The Mechanism of Titanium Hydride Growth on Grade‑2 Titanium under Simulated Crevice Corrosion Conditions

Abstract

ASTM grade-2 titanium (Ti-2) is widely used in industrial applications due to its excellent corrosion resistance. However, in these applications, crevice corrosion or galvanic coupling can cause the formation of titanium hydride (TiH _x ), potentially resulting in material failure through hydrogen-induced cracking (HIC). During the metal fabrication process, Fe impurities in Ti-2 can form Ti _x Fe intermetallic particles (IMPs) that can significantly alter the corrosion behavior of the material. This study investigates the relationship between TiH _x formation and Ti _x Fe IMPs in 0.5 M HCl at 60 °C by forming TiH _x galvanostatically for varying durations and subsequently analyzing the surface. We determined that TiH _x formation initiates at Ti _x Fe IMPs, causing swelling and creating a protrusion at these sites. Initially, the hydride grows in an arc profile around the Ti _x Fe IMP, reaching a thickness of ∼5 μm before beginning to form on the Ti matrix surface. Hydride growth into the metal occurred in preferential directions, attributed to favored H diffusion pathways. The dissolution of the Ti matrix and Ti _x Fe IMPs in the acidic solution increased with polarization time, presumably because of oxidation and/or chemical dissolution of the TiH _x formed. The potential measurements during galvanostatic polarization, XRD analysis, and electron microscopy observations all indicated that TiH _x grew rapidly over the first 4 h, followed by slower growth that increased the surface coverage.

1. Introduction

ASTM grade-2 titanium (Ti-2) is the most widely used commercially pure (CP) grade of Ti because of its combination of high strength, excellent ductility, and elite corrosion resistance across various environments. These applications include being used as components in flue-gas desulphurization plants, exhaust systems in the automotive industry, and various components in marine equipment. The excellent corrosion resistance is conferred by an inert oxide film on the surface that is primarily composed of TiO₂ that provides resistance to pitting, intergranular, and general corrosion.

Fe is inevitably present as an impurity in Ti-2 (up to a maximum of 0.3 wt %) due to challenges with the fabrication process. The microstructure of Ti-2 varies with the Fe content, and its distribution within the microstructure can vary significantly. For example, while a small amount of Fe can be distributed in solid solution within the α-grains, it is more commonly found along grain boundaries in the form of Fe-stabilized β-phase, leading to a smaller grain size for specimens with higher Fe content. When the Fe content surpasses approximately 0.05–0.07 wt %, Ti _x Fe intermetallic particles (IMPs) can precipitate along grain boundaries and at triple points, producing areas heavily enriched in Fe (16–20 at.%) relative to the grain interiors.

In industrial usage, Ti-2 is most susceptible to failure from crevice corrosion and hydrogen-induced cracking (HIC). When oxide-covered Ti is electrically coupled to a more active metal, galvanic coupling can occur. , In industrial applications, coupling is commonly observed with carbon steel, which is considerably less noble than spontaneously passive Ti. In such a scenario, Fe acts as the anode and undergoes dissolution, while Ti functions as the cathode. The cathodic reaction on titanium will involve either oxygen reduction or proton reduction, leading to H₂ evolution and the absorption of H, with the latter scenario potentially leading to HIC.

While galvanic couples can be avoided by proper industrial design, the occurrence of crevice corrosion for materials such as Ti-2 is not so readily prevented. Ti-2 becomes susceptible to crevice corrosion at temperatures above ∼60 °C due to the crystallization of the protective, passive oxide layer. This crystallization introduces grain boundary pathways in the oxide, allowing access of H₂O to the metal surface. When this is combined with the presence of occluded cell conditions, a drastic local decrease in both oxygen concentration and pH can lead to the initiation and propagation of crevice corrosion. The HER occurring inside active crevices on Ti has been shown to be the main driving force for crevice corrosion propagation on Ti-2, with only minimal support required from the reduction of O₂ on the surfaces external to the crevice.

