The significant impact of hot isostatic pressing on the corrosion behavior of Ti-6Al-4V alloy prepared by electron beam powder bed fusion

Additive manufacturing (AM) is a technology that can rapidly process the shape of complex workpieces with the aid of computer. At the same time, the benefits such as high flexibility, short production time, small material loss and zero mold cost can be used for fabricating biomedical devices. Ti-6Al-4V is the most widely used in the biomedical field. Extensive studies have shown that AM Ti-6Al-4V alloy exhibits excellent mechanical properties and biocompatibility^1,2,3,4,5.

It is well known^{6,7,8,9,10,11,12} that metallic materials would corrode during their service in corrosive environment, and therefore, corrosion resistance is a primary consideration in the selection of a material. In view of the different microstructures obtained by rapid solidification during AM process, the AM-fabricated titanium alloy component demonstrates different corrosion performance from the conventionally produced components. Therefore, extensive attention has been given to the influence of microstructure on the corrosion behavior of AM components. For example, Seo et al. ¹³ investigated the corrosion properties of Ti-6Al-4V alloy fabricated by directed energy deposition (DED), and their results indicate that pitting is mainly determined by acicular α‘ martensite in the alloy. Dai et al. ¹⁴ studied the influence of heat treatment on the corrosion behavior of Ti-6Al-4V samples prepared by laser powder bed fusion (L-PBF) and found that the corrosion rate decreases with the decrease of heat treatment temperature, and the passivation performance decreased resulting from the increase in grain size. Yang et al. ¹⁵ reported that the corrosion resistance of L-PBF-produced Ti-6Al-4V alloy significantly improved after appropriate heat treatment (e.g., heat treatment at 750 °C for 2 h) because of the decomposition of martensite into a fine lamellar mixture of α and β phase. Chandramohan et al. ¹⁶ also reported the improved corrosion and wear resistance of L-PBF manufactured Ti-6Al-4V alloy after heat treatment than prior heat treatment. It was believed that heat treatment exerted a more impact than the printing direction in determining wear resistance and corrosion resistance. Gong et al. ¹⁷ reported that the corrosion resistance of Ti-6Al-4V alloys fabricated by EB-PBF in 1 mol/L hydrochloric acid solution increased in the sequence of 45° < 90° < 55° < 0° (column axis deviating from the build direction), which was attributed to the higher content of β-phase and its better corrosion resistance. The corrosion occurred at grain boundaries, and grain orientation of the EB-PBF sample did not play a dominant role in influencing its corrosion resistance. In addition, defects induced by high thermal gradient and rapid cooling rate in AM process, i.e., pores, lack of fusion (LOF), cracks, and inclusions^18,19, may also influence the corrosion performance of AM components. Although significant endeavors have been made to study the effect of microstructure on the corrosion performance of AM titanium alloy components, the majority of studies investigated the effect of a certain microstructural characteristic on corrosion performance. The correlation between the effects of various characteristics, such as porosity, phase composition, grain orientation, and so on, on the corrosion performance still remains less understood.

To improve their fatigue properties for long-term applications, AM titanium alloys are usually subjected to post-processing treatments. Amongst them, hot isostatic pressing (HIP) refers to the process of sintering and densification of the sample by applying an isotropic pressure (up to ~200 MPa in most cases) to the sample in a closed vessel, while applying high temperature (>0.7 melting point). The application of HIP can close internal pore defects, accompanied by coarsening the morphology of AM titanium alloys. Since the microstructure of metals is an important factor affecting their corrosion resistance^18,20, the hole closure and lamellar coarsening caused by HIP are likely to impact the corrosion performance. To sum up, it is essential to investigate the effect of HIP treatment on the corrosion performance of additively manufactured titanium alloys for engineering applications. In this work, the influence of HIP treatment on the electrochemical behavior of Ti-6Al-4V alloy prepared by EB-PBF in artificial saliva solution was studied. The passivation mechanism of as-fabricated and HIPed EB-PBF Ti-6Al-4V alloy was elucidated in terms of the microstructure, composition, semiconductor properties, defect diffusion ability, formation mechanism of passive film.

Microstructure

The XRD patterns of the studied EB-PBF Ti-6Al-4V samples are shown in Fig. 1. The most characteristic peaks correspond to the α-phase and a small amount of β-phase. And it can be seen that the A0 and B0 samples have a higher content of β-phase than the A90 and B90 samples and HIP treatment led to an increase in β-phase content in B0 and B90 samples. In this work, the B90 sample was HIPed at a higher temperature (~940 ^oC) closed to β-α transition temperature and a pressure of 120 MPa. The high pressure may decrease the β-α transition temperature²¹ and high temperature provides the sufficient diffusion process for phase transformation, which may enhance the β phase formation in HIPed samples^22,23.

