Study of changes in composition and EELS ionization edges upon Ni4Ti3 precipitation in a NiTi alloy

Micron 37 (2006) 503–507 www.elsevier.com/locate/micron

Study of changes in composition and EELS ionization edges upon Ni4Ti3 precipitation in a NiTi alloy Zhiqing Yang*, Dominique Schryvers Electron Microscopy for Materials Science (EMAT), University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, Belgium Received 15 May 2005; revised 1 August 2005; accepted 9 August 2005

Abstract Concentration gradients surrounding Ni4Ti3 precipitates grown by appropriate annealing in a Ni51Ti49 B2 austenite matrix are investigated by electron energy loss spectroscopy (EELS) and energy filtered transmission electron microscopy (EFTEM). Concentration gradients of w1.0–2.0 at.% in Ni within the surrounding B2 matrix can be detected by both EELS and EFTEM, revealing a Ni depleted zone in the matrix. Besides the concentration gradients, the EELS integrated cross-section of the Ni L2,3 edges for the Ni-depleted region increased slightly, when compared with a matrix region away from the precipitate and not depleted in Ni. q 2005 Elsevier Ltd. All rights reserved. Keywords: NiTi; Composition; EELS; ELNES; EFTEM

1. Introduction Electron energy loss spectroscopy (EELS) in a transmission electron microscope (TEM) has been recognized as a powerful technique to probe microstructures at very high spatial resolution down to the nanometer level, allowing studies of precipitates, internal interfaces, grain boundaries and lattice defects (Keast et al., 2001; Muller et al., 2004). Inelastic scattering in a solid is sensitive to local electronic structure and elemental composition as well as crystallographic structure (Egerton, 1996). The integrated EELS intensity at the edges is proportional to the areal density of the element. The study of the integrated EELS cross-section at the edge can yield information on the total charge around a given element (Pearson et al., 1993), while the details of the electron energy loss near edge structure (ELNES) are related to the local density of states (DOS) above the Fermi level (Keast, et al., 2001; Potapov et al., 2004). NiTi alloys with near-equiatomic composition can exhibit shape memory and superelastic properties resulting from a temperature or stress induced austenite-martensite phase transformation (Otsuka et al., 2005). The behavior * Corresponding author. E-mail address: [email protected] (Z. Yang).

0968-4328/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.micron.2005.08.002

and characteristics of this transformation are strongly influenced by the presence of Ni4Ti3 precipitates in the B2 austenite matrix as obtained by appropriate annealing procedures. The formation of Ni4Ti3 precipitates not only introduces a strain field in the surrounding matrix (Tirry et al., 2005), it also affects the composition of the retained matrix since the precipitates are enriched in Ni with respect to the original material with a near-equiatomic composition (Yang et al., 2005). In this paper the focus is on the changes in local composition and the L2,3 ionization edges of Ni upon precipitation of Ni4Ti3 in a NiTi alloy with a nominal Ni composition of 51 at.%.

2. Experiments A binary Ni-rich NiTi alloy with a nominal composition of 51 at% Ni was used in the present study. Samples containing precipitates were prepared by first annealing at 950 8C for 1 h followed by water quenching and then aging for 4 h at 500 8C in vacuum. TEM specimens were prepared by mechanical grinding followed by twin-jet electropolishing. Solution annealing resulted in a homogenous large grain microstructure without formation of Ni4Ti3 precipitates. An aging treatment of 4 h at 500 8C resulted in a microstructure with a largely homogeneous distribution of Ni4Ti3 precipitates (Yang et al., 2005).

504

Z. Yang, D. Schryvers / Micron 37 (2006) 503–507

EELS and EFTEM experiments were carried out on an ultratwin Phillips CM30 field emission TEM equipped with a GIF2000 post column energy filter. Before each TEM session, the specimen was cleaned by ion mill to remove the possible oxide layer on the surfaces, and then plasma cleaning was carried out to reduce the contamination due to electron beam irradiation. The habit plane of the precipitates is parallel to the {111}B2 planes, and eight precipitate variants are possible, with an orientation relationship such  B2 // [1 0 0]H (Tadaki et al., as: (1 1 1)B2 // (0 0 1)H; [32 1] 1986). When acquiring EELS spectra and EFTEM images, a  B2 is chosen in zone orientation slightly off-axis from [101] order to have a minimal overlap between the lens shaped precipitate and the matrix and to reduce diffraction effects. EELS spectra are collected in diffraction mode with a camera length of 195 mm and an entrance aperture to the GIF system of 2 mm, which corresponds to a collection semi-angle of 3.4 mrad. EFTEM images were acquired using the three-window method. The post-edge window was positioned right at the threshold of the L3,2 edges of Ni and Ti. Relative drift between successive images was corrected by a standard cross-correlation technique when computing the elemental intensity map. Zero loss spectra were recorded both before and after the acquisitions of the core loss spectra to monitor the thickness of the sample. If a significant increase of the thickness after the core loss acquisitions was measured, indicating that the specimen was not clean enough, it should be cleaned by prolonged plasma cleaning.

