Introduction

The processing of bulk ultrafine-grained (UFG) materials through severe plastic deformation (SPD) has attracted significant interest over the past decade [1–6]. By introducing large plastic deformation to a metallic material under high pressure, SPD can lead to exceptional grain refinement without any significant change in the sample dimensions, thereby producing UFG structures containing a large fraction of high-angle grain boundaries. Among the available SPD processing techniques, equal channel angular pressing (ECAP) [4] and high-pressure torsion (HPT) [5] are the most developed and promising methods for grain refinement. In general, ECAP is easy to perform and potentially scalable for the production of large bulk samples. However, the HPT technique has the following advantages compared with ECAP: (i) it is a relatively simple processing technique, (ii) it is more effective in producing smaller grains [7–9], (iii) it produces a higher fraction of boundaries with high misorientation angles [9–11] and (iv) it provides the capability of advanced consolidation of fine particles and amorphous ribbons [12–14].

Severe plastic deformation has successfully produced UFG metallic materials with significantly improved hardness, strength and ductility [15–17]. However, previous work was focused on studying the relationship between mechanical properties and the microstructure produced by SPD and relatively little attention was given to the effect of grain refinement on the corrosion performance of as-processed metals and their alloys. Moreover, most interest has focused on the use of ECAP for studies of the effect of grain refinement on the corrosion resistance of aluminium alloys [18–21], copper [22], steel [23, 24], and Ti [25, 26]. An examination of these results reveals a general lack of consistency in the corrosion properties of different materials or even similar alloy classes [27, 28]. For example, Birbilis et al. [29] found that the corrosion resistance of ECAP-processed magnesium increased in a 0.1 M NaCl solution as the grain size decreased whereas Kutniy et al. [30] reported that the corrosion resistance of magnesium processed by ECAP decreased in a 1 % NaCl solution with decreasing grain size.

The excellent corrosion resistance and biocompatibility of titanium and its alloys, combined with a superior strength-to-weight ratio and high heat transfer capability, makes these materials of interest for high-performance applications such as marine and aerospace structures, chemical and petrochemical plants and biomedical devices. There are only a limited number of studies on the effect of grain refinement on the corrosion resistance of commercially pure (CP) titanium and there is at present no general agreement on the grain refinement effect on corrosion resistance [25–27, 31, 32]. Balyanov et al. [25] first investigated the corrosion behaviour of UFG Ti processed by ECAP and found that UFG Ti is over 30 % more resistant to corrosion in both HCl and H2SO4 solutions than its coarse-grained (CG) counterpart. Balakrishnan et al. [31] later reported on the corrosion resistance in a simulated body fluid of UFG CP Ti produced by ECAP and they observed that the corrosion resistance of the UFG Ti evaluated by the Tafel extrapolation method was 10 times higher compared to the unprocessed CP Ti. Conversely, Hoseini et al. [26] reported that there was no significant effect of grain size on the corrosion resistance in 0.16 M NaCl of UFG CP Ti processed by ECAP. In addition, Faghihi et al. [32] investigated the tribocorrosion behaviour of nanostructured CP Ti processed by HPT under an applied pressure of 6.0 GPa for 5 turns and reported that the resulting nanostructured samples showed lower corrosion resistance in a phosphate buffer solution compared to untreated CG samples but superior performance under the combined action of both wear and corrosion.

In general, the effect of grain size and microstructure on the corrosion behaviour is not yet understood for CP Ti. Accordingly, the objective of this research was to use HPT in order to produce CP Ti specimens having a range of grain sizes and microstructures and then to study the effect of these grain refinement parameters on the corrosion resistance in a 3.5 % NaCl solution.

