Volume 59, Issue 9, May 2011, Pages 3710-3719

a r . arc fo A surface layer with a depth-dependent microstructure was produced on a ferritic steel (P92) plate by means of surface mechanical attrition treatment (SMAT). The austenitization processes of ferrite and carbides in the surface layers with different average grain sizes were investigated by means of in situ X-ray diffraction, differential scanning calorimetry and transmission electron microscopy. Exper- especially in steels, are a central topic in physical metal- at high temperatures [6,7,12]. The case is more complicated in high-alloy steels because the carbides in the initial steel is relatively sluggish while growth of austenite is rapid, The on-heating evolutions, such as dislocation recovery, grain growth and carbide precipitation, as well as the resul- tant mechanical properties, have been studied in nano- structured or ultrafine-grained ferritic steels [20–22]. However, to the authors’ knowledge, there is no study on the austenitization process of nanostructured ferrite ⇑ Corresponding authors. Tel.: +86 24 2397 1508; fax: +86 24 2399 8660. E-mail addresses: zbwang@imr.ac.cn (Z.B. Wang), lu@imr.ac.cn (K. Lu). Available online at www.sciencedirect.com Acta Materialia 59 (2011) 3710–37 lurgy because of a combination of fundamental scientific interests and technological importance [1–4]. Among phase transformations in steels, the on-heating formation of face- centered cubic (fcc) austenite (c) from body-centered cubic (bcc) ferrite (a) matrix has been studied extensively and some investigations have dealt with the effects of alloying elements and initial microstructures on the austenitization process [5–13]. Experimental works on low-alloy steels have indicated that cementite particles provide nucleation sites for austenite, and austenite formation is very rapid and the dissolution of carbides occurs in the austenite but not in the ferrite. By studying the austenite formation and the carbide dissolution in Fe–8.2Cr–C alloys with dif- ferent C concentrations, Shtansky et al. [11] noticed that the mechanisms of austenite nucleation and growth depend on the composition, starting microstructure and austenitiz- ing temperature. Nanostructured materials have attracted intensive inter- est for several decades, due to their novel properties origi- nating from a large volume fraction of interfaces [15–19]. imental results showed that the onset temperature of the austenitization process decreases gradually with decreasing sizes of ferrite grains and carbide particles, being �120 K lower in the top SMAT surface layer compared with that in the original sample. In addition, the two- step austenitization process in the surface layers becomes a one-step one when the mean size of carbide particles is smaller than 20 nm. The effects of microstructure refinement on the accelerated austenitization processes were discussed in terms of thermodynamic and kinetic. � 2011 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Nanostructured; Surface mechanical attrition treatment; Ferritic steel; Grain size effects; Austenitization process 1. Introduction Solid-state phase transformations in metallic materials, structure are thermodynamically more stable than cement- ite [14]. For example, Lenel and Honeycombe [10] observed that nucleation of austenite in an Fe–10Cr–0.2C (wt.%) Grain size effects on the in a nanostructu L.M. Wang, Z.B Shenyang National Laboratory for Materials Science, Institute of Metal Rese Received 23 December 2010; received in revised Abstract 1359-6454/$36.00 � 2011 Acta Materialia Inc. Published by Elsevier Ltd. All doi:10.1016/j.actamat.2011.03.006 ustenitization process ed ferritic steel Wang ⇑, K. Lu ⇑ h, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, China rm 25 February 2011; accepted 5 March 2011 www.elsevier.com/locate/actamat 19 rights reserved. ater matrix. This might be related to the fact that significant growth of nanosized grains may occur upon heating prior to the phase transformation temperature being reached. At ambient temperature, c-Fe has been experimentally observed in nanostructured Fe when the grain size is small enough [23,24], and it was suggested to be thermodynami- cally stable by calculating the Gibbs free energies of inter- faces in nanostructured c and a grains [25]. Therefore, notable grain size effects on the austenitization behaviors in ferritic steels might be expected. By means of surface mechanical attrition treatment (SMAT), surface layers with a gradient grain size distribu- tion (ranging from nanometers, submicrons to microns) have been synthesized on various metallic materials [26–34] as a result of gradient variations of applied strains and strain rates with depth from the treated surface. Previ- ous studies showed that the nanostructures with enhanced Cr diffusivity in the SMAT surface layers of low-carbon steel [35] and H13 steel [36] are effectively stabilized by the formation of fine