Materials Science and Engineering A 445–446 (2007) 31–39

Electrodeposited nanocrystalline Co–P alloys: Microstructural characterization and thermal stability M. da Silva a , C. Wille b , U. Klement a,∗ , P. Choi b,1 , T. Al-Kassab b a

Department of Materials and Manufacturing Technology, Chalmers University of Technology, H¨orsalsv¨agen 7, SE-412 96 G¨oteborg, Sweden b Institut f¨ ur Materialphysik, Universit¨at G¨ottingen, Friedrich-Hund-Platz 1, 37073 G¨ottingen, Germany Received 4 April 2006; received in revised form 28 June 2006; accepted 9 July 2006

Abstract Nanocrystalline Co–P has a significantly increased thermal stability compared to other nanocrystalline Ni- and Co-based electrodeposits. By combining observations from transmission electron microscopy in situ annealing with differential scanning calorimetry and tomographic atom probe measurements, the microstructural development of nanocrystalline Co–1.1 at.% P and Co–3.2 at.% P is described. In the as-prepared state, both alloys consist of 10 nm sized hcp-Co grains, and the P-distribution is already inhomogeneous. Upon annealing, P-atoms segregate to the grain boundaries, and grain growth takes place. Initial grain growth is abnormal in both electrodeposits, but occurs slower and more homogeneously in Co–3.2 at.% P. When P-saturation of the grain boundaries is reached, P-rich precipitates form accompanied by rapid (normal) grain growth. Due to the higher initial P-concentration in Co–3.2 at.% P, saturation of P and precipitation occurs earlier leading to a slightly lower thermal stability. Apart from the different P-contents, the effect of impurities, allotropic phase transformation of Co and texture is discussed. © 2006 Elsevier B.V. All rights reserved. Keywords: Nanocrystalline material; Electrodeposit; Co–P alloys; Microstructural characterization; Segregation; Grain growth

1. Introduction Due to their large interfacial volume fraction, electrodeposited nanocrystalline materials commonly have exceptional mechanical, magnetic, electrical, and corrosive properties. High hardness, corrosion resistance, and wear resistance make nanocrystalline materials strong contenders for protective coating applications, replacing hard chromium layers which are harmful to the environment [1]. Furthermore, Co-based nanocrystalline alloys have higher saturation magnetization compared to Ni-based nanocrystalline alloys, which also makes them promising for soft magnetic applications [2]. To retain the exceptional macroscopic properties of nanocrystalline materials at elevated temperatures, it is essential to preserve a stable nanocrystalline microstructure upon annealing. However, in some nanocrystalline materials abnormal grain

∗

Corresponding author. Tel.: +46 31 7721264; fax: +46 31 7721262. E-mail address: [email protected] (U. Klement). 1 Present address: Korea Institute of Science and Technology, Nano-Materials Research Center, P.O. Box 131, Cheongryang, Seoul 130-650, South Korea. 0921-5093/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2006.07.069

growth occurs at temperatures much lower than expected for normal grain growth. This results in a bimodal microstructure, i.e. a microstructure consisting of grown grains in a nanocrystalline matrix. It has been shown that the microstructure of pure nanocrystalline electrodeposits such as Co and Ni becomes unstable at quite low temperatures. For example, electrodeposited nanocrystalline Co, with an initial grain size of 20 nm, is stable up to 423 K [3,4], and in nanocrystalline Ni abnormal grain growth sets in at temperatures as low as 353 K [5]. The thermal stability of nanocrystalline Ni has been raised up to 513 K by adding solutes such as P [6]. A similar improvement of the thermal stability can be expected for nanocrystalline Co when P is added. Additionally, the allotropic phase transformation (hcp Co → fcc Co) has to be considered as it may affect the microstructural development upon annealing. A careful analysis of the initial microstructure is required since thermal stability of the nanocrystalline electrodeposits strongly depends on initial grain size, texture, presence of solutes and second phases, etc. In this paper, the initial microstructures and their development upon annealing of Co–1.1 at.% P and Co–3.2 at.% P are described based on in situ annealing experiments in the transmission electron microscope (TEM) and by