In both the crevice corrosion and galvanic coupling scenarios, the HER occurring on the surface of Ti presents a challenge to the long-term integrity of the material due to the potential for HIC. The HER can occur through either the Volmer–Tafel (Reactions 1 and 2) or the Volmer–Heyrovsky mechanism (Reactions 1 and 3), with both mechanisms involving the electrochemical formation of surface-adsorbed H (H_Ads).

$${\mathrm{H}}^{+}+{\mathrm{e}}^{-}+\mathrm{Ti}\to {\mathrm{TiH}}_{\mathrm{Ads}}$$ 1

$${\mathrm{2TiH}}_{\mathrm{Ads}}\to \mathrm{2Ti}+{\mathrm{H}}_2(\mathrm{g}\uparrow )$$ 2

$${\mathrm{TiH}}_{\mathrm{Ads}}+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}\to \mathrm{Ti}+{\mathrm{H}}_2(\mathrm{g}\uparrow )$$ 3

$${\mathrm{TiH}}_{\mathrm{Ads}}\to {\mathrm{TiH}}_{\mathrm{Abs}}$$ 4

A fraction of this adsorbed H produced by Reaction 1 will absorb into the Ti as H atoms, which are able to diffuse within the material (Reaction 4). Once the solubility limit for H in Ti is exceeded locally, titanium hydride (TiH _x ) will precipitate, which can occur both on the surface of the metal and in the bulk material. A sizable volume expansion accompanies the precipitation of TiH _x , producing stresses on the material and rendering it susceptible to HIC if a tensile stress is present. , The deterioration of the mechanical properties caused by hydride formation can persist even if crevice corrosion or galvanic coupling has ceased.

The H absorption and hydride formation behavior of Ti-2 varies on the surface and in the bulk material due to microstructural features. Grade-2 Ti consists primarily of an α-Ti matrix with a H diffusion coefficient of 2.6 × 10^–10 cm²/s. However, along grain boundaries where Fe tends to segregate to form Fe-stabilized IMPs, the diffusion coefficient is several orders of magnitude higher and is comparable to that inpure Fe (9.1 × 10^–5 cm²/s).

Additionally, Ti _x Fe IMPs have been shown to play a large role in the kinetics of the HER, the formation of hydride, and the overall crevice corrosion behavior of Ti-2. , This makes H⁺ reduction kinetics critical in determining the corrosion behavior of Ti. For example, the corrosion resistance of Ti can be drastically increased by alloying with elements that are catalytic for the HER (e.g., Pd, Ru, Ni) by a process known as cathodic modification. In the event of a film breakdown, which could lead to active dissolution within a crevice, catalysis of H⁺ reduction (the kinetics of which is indicated by the exchange current density (i_o) for the HER) due to the presence of these alloying elements forces an increase in potential leading to repassivation. For example, the exchange current density (i_o) for the HER on Pd (which is alloyed in Ti-7 (0.2 wt %)) is 1.0 × 10^–3 A/cm². Fe exhibits a lower catalytic activity for the HER (i_o = 2.0–3.0 × 10^–6 A/cm²) , compared to noble metals such as Pd but surpasses that of Ti (i_o = 5.0–6.3 × 10^–9 A/cm²) , and is only slightly less catalytically efficient than nickel (i_o = 5.6–6.3 × 10^–6 A/cm²) , which is alloyed as a cathodic modifier to Ti-12 (0.8 wt % Ni, 0.3 wt % Mo).

While Ti _x Fe IMPs may not be directly comparable in catalytic activity to elemental Fe, it is likely that they will support a significant amount of cathodic current density for H⁺ reduction, due to the presence of Fe. Their ability to support cathodic reactions when negatively polarized has been demonstrated by the application of scanning electrochemical microscopy (SECM) using the redox mediator ferrocene methanol. While this demonstrates that the locations of Fe-containing IMPs are conductive and support cathodic reactions, it does not indicate that they are windows for H absorption. In addition to Fe, TiH _x is also more catalytic for proton reduction relative to Ti, with an exchange current density of 2.5–5 × 10^–7 A/cm².

