TEM Investigations of the Structural Evolution in a Pearlitic Steel Deformed by High-Pressure Torsion F. WETSCHER, R. PIPPAN, S. STURM, F. KAUFFMANN, C. SCHEU, and G. DEHM A fully pearlitic steel was deformed by high-pressure torsion up to very high strains, and the changes in the microstructure were determined by analytic and conventional transmission electron microscopy. The imposed strain leads to a fragmentation and an alignment of the cementite lamellae parallel to the shear plane. The electron energy-loss near-edge-fine structures of the Fe-L2,3-edge of the iron matrix and the cementite lamellae were measured with high spatial resolution. The results indicated that after high-pressure torsion, the iron matrix contains finely dispersed carbon-rich areas that do not show the electronic fingerprint of cementite. However, the refinement in microstructure leads to an enormous increase in mechanical strength.
I. INTRODUCTION
COLD deformation of pearlitic steels occurs in many technical applications, for instance during wire drawing or at the surface of rails during service.[1,2] Especially in the latter example, it is important to relate the deformation and the resulting microstructure to changes of mechanical properties and damage processes. A key feature is therefore the behavior of the cementite lamellae (Fe3C) under shear deformation. In most cases, methods of severe plastic deformation are used as a new route to produce ultrafine grained materials with improved mechanical properties.[3,4,5] In the current study, high-pressure torsion (HPT) is used to obtain clearly defined deformed specimens of a low alloyed steel with a initially pearlitic microstructure. The microstructural evolution was analyzed by different transmission electron microscopy (TEM) techniques. The aim of the study is to characterize the microstructure, especially of the cementite lamellae, resulting from the severe plastic shear deformation. II.
EXPERIMENTAL DETAILS AND MATERIALS
A pearlitic rail steel 260 (UIC 900A) was deformed by HPT to obtain severely deformed material. The chemical composition and the mechanical properties of the material are given in Table I. Details of the HPT technique are given elsewhere.[6,7] The samples for HPT had a diameter of 8 mm and a thickness t of 0.8 mm (0.7 mm after the deformation). These discs where deformed at room temperature under a hydrostatic pressure of 5.7 GPa. The number of turns n was calculated according to Eq. [1] to reach equivF. WETSCHER and R. PIPPAN are with the Erich Schmid Institute for Materials Science, Austrian Academy of Sciences, Leoben, Austria, and the CD-Laboratory for Local Analysis of Deformation and Fracture, Leoben, Austria. Contact e-mail:
[email protected] S. STURM is with the Max Planck Institute for Materials Research, Stuttgart, Germany. F. KAUFFMANN is with Materialpru¨fungsanstalt, University of Stuttgart, Germany. C. SCHEU is with the Department of Physical Metallurgy and Materials Testing, University of Leoben, Austria. G. DEHM is with the Department of Materials Physics, University of Leoben, Austria. Manuscript submitted November 17, 2005. METALLURGICAL AND MATERIALS TRANSACTIONS A
alent von Mises strains eeq of 2 and 8, calculated for a radius r of 3 mm. eeq ¼
2"p"n"r pffiffiffi t" 3
[1]
TEM specimens from the deformed samples as well as from undeformed material were prepared by cutting severalmillimeter-small sections of the deformed material parallel to the torsion axis at a radius of 3 mm. The material was mechanically polished to a final thickness of ;50 mm and further reduced in size to keep its magnetic volume as small as possible (Figure 1). Finally, several electron transparent regions of ;10 3 10 mm2 were made using a Zeiss XP 1540 focused ion beam microscope. The TEM samples were then glued onto a copper grid and subsequently analyzed with a JEOL 2000 FX equipped with a Gatan imaging filter and a VG HB 501 UX scanning TEM. The VG HB 501UX has a cold-field-emission gun and is equipped with a Gatan Enfina system. On the JEOL 2000 FX, bright-field micrographs and elemental distribution maps for carbon were taken at 200 kV. The carbon maps were measured by the three-window method.[8,9] The energy slit width was 20 eV and an acquisition time of 10 seconds was used for each image. With the VG HB 501 UX electron energy loss (EEL) spectroscopy measurements of the Fe-L2,3 edge using the ENFINA system were performed at 100 kV. The EEL spectra were recorded with a dispersion of 0.1 eV and an acquisition time of 20 seconds. The full width at half maximum (FWHM) of the zero loss was smaller than 0.8 eV. The electron energy-loss near-edge-fine structure (ELNES) of the Fe-L2,3 edge was evaluated by fitting the measured spectra after background subtraction with a Gaussian function (see Appendix) and deriving values for the peak areas, peak heights, and FWHM for the Fe-L2,3 edge of the matrix and the carbide phase. For the undeformed sample and the sample deformed up to eeq 5 2, the beam was centered exactly on a well-defined cementite lamellae. Hence, it is possible to distinguish between measurements in the cementite and in the ferritic phase. In the case of the specimen deformed up to eeq 5 8, VOLUME 37A, JUNE 2006—1963
Table I. C 0.76
Chemical Composition of the Pearlitic Steel
Si
Mn
P
S
Cr
0.35
1.0
0.014
0.017
0.04
Rm 900 MPa
The balance is Fe. Data are given in wt pct.
