdjumov-Sachs orientation relationship, the misorientation peaks at about 7.5 • correspond to sub-block boundaries, while the peaks at about 16 • , 52.5 • , and 59 • correspond to the packet and/or block boundaries [57]. The addition of B or B + Nb increased the sub-boundaries at about 7.5 • and decreased packet and/or block boundaries at about 52.5 • , as can be seen in Fig. 12b.

Non-metallic inclusions characteristics
Even though there is no variation in the total impurity level (see Table 1), there is a large difference in respect of the characteristics of NMIs within the investigated steels. This variation may be attributed to the differences in the heat types and the starting raw materials. The size distribution and the area fraction of the NMIs are shown in Fig. 14. Generally, the majority of NMIs were relatively small (<8 μm) with only few NMIs coarser than 8 μm. The 2.5Cr steel with the lowest total impurity level (see Table 1) showed the lowest number density and area fractions of NMIs for all size ranges compared to steels 2.5CrB and 2.5CrBNb.
The chemical compositions of the NMI types with their number density and area fractions are shown in Fig. 15. NMIs in the investigated steels have been classified into four classes: manganese sulfide (MnS), aluminum oxide (Al 2 O 3 ), manganese sulfide aluminate (MnS⋅Al 2 O 3 ), and others that include all unclassified inclusions. The EDS analysis of some unclassified inclusions showed the presence of N combined with other elements, i.e. Mn, S, Cr and O, and this presumably could be due to the formation of a complex inclusion containing MnS, CrN, Cr 2 O 3 and MnO (though not identified by the software). These types of inclusions comprise a part of the unclassified group since these do not fit in any of the pre-determined classes. The most common inclusions in the investigated steels were Al 2 O 3 and MnS⋅Al 2 O 3 , whereas MnS formed only a small fraction of all the inclusions. The unclassified inclusions were observed only in 2.5CrB and 2.5CrBNb, as can be seen from Fig. 15. Al 2 O 3 and MnS inclusions are quite typical for the laboratory cast steels that are usually manufactured without Ca-treatment. In industrial-scale trials, these inclusions would most likely be modified to CaO⋅Al 2 O 3 , CaS, and CaO⋅Al 2 O 3 ⋅CaS inclusions, which are generally considered less harmful in respect of both castability as well as mechanical properties of the steels, compared to the presence of Al 2 O 3 and MnS inclusions. Based on the TILs (see Table 1), size distribution, number density, and area fractions of the NMIs (see Figs. 14 and 15), 2.5Cr steel showed the best cleanness followed by 2.5CrBNb steel, whereas 2.5CrB steel exhibited minimum cleanness. Fig. 16 shows the spatial distribution of NMIs in the investigated steels. A few large stringers of Al 2 O 3 can be seen in steels 2.5CrB and 2.5CrBNb (see Fig. 16b and c).