XRD patterns of as-fabricated and HIPed EB-PBF Ti-6Al-4V samples.

Full size image

Fig. 2 shows the metallographic structure of the studied Ti-6Al-4V samples. The A0 and A90 samples are dominated by equiaxed and columnar grains, respectively. Lamellar α-phase and β-phase with small size between adjacent α-phase layers are observed in the A0 and A90 samples, and no α‘-phase is noted. After HIP treatment, the α-phase in B0 and B90 samples significantly coarsens, which is mainly due to the high heat treatment temperature (940 ± 20 °C) and low cooling rate after the heat treatment¹⁹.

The metallographic structure of the as-fabricated (a, b) and HIPed (c, d) EB-PBF Ti-6Al-4V samples.

Full size image

As shown by CT analysis in Fig. 3, a large number of pore defects are detected in the as-fabricated samples, and the A90 sample has slightly greater porosity than A0 sample. HIP treatment significantly decreases the number and size of pore defects in B0 and B90 samples (Fig. 3 and Table 1).

a A0, b A90, c B0 and d B90.

Full size image

The texture characteristics were analyzed using EBSD. In the process of data collection, it can be seen that the content of β-Ti increases after HIP by analyzing the content of α-Ti and β-Ti. The inverse pole figures (IPFs) of the studied samples are shown in Fig. 4. As seen from this figure, the grain orientation of A0 and B0 is most (0001), while the A90 and B90 samples have columnar crystals approximately parallel to the processing direction. After HIP treatment, more grains oriented close to the (0001) plane are observed in the B0 and B90 samples, but the columnar grains are significantly reduced.

a A0, b A90, c B0 and d B90.

Full size image

The α-phase polar figures of the studied samples are shown in Fig. 5, and it is clear that the A0 and A90 samples have strongest texture in the (0001) crystal plane orientation. After HIP treatment, the texture strength of B0 and B90 sample is enhanced but the nature of texture remains unchanged (Table 2). The data collection results of α-Ti and β-Ti in the studied samples also indicated that the content of β phase increases by about 40% after HIP treatment.

a A0, b A90, c B0 and d B90.

Full size image

Passive film formation

The formation of passive film was studied by monitoring the current density at constant potential as a function of time. Fig. 6 shows the log (current) – log (time) plots of Ti-6Al-4V samples at potentials of 0.6 to 1.0 VSCE. The trend of all curves was similar. The current density slowly decreases with time and eventually reaches a stable value, which means the passivation film grows and becomes compact on the sample surface. Sometimes, the passive film may dissolve. In a diffusion-controlled process²⁴, a dense inner oxide layer and a loose outer oxide layer can be formed on the EB-PBF Ti-6Al-4V alloy surface.

a–e Double-log plots of current-time, f the fitted slope of the plot.

Full size image

The growth kinetic equation of the passive film can be obtained from the MacDonald model as follows²⁵:

where i is the current density, t is the passivation time, and A and n are constants. The value of the slope n of the curve in Fig. 6a–e indicates the mass of the passive film. When the slope approaches −1, it indicates the formation of a dense protective passive film on the surface of the sample [26]. The slope of the curve was fitted for all the samples and summarized in Fig. 6f. The results show that at each applied potential, the fitted current density is very close, that is, a quasi-stable oxide film is formed on the sample surface at the applied potential. In addition, the density of the passive film on the surface of the as-fabricated sample is slightly higher, as the slope of the curves of the studied samples is closer to −1. So, the performance of the passive film on the surface of A0 and B0 samples is superior to that of A90 and B90 samples.

Semiconductor properties of passive film

Since the damage of the passive film mainly depends on the point defect, the point defect model (PDM)²⁶ is adopted in this work to describe the process of the growth and breakdown of the passive film, which can also explain the influence of the electrostatic field on the migration of the point defect. According to the PDM model, vacancy density and vacancy diffusion coefficient D_O are the two key parameters affecting the electrochemical properties and corrosion resistance of passive films, and since vacancies are point defects responsible for ion transport, the D_O controls the growth kinetics of passive films. A typical linear region with a positive slope on the M-S plots indicates that the passive film of the studied sample is a n-type semiconductor, i.e., the oxygen vacancy is the existing form of the electron donor.