Fig. 1—A section of the HPT sample was cut out from the deformed disk for TEM investigations. Electron transparent windows were thinned into the specimen using a focused ion beam microscope.
it was not possible to determine whether the electron spot was focused on a cementite lamellae or not. Therefore, line scans with a length of approximately 15 nm and a step size smaller than 1 nm were performed over a line that was crossing a deformed cementite lamellae. The beam diameter was nominally 1 nm for these measurements. III.
RESULTS
Fig. 2—Scanning electron microscopy micrographs (using secondary electrons) of (a) the undeformed material and (b) a sample after a deformation of eeq 5 8, in radial direction.
A. Microstructure and Elemental Maps The initial structure of the material consists of 10- to 20-mm-large pearlite colonies of ferrite lamellae ;300 nm and cementite lamellae ;20 nm wide, as can be seen in a scanning electron micrograph (Figure 2(a)). After deformation by HPT of eeq 5 8, pearlite colonies are no longer visible in scanning electron micrographs and a quite uniform lamellar structure with a decreasing lamellae spacing of deformed and broken cementite lamellae and ferrite is present (Figure 2(b)). The cementite fragments have predominately the form of plates with a maximum diameter of 1 mm.[10] All the fragments are now aligned parallel to the shear plane. This change of the microstructure also leads to an enormous increase in the mechanical strength, as can be seen in Figure 3, in terms of the in-situ measured torque to deform the sample (reported elsewhere[10]). At high deformations, the use of TEM is necessary to clearly resolve the microstructure. In Figure 4 a series of bright-field TEM micrographs and the corresponding element maps for carbon are depicted, showing the development of the microstructure as a function of the strain. In the undeformed sample the cementite lamellae with a width of about 20 nm are 1964—VOLUME 37A, JUNE 2006
Fig. 3—In-situ measured torque during the HPT deformation.
straight and continuous (Figures 4(a) and (d)). Figures 4(b) and (e) show a bright-field image and the corresponding carbon elemental map of a sample deformed to eeq 5 2. The cementite lamellae, which start to break up, are clearly resolved in the C-K map. At this degree of deformation the METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 4—TEM bright-field micrographs of (a) the undeformed sample, eeq 5 0, (b) the sample after a deformation of eeq 5 2, and (c) the sample after deformation of eeq 5 8. Corresponding carbon maps of the samples with (d) eeq 5 0, (e) eeq= 2, and (f) eeq 5 8.
values for the Fe-L2,3 peak ratio, the Fe-L2,3 area ratio, the Fe-L2 FWHM, and the Fe-L3 FWHM measured from the undeformed samples were used as a fingerprint and compared to the results of the deformed samples. In Figure 6 all data for the area ratio and the FWHMs are summarized. The mean values as well as the largest and smallest values obtained from the measurement of the initial microstructure are indicated by black and gray lines. It is obvious that only the initial structure of the material and the sample deformed up to eeq 5 2 reveal the characteristic values for the cementite phase. The values from the samples with a deformation of eeq 5 8 are all in the range of the ferritic phase.
Fig. 5—Measured EEL spectra of ferrite and cementite.
crystallographic cementite structure is still preserved, as found by selected area diffraction. Also in the case of the highest deformed sample, the lamellar structure that can be seen in bright-field images corresponds to carbon-enriched areas in the elemental map (Figures 4(c) and (f)). In this sample, the lamellae spacing and the lamellae thickness are markedly decreased down to a minimum value of approximately 25 nm and 2 nm, respectively. B. ELNES Measurements Figure 5 shows the measured EEL spectra of ferrite and cementite of the initial pearlitic material. A change in the Fe-L2,3 area ratio from 3.24 6 0.1 for a-iron to 2.82 6 0.12 for cementite, as well as in the FWHM for both peaks, is observed for the cementite phase compared to ferrite. The METALLURGICAL AND MATERIALS TRANSACTIONS A
IV.
