Evaluation of Surface Damage of Pd Using Cross-Sectional Electron Backscatter Diffraction Analysis

Abstract

Surface damage induced by mechanical polishing of cold-rolled and annealed Pd specimens was examined by cross-sectional electron backscatter diffraction (EBSD) measurements. Fine grains with high-angle grain boundaries were detected in the outermost layer in both specimens. Less granular but layered gradation of crystallographic orientation was detected in the sub-surface layer of the annealed specimen. In the cold-rolled specimen, a lot of elongated grains were detected in the entire inner layer. The formation of the sub-surface layer seemed to be prevented in the cold-rolled specimen by pre-introduced microstructures. In the annealed specimen, the depth of the surface damage layer was dependent on the crystallographic orientation of the matrix grain. This study clearly demonstrated the application of cross-sectional EBSD analysis for evaluating surface damage in metallic materials.

1. Introduction

Mechanical polishing of metallic materials is a common preparation step for numerous mechanical tests and microstructural analyses of surfaces. In the mechanical polishing process, shear stress is generated on material surface due to the friction between the material and the abrasive grains. As for materials with lower values of critical resolved shear stress (CRSS), shear stress on polished surface may induce dislocation slips in the crystal grains, leading to the formation of surface damage layer. The value of CRSS has a good correlation with the value of rigidity modulus G, and it is known that the actual measurement values of CRSS for high-purity metals are on the order of 10⁻⁴∼10⁻⁵ × G. In fact, the formation of surface damage layer has been confirmed for bulk material of Pd in the mechanical polishing process during our specimen preparation for hydrogen permeation tests. This fact indicates Pd is one of materials with the lower value of CRSS. Since the surface damage layer can be a major obstacle to obtaining reliable experimental data of hydrogen permeation, it is our urgent task to understand the formation status of the surface damage layer of Pd.

Various methods of evaluating surface damage have been proposed based on destructive and non-destructive measurements.^1–9) Non-destructive methods, including X-ray diffraction,^2,3) confocal microscopy,^4,5) and Rutherford scattering spectrometry,⁶⁾ may provide global information on the surface damage layer without time-consuming specimen preparation, but their detection accuracy requires significant improvement. By contrast, destructive methods including traditional taper-polishing,⁷⁾ cross-sectional microscopy,^8,9) and microhardness testing^8,10) are relatively mature with acceptable detection accuracies and are therefore widely used. However, consistent specimen preparation for most destructive methods requires time-consuming optimization of the numerous parameters in the chemical etching process. Optical and scanning electron microscopy can only detect the surface damage layer as the surficial unevenness derived from crack initiation and intrusion/extrusion of severely deformed surfaces. Therefore, only a limited number of papers^8,11) have dealt with the evaluation of surface damage layer thickness caused by the polishing process without severe plastic deformation of metallic materials in destructive methods.

Electron backscatter diffraction (EBSD) analysis can provide the distribution maps of grains in material surfaces, including their crystallographic orientations, grain boundaries, and misorientations. Even a slight deformation can be detected from the modification of crystallographic orientation in grain or the formation of new grain boundaries by using EBSD analysis. Furthermore, with respect to the specimen preparation for EBSD measurements, an ion milling method is so efficient as to adjust the observation surface with consistent results. In the present study, surface damage induced by mechanical polishing was examined by cross-sectional EBSD measurements in cold-rolled and annealed Pd specimens. The purpose of this study is to investigate the formation status of surface damage layer in Pd specimens with/without annealing treatment and to demonstrate the application of cross-sectional EBSD analysis for the evaluation of surface damage in metallic materials.