The electrical properties of the passive film were analyzed by the Mott-Schottky (M-S) method and the vacancy density was deduced. Fig. 7 shows the M-S curves in the ASS at 37 °C and 0.6 to 1.0 V_SCE for studied EB-PBF Ti-6Al-4V samples, respectively. Based on M-S theory²⁷:

where C is the capacitance of the passive film/electrolyte interface, C_H is the capacitance of the Helmholtz layer, ε is the compositional complexity of the passive film(the main component of the passive film on the sample surface is TiO₂, and the ε is taken to be 60²⁸), ε₀ is the vacuum dielectric constant, e is the electronic charge, N_D is the donor density in the passive film, E is the film-forming potential, E_FB is the flat-band potential, k is the Boltzmann constant and T is the absolute temperature. For n-type semiconductors, the donor concentration in the passive film can be calculated from the slope of the M-S plot. The N_D values of passive films at different film-forming potentials for the studied EB-PBF Ti-6Al-4V samples are listed in Table 3. The results show that N_D of all the samples decreases with the increase of E, which indicates that the corrosion resistance of the sample will increase with the decrease of the conductivity of the passivated film.

a A0, b A90, and c B0, d B90.

Full size image

Semiconductor properties of passive film

From the measured parameters in the M-S equation, the N_D can be calculated, and based on this, the oxygen vacancy diffusion coefficient (D_O) of the passivation film on the sample surface can be deduced, as follows²⁹:

where ({omega}_{2}=frac{{i}_{{rm{p}}}{RT}}{4{rm{e}}F{varepsilon}_{{rm{L}}}{D}_{{rm{O}}}},) thus ({D}_{{rm{O}}}=frac{{i}_{{rm{p}}}{RT}}{4{rm{e}}F{varepsilon}_{{rm{L}}}{omega}_{2}})where R is the gas constant (8.314 J∙mo⁻¹), F is the Faraday constant (96485 C∙mol⁻¹), ω₁, ω₂ and b are constants determined by the relationship between N_D and E. Fig. 8a, b shows the N_D of the studied EB-PBF Ti-6Al-4V samples in ASS, and the ω₂ values can be directly obtained from the Eq. 3. The calculated values of the studied samples are shown in the figure. The average electric field strength ε_L can be calculated from the relationship between the steady-state thickness L_ss and E, as follows:

where α is the polarizability of the solution/passive film (α = 0.5³⁰), B is a constant. L_ss is the thickness of the passivation film which is estimated from EIS measurements. Since electrochemical impedance produces almost pure capacitive behavior at frequencies above 1 kHz³¹, the L_ss value can be obtained from the equation:

where ε, ε₀ and A are constants. Thus, the L_SS of the passive film can be deduced by the capacitance C. Fig. 8c, d) shows the thickness of passive film of the studied samples at different film-forming potentials. In general, there is no significant difference in the thickness of passive film on the surface of the A0 and A90 samples, and the thickness increases with the increase in potential. HIP treatment slightly decreases the thickness of the passive film. The value of ε_L obtained from the slope of the fitted line is summarized in Table 4. It is seen from the table that the higher value of ε_L, the thinner is the passive film on the surface. The quasi-steady-state current-density i_p can be obtained from the i-t curves. Based on i_p, ε_L and ω₂ obtained above, D_O at different films forming potentials can be calculated, and the results are shown in Table 3. In general, D_O is related to the growth and breakdown of passive film, while the higher the D_O, the higher the breakdown sensitivity rate of the passive film³¹. Obviously, the D_O value of the passive film on the surface of the HIP samples is higher, which leads to faster growth of the passive film on its surface and easier breakdown.

a, b Donor densities and c, d thickness.

Full size image

Sine the oxygen vacancies and cation interstitials need to pass through the passive film to circulation and transport to achieve the growth and destruction of the passive film. Therefore, the film growth kinetics can be expressed by the following formula³²:

where Ω is the molar volume per cation, N_A is the Avogadro’s number, and J is the vacancy flux. According to Fick’s first law, the concentration gradient and the potential gradient can affect the diffusion rate of charged particles in an electric field, which is given by:

where J_C and J_P are the fluxes due to the concentration gradient and the potential gradient, respectively. Considering that the test electrolytes are in a similar steady state, J_C is treated as a constant, i.e., J is mainly determined by the J_P, and is given by^29,32:

where q_i is the charge of the species (q_i = 2e), K is defined as (K,=frac{{F}_{varepsilon }}{{RT}}), and ε is the dielectric constant of the passive film. Since q_i, F, R, and T are constants, the value of J_p is determined only by the N_D, the D_O, and the ε. Based on the above analysis, J_p was calculated from Eq. 8 and the results are listed in Table 3. Since the growth rate after the formation of the passive film is determined by the flux of oxygen vacancies, the higher diffusion flux would lead to a faster growth rate of the passive film, which in turn leads to a limited time for the atoms/ions to diffuse to the “right” location and more defects in the passivating film are created. In summary, the growth rate of the passive film on the surface of the sample is increased by HIP. However, since the passive film on the surface of the as-fabricated sample is denser and the corrosion rate is lower, the corrosion resistance of the as-fabricated sample is better.