DISCUSSION
The microstructures as can be seen in the bright-field micrographs are very similar to the microstructure resulting from wire drawing of pearlitic steels reported by Embury and Fisher[11] and Langford.[12] The lamellae spacing l in the undeformed microstructure was about 300 nm and decreased in the sample deformed to eeq 5 8 to about 25 nm. This is consistent with the calculated value for l ;23 nm (for n 5 0.55) due to the deformation according to Eq. [2], which indicates that the calculated ‘‘macrostrain’’ according to Eq. [1] is homogeneously distributed in the microstructure. The larger spacing in some areas can be explained by inhomogeneous or ‘‘deck-of-cards’’ deformation.[13] The increase of mechanical strength in terms of the macroscopic tensile stress resulting from the decreasing lamellae spacing can be estimated according to Dollar et al.,[14] by Eq. [3], where l is the interlamellar spacing given in mm: VOLUME 37A, JUNE 2006—1965
Fig. 6—Comparison of the EEL spectroscopy measurements of the starting material and samples with eeq 5 2 and eeq 5 8: (a) area ratios, (b) FWHM for the Fe-L3 peaks, and (c) FWHM for the Fe-L2 peaks. The mean values for ferrite and cementite of the initial microstructure are indicated by black lines, the largest and smallest values are indicated by gray lines.
lnew ¼
l"t 2"r"p"n
sE ¼
12:74 l0:5
[2]
[3]
For l # 25 nm the expected tensile stress is #2300 MPa, which is more than twice the initial strength of #900 MPa. 1966—VOLUME 37A, JUNE 2006
This increase is in good agreement with the measured increase in the microhardness[10] from 2 GPa for the undeformed sample to 4.2 GPa for a sample deformed to eeq 5 8 and the measured increase of the torque during the deformation (Figure 3). The elemental distribution maps of carbon for the samples with different deformation clearly show that the carbon is concentrated in the cementite lamellae. Hence, they confirm that the observed lamellar structures in the highest deformed samples are remains of the original pearlitic structure and the carbon is still concentrated near the remains of the cementite lamellae. The ELNES measurements of the undeformed material demonstrate that the differences of the Fe-L2,3 fine structure between the ferrite phase and the cementite phase in terms of the Fe-L2,3 peak ratio, the Fe-L2,3 area ratio, and the FWHMs are significant and allow us to postulate that a change in the chemical structure has occurred. The reason for that can be seen in the different chemical composition and consequently the different electronic structure.[15,16] No differences were detectable in the distance between the two peaks (measured by both the peak–peak distance and distance between the points of inflection), indicating a similar spinorbit splitting of the 2p states for ferrite and cementite.[17] The results for the sample deformed to eeq 5 2 are quite similar to the undeformed material. No significant deviation for the Fe-L2,3 values from ferrite and cementite were observed, and both phases are well distinguishable. Therefore, it can be concluded that there was no significant change in the chemical composition and structure in the cementite due to a deformation of eeq 5 2. In contrast, the electronic fingerprint for the highest deformed sample (eeq 5 8) lies in the range of the values for the ferritic phase. This means that even in the areas where the EFTEM micrographs showed an enrichment of carbon, this carbon is no longer present as a Fe3C carbide. Hence, at least in the measured areas, the condition of the cementite has changed markedly due to the severe plastic deformation by HPT at room temperature. Similar changes of cementite are especially known for cold-drawn pearlitic wires[18–22] and ball-milled pearlitic powders.[23] It is generally assumed that the cementite dissolves as a result of the plastic deformation. There is less agreement about the whereabouts of the carbon atoms after the dissolution. Saturations of the carbon atoms at dislocations,[22] at grain boundaries,[24] or in a fine-grained martensite[20] are discussed in the literature. The methods used in the literature were Mo¨ssbauer spectroscopy,[22] atom probe field ion microscopy,[19] three-dimensional atom probe methods,[21] and thermomagnetic methods.[24] The most commonly used explanation for the dissolution of cementite due to deformation is that carbon atoms are dragged out of the carbide by crossing dislocations. This might happen because the binding energy of the carbon atom to the dislocation is greater than the binding energy of the carbon atom in the carbide. A more detailed explanation of this process is given by Gavriljuk,[25] for instance. Ivanisenko et al.[24] reported a 40 pct decomposition of cementite due to deformation by HPT for a similar steel as used in this study after eeq 5 36 and a total dissolution after eeq 5 245 measured by a thermomagnetic method. In the current study a decomposition was observed already at METALLURGICAL AND MATERIALS TRANSACTIONS A
Fig. 7—(a) Comparison between a measured spectrum (ferrite, eeq 5 0) and the fit. (b) Plot of the individual functions used for the fit.