32

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

tomographic atom probe (TAP) analysis. The results are discussed and compared in order to explain the effect of P-content on the thermal stability of Co. Complementary experimental techniques such as differential scanning calorimetry (DSC), Xray diffraction (XRD), and field ion microscopy (FIM) have been applied. 2. Experimental Electrodeposited nanocrystalline Co–P material was provided by Integran Technologies Inc., Toronto, Canada. The electrodeposition technique is described elsewhere [7]. Fully dense material is supplied in sheet form with a thickness of 60 ␮m for the low P-content electrodeposit and 200 ␮m for the high P-content electrodeposit. The average P-content in the materials was determined by electron probe micro analysis (EPMA) and impurity sulphur (S) and carbon (C) contents were measured by the combustion LECO technique. XRD measurements were performed with a Philips X’Pert MPD X-ray diffractometer using a Co K␣ beam with a wavelength of λ = 0.179 nm. For TEM investigations, 3 mm diameter disks were punched from the material and prepared by usual dimple grinding and ion milling. In the case of the thicker Co–3.2 at.% P material, the sheet thickness was reduced to about 60 ␮m with a Buehler Minimet grinder-polisher prior to dimple grinding. Ion milling was carried out in a Gatan precision ion polishing system using an angle of incidence of 4◦ on each side and a beam energy of 4 keV. TEM investigations were performed using a Zeiss 912 Omega microscope with an inbuilt energy filter, operating at an acceleration voltage of 120 kV. A Gatan double-tilt heatingholder with a tantalum furnace, which can achieve heating rates up to 100 K/min was used for the in situ annealing experiments. DSC measurements were performed in a Perkin-Elmer DSC 7 with a heating rate of 10 K/min. Two consecutive runs were performed on each sample and the resulting curves subtracted from each other in order to separate reversible transformations (e.g. Curie temperature) from irreversible transformations such as grain growth. Samples for FIM and TAP analysis were produced by cutting strips of 200 ␮m in width and 10 mm in length from the sheets by spark erosion, followed by mechanical polishing to obtain square-shaped cross-sections. These strips were then electropolished to tips with a radius of curvature between 10 and 100 nm by applying a direct voltage of 3 V between the tip and a platinum electrode in a solution of 20% perchloric acid and 80% 2-butoxyethanol. Final sharpening of annealed Co–3.2 at.% P tips was performed by ion milling with a dual beam focused ion beam. Field ion micrographs were recorded at tip temperatures between 20 and 75 K using Ne as imaging gas with a partial pressure of 5 × 10−3 Pa. For TAP analysis of as-prepared Co–1.1 at.% P a pulse rate of 1000 Hz, pulse fraction of UP /UDC = 20%, and temperature of 60 K were chosen as acquisition parameters. Due to the higher grain boundary brittleness of Co–3.2 at.% P, the pulse rate and pulse fraction had to be adjusted to 2000 Hz and 17.5%, respectively, in order to lower the stress applied to the tip. Further details about the atom probe technique are described elsewhere [8,9].