XPS analysis of passive film

The passive film on the sample surface was characterized by XPS. Fig. 9 shows the spectra corresponding to Ti 2p, Al 2p, and V 2p ionization at 0, 2, 6, 10, 16, 22, and 28 nm depths for EB-PBF Ti-6Al-4V samples, and there is no significant difference in the composition of the passive film for the four samples. The XPS spectra at 0 nm clearly show that the outermost surface of the passive film of the four samples mainly contained ideal oxides (TiO₂ and Al₂O₃). With the sputtering time increases, the peak positions of Ti 2p and Al 2p spectra shift toward lower binding energies, which indicates the presence of a reduced valence state of the sub-oxides. At the sputtering depth of 10 nm, a peak corresponding to titanium metal (Ti°) with a value of 454.1 eV was detected in the A90 and B90 samples, but not by A0 and B0. The majority of aluminum in the passive film is in the form of alumina (Al³⁺) until the depth of 16 nm, where Al atoms appear. However, no peaks associated with oxides appeared in the V 2 d spectra. This is due to the fact that the -ΔG^Θ values of the three metal elements follow the order: Al > Ti > V. As a result, the formation of V oxides is less likely at low oxygen content. In summary, the surface passive film on the studied four samples is a mixed oxide, consisting mainly of TiO₂ and Al₂O₃ as well as some sub-oxides TiO_x and AlO_y (0 ≤ x ≤ 2, 0 ≤ y ≤ 1.5). Moreover, according to the XPS depth analysis, the film thickness of A0 and B0 surfaces is greater than 10 nm, and that of A90 and B90 is slightly less than 10 nm.

XPS spectrum for A0, A90, B0, and B90 samples at different sputtering depths (0,2,6,10,16,22,28nm) after 1800s of passivation at 0.6 V_SCE in artificial saliva solution at 37 °C.

Full size image

Since the surface passive film of Ti-6Al-4V sample prepared from EB-PBF is mainly composed of Ti oxide, the Ti 2p3/2 spectra of 0 nm, 2 nm, 6 nm, and 10 nm were quantitatively analyzed on the basis of the peaks of fitted XPS spectrum to derive the components in Fig. 10. It can be seen that the outermost layer is composed of TiO₂. As the sputtering depth is increased to 2 nm, the spectral peak of Ti 2p is decomposed into two peaks corresponding to TiO₂ and Ti₂O₃, respectively. Moreover, after HIP, the content of TiO₂ lowers slightly and the content of Ti₂O₃ increases. With the increase in sputtering time, the fraction of TiO₂ and Ti₂O₃ in the oxide film decreases and the TiO content increases, whereas Ti atoms appear in the A90 and B90 samples at the sputtering depth of 10 nm.

XPS spectrum of Ti 2p3/2 for as-fabricated state (A0.A90) and hot isostatic pressure state (B0,B90) at sputtering depth of 0 nm, 2 nm, 6 nm, and 10 nm of passive film, which were promoted potentiostatically at 0.6 V_SCE for 1800s in artificial saliva solution at 37 °C.

Full size image

Based on the above results, it is concluded that compared to A90 and B90 samples, A0 and B0 samples have a thicker passive film on their surface. After HIP treatment, the thickness of the passive film changes slightly, but the content of high valence oxides in the passive film decreases. TiO₂ in the passive film can effectively improve the crystallization characteristics of the passive film³³, thereby improving the protective properties of the passive film. It can be deduced that after HIP treatment, the corrosion resistance of the passive film formed on the surface of the sample decreases, which is consistent with the results obtained from M-S measurements after constant potential polarization in the previous section.