much lower strains. The reason for this might be the different experimental setup, especially the large shape change that is avoided in our HPT tool, which permits a defined shearing of the samples. Another reason might be the different techniques used to investigate the actual state of the carbon. In our case we used a technique with high spatial resolution enabling electronic structure measurements down to the nanometer level, while the observations of Ivanisenko et al.[24] were mainly based on integral techniques. The results from Sauvage[26] obtained by 3D atom probe of HPT-deformed pearlitic steel are in good agreement with the present observations. He measured a decrease of the initial carbon concentration of 25 at. pct of the cementite down to 2 to 10 at. pct after 5 turns of HPT, observing similar carbon-rich lamellae instead of the cementite present in the initial microstructure.
V. CONCLUSIONS The current TEM study found marked decreases of both the lamellae spacing and the lamellae thickness in the pearlitic steel due to severe plastic deformation by HPT. In the bright-field micrographs of the severely deformed sample, a lamellar structure was still observable, and the elemental distribution maps showed an enrichment of carbon in these thin lamellae. ELNES measurements were successfully performed at the samples and analyzed by using exponentially modified Gaussian functions. With this method, reliable ratios and peak width were obtained, which act as fingerprints of the iron atoms. The ELNES measurements of the carbon-rich areas in the most severe deformed sample did not show the characteristics of cementite, indicating that cementite is no longer in these areas. This leads to the conclusion that due to the severe plastic deformation by HPT, at least a partial dissolution of the carbon from the cementite has occurred. APPENDIX Fitting of the EELS spectra The measured EELS spectra consist of element specific ionization edges superimposed on a background METALLURGICAL AND MATERIALS TRANSACTIONS A
mainly arising from plural-scattering events of outer shell electrons.[8,17] This background must be removed to analyze the edges. For background subtraction, a power-law approximation: IBackground 5 A " E$r
[4]
was used.[17] The remaining signal was then analyzed with the software Mathematica from Wolfram Research. For the investigated Fe-L2,3, edge transitions occur from the 2p states into unoccupied 3d and 4s states.[17] The transition into the narrow unfilled 3d band leads to ‘‘white lines.’’ These structures are superimposed on a smooth background due to transitions into the extended 4s bands. To describe the white lines as a function of the energy E, two exponentially modified Gaussian functions: " 2 # B2 B1 $ E 1 2 B0 B3 " e 2 " B3 Iwhiteline 5 2 " B3 % " $ # E $ B1 B2 B3 pffiffiffi $ pffiffiffi 1 " Erf [5] jB3 j B2 " 2 B3 2 were used. The parameters of this function are: B0 5 Area of the deconvolved Gaussian, B1 5 Center of the deconvolved Gaussian, B2 5 Width, and B3 5 Distortion. This function is a mathematical convolution of a Gaussian and an exponential function. Due to its capability to describe asymmetric peaks, it is widely used in the field of chromatography. An overview is given by Jeansonne and Foley.[27] This function was used in our study because it gave much better agreement with the measured white lines than the normally used Lorentz function.[28] The smooth background intensity in the threshold region results from the transitions into the 4s band, which is free electron like.[17] Therefore, the density of states is proportional to the square root of energy, and the intensity was fitted with C and E0 as parameters: pffiffiffiffiffiffiffiffiffiffiffiffiffiffi I4s$states ¼ C " E $ E0 [6] The actual fitting was done using a least-squares method. After choosing reasonable starting parameters, pairs of parameters (e.g., B0 for the Fe-L2 peak and for Fe-L3) were calculated separately and used as constants for the next VOLUME 37A, JUNE 2006—1967
fitting step. After calculating all the parameters, the whole procedure was repeated. A third repetition of this gave no significant changes of the values for the parameters. In Figure 7(a) a measured spectrum is compared with the final fit. The separate graphs for the fitted functions are displayed in Figure 7(b). From the obtained functions, all other values, like the peak height, the peak area, and the FWHM for both peaks, as well as the corresponding ratios (Fe-L2,3 peak ratio, Fe-L2,3 area ratio), were calculated:
peak hight ratio 5
area ratio 5
MaximumðIWhiteline;FeL3 Þ MaximumðIWhiteline;FeL2 Þ
B1L3 R110 B1L3 $10
B1L2 R110 B1L2 $10
[6]
IWhiteline;FeL3 dE [7]
IWhiteline;FeL2 dE
ACKNOWLEDGMENTS The authors thank Dr. P. Pointer and Mr. R. Stock from voestAlpine Schienen GmbH for providing the material and their support of the CD-laboratory. The help of Mr. J. Thomas with the STEM measurements is gratefully acknowledged. REFERENCES 1. M. Zelin: Acta Mater., 2005, vol. 50, pp. 4431-47. 2. G. Baumann, H.J. Fecht, and S. Liebelt: Wear, 1996, vol. 191, pp. 13340.
