Strain effects on annealing behaviours of moderately deformed nanocrystalline nickel

Introduction

Thermal stability is an important property of nanocrystalline (nc, grain size d<100 nm) metals for many applications. In the last two decades, much attention has been paid to the thermal behaviours of nc metals, including film/foil, powder and bulk forms.1,2 Previous studies suggest that the nanograins of nc metals have a strong tendency to grow upon heating, and temperature has a significant effect on the kinetics of grain growth (GG). For example, several researchers have examined the thermal behaviour of nc nickel, and they have provided evidence and explanation for the exothermic and/or endothermic peaks from 300 to 700 K.^3–3 It is known that the nanograins should undergo normal or abnormal GG. In particular, the influence of impurity elements like sulphur, which should be enriched at grain boundaries (GBs) after heat treatment, is also taken into account during GG.^12–12 Owning to the concentration and segregation effect at GBs, the thermal stability of nc nickel has been seen to improve significantly.

Despite many investigations on the thermal behaviour and GG of nc nickel at low temperatures, by far little is known of the non-isothermal behaviours at relatively high temperatures. In view of this, in the present effort, we attempt to extend the temperature range up to 1273 K in the hope to close the loop in understanding the thermal behaviour of nc nickel. Furthermore, notwithstanding the fact that a number of investigations have examined GG and texture evolution at elevated temperatures of nc nickel, no consideration has been given to the effect of prior plastic strain (or crystalline defects, in other words) on GG and texture evolution. We believe that this is inappropriate as many studies on the thermal stability of nc metals have pointed to an apparent correlation between GG and pre-existing plastic strain.^18–18 Molinari and co-workers suggest that the grain size of nc alloy can be stabilised by plastic strain so that the GG process will not start until a certain amount of plastic strain has been released.²¹ Experimental results for nc copper show that with an increase in the microstrain produced by cold rolling, the onset temperature of GG is increased while the onset temperature of strain release is decreased.²⁰ Thus, the present study is motivated by the lack of quantitative knowledge about the role of imposed plastic strain in the thermal behaviour of nc nickel.

Experimental

Fully dense, electrodeposited nc nickel sheets were procured from Integran Technologies Inc. The as received sheets were 200 μm thick with an average grain size of ∼20 nm. Samples of nc nickel were rolled at room temperature to various von Mises equivalent strains ϵ_VM, calculated as , where δ is the rolling reduction. The thermal behaviour as well as the overall microstructural change of the samples was investigated by differential scanning calorimetry (DSC), X-ray diffraction (XRD) and electron backscatter diffraction (EBSD) techniques. The DSC measurements were performed on a Netzsch STA449C. Heating was carried out at a rate of 20 K min⁻¹ from room temperature to 1273 K. Pure argon (99·999 purity) was used to purge the sample at a flowrate of 80 mL min⁻¹. The XRD studies were performed on the as received and deformed samples on a Rigaku D/MAX 2500 PC diffractometer (18 kW) with Cu K_α radiation. Related microstructural parameters were obtained by X-ray line profile analysis.²² In order to achieve good surface quality required for EBSD examinations, each sample was electropolished in a solution consisting of 10 mL perchloric acid, 30 mL acetic acid and 40 mL alcohol. The EBSD scans were obtained using an FEI Nova 400 field emission gun scanning electron microscope operating at 20 kV. Step sizes of 0·08, 0·5 and 2 μm were used to collect orientation maps. The Channel 5 software was used to analyse the data. In the generated orientation maps, random high and low angle boundaries were shown as black and grey lines respectively. Different colours were used for high angle boundaries with a coincidence site lattice relationship. Σ3 twin (misorientation of 60° about the axis), Σ5, Σ7, Σ9 and Σ11 boundaries were shown as red, green, blue, pink and yellow lines respectively.

Results and discussion

Figure 1 presents the DSC profiles of the as received and deformed samples. The as received sample shows an obvious exothermic peak at 600 K believed to be from normal GG and well demonstrated as a result of GB activities. Quantitative analysis shows that the normal GG process involves an energy release of 3·05 J g⁻¹, causing an exothermic peak with an onset temperature of 553 K. Another broad exothermic peak is ∼771 K with 7·19 J g⁻¹ energy release, which is attributed to abnormal GG and annealing defects. In the subsequent heating to 900 K, GG should continue. Nevertheless, two exothermic peaks appear below 700 K in the deformed samples of ϵ_VM = 0·059, 0·116 and 0·161, which have been identified to originate from the recovery process. The first peak is mainly due to residual strain release, and the second (high temperature) peak is associated with the annealing of deformation induced defect such as dislocation.10,18 The increasing temperature leads to an exothermic peak at ∼960 K for all the samples. These peaks are not only related to recrystallisation but also to GG. Annealing of deformed nc metals leads to recrystallisation, which starts as soon as the recrystallisation temperature (775–935 K for nickel²³) is reached.^24–24 Stored energies in the form of lattice strain or defects in plastically deformed samples are the primary driving force for recrystallisation.²⁷ Details of the broad exothermic reaction within 900–1100 K in the ϵ_VM = 0·116 sample are shown in Fig. 1b. Its deconvolution via two subpeaks implies two distinct processes contributing to this peak. Presumably, at the end of the primary recrystallisation, the microstructure is not yet stable. As such, further annealing at higher temperatures leads to secondary recrystallisation (or abnormal GG), whose driving force is the reduction in GB population that lowers the total system energy. As mentioned previously, although in all cases the impurity element concentration and segregation effects on GG of nc nickel are present, it is assumed that their influence is similar in our case.