3. Results 3.1. As-prepared state 3.1.1. Solute and impurity content EPMA was used to determine the P-content in the two materials and to confirm a homogeneous P-distribution throughout the thickness and along the faces of the electrodeposits. Various measurements were performed on the cross-sections of the material sheets to measure if any concentration gradients exist, owing to for example composition fluctuations during the electrodeposition process. Using a step size of a few ␮m, a homogeneous concentration of P was measured through the thickness and along the faces of the electrodeposits. The Pcontents in the two electrodeposits amount to 1.1 ± 0.1 at.% P and 3.2 ± 0.1 at.% P, respectively. Impurity contents were determined by the sulphur/carbon combustion LECO analysis technique with an accuracy of ≤0.002 wt.%. In Co–1.1 at.% P the impurity contents are 0.085 wt.% S and 0.052 wt.% C, compared to <0.001 wt.% S and 0.004 wt.% C in Co–3.2 at.% P. 3.1.2. Grain size Fig. 1a shows the zero-loss bright field image of as-prepared Co–1.1 at.% P. The microstructure consists of grains in the nanocrystalline range with a narrow grain size distribution. Since the microstructure of Co–3.2 at.% P is quite similar, it is not shown here. The mean grain size was determined from TEM dark field images. Measuring 337 grains in total, a mean grain size of 13.6 ± 3.3 nm is obtained for Co–1.1 at.% P; the corresponding grain size histogram is shown in Fig. 1b. For asprepared Co–3.2 at.% P, a mean grain size of 11.9 ± 3.2 nm was determined by measuring 417 grains in total (histogram given in Fig. 1c). Generally, grain size measurements in Co-based nanocrystalline materials were found to be more difficult than in Ni-based materials owing to the high density of stacking faults in the grains. This characteristic appearance of nanocrystalline Co has also been observed by other authors [3,4,10]. During field evaporation in the FIM, grain sizes are observed that are in good agreement with the ones obtained by TEM. In FIM-images of asprepared Co–1.1 at.% P and Co–3.2 at.% P, regions can be found in which ring patterns are visible, each being an image of single nanocrystalline grains. By increasing the voltage applied at the FIM tip, a continuous field evaporation process is induced during which ring patterns with diameters up to 10 nm are observed. 3.1.3. Phosphorous distribution The spatial distribution of P-atoms in the microstructure, playing a key role for the microstructural development upon annealing, was determined by means of TAP. Fig. 2a illustrates the color-coded P-concentration in a cross-sectional plane of the analyzed volume in as-prepared Co–1.1 at.% P. Blue color (grayscale: white) represents areas with 0 at.% P and red color (grayscale: black) represents areas with a P-content higher than 6 at.%. Areas containing virtually no P are surrounded by P-rich areas in a net-like structure as shown in the sketch of the grain morphology in Fig. 2b. It is worth mentioning that some of the P-free areas are not imaged in their full extension in the

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

33

Fig. 1. (a) TEM zero-loss bright field image of as-prepared Co–1.1 at.% P, and grain size histograms for (b) Co–1.1 at.% P; and (c) Co–3.2 at.% P.

cross-section of the analyzed volume, making the grains look smaller than they really are. P-atoms are clearly inhomogeneously distributed in the analyzed volume, which can be verified by the concentration profile shown in Fig. 2c. Several P-peaks clearly exceed the 2σ-level (dashed line). The peak heights are around ∼5 at.% P, and occur in characteristic distances ranging from 6.5 to 8.5 nm. The characteristic distances between the high P-content peaks in the concentration depth profile in Fig. 2c, and the extension of the P-free areas in the analyzed volume in Figs. 2a and b correspond well with the measured sizes of the Co-grains. The average detected P-concentration is marked with a solid horizontal line in Fig. 2c, and is somewhat higher than previously measured by EPMA (1.5 at.% compared with 1.1 at.% P). The P-distribution in as-prepared Co–3.2 at.% P is also inhomogeneous. Enrichments with up to ∼25 at.% P are found in the analyzed volumes of as-prepared Co–3.2 at.% P, and the Gibbsian P-excess in the grain boundaries is estimated to (5.3 ± 2.6) × 108 m−2 . This value is comparable with the excess