2. Experimental Procedure

The high-purity Pd (99.9% purity) sheet used in this study was rolled from its initial 0.8 mm thickness to 0.3 mm at room temperature. This cold-rolled sheet was cut into specimen shapes 10 mm × 20 mm, which were then each mechanically polished by #800 and #1200 emery papers for 30 min and mirror finished by a choroidal silica suspension for 1 h to form cold-rolled (CR) specimens. Nominal abrasive grain sizes of #800, #1200 emery papers and choroidal silica suspension are 13, 8 and 0.06 µm, respectively. The rotation speed of the polishing machine was 100 rpm, and the specimen was rotated frequently so as for polishing direction to be at random in all the polishing processes. As-annealed specimens were formed from CR specimens that were subsequently annealed at 1273 K for 3 h, followed by furnace cooling. As-annealed specimens that were then mechanically polished and mirror finished formed the annealed (AN) specimens, where the procedures of mechanical polishing and mirror finishing were the same as those of the CR specimens. After polishing by #800 and #1200 emery papers for CR and AN specimens, average surface roughness (Ra) and maximum roughness (Rz) were measured in 5 areas of 50 × 50 µm² by using an atomic force microscope (Hitachi, AFM5500), and the change of specimen thickness (Δt) was measured at 3 points by a digital thickness gauge (Mitutoyo Digimatic indicator S43-500). Surface roughness and change of specimen thickness after polishing for CR and AN specimens were summarized in Table 1. The measurements of EBSD were performed on surfaces of CR, as-annealed, and AN specimens using a field-emission scanning electron microscope (FE-SEM, JEOL-JSM7900F) equipped with a diffraction detector (EDAX, DigiView V EBSD camera).

Table 1 Surface roughness and change of specimen thickness after Emery paper polishing.

Schematic procedures of cross-sectional EBSD measurements for Pd specimens are shown in Fig. 1. Specimens were erected vertically with spring clips and secured with resin (Fig. 1(a)), with each specimen’s cross-section mechanically polished to a mirror finish (Fig. 1(b)). Finally, cross-sections were ion-milled with 4 kV Ar+ at a beam current of 120 µA and an incident angle of 60° for 15 min (Fig. 1(c)), using an ion-milling tool (Hitachi-Hightec, IM4000Plus), followed by EBSD measurements of the cross-sections.

Fig. 1

Schematics of the EBSD measurement procedure on Pd cross-section specimens; (a) Securing with resin, (b) Mechanically polishing, (c) Ar-ion milling.

3. Results and Discussion

Figure 2 shows maps of (a) the inverse pole figures (IPFs) and (b) the grain boundaries for the polished surfaces of CR and AN specimens and the unpolished surface of an as-annealed specimen. When describing these IPF maps, the normal direction (ND) was set normal to the mechanically polished surface. The color in the IPF map represents the crystal orientation when viewed from ND. Grain boundaries with a misorientation ≥15° or <15° were defined as high-angle or low-angle boundaries, respectively. The average grain sizes and the percentage of high-angle boundaries on the surfaces of all specimens are summarized in Table 2, where grains outlined only with high-angle boundaries were accounted for when calculating average grain size. Fine grains in the CR specimen had an average grain size of 0.19 µm and the percentage of high-angle boundaries was 70%. During annealing, this fine grain structure appeared to recrystallize into the structure with an average grain diameter over 200 µm in the as-annealed specimen. This coarse structure of as-annealed specimen was refined during mechanical polishing and mirror finishing into the structure with an average grain size of 0.20 µm in AN specimen. Thus, the average grain size was similar in both CR and AN specimens, although the percentage of high-angle boundaries in the CR specimen was larger than in the AN specimen. Figure 3 presents IPF maps of CR and AN specimens, designating subsequent EBSD measurement areas of (a) Area A and (b) Area B. These measurement areas were set at the top end of the cross-sections, at the interface of the polished surface and the resin. IPF and grain boundary maps of the cross-section in Area A of the CR specimen and Area B of the AN specimen are shown in Fig. 4. The rolling direction was described in Figs. 3 and 4 for CR specimen. For the IPF maps in Fig. 3 and Fig. 4, the color in the map represents the crystal orientation when viewed from ND. As indicated in Fig. 4, fine grains with high-angle boundaries were detected in the outermost layer in both areas. In Area B of the AN specimen, less granular but layered gradation of the crystallographic orientation was observed in the sub-surface layer. In comparison, there was no sub-surface layer in Area A of the CR specimen, but the entire inner area was filled with a lot of elongated grains. Thus, the surface damage layer comprised the outermost and sub-surface layers of the AN specimen, but only comprised the outermost layer of the CR specimen. The depths of outermost and sub-surface layers were determined from the grain boundary map in the deepest range where granular high-angle boundaries and layered low-angle boundaries were detectable, respectively. Table 3 summarizes the depth of the surface damage layer and the percentage of high-angle boundaries on cross-sections in Areas A and B. The depth of the outermost layer was estimated at approximately 1 µm in both areas, and the larger sub-surface layer was detected in Area B. The percentage of high-angle boundaries in the outermost layer in Area A was higher than that in Area B, and the percentage of high-angle boundaries was only 3.5% in the sub-surface layer in Area B. As indicated in Fig. 3 and Fig. 4, a lot of elongated grains in the inner area appear to be oriented in multiple directions such as (111), (112), (123) and (110) planes for the CR specimen. The rolling texture mainly composed of {112}, {123} and {110} planes induced by multiple cross-slips on the {111} plane has been reported for high-purity face-centered cubic metals with higher stacking fault energy.¹²⁾ Since dislocation slips on (111) plane are considered to play an important role in the formation of elongated grains in the inner layer for the CR specimen, they could be also involved with the formation of surface damage layer, in particular sub-surface layer, for the AN specimen. Further measurements of the AN specimen’s outermost and sub-surface layers were conducted for other matrix grains with different crystallographic orientation as a function of deviation angles from the (111) plane. Figure 5 shows the assignment procedure of a plot in the matrix grain on (a) an IPF map, (b) the orientation legend for IPF map, and (c) a diagram of the deviation angle from the (111) plane. For the IPF maps in Fig. 5, the color in the map represents the crystal orientation when viewed from ND. Figure 6 presents a scatter diagram of the depths of outermost and sub-surface layers as a function of deviation angle from the (111) plane of the AN specimen. The depth of the sub-surface layer was larger than that of the outermost layer in each matrix grain, and the depths of both layers decreased in a matrix grain with increasing deviation angle from the (111) plane.