dispersive Cr compound particles during chromizing at 873 K, resulting in the growth of much thicker chromized surface layers than that on the coarse-grained samples after chromizing treatments at tem- peratures as high as 1323 K. This means that the gradient microstructure of the SMAT surface layer on steels is sta- ble at elevated temperatures with dispersive precipitates. Such a kind of gradient nanostructured surface layer pro- vides a unique opportunity to study the grain size effect on austenitization behavior on the nanometer scale in one sample. In this work, a nanostructured surface layer with a depth-dependent microstructure is synthesized on a com- mercial ferritic steel plate by means of SMAT. Thermal sta- bility and austenitization process in the surface layer are characterized with respect to the microstructure. 2. Experimental 2.1. Sample preparation The studied ferritic steel (P92) was supplied by Wyman Gordon Forgings Inc., with the chemical composition (in wt.%) of 0.11 C, 8.75 Cr, 0.40 Mo, 1.75 W, 0.18 V, 0.05 Nb, 0.05 N and balance Fe. The initial material was in an austenitized and tempered condition (1 h at 1343 K fol- lowed by 2.5 h at 1048 K). The plate sample (100 � 50 � 4.0 mm3 in size) of the as-received steel was submitted to SMAT, the set-up and procedure of which have been described previously [26–28]. In brief, a large number of hardened steel balls of 8 mm diameter were placed at the bottom of a cylinder-shaped chamber and vibrated at a high frequency by a generator. The sample to be treated was fixed at the upper side of the chamber and impacted by flying balls repeatedly and multidirection- ally. Because the sample surface was plastically deformed L.M. Wang et al. / Acta M with high strains and high strain rates, grains in the surface layer were effectively refined, and a depth-dependent microstructure was consequently generated. In this work, the sample was treated for 60 min in vacuum at ambient temperature at a vibrating frequency of 50 Hz. No detect- able contamination was introduced into the surface layer during the SMAT process. 2.2. Microstructure characterization The microstructure evolutions with depth in both the as- SMAT and the annealed-SMAT samples were character- ized using a JEM-2010 transmission electron microscope (TEM) operated at a voltage of 200 kV. TEM foils of the topmost layers were cut by the electro-spark discharge technique, then mechanically polished, dimpled and finally ion-milled from the untreated side. In addition, TEM foils of different subsurface layers were cut, mechanically pol- ished, dimpled and finally electropolished from the untreated side of the SMAT samples after the removal of surface layers of different thicknesses. The electropolishing was carried out at 253 K with an electrolyte of 5 vol.% per- chloric acid and 95 vol.% alcohol. A short-period milling process at a low angle (at 4–5� for �10 min) from both sides was applied to clean the foil surfaces before the TEM observation. The grain/cell sizes were averaged from a few hundred grains/cells selected randomly from TEM images. In the case that not enough grains could be counted in a TEM image at the greater depths, an FEI Nova-nano scanning electron microscope (SEM) with a lower magnification was applied to observe the microstruc- ture at the corresponding region of a cross-sectional sample. The phase constitution of the original sample and the SMAT surface layer at room temperature was identified by X-ray diffraction (XRD) analysis using a Rigaku D/max 2400 X-ray diffractometer (7.5 kW) with Cu Ka radiation, with a step size of 0.02�. 2.3. Phase transformation measurements The surface layers at different depths of the SMAT sam- ple were cut by electro-spark discharge technique and then mechanically polished from the untreated side to �10 lm in thickness. Thermal analyses of the prepared foil samples (�15 mg in weight) were conducted on a Netzsch differen- tial scanning calorimeter (DSC 404C). The experiments were carried out from room temperature to 1373 K at a heating rate of 20 K min�1, in a flowing Ar atmosphere with a gas flow rate of 50 ml min�1. The temperature was calibrated by the melting points of pure In, Au and Ni. In addition, the baseline in the region of each peak was constructed by a polynomial (or an apparent transforma- tion function) through the tangents at the left and right sides of each peak at the evaluation limits [37]. The DSC curves shown in the present work are the subtraction results of the measured curves to the baselines. According ialia 59 (2011) 3710–3719 3711 to the standards of the International Confederation for Thermal Analysis, the onset temperature of each reaction ter (To) was determined as the intersecting point of the base- line with the tangential line from the point with the maxi- mum slope of the DSC curve at the left side of the peak. In order to understand the phase transformation process of the SMAT sample, an in situ XRD experiment was con- ducted using a Brueker D8 Discover XRD (12 kW) equipped with a high-temperature attachment. The temper- ature was calibrated by the melting point of pure Al. The measured sample was heated to preset temperatures at a rate of 60 K min�1 and held for 1 min before collecting the XRD profiles from 41.6 to 46�, with a step size of 0.02� and a scanning rate of 4� min�1. The sample temper- ature was monitored by using a PtRh–Pt thermocouple, of which the accuracy is ±2 K. All analyses were carried out in vacuum (2 � 10�3 Pa). 