Experimental material and procedures

Specimen preparation by high-pressure torsion

Commercially pure titanium (ASTM Grade 2) was used in this investigation with the following impurities in wt%: 0.015 % H, 0.1 % C, 0.25 % O, 0.03 % N, and 0.3 % Fe. Prior to HPT processing, the as-received material was annealed at 700 °C for 2 h to release the residual stress introduced during the production stage. In the annealed condition, the microstructure was equiaxed with a grain size of ~10.5 μm. All specimens were sliced from the annealed billet with a diameter of 10 mm and their surfaces ground with silicon carbide abrasive papers to a final thickness of ~0.8 mm. Quasi-constrained HPT processing was used so that there was some restricted outflow of material around the periphery of the sample during the processing operation [33, 34]. The HPT processing was conducted at room temperature with an apparatus having two anvils with circular flat-bottom depressions at the centres [35, 36]. These depressions had depths of 0.25 mm and diameters of 10 mm. Each disc was placed within the depressions, a pressure of 6.0 GPa was applied and the disc was then torsionally strained through rotation of the lower anvil. Various strains were imposed on the discs by processing at a constant speed of 1 rpm through 1, 5 or 10 revolutions.

Characterization of microstructure and microhardness

Measurements of the Vickers microhardness, Hv, were taken along randomly selected diameters on the surfaces of the Ti discs using a Matsuzawa Seiki MHT-1 microhardness tester. These values were recorded under a load of 1000 g and a dwell time of 15 s with a distance of 0.3 mm between each indentation. The value of Hv for each indentation was estimated from the average of four hardness measurements.

The microstructures of the specimens were characterized by optical microscopy (OM) and transmission electron microscopy (TEM). The specimens for OM were polished and etched with a solution of HF(48 %):HNO3(70 %):H2O at a 2:7:41 volume ratio. The TEM foil samples were produced by twin-jet polishing at a temperature of −30 °C in a solution of 5 % perchloric acid, 35 % butanol and 60 % methanol.

Corrosion testing procedure

Prior to corrosion testing, the processed Ti specimens, as well as annealed CG disc specimens, were sequentially wet ground with 120, 800 and 1200 grit silicon carbide abrasive papers, polished with 6 and 1 μm diamond pastes and then degreased with distilled water and acetone followed by air drying. The disc specimens were placed in a Teflon sample holder with a 6 mm in diameter window exposed to the solution. Electrochemical measurements were performed in a 3.5 % NaCl solution using a Gamry Reference 600 potentiostat within a Faraday cage at room temperature. A standard reference electrode of Ag/AgCl (3.5 M KCl) was used for all electrochemical measurements with a graphite rod as the counter electrode.

After immersion into the 3.5 % NaCl test solution, the Ti disc specimens were held for 1 h to achieve a steady open-circuit potential (OCP). Electrochemical impedance spectroscopy (EIS) was performed at OCP over a frequency range from 100 kHz to 0.01 Hz using a sinusoidal AC voltage amplitude of ±10 mV. Subsequently, potentiodynamic polarization was carried out in the range of −0.2 to +1.8 V with respect to the OCP at a scan rate of 0.15 mV s−1. In order to maintain a high statistical accuracy, all electrochemical measurements were repeated at least three times.

Results and discussion

Microstructure and hardness of HPT-processed CP Ti

The microstructure of the CP Ti in the initial annealed condition is shown in the OM image in Fig. 1a where the grain size is ~10.5 μm. After HPT processing, the grain size of all specimens was significantly reduced. The bright-field TEM image in Fig. 1b shows a specimen processed for 10 turns where there is a high density of dislocations, evidence for high internal stresses within the grains as demonstrated by the extinction contours and ill-defined grain boundaries. The shape of the grains is more easily recognized in the corresponding dark-field TEM image in Fig. 1c where the grain size was estimated as ~110 ± 20 nm which is similar to a value reported earlier in HPT processing [37] but smaller than the minimum grain size of ~260 nm reported for Ti when using ECAP [26].

Fig. 1
figure 1

Microstructure of CP titanium (Grade 2): a optical microscopic image of CG Ti; b bright-field and c dark-field TEM images of HPT-processed Ti under applied pressure of 6 GPa for 10 turns

Full size image

Microhardness measurements across the disc surfaces provide an efficient tool for studying microstructural evolution as used earlier in studies of Ni [7], Ti [37] and Al [36]. The microhardness of the CP Ti was significantly enhanced by HPT processing such that the average hardness increased from 182 ± 5 Hv for the annealed CG specimen to 296 ± 25 Hv for a specimen processed through 1 turn (henceforth designated 6 GPa 1T). However, the average hardness thereafter increased only slightly to 314 ± 13 Hv and 319 ± 6 Hv upon HPT processing for 5 turns (6 GPa 5T) and 10 turns (6 GPa 10T), respectively. In practice, the microhardness values vary across each disc as shown by plotting microhardness against the position on the disc surface as in Fig. 2: the lower broken line denotes the initial annealed condition.