Precipitate characteristics
STEM micrographs and EDS analyses of the common precipitates in the investigated steels are shown in Fig. 17. A few M 23 C 6 precipitates were observed in steels 2.5Cr and 2.5CrB with mean equivalent circular diameters (ECD ppt ) of 19 nm and 17 nm, respectively. However, the presence of Nb in the case of 2.5CrBNb steel led to the formation of a large number of small Nb(C,N) precipitates in addition to formation of a few M 23 C 6 precipitates with mean ECD ppt 9 nm (see Fig. 17c). It is well known that the formation of M 23 (B,C) 6 precipitates on the austenite grain boundaries leads to the deterioration of the toughness [19,39].
It is obvious that there is a difficulty in the analysis of the light elements i.e., B, C, N, and O because of their low photon energy resulting in a low yield of x-ray and low energy peaks. In the current study, as C and B are light elements and due to the possibility of overlapping between their characteristic peaks, it may not be possible to detect any B in the M 23 C 6 precipitates due to the high carbon peaks from the replica itself. Therefore, FESEM study, combined with EDS analysis, was carried out to systematically examine the microstructures of the samples to see if there were any B-containing precipitates formed at the prior austenite grain boundaries or elsewhere. As expected, there were a number of large M 23 (B,C) 6 precipitates observed within the microstructure of 2.5CrB and 2.5CrBNb steels, which did not detach on the replica, see Fig. 18. However, a complex Al 2 O 3 . M 23 (B,C) 6 with traces of B were observed in 2.5Cr steel, presumably nucleation and growth of M 23 (B,C) 6 on a preexisting Al 2 O 3 particle from melting stage. Table 2 lists the chemical compositions of the B-containing precipitates in the investigated steels. This also clarifies the drawbacks of carbon replicas in respect of extraction of the large precipitates and the detection of B within the precipitates. The formation of large M 23 (B,C) 6 precipitates inhibit the anticipated effect of B addition on the hardenability of the steels and accordingly, hardness is affected in accord with the phase composition, as described above. Also, it is construed that the formation of M 23 (B,C) 6 precipitates adversely influences the tensile and impact toughness properties, as discussed later. The formation of large M 23 (B,C) 6 precipitates in steels 2.5CrB and 2.5CrBNb could be because of the addition of B by an amount (25 and 22 wt ppm, respectively) that presumably enhanced the propensity to form boro-carbides, as the concentration can vary from location to another at the grain boundaries. Several studies have reported that exceeding the required optimum amount of B does not improve the hardenability, but instead it might deteriorate the toughness properties because of the formation of M 23 (B,C) 6 or BN [19].

Mechanical properties
A summary of the tensile properties, mean hardness, and the 35 J/ cm 2 transition temperatures (T35J) estimated from the Charpy V-notch impact toughness data is listed in Table 3. The impact transition curves are shown in Fig. 19. The three investigated steels exhibited better than the targeted 700 MPa yield strength (YS), which is a common strength requirement in several structural applications [59].
From Table 3, it is clear there is no significant difference between 2.5Cr and 2.5CrB steels in respect of hardness, YS, ultimate tensile strength (UTS), and elongation to fracture (El) values. However, the toughness properties deteriorated and T35J increased significantly for steel 2.5CrB (− 54 • C) compared to steel 2.5Cr (− 124 • C). This could be explained by the slight compositional difference between 2.5Cr and 2.5CrB steels in weight percentage of C, Mn and Al ( Table 1) that counteracted the expected effect of B on the hardness, UTS, YS, and elongation to fracture values. Based on the microstructural examinations, the deterioration of impact toughness and the variation in the transition temperatures can also be explained in respect of the cleanness i.e., TIL and NMIs and in the term of PAGS, effective grain size, and coarsest grains (D80%). As discussed above, even with the slight difference in the TIL, there exists a high number density and area fraction of the NMIs in steel 2.5CrB compared to 2.5Cr steel. Moreover, the presence of complex inclusions (containing MnS, CrN, Cr 2 O 3 and MnO) in the case of 2.5CrB and 2.5CrBNb steels which is not observed in 2.5Cr steel. This could be one of the reasons for the lower toughness and the higher transition temperature of steel 2.5CrB compared to 2.5Cr steel. However, the main reason is the formation of large M 23 (B,C) 6 precipitates, as several studies reported that the presence of M 23 (B,C) 6 borocarbides on the grain boundaries led to the deterioration of the toughness as a consequence of the decrease in the upper shelf energy or increase in the transition temperature, cf [19].
In the case of steel 2.5CrBNb, the YS and UTS were improved slightly as can be seen from Table 3. However, it had a lower toughness and higher T35J transition temperature (− 77 • C) compared to that of steel 2.5Cr (− 124 • C), but marginally better than in steel 2.5CrB (− 54 • C). This could be attributed to i) the higher number density of NMIs and their area fractions compared to those of steel 2.5Cr, but lower than those of steel 2.5CrBNb and ii) the formation of large M 23 (B, C) 6 precipitates despite the reduction in the PAGS, D80% and effective grain size in the case of 2.5CrBNb steel compared to 2.5Cr and 2.5CrB steels [60].