The ability of both Ti _x Fe IMPs and TiH _x to catalyze proton reduction has been shown to be beneficial for the corrosion behavior of Ti in certain scenarios. For example, during the crevice corrosion of Ti-2, the presence of Ti _x Fe IMPs resulted in a decrease in corrosion damage in several cases. ,, In other studies, it has been claimed that TiH _x was responsible for enforcing repassivation through the cathodic modification mechanism, a process postulated to initiate at Ti _x Fe IMPs. ,

Given the extreme applications of grade-2 titanium and the potential of TiH _x to influence corrosion behavior both positively and negatively, it is crucial to enhance our understanding of TiH _x formation and distribution on and within Ti-2. There is evidence that TiH _x formation initiates at Ti _x Fe IMPs due to their ability to catalyze proton reduction. , However, a clear mechanistic study that investigates time points spanning across the initiation and propagation stages of hydride formation on Ti-2 is lacking. This study conducted a detailed mechanistic study of hydride formation on Ti-2 under simulated crevice corrosion conditions. This was accomplished by growing TiH _x galvanostatically and then conducting surface analyses at various time points to provide insight into the initiation and propagation stages of TiH _x formation. During these experiments, Ti _x Fe IMPs were tracked to elucidate their role in hydride formation. The goal of this research is to develop a relationship among the Fe content, microstructure, and hydride formation, leading to improved material selection criteria, alloy development, and industrial performance.

2. Experimental Section

2.1. Materials

Specimens cut from a plate of commercially pure ASTM grade-2 Ti supplied by RMI Titanium Co. (Niles, Ohio, USA) were used for all experiments. The nominal composition of this material is given in Table . The electrodes were fabricated by cutting 1 cm × 1 cm coupons using electrical discharge machining (EDM). A 3-48 thread was machined into the back of each specimen to provide an electrical connection to a steel rod connected to external electrochemical circuitry. The steel rod was insulated with Teflon tape to avoid contact with the solution during experiments. Each electrode was mounted in epoxy resin (HYSOL EE4190) and left in air for 24 h to allow the epoxy to harden.

1. Nominal Composition (wt %) of the Grade-2 Ti Specimen from the Mill Report Provided by the Manufacturer .

	Ti	N	O	C	Fe	Pd	Ni	Mo
Ti-2	Bal.	0.014	0.13	0.01	0.12	ND	ND	ND

^aThe balance of the material is Ti.

^bND = Not detected.

2.2. Surface Preparation

Specimens were prepared for electrochemical experiments by sequential grinding with P180, P320, P600, P800, and P1200 SiC paper using water as a lubricant. Each specimen was then sonicated in MeOH for 3.5 min, rinsed with Type 1 water (18.2 MΩ cm), and then left in air for 24 h prior to an experiment to ensure a reproducible initial oxide condition.

For experiments involving the study of IMPs using FE-SEM and XRD, the specimen surfaces were prepared by grinding with P180, P320, P600, P800, P1200, P2500, and P4000 SiC paper, using water as a lubricant. A final polish was applied using a suspension of 0.02 μm colloidal silica (Buehler MasterMet 2) and 3% hydrogen peroxide. Subsequently, each specimen was sonicated in MeOH for 3.5 min, rinsed with Type 1 water, and then left in air for 24 h prior to an experiment to ensure a reproducible oxidized surface. During this 24 h period, microindentations were made on the specimen surface by using a LECO LM-100 Microhardness Tester. These microindentations served as markers, enabling changes in specific IMPs to be tracked. The IMPs were imaged by using FE-SEM and characterized by EDX before and after each electrochemical experiment.