Based on the electrostatic potential polarization and XPS results, combined with the PDM model, Fig. 11 shows the formation and destruction of the passive film of the Ti-6Al-4V samples prepared by EB-PBF in ASS solution. In Fig. 11, the various defect symbols in the process are represented separately according to the Kröger-Vink notation^6,34: Ti_Ti and Al_Ti are titanium ions and aluminum ions in the passive film, V_O, V_TM and V_AM are the oxygen vacancy, titanium ions and aluminum ions in the passive film, Oo is the oxygen ions on the anionic sites of the passive film, V_Ti is the vacancy in the sample, ({{rm{Ti}}}_{({aq})}) and ({{rm{Al}}}_{({rm{aq}})}) are the titanium ions and aluminum ions in solution.

Schematic illustration of the formation and dissolution of passive film formed on Ti-6Al-4V samples prepared by EB-PBF in ASS solution in terms of point defect model.

Full size image

Four main reactions occur during corrosion of titanium alloys in corrosive solutions in the following order:

1. (I)

   Ti/Al atoms in the sample enter the passive film due to passivation and oxygen uptake reactions and become Ti_Ti/Al_Ti. At the same time Ti/Al atoms enter the passive film by exchanging with ({{rm{V}}}_{{rm{Ti}}}^{cdot })/V_Ti and produce Ti_Ti/Al_Ti in the passive film. Both reactions occur at the metal/passive film interface.

2. (II)

   Ti_Ti or Al_Ti in the outer layer of the passive film enters the solution as Ti_(aq) and Al_(aq) because of thermal motion. At the same time, V_Ti is generated at the original position of Ti_Ti or Al_Ti.

3. (III)

   As the thickness of passive film increases, oxygen at the metal/film interface is reduced, which leads to incomplete oxidation of the metal ions. As a result, a large amount of V_O is generated within the passive film and migrates to the film/solution interface to react with water to form H⁺ and O_O.

4. (IV)

   TiO₂ and Al₂O₃ in the outer layer of the film react with H⁺ in solution to generate Ti_(aq) and Al_(aq) and water.

Amongst the above listed reactions, reaction (I) represents the formation of the passive film, reaction (IV) shows the dissolution of the passive film, and reactions (II) and (III) involve the movement of ions across boundaries during passivation. Similar process of passive film formation on titanium and titanium alloys is reported in the literature^6,35.

Since the formation of the passive film is a diffusion-controlled process³⁶, and the diffusion of V_O is obviously faster than V_M⁶, a large number of V_M aggregated at the metal/passive film interface, which further accelerated the formation of the passive film (I). In addition, since the charge of Al³⁺ in the passive film is smaller than that of Ti⁴⁺, oxygen vacancies are inevitably generated in the passive film to maintain local electrical neutrality³⁷. Thus, Al₂O₃ within the passive film also indicates a continuous transition of Al³⁺ and electron vacancy pairs of metallic Al at the metal/passive film interface (I). This results in an increase of the concentration gradient of V_O and, thus, the oxygen diffusion coefficient. Therefore, the pristine state samples generally have high oxygen diffusion coefficients at different potentials (Table 3). It is suggested that the high content of TiO₂ in metallic titanium passive film has significant crystallinity, which improves the protective properties of the passive film³⁸. In addition, Al has a higher affinity compared with Ti, and Al³⁺ will stay in the reduced TiO₂ at the adjacent site and combine with one or two Ti³⁺ at the adjacent site, so Al³⁺ ions will replace Ti lattice^39,40, and is the underlying reason for the some penetration of Al₂O₃ in the TiO₂ matrix.