value estimated for Co–1.1 at.% P annealed for 1 h at 673 K [11]. In Fig. 3a the color-coded cross-sections show the distribution of P-atoms in the material. To demonstrate that the P-enriched and P-depleted zones are three dimensional, two perpendicular cross-sections in the analyzed volume are given. In Fig. 3b and c, ladder diagrams show the total amount of P-atoms accumulated along the cylinder marked in Fig. 3a passing through enrichments of P-atoms. From the slope of the lines in the ladder diagrams, the local P-content can be estimated. In Fig. 3b and c, the Co matrix is found to contain ∼5 at.% P. Furthermore, P-contents of up 16 and 26 at.% are observed, which is consistent with values given by concentration profiles of Co–3.2 at.% P. This is exceptionally high compared to the ∼6 at.% P in as-prepared Co–1.1 at.% P (Fig. 2a). 3.2. Annealed material A combination of DSC, TEM in situ annealing, and TAP analysis of ex situ annealed samples is applied to investigate

34

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

Fig. 2. (a) Cross-section of the analyzed volume of as-prepared Co–1.1 at.% P (front view), and (b) sketch of the corresponding grain morphology. (c) Concentration depth profile of P in as-prepared Co–1.1 at.% P. The solid horizontal line marks the mean P-concentration detected by TAP, and the dashed line marks the upper 2σ-level.

the thermal stability of the Co-P electrodeposits. DSC is used to obtain heat-release temperatures, which indicate the occurrence of microstructural changes such as grain growth. In Fig. 4, the DSC-curves of Co–1.1 at.% P and Co–3.2 at.% P are shown. A

main heat-release peak is observed for both materials, i.e. at 753 K for Co–1.1 at.% P and at 723 K for Co–3.2 at.% P. In the DSC-curve of Co–1.1 at.% P a pre-peak is visible, i.e. a low energy exotherm, setting in at 733 K. The lower onset tempera-

Fig. 3. (a) P-concentration in as-prepared Co–3.2 at.% P illustrated with two perpendicular cross-sections in the analyzed volume, (b and c) ladder diagrams corresponding to the two enrichments of P-atoms marked with a cylinder in (a).

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

Fig. 4. DSC-curves of Co–1.1 at.% P and Co–3.2 at.% P; heating rate 10 K/min.

ture for heat-release in Co–3.2 at.% P (713 K compared to 733 K in Co–1.1 at.% P) suggests that this alloy has a lower thermal stability. 3.2.1. Allotropic phase transformation In the as-prepared Co–1.1 at.% P material, all peaks are assigned to the hcp-Co phase, and the increased intensity of the hcp (0 0 0 2)-peak hints at a preferential orientation of the close-packed (0 0 0 1)-planes parallel to the sample surface [11]. The (2 0 0)-peak of fcc-Co is detected at 733 and 753 K, indicating that the allotropic phase transformation from hcp-Co to fcc-Co sets in at a temperature between 673 and 733 K in this material. In Fig. 5, the X-ray diffractograms of as-prepared and

Fig. 5. X-ray diffractograms of as-prepared and annealed Co–3.2 at.% P for 1 h at 573, 673, 723, and 753 K. The powder standard intensities of hcp-Co and fcc-Co are indicated with vertical lines.