from well-defined standards, which is more accurate than the theoretically calculated results. Additionally, the accuracy of the calculated cross-sections is often uncertain within small energy windows due to white lines at the edge onset for transition metals. However, small energy windows are often required for EFTEM elemental mapping to reduce effects of chromatic aberration. Therefore, in the present study, the relative quantification method, for which no calculation of cross-section for each element is required, will be utilized. The stoichiometric proportion for the precipitate is considered to be NNi/NTi Z4/3, and can be used as a reference standard. The k-factor can thus be determined from the precipitates in the samples and the elemental concentration for each element in the matrix can then be deduced from the measured Ni/Ti atomic ratio. This approach has the advantage that diffraction and thickness variation effects are largely eliminated as all references and measurements are obtained from regions very close to one another (Hofer et al., 1997). Fig. 1 shows a TEM image (a) and the corresponding EELS analysis results (b) on a precipitate and the neighboring matrix. The dark spots on the TEM image are contamination cones of carbon because of prolonged nanoprobe electron illumination and can be used as markers for the probed positions. The relative intensity of the plasmon peak indicates that the thickness is in the range of 0.60–0.75 times the inelastic scattering mean free path, which is suitable for (a)

3. Results and discussion The number of atoms per unit area NA for an element A in the material can be calculated from the EELS spectra by (Egerton, 1996), IA ðb; DÞ Iz ðb; DÞ$sA ðb; DÞ

(1)

where IA(b, D) is the measured ionization edge intensity integrated within an energy range D and inside a collection semi-angle b, Iz(b, D) is the intensity of a window of equal b and D containing the zero loss, and sA(b, D) is the partial ionization cross-section. Eq. (1) can also be used in EFTEM analysis. In the case of the binary material the atomic ratio NA/NB is thus determined as NA I ðb; DA Þ$sB ðb; DB Þ I ðb; DA Þ Z kAB $ A Z A IB ðb; DB Þ$sA ðb; DA Þ IB ðb; DB Þ NB

1.8

(b) Ni, at%

60

1.6 51

1.4

50 interface

NA Z

40

(2)

where kABZsB(b, DB)/sA(b, DA) is the so-called k-factor. Calculations based on Eqs. (1) and (2) are referred to as absolute and relative quantification, respectively. In the case of absolute quantification, the key parameter is the partial ionization cross-section. The cross-section can principally be calculated based on the theory of atomic physics, which provides rough quantification results, especially for heavier elements; or it can be derived from EELS spectra measured

1.2 1.0 0.8

Ni /Ti atomic ratio

30

0.6 –50

0

50

100 150 200 Position (nm)

250

300

Fig. 1. (a) TEM image with contamination dots indicating the place of measurement in and nearby a precipitate, (b) EELS results showing the Ni/Ti atomic ratio, and averaged Ni concentrations. The dashed line shows the Ni concentration of 51 at.%.

Z. Yang, D. Schryvers / Micron 37 (2006) 503–507

EELS analysis. Actually, the background before the Ti L2,3 edges could be fitted very well, demonstrating that the effect of the carbon contamination could be neglected. Moreover, using the same type of analysis no significant difference in thickness between the matrix and the precipitate is observed. Still, a foil thickness of less than 0.5 times the inelastic scattering mean free path would have been better, but bending of such thin sample edges due to the existing strong strain make them unsuitable for EELS and EFTEM measurements. Since the thickness of the contamination layers on the precipitate and the other probed positions was almost the same, and the precipitate was used as a reference to quantify the atomic concentration of the surrounding matrix, the contamination did not affect the results of the Ni/Ti atomic ratio for the surrounding matrix. The reported data in Fig. 1(b) are averaged from 3 measurements per single distance from the precipitate-matrix interface, thus each time containing three measurements. The calculated Ni/Ti atomic ratio, with a minimum value of 0.95, reveals a region depleted in Ni within a range of about 150 nm from the precipitate-matrix interface and which should be compared with the nominal value of 1.04 for the original Ni51Ti49 matrix. The standard deviation for the measurements inside the precipitate is about 1.0%, (although an absolute accuracy (a)