In this work, the corrosion behavior of as-fabricated and HIPed Ti-6Al-4V samples prepared by EB-PBF was explored. Due to the high chemical activity of Ti⁴¹, Ti-6Al-4V samples prepared by EB-PBF show typical passivation in ASS solution. Since the passive film formed on all the studied samples shows similar chemical and phase compositions, it implies that the samples have similar electrochemical processes during corrosion. Therefore, the quality of the passive film on the surface of the samples plays a major role in their corrosion resistance. The quasi-steady-state current density is related to ion migration in the passive film, and is affected by the J_C and the J_P during diffusion of oxygen vacancies⁴². It is known that the concentration of oxygen vacancies is close to the parts of Ti³⁺ and Ti²⁺ in passive film⁶. Since the concentration of oxygen vacancies in the passive film is close to the sum of the concentrations of Ti³⁺ and Ti²⁺, the value of J_C can be evaluated from the concentration content of Ti³⁺ and Ti²⁺ at different depths. According to the principle of electrically neutral principle, a Ti³⁺and Ti²⁺ ion can generate (or stabilize) one and two V_O at a location⁴³. At identical depths and the same angle, the number of parts of Ti³⁺ and Ti²⁺ of the as-fabricated samples is not significantly different from the HIPed samples, so the J_C difference between the as-fabricated and the HIPed samples is not obvious. Thus, J_P dominates ion migration in the passive film, and the value of J_p is directly proportional to the N_D and the D_O. In this work, A0 and A90 have a large N_D and a smaller D_O (Table 3), and from Eq. 8, it is deduced that the as-fabricated sample possesses a smaller J_P. That is, the current density required to form and maintain the passive film on the surface of the as-fabricated sample is smaller, which also indicates lower passive film dissolution rate. The experimental data is consistent with the above calculation results and the as-fabricated samples have a low quasi-steady-state current density in electrostatic potential polarization tests and a low passive current density in potentiodynamic polarization tests. In general, the quasi-steady-state current can be regarded as the ratio of the applied potential to the resistance on the passivation film, so that lower current densities tend to correspond to larger resistances⁶. As a result, a thinner, more compact, and larger resistance passive film was formed on the surface of the as-fabricated samples compared to the HIP samples. Thus, the corrosion resistance of the EB-PBF samples is reduced after HIP treatment. After HIP, the α-phase of EB-PBF sample is coarsened, the content of β-phase increases (Fig. 1), the size and number of the internal holes of the samples decrease (Fig. 2), and the type of the texture remains unchanged but the strength of the texture at the (0001) plane increases (Fig. 4). All these variations may influence the corrosion performance of the studied samples. The internal relationship of these factors on the corrosion performance of HIPed samples is discussed as follows:

Pore defects

As we all know, the passive film at the defect is weak, and crack initiation will occur preferentially, which leads to a decrease in the corrosion resistance of the sample^44,45. In this work, after HIP treatment, the internal defects of the samples are reduced and the size becomes smaller (Table 1), but their corrosion resistance is not significantly improved by the elimination of these defects. The HIPed sample shows a lower corrosion resistance than the as-fabricated sample. Therefore, the change in pores defects by HIP treatment is not the main factor that influences the corrosion performance of HIP samples.

β-phase

In the Ti-6Al-4V alloy, the α-phase and β-phase can produce an electric potential difference in the corrosion medium due to their different composition⁷. When electrochemical corrosion occurs, the α-phase at the α/β-phase interface is preferentially dissolved, and the passive film formed on the β-phase is more stable than that formed on the α-phase. Meanwhile, the higher density of β grain boundaries can effectively improve the stability of the passive film⁴⁶. However, when the fraction of β-phase is low and the distribution is discontinuous, it can also play an important role in reducing the corrosion resistance as a cathode, and the corrosion is more likely to occur at the grain boundaries¹⁷. In this work, A0 and B0 samples contain a higher β-phase content (Fig. 1), demonstrating superior corrosion resistance compared with A90 and B90 samples. After HIP treatment, the B0 and B90 samples have higher β-phase content than A0 and A90 samples, respectively. However, they do not show better corrosion resistance (Fig. 1). This suggests that other factors instead of β-phase may play a dominant role in influencing the corrosion properties of HIPed sample.

Texture

It was reported that the dissolution activation energy of dense surfaces is higher than that of loose surfaces⁴⁷, which requires higher activation energy to dissolve and is, therefore, more resistant to corrosion. Some studies reported that the dense substrate in the bulk material does have a lower activation energy, but this is not true for the surface of the material. Corrosion rate should be more focused on surface energy. Ren et al. ⁴⁸ found that the (0001) plane of CP-Ti is more corrosion resistant than the ((10bar{1}0)) and ((11bar{2}0)) planes. Fu et al. ⁴⁹ calculated the surface energy of HCP-metals and found that for titanium, the surface energy increases in order of (011), (010), (011) and (110) planes. Hoseini et al. ⁵⁰ confirmed that texture is the main reason for controlling the corrosion properties of commercial pure titanium. Amirnejad et al. ⁵¹ found that the effect of grain orientation on corrosion performance is still applicable to the rolled Ti-6Al-4V alloy, and the surface along the parallel rolling direction has better corrosion resistance. The lower the surface energy, the more stable the surface state and the better the corrosion resistance of the material. Since the surface energy of (0001) plane is lower than that of ((10bar{1}0)) and ((11bar{2}0)) planes, in this work, A0 and B0 samples with more (0001) crystal faces have better corrosion resistance. Therefore, the A0 and B0 samples demonstrate better corrosion resistance. After HIP, the texture strength of the two samples at the (0001) plane increases, but the corrosion resistance decreases. Therefore, the texture has a great influence on the corrosion performance, but it is not a main factor for reduced corrosion resistance of the HIP samples.