Fig. 2
figure 2

Variation of Vickers microhardness measured at the surface with distance from the centre of the CP Ti specimen disc after HPT processing under an applied pressure of 6 GPa through 1, 5 and 10 turns

Full size image

Several important conclusions can be drawn from Fig. 2. Firstly, the microhardness of all Ti specimens increases significantly through HPT processing by comparison with the annealed CG specimen. Secondly, the HPT-processed Ti specimens processed through 1 and 5 turns have much lower hardness values in the centre areas compared to the edges of the specimens. This is due to the variation in strain across each disc during HPT processing because the shear strain, γ, at different positions of the disc is estimated by [38]:

$$ \gamma = \frac{2\pi NR}{h}, $$
(1)

where N is the number of revolutions, R is the radial distance from the centre of the disc and h is the disc thickness. Thirdly, an essentially uniform microstructure is achieved after 10 turns of HPT processing. Thus, for the specimen processed through 10 turns, the difference in microhardness between the centre and edge areas is very small and the microhardness values obtained at all distances from the disc centre are narrowly distributed with small error bars. These results confirm the potential of using HPT processing to achieve reasonably homogeneous microstructures.

Corrosion behaviour of HPT-processed CP Ti

In order to investigate the effect of HPT processing on the corrosion behaviour of CP Ti, three different electrochemical techniques were employed. Thus, OCP measurements with immersion time, EIS and polarization curves were used to characterize the corrosion properties of HPT-processed specimens in a 3.5 % (wt%) NaCl test solution. All results for the unprocessed and HPT-processed specimens are summarized in Table 1.

Table 1 Corrosion performance of HPT-processed CP Ti (Grade 2) in a 3.5 % NaCl solution
Full size table

All three sets of tests show that the corrosion resistance of the HPT-processed Ti specimens is significantly lower in the 3.5 % NaCl solution than for the annealed CG specimen but the corrosion susceptibility of the HPT-processed specimens decreases with increasing torsional strain.

It is well known that the change in OCP with immersion time can be used to monitor the corrosion behaviour of metallic alloys. A rise of potential in the positive direction indicates the formation of a passive film and a steady potential indicates that the film remains intact and protective. A shift of potential in the negative direction indicates breaks in the film, dissolution of the film or no film formation. As shown in Fig. 3, the OCPs for all HPT-processed Ti specimens shift positively to achieve a steady-state value within 1 h immersion and they exhibit more negative values than for the annealed CG specimen. This shows that all processed Ti specimens form passive protective TiO2 films but their corrosion resistance performance deteriorates upon HPT processing. However, an apparently different behaviour was observed for the specimen processed under 6.0 GPa through only 1 turn. This specimen exhibited several abrupt fluctuations in the OCPs within the initial 0.5 h immersion which suggests that some surface defects or pits may contribute to pitting corrosion during the immersion process. This result also demonstrates that the inhomogeneous microstructure produced by HPT processing at the smallest number of torsional strains significantly influences the corrosion performance of the processed specimens.

Fig. 3
figure 3

OCP versus exposure time for the annealed coarse-grained and processed CP Ti (Grade 2) specimens by HPT under an applied pressure of 6 GPa

Full size image

Potentiodynamic polarization is usually used as a tool to evaluate the corrosion properties of an alloy in a given environment as well as to determine its corrosion rate. Figure 4 shows the typical polarization curves for the HPT-processed and unprocessed Ti specimens and they clearly demonstrate that all HPT-processed and unprocessed specimens exhibit similar passive corrosion behaviour in a 3.5 % NaCl solution except that the passivation current densities observed on the processed specimens are about one order of magnitude larger than for the unprocessed specimen. In addition, it is evident that the corrosion potentials of the processed specimens are more negative than the unprocessed specimen. The corrosion current density was evaluated by using a Tafel slope fitting from the polarization curves and thus the corrosion rates were determined as listed in Table 1. It is apparent that the corrosion rates increase significantly from ~0.24 μm year−1 for the unprocessed specimen to ~3.17 μm year−1 for the specimen processed under 6.0 GPa for 1 turn and then slightly decreases to ~0.68 μm year−1 as the number of HPT turns increases to ten.