2.3. Electrochemical Experiments

All electrochemical experiments were conducted in a conventional three-compartment electrochemical cell. A Pt mesh connected to a Pt wire was used as a counter electrode, and a saturated calomel electrode (SCE) was used as the reference electrode. All potentials are reported relative to the SCE potential (0.242 V vs standard hydrogen electrode). Before each experiment, the potential of the reference electrode was verified against a “master” reference SCE electrode to ensure its accuracy (no more than 5 mV difference). Solutions of 0.5 M (±0.012 M) HCl were prepared using ACS reagent-grade HCl (Sigma-Aldrich) and Type 1 water. Prior to the start of each electrochemical experiment, the solution was sparged with ultrahigh-purity (UHP) Ar gas for a minimum of 20 min to remove oxygen. The solution was also heated to 60 °C (±1 °C) using a heated circulating bath (VWR) connected to the electrochemical cell’s water jacket. Electrochemical potential monitoring and current application were performed using a Solartron Analytical Modulab with the XM-Studio-ECS software.

2.4. Surface Analysis

Specimens were analyzed using a Hitachi SU8230 Regulus Ultra High-Resolution Field Emission SEM instrument equipped with a Bruker X-Flash FQ5060 Annular Quad EDX detector for energy-dispersive X-ray spectroscopy (EDX) analysis. Tilted and cross-sectional images were obtained by using a LEO (Zeiss) 1540XB FIB/SEM system. The cross-sections were sputter-milled using a Ga ion beam.

X-ray diffraction (XRD) was performed using a Rigaku SmartLab X-ray Diffractometer with a Cu Kα radiation source. The specimen was scanned over a 2θ range of 30° to 80° at a rate of 2°/min. A glancing incidence configuration was used at an incident angle of 3° to enhance the surface sensitivity. Diffraction peaks in the experimental XRD profiles were identified using the PDF-4+ database. The Supporting Information section contains the XRD patterns for the reference data discussed in this study. The XRD analyses were conducted within ∼20 min following the conclusion of each electrochemical experiment to avoid the reaction of TiH _x with the ambient atmosphere.

3. Results and Discussion

3.1. The Electrochemistry of TiH _x Formation

The galvanostatic polarization behavior of Ti-2 for various cathodic current densities is displayed in Figure . The applied cathodic current densities were selected because they are in the range observed for a propagating Ti-2 crevice. , The HER is the predominant cathodic reaction occurring under these conditions, and changes in potential can be attributed to changes in HER behavior.

1.

Galvanostatic polarizations on Ti-2 in deaerated 0.5 M HCl at 60 °C using various current densities. The Tafel slope value (in mV/decade) at 20 min, 40 min, 1 h, 4 h, 10 h, 16 h, and 24 h is noted at the top of the figure.

Several trends were observed. First, as the magnitude of the applied cathodic current increased, the initial potential of the Ti-2 electrode first established upon applying the current decreased since the surface required a higher overpotential to support a higher current. Despite the difference in starting potential, the behavior of the polarization curves was similar for all current densities. During the first 2–4 h, the potential sharply increased (i.e., the surface supported the same cathodic current at a lower overpotential), indicating a depolarization of the HER, likely due to the formation of a TiH _x phase, which could catalyze proton reduction. After this sharp increase, either the potential reached a plateau or the rate of increase slowed until the end of the experiment after 24 h. This indicates that the catalytic nature of the surface achieved a kinetic steady state for each current density. Overall, this behavior has been observed by various other researchers during the cathodic galvanostatic polarization of Ti. ,,