3. Results and discussion 3.1. Microstructure of the SMAT sample The initial microstructure of the sample before SMAT is shown in Fig. 1a. A martensitic structure is observed with numerous rod-like precipitates at lath/subgrain boundaries in the austenitized and tempered ferritic steel. The width of the laths is typically �540 nm and the length is �13 lm. Subgrains are formed within laths during tempering to reduce the density of dislocations formed by austenitizing [38]. The precipitates are determined to be mostly of the type M23C6 (M = Cr, Fe) according to the corresponding selected area electron diffraction (SAED) pattern (see the inset in Fig. 1a), with average dimensions of �80 nm along the short axis and �160 nm along the long axis. The total amount of precipitates is estimated to be �3 vol.% in the original sample. In addition to M23C6 precipitates, various much finer M(C, N) precipitates (�16 nm) were also detected in the same material with the same heat treatment by observing extraction double replicas [38,39]. Clear evidence of microstructure refining induced by plastic deformation is observed in the SMAT surface layer of �100 lm thickness. As shown in Fig. 1b, a high density of dislocations are formed within the martensite (or ferrite; the same in this work) laths and various dense dislocation walls (DDWs) develop mostly parallel or perpendicular to the lath directions. These are subdivided into smaller grains/cells at a depth of �60 lm from the treated surface. The grain/cell size is �300 nm along the short axis and �600 nm along the long axis, respectively. Meanwhile, pre- cipitates are also slightly refined to sizes of �60 nm along the short axis and �120 nm along the long axis. Further TEM observations indicate that more and more dislocations and DDWs develop in the martensite laths and the resulted grain/cell size decreases gradually with decreasing depth from the treated surface, due to increas- ing strain and strain rate. Moreover, the refined ferrite grains/cells appear to be equiaxed when their size is below 3712 L.M. Wang et al. / Acta Ma �35 nm (at a depth of <30 lm). As for the precipitates (M23C6), the sizes along both axes also decrease gradually with decreasing depth, and equiaxed particles are formed when the size is below �20 nm (at a depth of <40 lm). Due to the very high strain and strain rate (102–103 s�1 [27,28]), extremely fine equiaxed ferrite grains with random crystallographic orientations are formed in the top surface layer, as revealed by TEM observations and the corre- sponding SAED pattern in Fig. 1c and d. The average size of ferrite grains is estimated to be �8 nm from a number of dark-field images taken from the (1 1 0) diffraction of a-Fe. In addition, significant decreases in both the size and vol- ume fraction of precipitates are observed in the top surface layer. The mean size of M23C6 particles is refined to be �4 nm and the volume fraction is reduced to be �1%, according to the estimated results from a number of dark-field images taken from the (3 1 1) diffraction of M23C6 phase (see Fig. 1e). Microstructure observations of the SMAT surface layer indicate that both ferrite and precipitate grains in the pres- ent ferritic steel are refined by dislocation activities, which are similar to those observed previously in AISI 52100 steel [29] and AISI H13 steel [40] during SMAT. In brief, the refinement mechanism of ferrite grains involves formation of DDWs and dislocation tangles in both the original grains and the refined cells (under further straining), trans- formation of these microstructures into subboundaries with small misorientations, and evolution of subboundaries to highly misoriented grain boundaries [28,29]. When fer- rite grains are refined to a certain size, plastic deformation occurs in precipitate particles due to grain refinement strengthening of the ferrite matrix. Therefore, the particles are also progressively refined into smaller particles and/or dissolved into the ferrite matrix, as suggested by significant decrements in both the size and volume fraction of M23C6 particles in the SMAT surface layer. The measurement results of average sizes of ferrite grains/cells and M23C6 particles are summarized in Fig. 2 as a function of depth from the SMAT surface. It is clear that the top 40 lm surface layer is nanostructured (with grain sizes below 100 nm) and the average sizes of both phases increase with