Ni

for one measurement generally could be considerably worse than 1.0% in an absolute quantification by Eq. (1)), which shows the advantage of the relative quantification method using the precipitate with a known atomic concentration as a well-defined standard. Additionally, the Ni concentration is calculated and shown in Fig. 1(b). The standard error for the measurements in the adjacent matrix is more than three times larger than in the case of the precipitate, which indicates a certain composition variation among the three averaged positions. To check the balance of Ni after the precipitation of Ni4Ti3, a calculation was carried out based on the size of the precipitate and symmetric Ni-depleted regions in the adjacent matrix on both sides of the precipitate. For the purpose of simplicity, a mean value of Ni concentration in the depleted region was calculated without considering the possible gradient. The calculated Ni concentration is 49.5 at.%, which is quite close to the measured value 49.6 at.%. Therefore, it can be concluded that all Ni required to form the Ni4Ti3 precipitate is obtained from the matrix region surrounding it, which also confirms the high precision of the measurements and quantification method used. Fig. 2 shows the EFTEM elemental intensity map for Ni (a), Ti (b), Ni/Ti ratio map (c) for a region with two precipitates on crossing {111} planes of the B2 matrix (b)

50nm (c)

505

Ti

50nm

Ni/Ti

(d) 1.4

Ni/Ti atomic ratio

1.3 1.2 1.1

1.04

1.0

(

0.9 0

50

100 150 Position (nm)

200

250

Fig. 2. (a) Ni map, (b) Ti map, (c) Ni/Ti atomic ratio map, (d) profile for the rectangular region in (c). The dashed line shows the atomic ratio for Ni51Ti49.

506

Z. Yang, D. Schryvers / Micron 37 (2006) 503–507

and the profile (d) for a rectangular region across one precipitate in (c). The different contrast between the two precipitates in the elemental intensity maps (a) and (b) is mainly the result of the diffraction effect. The effect of diffraction contrast is eliminated in the ratio map (c). A statistical analysis on the atomic ratio map yields a Ni/Ti ratio of 1.33G0.03 for the precipitates. Profile analysis of the selected rectangular region indicates that there are Nidepleted regions on both sides of the precipitate. The left side (with a mean Ni/Ti atomic ratio of 0.97 in a region stretching out over 60 nm) is more depleted in Ni than the right side (with a mean Ni/Ti atomic ratio of 0.98 over a region of 40 nm). The different gradients on both sides of the precipitate can be understood by the fact that the measured region on the left side of the precipitate could more strongly be depleted by the formation of both precipitates, while the right side was depleted by the formation of only one. Fig. 3 shows the experimental EELS spectra for Ni L3,2 edges from the Ni-depleted region and a matrix region away from the precipitate and not depleted in Ni. For the purpose of quantitative comparison, the spectra were normalized to a window of 10 eV located between 890 and 900 eV after background subtraction and deconvolution (Pearson et al., 1993; Potapov et al., 2003). The intensity of the first peak in the ELNES from the Ni-depleted region is slightly higher than that from a matrix region away from the precipitate and not depleted in Ni. This indicates that the unoccupied DOS of Ni in the Ni-depleted region is slightly larger than those in the region not depleted in Ni. In an earlier study it was demonstrated by EELS that a small amount of electrons (!0.1 electron/atom) was transferred from Ni to Ti when forming NiTi compounds, by comparing the ELNES from alloys and pure Ni (Potapov et al., 2001). The depletion of

Ni could thus be one of the factors resulting in more electron transfer from Ni to Ti in the Ni depleted region, which increased the unoccupied DOS above the Fermi level of Ni. However, besides the difference in composition, highresolution electron microscopy investigations demonstrated that there was an expansion strain field in the B2 matrix surrounding the precipitates due to lattice mismatch between the two phases (Tirry et al., 2005). Preliminary ELNES first principle calculations indicate that the same effect of an increased first peak in the Ni L2,3 edge can also be due to a lattice expansion of about 2% (Kulkova, 2005). Moreover, it has been claimed that the intensity of the Ni L2,3 white lines could be increased upon a martensitic transformation in a NiTi alloy (Murakami et al., 1998) which also clearly changes the lattice structure. To ascertain more clearly the changes in ELNES of Ni upon the Ni4Ti3 precipitation, further first principle calculations are now under way.