Tafel slopes were determined at specific times during the galvanostatic polarization period to assess the changes in the surface’s capability to catalyze proton reduction as hydride formation progressed. These values are presented at the top of Figure . The Tafel slopes were calculated by constructing log (i) vs E plots using the potentials recorded at the specific time points on the galvanostatic curves. For example, the Tafel plot for the 10-h mark is shown in Figure . The Tafel plots for all of the time points indicated in Figure can be found in the Supporting Information section. For the time points of 20 min, 40 min, and 1 h, the Tafel slope remained at a relatively constant value, close to −140 mV/decade. A Tafel slope of −120 mV/decade indicates that the rate-controlling step in the HER was the Volmer reaction (Reaction 1). However, the observed Tafel slope of ∼−140 mV/decade indicates not only that the Volmer reaction was rate-controlling but that it was somewhat inhibited. This slight inhibition of the HER has been observed by other researchers who have extracted a Tafel slope on Ti-2 in deaerated solutions. ,, Noël calculated a Tafel slope of −140 mV/decade in a 1 M HCl solution at 25 °C, Thomas and Nobe obtained a slope of −150 mV/decade in H₂SO₄ solutions at 25 °C, and Yan et al. a slope of −128 ± 6 mV/decade in a solution of 0.27 M NaCl + 0.10 M HCl at various temperatures.

2.

Tafel slope analysis after 10 h of galvanostatic polarization for varying current densities. The data points were taken from the polarizations in Figure .

In all experiments, galvanostatic polarization commenced only after the electrode potential had remained in the active region for 15 min at the OCP conditions. Even in this active potential region, there was likely oxygen on the surface in the form of a monolayer of oxygen-containing species or an incompletely removed native oxide film, ,, suggesting a reason why the first electron step was rate-controlling and slightly inhibited. After 4 h of galvanostatic polarization, the magnitude of the Tafel slope decreased to −109 mV/decade, indicating an increase in proton reduction kinetics. A value of −30 or – 40 mV/decade would indicate rate control by the Tafel (Reaction 2) or Heyrovsky (Reaction 3) steps of the mechanism, respectively. This intermediate Tafel slope of −109 mV/decade could have stemmed from the formation of catalytic hydride only at certain preferred locations on the surface, resulting in a distribution of rates and mechanisms across the surface. Therefore, it is not possible to definitively determine the rate-controlling step of the HER at the 4 h point, and the only reliable observation is that there was an increase in the proton reduction kinetics. Beyond 4 h, the Tafel slope continued to become steeper, reaching a value of −57 mV/decade by the end of the 24-h experiment. The increase in the proton reduction ability was further supported by the slow potential increase over time for most of the current densities tested. This behavior was likely due to a gradual increase in the surface coverage of the catalytic hydride over time.

3.2. The Relationship between Microstructure and Hydride Formation

3.2.1. Characterization of Ti _x Fe IMPs

FE-SEM images of a number of IMPs in the as-polished condition are presented in Figure . The corresponding EDX maps showing Ti (yellow) and Fe (purple) are also shown. In the FE-SEM images, the Ti _x Fe IMPs can be distinguished as light-contrast spots on the surface. This is supported by the EDX analyses that indicated an enrichment in Fe and a corresponding depletion in Ti in these locations. All IMPs showed no signs of preferential damage in the as-polished condition and typically measured a few micrometers in length. Additionally, several images contain bright white features on the surface of the metal, which were determined to be SiO₂ or SiC embedded in the surface during the polishing and/or grinding procedure. Since only Fe and Ti were considered in the EDX analyses, these SiO₂ or SiC spots appear black on the maps.

3.

(A, C, E, G, I, K) FE-SEM images and (B, D, F, H, J, L) EDX maps of the relocatable TixFe IMPs in the as-polished condition.

After characterizing a number of relocatable Ti _x Fe IMPs by FE-SEM (Figure ), we conducted a series of experiments on the same specimen at a current density of −1 mA/cm². To investigate the stages of Ti hydride growth and the role of IMPs during this process, the duration of polarization was the sole variable in each experiment.