Fig. 2

(a) Inverse pole figure (IPF) and (b) grain boundary maps of the polished surfaces of cold-rolled (CR) and annealed (AN) specimens and the unpolished surface of an as-annealed specimen.

Table 2 Average grain diameter and the percentage of high-angle grain boundaries in the CR and AN specimens and the unpolished as-annealed specimen.

Fig. 3

IPF maps of CR and AN cross-sections, designating subsequent EBSD measurement areas of (a) Area A and (b) Area B.

Fig. 4

(a) IPF and (b) grain boundary maps of cross-sections in Area A of the CR specimen and Area B of the AN specimen.

Table 3 Depth of surface damage layer and the percentage of high-angle grain boundaries on cross-sections in Area A (CR specimen) and B (AN specimen).

Fig. 5

Assignment procedure of a plot in the matrix grain on (a) an IPF map, (b) the orientation legend for the IPF map, and (c) a diagram of the deviation angle from the (111) plane.

Fig. 6

Scatter diagram of the depths of outermost and sub-surface layers as a function of deviation angle from the (111) plane of the AN specimen.

The development of surface damage with the processing conditions for Al has been divided into four stages (Stages I–IV) by some studies.^13,14) In Stage I, small deformation induces dislocation slips and dislocation tangles owing to their intersection, leading to the formation of cellular sub-grain structures. In this stage, the angles of misorientation between individual sub-grains are small. The increase in deformation enhances the accumulation of dislocation tangles along dislocation slip planes, leading to the formation of elongated grains with well-defined parallel boundaries (Stage II). Further deformation promotes the agglomeration of dislocation tangles within the grains into the fragmentation of elongated grains and the subsequent formation of fine sphere-like grains (Stage III). In Stage IV, severe deformation causes the rotation of sphere-like grains, followed by the widespread transformation of low-angle to high-angle boundaries. Although a further increase of deformation in Stage IV does not change microstructures significantly, it promotes the occurrence of dynamic recrystallization¹⁴⁾ involved with grain boundary migration to increase grain size at the end of this stage.