35

annealed Co–3.2 at.% P for 1 h at 573, 673, 723, and 753 K are shown. In the as-prepared state, all peaks have intensities corresponding to those of the standard hcp-Co powder, indicating a random orientation of hcp grains. XRD measurements were performed on both sides of the sheets, i.e. on the side facing the substrate during electrodeposition and on the upper side of the electrodeposit. Different results were obtained: on the upper side, a random texture was obtained as seen in Fig. 5, whereas on the substrate side the intensity of the (1 0 −1 0)-peak is higher than expected from the Co powder reference. This is the case as first grains deposited on a substrate often have a preferential orientation, resulting in a pronounced texture [12]. However, as the thickness of the electrodeposit increases, texturing decreases and finally results in random grain orientations. By comparing the diffractograms, it is seen that the fcc-Co(2 0 0)-peak appears after annealing for 1 h at 723 K, and becomes more prominent after annealing at higher temperature. This indicates that the allotropic phase transformation from hcp-Co to fcc-Co sets in between 673 and 723 K in Co–3.2 at.% P. 3.2.2. Grain growth TEM in situ annealing experiments are performed in successive steps. Holding temperatures for isothermal annealing are chosen based on heat-release temperatures in the DSC curves. Figs. 6 and 7 show the microstructural developments as observed by TEM in situ annealing of Co–1.1 at.% P and Co–3.2 at.% P, respectively. In Co–1.1 at.% P, annealing for 30 min at 713 K results in a microstructure virtually unchanged compared to the as-prepared state (compare Fig. 6a with Fig. 1a), whereas holding for 25 min at 733 K leads to first grown grains (Fig. 6b). Annealing for 15 min at 753 K yields a microstructure which mostly consists of grown grains with diameters up to 400 nm. Fig. 7a–c shows TEM micrographs of annealed Co–3.2 at.% P at the temperatures 673, 713, and 753 K. No grain growth is observed at 673 K (Fig. 7a), and annealing for 30 min at 713 K yields a microstructure consisting of a few slightly grown grains in a still nanocrystalline matrix (Fig. 7b). Annealing for 30 min at 753 K yields a microstructure with grown grains of diameters up to 200 nm (Fig. 7c). Even though initial grain growth also occurs abnormally in Co–3.2 at.% P, it is slower compared to Co–1.1 at.% P. 3.2.3. Phosphorous distribution TAP analysis of annealed Co–1.1 at.% P for 1 h at 673, 733, and 753 K show that P-atoms segregate to the grain boundaries upon annealing [11]. The grain boundary P-excess is higher after annealing at 673 K compared to the as-prepared state, and increases further at 733 K [11]. At 753 K, precipitates of both the equilibrium Co2 P-phase and the non-equilibrium CoP-phase are found in the material, and the grain boundary P-excess is significantly lowered [11]. Performing TAP analysis on Co–3.2 at.% P is more difficult compared to Co–1.1 at.% P due to the increased grain boundary brittleness. Best results are obtained when final tip sharpening is performed by use of the dual beam FIB technique. In addition, a higher pulse rate (2000 Hz) and lower pulse fraction (17.5%) have to be chosen to avoid tip rupture (flash) during TAP analysis. However, these acquisition parameters

36

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

Fig. 6. TEM micrographs of in situ annealed Co–1.1 at.% P for (a) 30 min at 713 K; (b) 25 min at 733 K; and (c) 15 min at 753 K.

affect the field evaporation rate of the elements present in the analyzed material. Due to the lower evaporation field strength of Co compared to P, the detected average P-concentration by TAP is higher than expected. Moreover, the higher pulse rate and in particular the lower pulse fraction used for TAP analysis contribute to an even higher preferential Co evaporation (also between the pulses) during acquisition. Fig. 8a shows two crosssectional planes in the analyzed volume of a sample annealed for 1 h at 673 K. As can be seen, a precipitate with a P-content of around 50 at.% is included in the analyzed volume. Fig. 8b shows the ladder diagram corresponding to the cylinder marked in Fig. 8a, where it is seen that the P-content in the precipi-

tate is ∼50 at.%. Only CoP-precipitates are found in the TAP measurements of annealed Co–3.2 at.% P. 4. Discussion The mean P-concentrations measured by EPMA in the asprepared Co-P electrodeposits are 1.1 at.% P and 3.2 at.% P, respectively, but higher values are given by TAP (1.5 at.% P and 6.9 at.% P, respectively). Energy dispersive X-ray spectrometry (EDX) measurements in the TEM verify the EPMA results, hence the high P-concentration detected by TAP is attributed to the preferential field evaporation of Co (especially

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

37

Fig. 7. TEM micrographs of in situ annealed Co–3.2 at.% P for (a) 30 min at 673 K; (b) 30 min at 713 K; and (c) 30 min at 753 K.