A current density of −1 mA/cm² was chosen as it represents an intermediate value within the range used to record the potential profiles in Figure . For all experiments, prior to applying polarization, the open-circuit potential was measured until a stable active-region potential was maintained for 15 min. The purpose of this was to provide controlled surface conditions with minimal residual oxygen prior to polarization. The control specimen underwent only this control condition and not the galvanostatic polarization. For the remaining specimens, this open-circuit period was followed by galvanostatic polarization at −1 mA/cm² for various times. Following each experiment, Ti _x Fe IMPs were relocated, and the state of hydride formation was analyzed using FE-SEM, EDX, FIB cross-sectioning, and XRD. The Supporting Information section contains the electrochemical data and individual element EDX maps recorded for these experiments.

3.2.2. Surface Topography

Tilted FE-SEM images of the surface of the control, 20-min, and 1-h specimens are displayed in Figure . The images for the control (Figure A) and 20 min (Figure B) specimens were taken at the location of the relocatable IMP. However, since the tilted FE-SEM image for the specimen that underwent 1 h of polarization (Figure C) was taken after a cross-section had been ion-milled at the relocatable Ti _x Fe IMP, a random location on the surface was imaged. The images in Figure show that during the first hour of cathodic galvanostatic polarization, the rapidly increasing potential was accompanied by significant changes in surface morphology. The control specimen showed a relatively flat surface with signs of preferential corrosion at the location of the IMPs and some areas on the Ti matrix (Figure A). For the 20-min specimen (Figure B), a wider range of topographic diversity was observed, with a protrusion developing around the Ti _x Fe IMP and at other areas on the surface. After 1 h of polarization, numerous areas on the surface displayed noticeable protrusions, ranging in length from ∼5 to 15 μm. The dimensions, density, and distribution of these features were consistent with those expected for IMPs on the surface of Ti-2 (Figure ), suggesting they developed around each Ti _x Fe IMP. The development of these protrusions is presumed to be due to a ∼15–21% volume increase that occurs upon TiH _x precipitation. , These topographic images suggest that TiH _x formation initiated at the Ti _x Fe IMPs.

4.

Tilted FE-SEM images of the surface for the (A) control condition or the control condition followed by (B) 20 min or (C) 1 h of galvanostatic polarization at −1 mA/cm² in 0.5 M HCl at 60 °C.

3.2.3. Detection and Semi-Quantification of TiH _x Using X-ray Diffraction

X-ray diffraction was utilized to detect TiH _x on the control and polarized specimens (Figure ). The figure shows the XRD patterns obtained from the specimens that were subjected to a variety of cathodic charges. Due to technical issues with the XRD instrument, the 10-h specimen was exposed to air for ∼3 h prior to XRD analysis. The XRD data indicated that this specimen was an outlier, as it showed extremely low levels of TiH _x , likely due to prolonged air exposure. While oxidation of TiH _x is typically a greater concern at elevated temperatures, it is well known that TiH _x will oxidize in air even at lower temperatures to form a surface film of TiO₂. , It is likely that a significant portion of TiH _x converted to TiO₂, which resulted in the weakening of the characteristic TiH _x diffraction peaks. Consequently, the XRD data for the 10-h specimen were excluded from the analysis. The peaks in Figure are denoted using standard reference patterns as originating from Ti metal (diamond symbol), TiH _x (+), or a combination of metal and hydride due to peak overlap (i.e., at 2θ = 35° and 71°).

5.

XRD analysis of the control specimen and specimens that were subjected to varying periods of galvanostatic polarization at −1 mA/cm² in 0.5 M HCl at 60 °C. The peaks corresponding to titanium metal or titanium hydride were identified using standard reference data.