The formation of fine grain structure on the surface of Pd produced by severe plastic deformation through high-pressure torsion (HPT) has been reported by Iwaoka et al.¹⁵⁾ The grain structure of a HPT processed surface was evaluated by EBSD measurements to have an average grain size of 0.35 µm and 77% in ratio of high-angle grain boundaries, with misorientation of 15° or more. In the present study, Table 2 shows the fine-grained crystal structures measured with an average grain size of 0.19 µm and 0.20 µm on the polished surfaces of the CR and AN specimens, respectively. This indicates that the development of surface damage during the polishing process of Pd corresponds to Stage IV. Similarly for the HPT process, a slight increase of average grain size on the HPT processed surface may be attributed to the dynamic recrystallization occurring in Stage IV. Table 3 shows that the depth of outermost layer was approximately 1 µm in CR and AN specimens. Because the surface roughness of both specimens did not reach 1 µm after #800 and #1200 polishing (see Table 1), it is obvious that the accumulation of dislocation slips induced by shear stress on polished surface was responsible for the development of the outermost layer in both specimens. As shown in Table 1, surface layer was removed during #800 polishing by roughly 1 and 4 µm in CR and AN specimens, respectively, whereas minimal during #1200 polishing in both specimens. This implies the approximate completion of outermost layer by the end of #1200 polishing in both specimens. Thus, the formation of similar outermost surface layer in CR and AN specimens suggests some specific characteristics of Pd as compared with other materials. First, besides of the smaller value of CRSS, the value of work hardening coefficient should be smaller. Although the value of CRSS increases by cold rolling, the formation of outermost layer in the CR specimen indicates that the increase of CRSS by cold rolling is not enough to exceed shear stress induced by polishing. Further increase of CRSS would be accomplished by the development of outermost layer and then balanced by the end of #1200 polishing in both specimens (see Table 1). Therefore, grain structure and depth of outermost layer were similar in both specimens. Furthermore, the value of stacking fault energy should be higher, because multiple cross-slips are involved with the formation of fine-grain structures in outermost layer. These characteristics would be found in some of high-purity face-centered cubic metals including Pd. As shown in Fig. 4 for the AN specimen, grains were less granular with low-angle boundaries in the sub-surface layer, which seem to be related to the formation of cellular sub-grain structures in Stage I. The surface-polishing process would not only promote grain refinement in the outermost layer, but also induce dislocation slips in the sub-surface layer of the AN specimen. By comparison, a lot of elongated grains with layered high-angle grain boundaries were detected in the entire inner area of the CR specimen. Because the elongated grain structures introduced by cold rolling correspond to Stage II, the pre-existence of elongated grains in the inner area would prevent the formation of sub-surface layer during the surface-polishing process of the CR specimen. In the AN specimen, the typical slip plane seems to affect the depth of surface damage. Figure 6 presents that the depths of both the outermost and sub-surface layers in a matrix grain decrease with increasing deviation angle from the (111) plane of the AN specimen. A diagram of slip system of face-centered cubic materials induced by shear force on polishing plane is shown in Fig. 7, where θ (0 ≦ θ ≦ 90°) is the deviation angle between the normal of polishing plane and the normal of slip plane, F and S are shear force and area of polishing plane, respectively. When F′ is resolved force of F in slip direction on slip plane, S′ is area of slip plane and φ is the deviation angle of polishing direction from the direction in which slip direction is projected on the polishing plane, the following formulas are established.   

\begin{equation} \mathrm{F}' = \mathrm{F}\cos\theta\cos\varphi \end{equation} (1)

  

\begin{equation} \mathrm{S}' = \mathrm{S}/\cos\theta \end{equation} (2)

From the formulas (1) and (2), the following equation is established between the shear stress of σ on polishing plane and the resolved shear stress of σ′ in the slip direction on the slip plane.   

\begin{equation} \sigma' = \mathrm{F}'/\mathrm{S}' = \mathrm{F}/\mathrm{S}\cos^{2}\theta\cos\varphi = \sigma\cos^{2}\theta\cos\varphi \end{equation} (3)

Here, φ is undefined due to random polishing direction, but there is no difference in polishing direction among the crystal grains on the polishing plane. Therefore, the contribution of φ can be ignored, and the following formula is established.   

\begin{equation} \sigma'\propto \sigma\cos^{2}\theta \end{equation} (4)

From this formula, it can be seen that the larger θ, the smaller the resolved shear stress on the slip plane. Although the depth of surface damage layer cannot be calculated from the resolved shear stress on the slip plane (111), the depth of surface damage layer in a matrix grain dependent on deviation angle from (111) plane (= θ) in the AN specimen (see Fig. 6) could be interpreted by the resolved shear stress dependent on θ.

Fig. 7

Slip system of face-centered cubic materials induced by shear force on polishing plane.