for Co–3.2 at.% P due to the lower pulse fraction used). Regardless the higher absolute P-concentration values, the distribution of P in the materials can be evaluated. As demonstrated in Figs. 2 and 3, the P-distribution in as-prepared Co–1.1 at.% P and Co–3.2 at.% P is inhomogeneous. The characteristic distances between the high P-content peaks in the concentration depth profiles (Fig. 2b for Co–1.1 at.% P—not shown for Co–3.2 at.% P), and the extension of the P-free areas in the analyzed volumes (Figs. 2a and 3a) correspond well with the measured sizes of the Co-grains. A similar enrichment of P at the grain boundaries has been found by use of TAP analysis in as-prepared chemically deposited nanocrystalline Ni–P alloys [13]. Combining the

observations of the morphology of P-enriched and P-free zones with the measured grain size, it is concluded that the P-free zones are hcp-Co grains and that P is mainly situated in the grain boundaries. TAP measurements show that the grain boundary P-content is significantly higher in as-prepared Co–3.2 at.% P, suggesting that the grain boundaries in as-prepared Co–1.1 at.% P are not saturated. Nanocrystalline Co–P has a significantly increased thermal stability compared to other investigated nanocrystalline electrodeposits such as Ni [4,5,14], Co [3,4], Ni–P [13], Ni–Fe [14,15], and Ni–Co [4]. DSC-curves of these materials have the same qualitative appearance, but the low energy exotherm

38

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

Fig. 8. (a) P-concentration in annealed Co–3.2 at.% P for 1 h at 673 K, and (b) ladder diagram corresponding to the cylinder marked in (a).

is generally more pronounced and extends over a wider temperature range compared to Co–P. TEM in situ annealing experiments of Co–1.1 at.% P show that the nanostructure is stable up to 733 K (onset temperature of the low energy exotherm in the DSC-curve), at which abnormal grain growth sets in. Above the main heat-release peak temperature (753 K) normal grain growth takes place. These results are consistent with previous observations in nanocrystalline Ni, Co, Ni–Fe, and Ni–P [3–5,13–15]. TAP analysis of annealed Co–1.1 at.% P shows that the grain boundary P-concentration increases. When saturation of the grain boundaries is reached, i.e. at ∼753 K, P-rich precipitates are formed causing the grain boundary P-excess to decrease. According to the equilibrium phase diagram of cobalt–phosphorous, the CoP-phase is a non-equilibrium phase at this alloy composition [16]. However, as the local P-concentration in the grain boundaries is much higher than the mean alloy concentration, precipitation of both Co2 P and CoP occurs. The presence of enrichments of P-atoms with up to ∼25 at.% in as-prepared Co–3.2 at.% P, indicates that the grain boundaries are already saturated in the as-prepared state. Therefore, it is not surprising that precipitation and grain growth set in at lower temperatures compared to the alloy with lower P-content. As seen in the DSC curves (Fig. 4), the heat release sets in at a slightly lower temperature for the high P-content alloy, which is consistent with the TEM in situ annealing observations (Fig. 7). In spite of the absence of a low energy exotherm in the DSCcurve of Co–3.2 at.% P, the microstructural development in this alloy is similar to that in Co–1.1 at.% P. Initial grain growth is