4. Conclusions The present work shows that the Ni4Ti3 precipitates influence the concentration in nanoscale regions of the surrounding matrix. Small concentration gradients with Nidepleted zones in the matrix near the precipitate-matrix interface can be detected by EELS and EFTEM. The size of the Ni-depleted zones ranges from tens of nm to more than 100 nm depending on the size of the given precipitate and distances between neighboring precipitates. The intensity of Ni L2,3 ELNES in the Ni-depleted regions in the matrix surrounding the precipitates is higher than that in regions far away from the precipitates and not depleted in Ni, for which further study is required to distinguish the effect of concentration variation and lattice expansion.

0.10 Nominal composition and unstrained Ni-depleted region and strained

Normalized intensity (a.u.)

0.08

Acknowledgements ZQ Yang is supported by the GOA project ‘Characterization of nanostructures by means of advanced EELS and EFTEM’ of the University of Antwerp.

0.06

0.04

References 0.02

0.00 850

860

870

880

Energy loss (eV) Fig. 3. EELS spectra from the Ni-depleted and strained region and matrix region away from the precipitate and not depleted in Ni.

Egerton, R.F., 1996. Electron Energy-Loss Spectroscopy in the Electron Microscope. Plenum Press, New York. Hofer, F., Grogger, W., Kothleitner, G., Warbichler, P., 1997. Quantitative analysis of EFTEM elemental distributrion images. Ultramicroscopy 67, 83–103. Keast, V.J., Scott, A.J., Brydson, R., Williams, D.B., Bruley, J., 2001. Electron energy-loss near-edge structure - a tool for the investigation of electronic structure on the nanometre scale. Journal of Microscopy 203, 135–175. Kulkova, S., 2005. Private Communication 2005.

Z. Yang, D. Schryvers / Micron 37 (2006) 503–507 Muller, D.A., Nakagawa, N., Ohtomo, A., Grazul, J.L., Hwang, H.Y., 2004. Atomic-scale imaging of nanoengineered oxygen vacancy in SrTiO3. Nature 430, 657–661. Murakami, Y., Shindo, D., Kazuhiro, O., Tesuo, O., 1998. Electronic structure changes associated with a martensitic transformation in a Ti50Ni48Fe2 alloy studied by electron energy loss spectroscopy. Journal of Electron Microscopy 47, 301–309. Otsuka, K., Ren, X., 2005. Physical metallurgy of NiTi-based shape memory alloys. Progress in Materials Science 50, 511–678. Pearson, D.H., Ahn, C.C., Fultz, B., 1993. White lines and d-electron occupancies fro the 3d and 4d transition metals. Physical Review B 47, 8471–8478. Potapov, P.L., Kulkova, S.E., Schryvers, D., Verbeeck, J., 2001. Structrual and chemical effects on EELS L2,3 ionization edges in Ni-based intermetallic compounds. Physical Review B 64, 184110.

507

Potapov, P.L., Kulkova, S.E., Schryvers, D., 2003. Study on changes in L23 EELS ionization edges upon formation of Ni based intermetallic compounds. Journal of Microscopy 210, 102–109. Potapov, P.L., Jorssen, K., Schryvers, D., Lamoen, D., 2004. Effect of charge trasfer on EELS integrated cross sections in Mn and Ti oxides. Physical Review B 70, 045106. Tadaki, T., Nakata, Y., Shimizu, K., Otsuka, K., 1986. Crystal structure, composition and morphology of a precipitate in an aged Ti-51at%Ni shape memory alloy. Transactions of the Japan Institute of Metals 27, 731–740. Tirry, W., Schryvers, D., 2005. Quantitative determination of strain fields surrounding Ni4Ti3 precipitates in NiTi. Acta Materialia 53, 1041–1050. Yang, Z.Q., Tirry, W., Schryvers, D., 2005. Analytical TEM investigations on concentration gradients surrounding Ni4Ti3 precipitates in Ni-Ti shape memory material. Scripta Materialia 52, 1129–1134.