The formation of surface damage has previously been examined by transmission electron microscopy (TEM) and ultra-microhardness testing on polished ferrite surfaces of single crystal Mn–Zn.⁸⁾ Although the presence of the sub-surface layer could not be confirmed, the surface damage layer of several microns in depth which was dependent on the crystallographic orientation in the matrix grain was implied in the study. However, time-consuming processes of precise back thinning and step-by-step chemical etching were required to prepare the specimens for TEM and ultra-microhardness tests, respectively. Therefore, extensive efforts to more efficiently accumulate experimental data are needed to improve the understanding of the surface damage induced by mechanical polishing in various metallic materials with the lower values of CRSS. The application of cross-sectional EBSD analysis for evaluating surface damage in the present study clearly demonstrates the effectiveness of this technique to promote these research activities.

4. Conclusion

The microstructure evolution caused by the surface polishing process was examined by EBSD for AN and CR Pd specimens. From the experimental results, the following conclusions (1)–(4) can be drawn:

1. (1)    The formation of fine grain structure with high-angle grain boundaries was detected in the outermost surface layer in both AN and CR specimens.
2. (2)    Less granular but layered gradation of crystallographic orientation was detected in the sub-surface layer of the AN specimen. By comparison, a lot of elongated grains were detected in the entire inner area of the CR specimen.
3. (3)    The dependence of the depths of the outermost and sub-surface layers in a matrix grain on the deviation angle from (111) plane in the AN specimen would be interpreted by the resolved shear stress on slip plane (111) dependent on the deviation angle.
4. (4)    The application of cross-sectional EBSD analysis for the evaluation of surface damage in metallic materials was demonstrated.

Acknowledgments

This paper is based on results obtained from a project commissioned by the New Energy and Industrial Technology Development Organization (NEDO). We thank Dr. Taro Yakabe and Dr. Chikashi Nishimura, NIMS, for providing materials and technical suggestion for specimen preparation.

REFERENCES

• 1)   J.  Yin,  Q.  Bal and  B.  Zhang: Chin. Mech. Eng. 31 (2018) 41.
• 2)   C.  Suba,  M.L.  Varsanyi,  A.  Miko,  L.  Kovacs,  N.  Velich,  B.  Kadar and  G.  Szabo: Mater. Sci. Eng. A 447 (2007) 347–354.
• 3)   S.  Storm,  W.  Neumann and  H.  Niehus: J. Crys. Growth 197 (1999) 517–522.
• 4)   Q.  Yang,  W.  Zhou,  Z.  Niu,  X.  Zheng,  Q.  Wang,  X.  Fu,  G.  Chen and  Z.  Li: Surf. Coat. Technol. 349 (2018) 1098–1106.
• 5)   J.C.P.  Zunega,  M.G.  Gee,  R.J.K.  Wood and  J.  Walker: Tribol. Int. 54 (2012) 77–86.
• 6)   Y.F.  Lu and  M.  Takai: J. Appl. Phys. 73 (1993) 158–162.
• 7)   Z.  Bi,  H.  Tokura and  M.  Yoshikawa: J. Mater. Sci. 23 (1988) 3214–3224.
• 8)   A.  Okada,  Y.  Yamamoto,  T.  Yoshiie,  I.  Ishida,  K.  Hamada and  E.  Hirota: Mater. Trans., JIM 34 (1993) 343–350.
• 9)   W.J.  Nowak,  D.  Serafin and  B.  Wierzba: J. Mater. Sci. 54 (2019) 9185–9196.
• 10)   X.  Feng,  H.  Liu and  J.C.  Lippold: Mater. Charact. 82 (2013) 97–102.
• 11)   P.  Zeng,  A.  Rana,  R.  Thompson and  W.M.  Rainforth: Wear 332–333 (2015) 650–661.
• 12)   K.  Morii and  Y.  Nakayama: Bull. Japan Inst. Met. 25 (1986) 404–410.
• 13)   K.  Bryla and  K.  Edalati: Mater. Trans. 60 (2019) 1553–1560.
• 14)   S.  Erbel: Met. Technol. 6 (1979) 482–486.
• 15)   H.  Iwaoka,  M.  Arita and  Z.  Hirota: Acta Mater. 107 (2016) 168–177.