1968—VOLUME 37A, JUNE 2006
3. R.Z. Valiev, R.K. Islamgaliev, and I.V. Alexandrov: Prog. Mater. Sci., 2000, vol. 45, pp. 103-89. 4. R.Z. Valiev and I.V. Alexandrov: Ann. Chim. Sci Mater., 2002, vol. 27, pp. 3-14. 5. V.Y. Gertsman, R. Birringer, R.Z. Valiev, and H. Gleiter: Scripta Metall. Mater., 1994, vol. 30, pp. 229-34. 6. T. Hebesberger, H.P. Stu¨we, A. Vorhauer, F. Wetscher, and R. Pippan: Acta Mater., 2005, vol. 53, pp. 393-402. 7. O. Dimitrov: Ann. Chim. Sci Mater., 2002, vol. 27, pp. 15-24. 8. D.B. Williams and C.B. Carter: Transmission Electron Microscopy: A Textbook for Materials Science, Plenum Press, NY, 1996. 9. F. Hofer, W. Grogger, G. Kothleitner, and P. Wasbichler: Ultramicroscopy, 1997, vol. 67, pp. 83-103. 10. F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan: Mater. Sci. Eng. A, 2004, vol. 387–389, pp. 809-16. 11. J.D. Embury and R.M. Fisher: Acta Metall., 1966, vol. 14, pp. 147-59. 12. G. Langford: Metall. Trans., 1970, vol. 1, pp. 465-77. 13. G. Langford: Metall. Trans., 1977, vol. 8A, pp. 861-75. 14. M. Dollar, I.M. Bernstein, and A.W. Thompson: Acta Metall., 1988, vol. 36, pp. 311-20. 15. R.D. Leapman and L.A. Grunes: Phys. Rev. Lett., 1980, vol. 45, pp. 397-401. 16. R.D. Leapman, L.A. Grunes, and P.L. Fejes: Phys. Rev. B: Condens. Matter Mater. Phys., 1982, vol. 26, pp. 614-35. 17. R.F. Egerton: Electron Energy-Loss Spectroscopy in the Electron Microscope, 2nd ed., Plenum Press, NY, 1996, pp. 269-74. 18. J. Languillaume, G. Kapelski, and B. Baudelet: Acta Mater., 1997, vol. 45, pp. 1201-12. 19. H.G. Read, W.T. Reynolds, Jr., K. Hono, and T. Tarui: Scripta Mater., 1997, vol. 37, pp. 1221-30. 20. K. Hono, M. Ohnuma, M. Murayama, S. Nishida, A. Yoshie, and T. Takahashi: Scripta Mater., 2001, vol. 44, pp. 977-83. 21. F. Danoix, D. Julien, X. Sauvage, and J. Copreaux: Mater. Sci. Eng. A, 1998, vol. 250, pp. 8-13. 22. W.J. Nam, C.M. Bae, S.J. Oh, and S.-J. Kwon: Scripta Mater., 2000, vol. 42, pp. 457-63. 23. M. Umemoto: Mater. Trans. JIM, 2003, vol. 44, pp. 1900-11. 24. Y. Ivanisenko, W. Lojkowski, R.Z. Valiev, and H.-J. Fecht: Acta Mater., 2003, vol. 51, pp. 5555-70. 25. V.G. Gavriljuk: Mater. Sci. Eng. A, 2002, vol. 345, pp. 81-89. 26. X. Sauvage: Mater. Sci. Forum, 2006, vol. 503–504, pp. 433-38. 27. M.S. Jeansonne and J.P. Foley: J. Chromatogr. A, 1992, vol. 594, pp. 1-8. 28. T. Manoubi, M. Tence´, M.G. Walls, and C. Colliex: Microsc. Microanal. Microstruct., 1990, vol. 1, pp. 23-39.
METALLURGICAL AND MATERIALS TRANSACTIONS A