OPEN IN VIEWER

Figure 1

a typical DSC curve of as received and deformed nc nickel samples with different strains (as indicated) at heating rate of 20 K min⁻¹ and b DSC curve of ϵ_VM = 0·116 sample: peaks A and B (open curves) represent least square fit of original experimental results

For conventional polycrystalline metals, a small plastic strain promotes the onset of abnormal GG after primary recrystallisation.²³ To determine such effect or the lack thereof in nc nickel, the characteristic temperatures and relative energy release for samples predeformed to various strains are listed in Table 1. The total recrystallisation heat releases are 35·5, 36·5 and 53·0 J g⁻¹, corresponding to ϵ_VM = 0·059, 0·116 and 0·161. For the least deformed sample (ϵ_VM = 0·059), the thermal effects from primary and secondary recrystallisations are ∼29·8 and ∼5·7 J g⁻¹ respectively. Compared with the other two deformed samples, its heat release is clearly larger during primary recrystallisation but smaller during secondary recrystallisation. The difference in plastic strain can account for this observation. It is also noted that the thermal effect due to secondary recrystallisation increases with increased plastic strain, which can be explained as follows: increased strain makes more sites available to nucleation of strain free grains; as such, at a given probability of abnormal GG, an increased plastic strain results in an increased tendency to secondary recrystallisation. In addition, primary recrystallisation starts at ∼898, ∼913 and ∼918 K for ϵ_VM = 0·059, 0·116 and 0·161 respectively. That is, the onset recrystallisation temperature seems to have been delayed with increased ϵ_VM, indicating that a small plastic strain enhances the thermal stability of nc nickel by postponing recrystallisation to higher temperatures.

OPEN IN VIEWER

Table 1

Data from DSC analysis for deformed nickel with various strains ϵ_VM, including onset temperature T_o, peak temperature T_p, end temperature T_e and relative heat release ΔH

ϵ VM	Peak A				Peak B
	To/K	Tp/K	Te/K	ΔH/J g−1	To/K	Tp/K	Te/K	ΔH/J g−1
0·059	898	958	…	29·8	…	1030	1108	5·7
0·116	913	960	…	20·7	…	1040	1085	15·8
0·161	918	977	…	25·5	…	1089	1176	27·5

As is well known, temperature has a significant influence on the evolution of the microstructure and the grain orientation of nc metals. Note that the temperature dependence of the coarsening behaviour is well recognised. Major attention has been paid to grain orientation during GG. Figure 2 shows two examples of XRD results for the as received and deformed samples (ϵ_VM = 0·161) after DSC measurement at various temperatures. For the as received sample, it can be seen from Fig. 2a that there is a significant preferred (200) orientation at 300 K. However, at 500 K, the (200) preferred orientation still exists but is weakened. As the temperature is raised, it appears that there is no obvious (200) preferred orientation at 800 K, but such preferred orientation is recovered to some extent at 1000 K. Quantitative analyses of the full width at half maximum (FWHM) of (111) and (200) peaks indicate that there are three stages in which the FWHM decreases. In the first stage, at low temperatures (300–500 K), very little GG can occur, leading to a slight change in FWHM. At the intermediate temperature range (500–800 K), there is a relatively large decrease in FWHM, corresponding to rapid GG. After the grain sizes reach several micrometres, the FWHM remains nearly constant or decreases a little with increasing temperatures from 800 to 1000 K.