Fractography
SEM fractographs of the fracture surfaces of CVN impact toughness samples after testing at − 40 • C and − 80 • C are shown in Fig. 20. The fractographic analysis of 2.5Cr steel tested at − 40 • C showed that it comprised ductile fracture marked by dimpled rupture (Fig. 20a), while the other investigated steels 2.5CrB (Figs. 20c) and 2.5CrBNb (Fig. 20e) displayed a mixture of brittle cleavage fracture along with some dimples (ductile fracture). At − 80 • C, the fractographic analysis of 2.5Cr steel (Fig. 20b) showed a very small fraction of ductile rupture alongside nearly complete brittle fracture, while other steels 2.5CrB (Fig. 20d) as well as 2.5CrBNb (Fig. 20f) showed essentially brittle fracture marked by beack markings. The variation in the degree of cleanness i.e. NMIs and TILs of the investigated steels and the formation of large M 23 (B,C) 6 precipitates attributed to the change in the fracture mode from ductile to brittle behaviour. The fractographic analysis results are in line with other results discussed above.

Conclusions
Three experimental 2.5 wt% chromium-containing steels were designed with 0.04C-0.2Si-1.0Mn (in wt.%) as the base without (2.5Cr) or with the addition of B (2.5CrB) as well as B combined with Nb (2.5CrBNb). The as-cast VIM ingots of the steels were processed through laboratory hot rolling followed by direct quenching. The influence of B and B with Nb additions on the continuous cooling transformation (CCT) and deformation continuous cooling transformation (DCCT) diagrams were studied using a Gleeble 3800 thermomechanical simulator to determine the behaviour of microalloying elements (B, Nb) both in nondeformed and deformed conditions of austenite. Also, the effects of these additions on the evolved microstructures and corresponding mechanical properties of these hot-rolled and direct-quenched materials were evaluated. Moreover, the probable factors that could inhibit their   beneficial effects were studied. The conclusions can be drawn as follows: 1. The addition of B and B with Nb led to an increase in the critical transformation temperatures (A C1 and A C3 ). However, the formation of coarse M 23 (B,C) 6 precipitates and the slight variation in the chemical composition i.e., C, Mn, and Al resulted in the reduction of their efficacy on the hardenability, besides a slight decrease in the bainite start transformation temperature (B s ). 2. In both the non-deformed and deformed cases, B addition had only a slight effect on the hardness values, mainly as a consequence of combining of a significant fraction of B in the formation of large M 23 (B,C) 6 precipitates. The addition of B combined with Nb, however, led to an increase in the hardenability and as a consequence, the hardness was found to be quite improved, even at low cooling rates. 3. Despite significant consumption of B in the formation of large M 23 (B, C) 6 precipitates at the grain boundaries, the difference in hardness values between the non-deformed and deformed cases decreased with the addition of B and/or B with Nb due to the enhancement of hardenability and suppression of the high-temperature transformation products (ferrite and pearlite). 4. Addition of just B or B combined with Nb led to the reduction in the volume fraction of the coarse bainite and hence, increased the volume fraction of lower and upper bainite. This in turn led to the refinement of effective grain size and D80% in the hot-rolled plates. 5. Prior austenite grain size decreased as a result of addition of B with Nb, owing to the retarding effect of Nb on the recrystallization  kinetics via solute drag on the grain boundaries at high temperatures, and the pinning effect of a large volume fraction of Nb(C,N) precipitates formed subsequently. 6. The beneficial effects of B on the hardness, UTS, YS, elongation, and the toughness properties were counteracted as a result of the variation in the cleanness (i.e., TILs and NMIs), in addition to the formation of large M 23 (B,C) 6 precipitates and the slight compositional difference in the percentage of C, Mn, and Al. Though refinement of PAGS and effective grain size, and D80% should normally improve both the strength and toughness concomitantly, the formation of large M 23 (B,C) 6 precipitates can be counterproductive. 7. The addition of Nb along with B improved the YS and UTS. However, the variation in the cleanness and the formation of large M 23 (B,C) 6 precipitates led to the deterioration of the toughness properties, though these were still better compared to the steel with only B addition.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.