Fig. 4
figure 4

Typical polarization curves of HPT-processed and unprocessed CP Ti (Grade 2) samples obtained in a 3.5 % NaCl solution at scan rate of 0.15 mV s−1

Full size image

Although polarization curves have been used widely for an evaluation of corrosion potential and corrosion rate, the accuracy of the estimated results may be compromised due to possible interference from a preceding cathodic reaction on the sample surface. EIS is a non-destructive and sensitive technique widely used for the characterization of the electrochemical interface between electrodes and electrolytes. The nature of the electrochemical interface, such as its impedance and capacitance, can be obtained by analysing the EIS spectra [26, 32, 39, 40]. Therefore, the EIS technique was used to further investigate the corrosion performance of the HPT-processed CP Ti specimens.

Figure 5 shows representative Nyquist and Bode plots obtained at OCP in the 3.5 % NaCl solution for CP Ti specimens processed under 6.0 GPa through 0, 1, 5 and 10 turns. Two distinct time constants are observed in all EIS spectra. In order to obtain information on the corrosion behaviour of these specimens, different equivalent circuits (ECs) were used to analyse the EIS spectra but only a model of two time constants in series provided a best fit with χ 2 values in the range of 3 × 10−4–6 × 10−4 for all EIS experimental data.

Fig. 5
figure 5

Nyquist (a) and Bode amplitude (b) and phase angle (c) representation plots of EIS spectra obtained in 3.5 % NaCl solution for Ti specimens without and with HPT processing under 6 GPa through 1, 5 and 10 turns. Inset in a shows that the EC model produced the best fitting for the experimental data. Experimental data are shown by symbols while solid lines represent the best fitting results with the EC model

Full size image

The corresponding values obtained from a simulation with the two time constants in series EC model are summarized in Table 2. As shown in the inset to Fig. 5a, the EC model consists of a solution resistance (R u), two constant phase elements (Øout, Øin) and two charge transfer resistances (R in, R out), respectively. One time constant is attributed to the resistance (R in) and capacitance (C in) parallel combination across the inner TiO2 oxide protective film and the second is attributed to the charge transfer resistance (R out) and capacitance (C out) parallel combination in the outer porous layer or surface defects. A constant phase element (CPE, Ø) was used instead of a pure capacitance in order to address the non-ideal behaviour of the capacitive element due to different physical phenomena, such as surface heterogeneities, which may result from surface roughness, impurities, dislocations and grain boundaries [26, 39–41]. The CPE impedance is defined as

$$ Z_{\text{CPE}} = \, [Y(j\omega )^{\alpha } ]^{ - 1}, $$
(2)

where Y is the admittance with dimensions of Ω−1 cm−2 sα, j is the imaginary number, ω is the angular frequency and α is an empirical exponent of the CPE, with a value of −1 ≤ α ≤ 1, associated with the non-uniform distribution of current as a result of surface roughness and defects. The CPE can represent a circuit parameter with limiting behaviour as an ideal capacitor for α = 1, a resistor for α = 0 and an inductor for α = −1.

Table 2 Parameters of EC obtained from simulation of the experimental EIS spectra
Full size table

It should be noted that in this case the polarization resistance R p, representing the overall corrosion resistance of the material, is the sum of R in and R out. Thus, the evaluated polarization resistances R p from the EC simulation are summarized in Table 1. As seen in Tables 1 and 2, the polarization resistances of CP Ti specimens processed by HPT are reduced significantly by comparison with the unprocessed specimen whereas there is a rising tendency of polarization resistance for the Ti specimens processed by HPT as the number of HPT processing turns increases.