abnormal, but from Figs. 6 and 7 it is seen that grain growth sets in at a slightly lower temperature (713 K) compared to the alloy with lower P-content. Also, during TEM in situ annealing it was observed that grain growth occurs more homogeneously in Co–3.2 at.% P, i.e. a larger amount of grains are growing and the growth appears to be slower compared with Co–1.1 at.% P. TAP analysis of annealed Co–3.2 at.% P, reveals that also in this alloy precipitation of the non-equilibrium CoP-phase occurs when the temperature is raised. However, this takes place at a lower temperature compared with the low P-content alloy. Moreover, no Co2 P precipitates have been observed in the analyzed volumes. Hence, it is concluded that saturation is reached earlier in Co–3.2 at.% P due to the higher initial P-excess in the grain boundaries, and precipitation takes place leading to grain growth at lower temperature compared to Co–1.1 at.% P. In order to estimate the reduction of grain boundary mobility predicted by the solute drag theory [17], or the grain boundary energy decrease due to segregation as proposed by Kirchheim [18], the assumption of a homogeneous solute segregation in the grain boundaries is a prerequisite. However, as demonstrated in this work, the P-distribution is highly inhomogeneous in the investigated nanocrystalline Co–3.2 at.% P material. Thus, such theoretical approaches are not valid in this case. Additional effects such as impurity content, texture, and allotropic phase transformation have also to be considered. Apart from the different P-contents in the investigated alloys, the C and S-impurity levels and textures in the as-prepared state are different. The alloy with lower P-content has a higher amount of C and S-impurities compared with the high P-content material. The influence of C and S-impurities on thermal stability in pure nanocrystalline Co has been previously investigated by Hibbard [4,19]. C-impurities were not found to influence thermal stability noticeably, but a strong correlation between the S-content and heat-release peak temperature in the DSC curves of Co was established. Comparing the impurity contents obtained in this work, a change in S-concentration from ∼0 to 0.085 wt.% would correspond to an increase in peak temperature of about 60 ◦ C in pure Co [4,19]. Therefore, the higher S-impurity content in Co–1.1 at.% P can be expected to contribute to the increased thermal stability of this alloy. In both electrodeposits, the allotropic phase transformation of Co was found to set in between 673 and ∼733 K (XRD), which is the temperature range where abnormal grain growth sets in. This is in good agreement with investigations of pure Co, where the combined reaction of allotropic phase transformation and grain growth was investigated and found to occur simultaneously at ∼120 ◦ C lower than the equilibrium phase transformation of Co [20]. In the present study, the reactions also occur simultaneously, but at a higher temperature due to the impurity and solute effects. Furthermore, the different textures should also affect the microstructural development upon annealing. Some grain orientations may be thermodynamically favored to grow, so that different textures result in different grain growth behaviors. Co–1.1 at.% P was found to have a weak (0 0 0 2)-texture, while Co–3.2 at.% P has a random texture in the as-prepared state. Hence, the more homogeneous grain growth in the high P-content alloy compared to the low P-content alloy may also be influenced by the texture difference.

M. da Silva et al. / Materials Science and Engineering A 445–446 (2007) 31–39

5. Conclusions Both investigated Co–P electrodeposits consist of 10 nm hcp-Co grains, and the P-distribution is highly inhomogeneous already in as-prepared state. In as-deposited Co–1.1 at.% P, Pcontents of up to ∼6 at.% are found in the grain boundaries, compared to ∼25 at.% P in Co–3.2 at.% P. Upon annealing, P-atoms segregate to the grain boundaries, and grain growth takes place. Initial grain growth is abnormal in both alloys, but is observed to occur slower and more homogeneously in Co–3.2 at.% P compared to Co–1.1 at.% P. Both alloys show a significantly increased thermal stability compared to other nanocrystalline Ni- and Co-based electrodeposits. When saturation is reached in the Co–1.1 at.% P grain boundaries, precipitation of Co2 P and the thermodynamically unstable CoPphase occur, followed by more rapid (normal) grain growth. Due to the higher initial P-concentration in Co–3.2 at.% P, saturation of P in the grain boundaries and precipitation of P-rich phases occurs earlier leading to a slightly lower thermal stability compared to Co–1.1 at.% P. Apart from the different P-contents, the effect of S and C-impurities, allotropic phase transformation of Co, and texture is discussed. The higher impurity content in Co–1.1 at.% P may contribute to the increased thermal stability of this alloy. In both electrodeposits, the allotropic phase transformation of Co was found to set in together with abnormal grain growth between 673 and ∼733 K. Furthermore, some grain orientations may be thermodynamically favored to grow, so that different textures result in different grain growth behaviors. Thus, the more homogeneous grain growth in the high P-content alloy may be due to the initial random texture. Acknowledgements The authors would like to thank Prof. U. Erb and Prof. G. Hibbard, Department of Materials Science and Engineering, University of Toronto, and Integran Technologies Inc., Toronto, Canada, for providing the material. Dr. Fabi´an P´erezWillard at Laboratorium f¨ur Elektronenmikroskopie, Universit¨at Karlsruhe, is acknowledged for performing final tip sharp-