Morphology of α-phase

Previous studies indicated that the corrosion resistance of AM titanium alloy components is closely related to the morphology of the constituent phase. The corrosion resistance of equiaxed grains is slightly greater than columnar grains, and the columnar grain boundaries preferentially dissolve compared to the equiaxed grain boundaries⁵². In addition, the majority of reported results indicated that grain refinement may improve the corrosion resistance of titanium alloys. This is because of the high grain boundary density by grain refinement, which increases the nucleation rate of the passive film and the growth rate of the ideal oxide TiO₂. The higher donor density and oxygen diffusion coefficient lead to the superior quality of the passive film on the surface of fine-grained titanium alloy, thus hindering the dissolution of the film^52,53,54. In this work, the A0 and B0 samples display equiaxed grains, and the A90 and B90 samples have columnar grains. Therefore, the A0 and B0 samples are more resistant to corrosion than the A90 and B90 samples. After HIP treatment, the grain size increases significantly (Fig. 1), so that the as-fabricated samples with fine α-lamellar has better corrosion resistance than the HIPed samples. Based on the above results, it is concluded that the morphology of α-and β-phase is the dominant factor for the decrease in the corrosion resistance of the HIPed sample.

The above analyses show that, although the texture, pore defects, phase composition, and morphology all influence the corrosion performance of EB-PBF Ti-6Al-4V alloy samples, the morphology has more influence than other factors. The solidification process of AM is complicated, which leads to the unique microstructure and defect characteristics of the component. On analyzing the corrosion performance of AM components, only considering a certain factor is considered may lead to incorrect or even contradictory results. For example, Dai et al. ¹⁴ investigated Ti-6Al-4V prepared by L-PBF and found that the corrosion current density and corrosion rate increased with the increase of heat treatment temperature, but the corrosion resistance decreased. However, Hemmasian et al. ⁵⁵ concluded that the L-PBF Ti-6Al-4V alloy, after similar heat treatment, demonstrates comparable corrosion resistance to the commercial grade. They also found that the L-PBF-prepared samples have high pitting density, which decreases as increasing the heat treatment temperature. Therefore, in order to accurately analyze the corrosion resistance of AM titanium alloy, the discussed factors should be considered comprehensively to analyze their cumulative effect on the corrosion performance of AM titanium alloy.

In conclusion, the formation and properties of the passive film on EB-PBF Ti-6Al-4V alloy surface were investigated to understand the corrosion mechanism. The conclusions are drawn as follows:

1. (i)

   After HIP treatment, the α-phase of the EB-PBF Ti-6Al-4V sample was coarsened, and the β-phase content was increased, while the size and number of pores in the alloy were decreased. The proportion of grains with (0001) plane increased significantly, but the texture type remained unchanged.

2. (ii)

   The surface passive film of as-fabricated Ti-6Al-4V samples prepared by EB-PBF mainly consisted of ideal oxides (TiO₂ and Al₂O₃), and some sub-oxides TiOx and AlOy (0 ≤ x ≤ 2, 0 ≤ y ≤ 1.5). After HIP treatment, the TiO₂ content decreased, and the growth rate of the surface passive film increased, resulting in an increase in the internal defect density of passive films.

3. (iii)

   After HIP treatment, the coarsening of α-grain decreased the growth of passive film and content of TiO₂ because of lower grain boundary density, which led to inferior quality of the passive film on the sample surface and decreased the corrosion performance of the EB-PBF processed alloy. Compared with the variation of texture, pores, and phase content by HIP treatment, the coarsening of α-grains seems to play a dominant role in determining corrosion performance.

Materials

The EB-PBF Ti-6Al-4V samples with dimensions of 12 × 12 × 12 mm³ blocks were fabricated by ArcamA1 system (Sweden). The EB-PBF preparation was carried out in vacuum at 10⁻⁴ mbar with a voltage of 60 kV, an electron beam size of 200 μm, and helium as the regulating gas control. For each layer, the powder bed was scanned 11 times using a beam with a current of 30 mA and a scan speed of 1500 mm s⁻¹ and heated up to 730 °C before melting. The 3D model was designed by Magics 17.0 software (Materialize, Leuven, Belgium), and then the melt scanning to print the component layer by layer. Table 5 shows the chemical compositions of as-fabricated EB-PBF samples. The rectangular-shaped plates with the size of 10 mm × 10 mm × 1 mm were cut by wire from the block along parallel (90^o) and perpendicular (0^o) to the build direction for the electrochemical measurements, as shown in Fig. 12. Some of the plates were HIPed using a QIH-15 hot isostatic pressure furnace at a temperature of 940 ± 20 °C and pressure of 120 MPa for 3 h. The plates were divided into two groups, the as-fabricated samples were denoted as A0 and A90, and the HIPed samples were denoted as B0 and B90.