OPEN IN VIEWER

Figure 2

Temperature dependence of XRD patterns for a as received and b deformed (ϵ_VM = 0·161) nc nickel samples: corresponding (111)/(200) intensity ratio and FWHM of (111) and (200) reflections are shown in right side

In the case of the deformed sample, as shown in Fig. 2b, the prevalence of the (200) orientation becomes apparent with increased annealing temperature. Annealing at 500 K produces no obvious change in the XRD result except for a slight increase in the intensities of the (111) and (200) peaks. Interestingly, annealing at 800 K leads to sharp decreases in both intensity and width of the (111) peak. However, annealing at 1000 K almost renders vanishing of the (220) and (311) peaks and much weakened (111) peak. A larger decrease in FWHM is also found in the temperature range from 500 to 800 K, where primarily the reduction in internal strain occurs. Earlier investigations suggest that elevated temperatures help to recover most defects inside the grains but may not be able to cause remarkable change in grain size if the temperature is insufficient. Meanwhile, new types of defects may form during annealing, depending on the specific material. In situ TEM observations at 498–676 K of nc nickel show abnormal grains containing various defects, including dislocations, stacking faults, twins and even stacking fault tetrahedra from vacancies.²⁸ A more recent effort on abnormal GG in nc nickel reveals a metastable hcp phase.²⁹

Figure 3 shows the misorientation angle distributions of the as received and deformed samples (ϵ_VM = 0·161) after DSC measurement at 800 and 1000 K respectively. For the annealing at 800 K, in both as received and deformed samples, the distribution has a smaller proportion in the 0–15° range and conversely a greater proportion of misorientations in the 15–50° range. Moreover, it is observed that there is a higher proportion of misorientation angle distribution at 60° (Σ3) for the as received sample. Nevertheless, there are clearly very pronounced differences in the misorientation distributions of the as received and deformed samples after DSC measurement at 1000 K. In the case of the as received sample, the proportion of the 60° misorientation angle increases sharply, indicating that the high temperature annealing behaviour is obviously dominated by a 60° misorientation angle. However, in the deformed sample, the distribution has a greater proportion in the 0–15° range but a smaller proportion in the 15–50° range. It should also be noted that the 60° misorientation angle is somewhat enhanced. Continuing the analysis of the 60° misorientation angle, we follow the proportions of Σ3 boundaries. For the as received sample, the proportion of Σ3 boundaries is 5·2 and 25·8 for the treatment at 800 and 1000 K respectively. As expected, the higher the temperature, the more extensive GG and the appearance of annealing twins.^30–30 The large number of boundaries with misorientation angle near 60°, as seen in Fig. 3b, is indicative of the presence of many twin boundaries in the structure. However, in the case of the deformed sample, the proportion of Σ3 boundaries is 1·1 and 3·7 for the treatment at 800 and 1000 K respectively. It is believed that strain has an influence on the thermal behaviour. For a deformed sample, the sample should undergo recovery and recrystallisation before GG at elevated temperatures.

OPEN IN VIEWER

Figure 3

Misorientation angle distributions of as received and deformed (ϵ_VM = 0·161) samples after DSC measurement at 800 and 1000 K respectively

In many studies, it has been shown that the majority of annealing twins can be formed during GG. Results from previous work on GG and twinning have shown clearly that almost all of the Σ3s are new annealing twins, which is typified by the long and straight morphology.30,31 As illustrated in Fig. 4a, in the as received sample at 1000 K, most of the boundaries are of high angle type, many of them have Σ3 twin relationship and the fraction of low angle boundaries is very low. Conversely, in the deformed sample, it can be seen that the frequency of annealing twins is obviously lower, as shown in Fig. 4b. Moreover, a relatively large fraction of low angle boundaries is observed. Before annealing, predeformation induces a high density of dislocations in the grain interiors and GBs. However, with the progress of annealing, the recovery process involves dislocation rearrangement to lower their energy by the formation of low angle GBs and the annihilation of intragranular dislocations.

OPEN IN VIEWER

Figure 4

Maps (EBSD) for a as received and b deformed (ϵ_VM = 0·161) samples after DSC measurement at 1000 K: random high and low angle boundaries were shown as black and grey lines respectively. Σ3, Σ5, Σ7, Σ9 and Σ11 boundaries were shown as red, green, blue, pink and yellow lines respectively