increasing depth in the top 100 lm surface layer. Both ferrite and M23C6 grains appear to be equiaxed in the top surface layer, but anisotropic morphol- ogies are observed in the subsurface layer at a depth larger than �30 lm. 3.2. Austenitization process of the SMAT ferritic steel 3.2.1. Phase transformation in the top surface layer during heating A typical DSC curve of the top surface layer (�10 lm in thickness) of the SMAT sample is given in Fig. 3, in com- parison with the DSC curve of the original sample without SMAT. An exothermic peak (CP) between 800 and 870 K is observed on the DSC curve of the SMAT sample, while no such a peak appears on the curve of the original sample. ialia 59 (2011) 3710–3719 A series of TEM observations and XRD analyses of micro- structure evolutions of ferrite and precipitates across this Fig. 1. Typical bright-field TEM images of (a) the original P92 steel sample and (b) at a depth of �60 lm of the SMAT sample. The inset in (a) shows the corresponding SAED pattern of M23C6. (c) A typical bright-field TEM image of the top SMAT layer. (d and e) Dark-field TEM images of ferrite grains and M23C6 particles, taken from diffractions of (1 1 0)a and (3 1 1)M23C6, respectively, as indicated on the SAED pattern (inset in (c)). 0 50 100 150 1 10 100 1000 Depth from surface (μm) Si ze (n m ) Ferrite grain/cell (short axis) Ferrite grain/cell (long axis) M23C6 particle (short axis) M23C6 particle (long axis) Fig. 2. Variations of average grain/cell sizes of ferrite and M23C6 with depth from the treated surface of the SMAT sample. 800 900 1000 1100 1200 1300 1019 K AI AIFM FM CP H ea t F lo w Temperature (K) Original0.1 W/g 1122 K AII SMAT Fig. 3. DSC curves of the original sample and the top surface layer of the SMAT sample, at a heating rate of 20 K min�1. L.M. Wang et al. / Acta Materialia 59 (2011) 3710–3719 3713 (b)γ(111) (a) γ(111) F nd le w 3714 L.M. Wang et al. / Acta Mater temperature range in an ongoing study (Wang et al., in preparation) confirmed that this peak is mostly induced by the significantly coarsening process of ferrite grains, as well as the reforming and/or coarsening processes of pre- cipitates from the broken and/or dissolved precursors dur- ing SMAT. The endothermic peaks FM on the DSC curves of both the SMAT sample and the original are confirmed to be induced by the magnetic transition from a ferromag- netic state to a paramagnetic state. No phase transition is detected across the peaks and they are completely revers- ible during DSC scans when the samples are heated to tem- peratures below the second (AI) peaks and then cooled down. Moreover, the peak positions of FM on the mea- sured DSC curves agree very well with the reported Currie point (�1018 K) of a reduced activation ferritic–martens- 42 44 46 48 42 44 46 48 In te ns ity (a .u. ) SMAT α(110) 1073 K 1023 K 1153 K 963 K 973 K 2 Theta (deg.) 1163 K Original α(110) 1173 K 1123 K 1263 K 1083 K 1093 K 1273 K Fig. 4. In situ XRD profiles of (a) the topmost layer of the SMAT sample a ratio F (derived by Eq. (1)) of the top SMAT layer and the original samp itic steel with a nominal composition (wt.%) of 9Cr– 0.09C–1W [8]. Comparing the DSC curves of both samples after the ferromagnetic transition, one can see that the onset temper- ature of the endothermic peak AI on the curve of the SMAT sample (1019þ26�2 K 1) is significantly decreased rela- tive to that of the original sample (1122 ± 2 K). In addi- tion, the broad endothermic peak AII observed in the original sample disappears in the SMAT sample. In situ XRD analyses were carried out on the SMAT and the original samples from room temperature to 1373 K, as shown in Fig. 4a and b, respectively. A decrease in the diffraction intensity of ferrite accompanied by an increase in the diffraction intensity of austenite is observed across the temperature intervals of peak AI for the SMAT sample and of peaks AI and AII for the original sample. While a quantitative determination of the volume fractions 1 Due to the overlap of FM and AI peaks, the baseline within this temperature range could not be accurately determined. The uncertainty of the onset temperature of AI in the SMAT sample is determined by using baselines before the peak FM and after the peak AI, respectively. of a and c phases during the in situ XRD measurements is difficult, a factor F is defined to indicate the variation of relative amount of c with temperature (see Fig. 4c), i.e. F ¼ Icð111Þ Icð111Þ þ Iað110Þ ð1Þ where Ic(111) and Ia(110) are the diffraction intensities of c (1 1 1) and a (1 1 0) peaks, respectively. Austenite is first detected in the SMAT sample at the temperature of 973 K, its volume fraction increases with increasing tem- perature and the sample is completely composed of austen- ite at 1163 K. A gradual decrease in the fraction of a accompanies the increase in c within this temperature range. That is to say, the austenitization process in the SMAT ferritic steel starts at