Schematic diagram of the printing direction and sampling position of electrochemical corrosion test for Ti-6Al-4V alloy with different stacking orientations prepared by EB-PBF.

Full size image

Microstructural characterization

The phase constitutions were characterized through X-ray diffraction (XRD, D/max 2400 diffractometer, Japan) using a Cu-Kα radiation source with an accelerating voltage of 40 KV and a current of 250 mA. The XRD analyses were carried out in the 2θ range from 30° to 80° using the Jade 5.0 software to analyze the phase composition, respectively, according to the corresponding JCPDS Nos.44-1294 and 44-1288. The microstructures were investigated using a scanning electron microscope (SEM, Japan, JSM-6301F). The samples for SEM analysis were mechanically polished, ultrasonically cleaned sequentially in acetone, ethanol, and distilled water, and then etched in a solution consisting of 8 vol.% HF, 15 vol.% HNO₃, and 77 vol.% H₂O. A computed tomography (CT) scan was performed to detect the internal porosity and the maximum pore size of the samples. The samples for electron backscatter diffraction (EBSD) analysis were further electrolytically polished after mechanical polishing with an electrolytic polishing solution composed of 6 vol.% HClO₄, 35 vol.% n-butanol, and 59 vol.% MeOH. Then, the EBSD analysis was performed by Zeiss field emission SEM equipped with HKL – EBSD attachment, and the data were analyzed by the Channel 5 software. The pore volume fraction was analyzed using a 3D X-ray microscope (Xradia 515 Versa, Zeiss) operated at 150 kV and a power of 10 W, which achieved a voxel size of 2 µm across a field of view of 5 mm.

Electrochemical measurements

The EB-PBF samples of dimensions 10 mm × 10 mm × 1 mm were connected to a Cu wire by conductive glue and sealed with epoxy resin. An area of 10 × 10 mm² was left as the working electrode surface. The surface was polished to 2000 grits with silicon carbide paper, followed by ultrasonically cleaning in ethanol and drying in air. The potentiostat (EG & G Princeton Applied Research, model 2273, USA) was used to perform the electrochemical measurements of a three-electrode system, which the Ti-6Al-4V sample, Pt foil and saturated calomel electrode (SCE) corresponding to the working, counter electrode and reference electrode, respectively. The test electrolyte was artificial saliva solution (ASS) with the specific chemical composition shown in Table 6, and the experimental temperature was maintained at 37 ± 1 °C.

In this work, electrochemical impedance spectroscopy (EIS) was used to analyze the structural characteristics of the electrochemical system, potentiodynamic polarization (PD) was used to understand the kinetic characteristics of the electrochemical reaction, and potentiostatic polarization (PS) and Mott-Schottky method (MS) was adopted to study the semiconductor characteristics of the passivated film on the sample surface. The working electrode was immersed in ASS for 2 h before each test to approach the stabilized potential. After obtaining stable open circuit potential (OCP) values, EIS measurements were performed at 10 mV AC amplitude in the frequency range of 10⁵ Hz to 10⁻² Hz. The EIS results were analyzed using ZsimpWin software. The PD was performed between −0.5 V and 2.5 V (vs. OCP) with a scan rate of 0.5 mV s⁻¹. On this basis, PS tests were performed at 0.6, 0.7, 0.8, 0.9, and 1.0 V_SCE potentials. The cathodic pre-polarization was performed at −1.0 V_SCE for 600 s to remove the oxide formed in the air. For each passivation potential, the electrostatic potential passivation process at the sample surface was for the 2000s, and i-t curves were obtained. After PS treatment, EIS was re-measured, and Mott-Schottky curve was measured at a rate of 10 mV/s and a frequency of 1 kHz immediately after the end of EIS. To ensure the accuracy of the data, at least four parallel experiments were performed for each set of experiments in this work.

Surface characterization

X-ray photoelectron spectroscopy (XPS) analyses were performed using ESCALAB 250 (ThermoVG). Al Ka X-rays were used as the source at 2 kV and 2 μA (hv = 1486.6 eV). The surface was etched from 0 to 280 s with Ar+ ions. The XPSPEAK4.1 software was used for peak fitting XPS spectra were tuned by combining the C1s standard peak with an energy of 284.6 eV.