Furthermore, plastic deformation also plays an important role in texture evolution and GB characters during annealing of nc nickel since recrystallisation involves GB migration and its kinetics is affected by grain orientations. There is evidence that the overall recrystallisation rate may be dependent on grain orientation. It has been reported that GG is not only a coarsening process but also an orientation selective process. Thus, there has been interest in the strain effect on grain orientation (or texture change). To clarify the microstructural change during thermal treatment, especially to compare with texture change during abnormal GG at low temperatures, we have performed texture analysis on the as received and deformed samples before and after DSC measurements. Grain orientation of nc nickel during cold rolling deformation has confirmed that the neighbouring grains exhibiting similar orientation can be enhanced by increasing plastic strain. Figure 5a shows that the (200) preferred orientation can become stronger with plastic strain, consistent with previous investigations.27,33 Messina et al. have reported in drawn copper wire that (200) oriented grains prevail after recrystallisation but abnormal GG at higher temperatures changes it back to the (111) orientation.³⁴ In all the post-DSC samples of plastically deformed nc nickel (Fig. 5b), (200) becomes the predominant texture, particularly for ϵ_VM = 0·116 and 0·161 deformed samples, where the (111) peaks almost vanish, evidently different from that of the as received sample whose (111)/(200) intensity ratio (I111/I200) is 0·97 before but 1·67 after DSC. That is, in the as received sample, the (111) oriented grains have overtaken the (200) oriented grains in XRD, similar to the low temperature annealing of nc nickel.35,36 The low temperature GG behaviour of nanograins with different orientations indicated that the (111) oriented grains should grow at the expense of the (200) oriented grains since the (111) oriented grains have the lowest energy compared to the (200) oriented grains and grains oriented in other directions on the basis of the GB energy distribution. Therefore, following energy consideration, when thermal energy is provided, growth of the (111) oriented grains should prevail to cause total energy minimisation. As a consequence, it is not surprising to observe an increase in the I₁₁₁/I₂₀₀ ratio of the XRD peaks of the as received sample after DSC measurement.

OPEN IN VIEWER

Figure 5

X-ray diffraction patterns of as received and deformed nc nickel samples a before and b after DSC measurement at 1273 K

The texture is strikingly different in the post-DSC samples with plastic deformation before DSC. The XRD results of samples of ϵ_VM = 0·116 and 0·161 imply some other mechanisms involved in the abnormal GG of plastically deformed nc nickel, especially at relatively high temperatures. A corollary is that high density lattice defects have been stored during cold rolling. Such defects may exist as stacking faults, deformation twinning, Lomer–Cottrell locks or even body centred cubic nickel,^37–37 which collectively increase the system free energy. The excess free energy then serves as the driving force for recrystallisation. Following the strain energy release maximisation model for the recrystallisation texture of face centred cubic metals,⁴⁰ the recrystallised grains are clustered around {100}, which entails predominance of the (200) oriented grains after high temperature annealing. To further our understanding of the contribution of recrystallisation to texture evolution in plastically deformed nc nickel, we have examined coarse grained nickel control samples in two different conditions. The first is 72 h vacuum annealed, and the second is deformed by cold rolling. In the case of the fully annealed sample, recrystallisation is not expected during high temperature thermal treatment. Naturally, no remarkable change in the (111)/(200) diffraction intensity ratio has been detected in the annealed samples before and after DSC measurements. On the contrary, there is an obvious change in the I₁₁₁/I₂₀₀ ratio in the cold rolled sample, which is believed to be from recrystallisation.

We also believe that the density and type of crystal defects, such as stacking faults, twin faults and sessile dislocations, should be considered to understand the difference in annealing behaviours of the differently deformed samples. Previous results show that twinning is kinetically favoured over the growth of (111) oriented grains.⁴¹ However, the role of twinning is not yet clear. Our quantitative XRD results on the planar fault probabilities of various samples are shown in Fig. 6. The twin fault probability of ϵ_VM = 0·059 sample is 3·9×10⁻⁴, smaller than those of the other two deformed samples (1·1×10⁻³ and 1·6×10⁻³ respectively). To account for the relation between twin and stacking faults, the stacking fault probability is also measured and presented in Fig. 6. We thus surmise that the difference in grain orientations among those deformed samples is directly correlated with the probability of planar defects. Twinning is regarded as yet another favoured energy minimisation mechanism during recrystallisation and GG.³⁰ The ϵ_VM = 0·059 sample exhibits weak (111) and (311) peaks, possibly related to GB energy, whose decrease is a key factor in determining the twinning events and stabilises the grains against further twinning.²³ Misorientations between grain clusters and the abnormally grown grains are mostly of low angles or of twin nature. Total energy minimisation requires that in the absence of sufficient twins to provide GB mobility, the (111) or (311) orientated grains tend to grow, giving rise to the weak (111) and (311) XRD peaks in this sample.

OPEN IN VIEWER

Figure 6

Probabilities of planar defect (twinning and stacking faults) in samples after DSC measurement

Conclusions

The thermal behaviours of nc nickel in the as received condition and after plastic deformation to different plastic strains (ϵ_VM = 0·059, 0·116 and 0·161) are investigated by means of DSC, XRD and EBSD technologies. From room temperature up to 1273 K, the deformed samples are subjected to recovery, primary and secondary recrystallisations. The experimental results suggest that both primary and secondary recrystallisations are enhanced by the imposed plastic strain. Analysis of texture evolution shows that recrystallisation plays an important role in the final grain orientations. Furthermore, grain orientation indicates that twinning also plays an important role in the recrystallisation, which is expected to provide progressively more mobile boundaries.