Effect of grain refinement by HPT on corrosion resistance of CP Ti

Grain refinement by SPD not only alters the grain size of both the bulk and the surface of a material but also leads to changes in crystallographic texture and residual stress and possibly also in impurity segregation and second-phase distribution. In addition to affecting the mechanical properties, the corrosion resistance of a material may be affected by these surface changes. There is no unified theory that adequately explains the effect of grain refinement on corrosion susceptibility. Thus, the corrosion response to the increased surface reactivity is dependent on a combination of environmental, processing and material parameters [27]. For Ti, Balakrishnan et al. [31] suggested that a reduced gain size may contribute to an increase in the activity of electrons near grain boundaries, rendering the surface more reactive and prone to form a stable passive oxide film. Similarly, Balyanov et al. [25] partially attributed the improvement of the corrosion resistance of ECAP-produced fine-grained Ti to the rapid formation of passive films at surface crystalline defects including grain boundaries and dislocations. Hoseini et al. [26] suggested that only specific crystallographic orientations can improve the corrosion performance of ECAP-processed CP Ti in a 0.16 M NaCl solution and thus the grain size has no effect on the corrosion resistance. By contrast, Faghihi et al. [32] observed a negative effect of HPT processing on the corrosion resistance of CP Ti in a phosphate buffer solution and attributed this effect to a higher surface activity of the HPT-processed UFG Ti and a difference in composition and structure between the UFG and CG samples.

The present electrochemical measurements clearly demonstrate a complicated influence of grain refinement on the corrosion resistance of CP Ti in a 3.5 % NaCl solution with a combination of both negative and positive effects. The negative effect arises because the corrosion resistance appears to significantly decrease upon HPT processing as demonstrated by comparing the corrosion rate and polarization resistance of all the HPT-processed Ti specimens with those of the unprocessed Ti. However, a positive effect is demonstrated since the corrosion resistance of the HPT-processed Ti increases as the number of HPT turns increases from 1 to 10. The microstructural analysis of the HPT-processed Ti specimens shows clearly not only a significant reduction in the grain size but also a high density of dislocations and large internal stresses within the grains which would enhance the surface reactivity of the processed samples and improve the corrosion performance by rapid formation of passive protection films. The microhardness measurements demonstrate that there is an initial inhomogeneous microstructure upon HPT processing and this will deteriorate the overall corrosion performance. With increases in the numbers of HPT turns, the homogeneity of the microstructure is improved. Nevertheless, for the specimen processed through 10 turns which exhibits a reasonably homogenous microstructure, the overall corrosion resistance remains lower than for the unprocessed CP Ti. This result suggests that the contribution of the microstructural characteristics to the corrosion performance may outweigh the effect of grain size.

Finally, it should be noted that there is extensive experimental evidence for the occurrence of an alpha to omega phase transition in pure Ti when processing by HPT under a pressure of 5 GPa [42] and other experiments demonstrated this transition when processing by HPT at pressures of at least 3 GPa [43, 44] or 4 GPa [37]. The present experiments were conducted at 6.0 GPa and accordingly it is reasonable to anticipate that a phase transition may occur during the processing by HPT. Nevertheless, as described in detail elsewhere [45], an ω-phase was not detected in the present material even after 10 turns of HPT. The reason for the absence of an ω-phase in these experiments probably relates to the relatively high oxygen content of ~0.25 wt% in the CP Ti because it was shown in an earlier study that the presence of oxygen tends to delay the phase transition in Ti [42]. Thus, it appears that the complex nature of the results cannot be attributed to the occurrence of a phase transition.

Summary and conclusions

  1. (1)

    A range of grain sizes and microstructures was obtained in commercial purity titanium (Grade 2) by HPT using an applied pressure of 6.0 GPa and processing through 1, 5 and 10 turns. The results show that the homogeneity of the microstructure is improved with increasing numbers of HPT turns.

  2. (2)

    The corrosion behaviour of HPT-processed CP Ti samples was investigated in a 3.5 % NaCl solution by electrochemical techniques including OCP, EIS and polarization curves. The electrochemical results show that the corrosion resistance after HPT processing is lower than for the annealed coarse-grained unprocessed specimen but the corrosion susceptibility of the HPT-processed specimens decreases with increasing torsional strain.

  3. (3)

    There is a complicated relationship between grain refinement and corrosion resistance of the HPT-processed Ti and this is interpreted in terms of a competition between the negative effect of the inhomogeneous microstructure and the positive effect of the grain size reduction.