39

ening of TAP specimens. Financial support from Deutsche Forschungsgemeinschaft—the German Research Foundation (Kl 230/28-1) and Kungliga Vetenskapsakademien—the Royal Swedish Academy of Sciences are gratefully acknowledged. References [1] C. Cheung, D. Wood, U. Erb, in: C. Suryanarayana, J. Singh, F.H. Froes (Eds.), Processing and Properties of Nanocrystalline Materials, Minerals, Metals and Materials Society, Warrendale, 1995, pp. 479–489. [2] M.J. Aus, C. Cheung, B. Szpunar, U. Erb, J. Szpunar, Mater. Sci. Lett. 17 (1998) 1949–1952. [3] G.D. Hibbard, K.T. Aust, G. Palumbo, U. Erb, Scripta Mater. 44 (2001) 513–518. [4] G.D. Hibbard, PhD Thesis: Microstructural Evolution during Annealing in Nanostructured Electrodeposits, Department of Materials Science and Engineering, University of Toronto, 2002. [5] U. Klement, U. Erb, A.M. El-Sherik, K.T. Aust, Mater. Sci. Eng. A 203 (1995) 177–186. [6] S.C. Mehta, D.A. Smith, U. Erb, Mater. Sci. Eng. A 204 (1995) 227–232. [7] U. Erb, A.M. El-Sherik, US Patent #5,352,266 (1994). [8] D. Blavette, B. Deconihout, A. Bostel, J. Sarrau, M. Bouet, A. Menand, Rev. Sci. Instrum. 64 (1993) 2911–2919. [9] T. Al-Kassab, H. Wollenberger, G. Schmitz, R. Kirchheim, in: M. R¨uhle, F. Ernst (Eds.), High-Resolution Imaging and Spectrometry of Materials, Springer, Berlin, 2003, pp. 271–320. [10] A.A. Karimpoor, U. Erb, K.T. Aust, G. Palumbo, Scripta Mater. 49 (2003) 651–656. [11] P. Choi, M. da Silva, U. Klement, T. Al-Kassab, R. Kirchheim, Acta Mater. 53 (2005) 4473–4481. [12] J.R. Roos, J.P. Celis, M. De Monte, in: R.W. Cahn, P. Haasen, E.J. Kramer (Eds.), Materials Science and Technology, vol. 15, VCH Publishers Inc., New York, 1987, pp. 481–537. [13] T.H. Hentschel, D. Isheim, R. Kirchheim, F. M¨uller, H. Kreye, Acta Mater. 48 (2000) 933–941. [14] M. da Silva, U. Klement, Z. Metallkd. 9 (2005) 1009–1014. [15] F. Czerwinski, H. Li, M. Megret, J.A. Szpunar, D.G. Clark, U. Erb, Scripta Mater. 37 (1997) 1967–1972. [16] K. Ishida, T. Nishizawa, in: T.B. Massalski (Ed.), Binary Alloy Phase Diagrams, second ed., ASM International, Materials Park, Ohio, 1990, p. 1217. [17] J.W. Cahn, Acta Metall. 10 (1962) 789–798. [18] R. Kirchheim, Acta Mater. 50 (2002) 413–419. [19] G.D. Hibbard, K.T. Aust, U. Erb, Acta Mater. 54 (2006) 2501–2510. [20] G.D. Hibbard, G. Palumbo, K.T. Aust, U. Erb, Philos. Mag. 86 (2006) 125–139.