The standard reference XRD pattern used for Ti metal (ICDD PDF Card No: 00-044-1294) corresponds to a material with a chemical formula of Ti. However, the identification of TiH _x is not straightforward, as several reference patterns in the XRD database can be used to identify the peaks corresponding to TiH _x of various compositions, such as TiH_1.5 (ICDD PDF Card No: 04-008-0303) and TiH₂ (ICDD PDF Card No: 03-065-0708), which have identical peak locations. All peaks in the experimental patterns could be accounted for by using standard reference profiles of Ti in combination with those for TiH_1.5 or TiH₂, except for a low-intensity peak at 36°. This peak, which was dominant for the 16-h specimen and apparent for the 4- and 1-h specimens, is likely attributable to TiH_1.7, although the reference pattern (ICDD PDF Card No: 01-078-2215) is of low quality and may not be definitive. In a study of galvanostatically formed TiH _x , Liu et al. distinguished between TiH_1.5 and TiH_1.7 using the peak at 36°, as this is the only peak for TiH_1.7 that is both significant in intensity and does not overlap with any TiH_1.5 or TiH₂ peaks.

Upon examination of Figure , it is evident that as the polarization time increased, the intensity of the hydride peaks increased relative to the metal peaks. However, for the 16-h specimen, there was a noticeable increase in intensity for the peak at 38.5° (corresponding to Ti metal, 002 plane). A small amount of hydride was detected on the control specimen after only 15 min on open circuit, indicating that active corrosion cathodically supported by the HER leads to hydride formation. Moreover, these data indicate that TiH_1.5 and TiH₂ phases are preferentially formed, while TiH_1.7 accounts for only a minimal fraction of the hydrides on the surface. These observations are consistent with those of Liu et al. However, the presence of TiH_1.7 cannot be confirmed in this study due to the low quality of the TiH_1.7 reference spectrum.

The semiquantification of TiH _x can be assessed using the intensity ratio of a hydride diffraction peak to a metal one (Figure ). For the hydride, the peak at 59° was selected since all TiH _x phases exhibit a peak at this location; thus, it is a good choice to represent the total amount of hydride formation. The most intense peak at 40.3° was used for the metal.

6.

XRD peak intensity ratio between TiH _x and Ti metal (squares, left y-axis) and the potential readings during galvanostatic polarization (lines, right y-axis).

Figure shows that the TiH _x /Ti ratio (squares) increased sharply over the first 4 h, followed by a very gradual increase from 4 to 16 h. This trend in hydride growth over time resembles that of the potential-time profiles (lines). Both the quantity of TiH _x and the potential increased sharply over the first 4 h, followed by a slower rise to an eventual plateau. This suggests that the sharp increase in potential during the initial stages of cathodic galvanostatic polarization and the significant changes in surface topography during this time (Figure ) were due to the formation of TiH _x .

3.2.4. Surface Analysis of Ti _x Fe IMPs after Hydride Growth

3.2.4.1. FE-SEM and EDX Analysis

Figure displays the FE-SEM images and EDX maps of the relocatable Ti _x Fe IMPs postexperiment. Direct comparisons between Figures and can be made by matching the panel letters from each figure (e.g., 3A and 7A). The EDX maps in Figure are also overlaid with the electron microscopy image to view the surface morphology in relation to the Ti and Fe signals.

7.

(A, C, E, G, I, K) FE-SEM images and (B, D, F, H, J, L) EDX maps of the relocatable TixFe IMPs for the control specimen and specimens that were subjected to varying periods of galvanostatic polarization at −1 mA/cm² in 0.5 M HCl at 60 °C.

The control specimen showed slightly preferential corrosion at the location of the IMP, which appears darker and lower than the surrounding Ti matrix in Figure A. Furthermore, slight etching and corrosion damage were observed on the Ti matrix. A strong Fe signal can still be seen in the EDX map, indicating that Fe remained in the IMP (Figure B). Similar results were observed for specimens cathodically polarized for 20 min and 1 h: the Ti _x Fe IMP appears to have been preferentially corroded, there was a slight etching of the Ti matrix, and the EDX maps show that a strong Fe signal was still associated with the Ti _x Fe IMP location. The EDX map for the 20-min specimen (Figure D) displays a dark region located around the periphery of several Ti _x Fe IMPs, indicating a relative decrease in the Ti content at these locations. This dark region expanded across the surface as the polarization time increased to 1 (Figure F) and 4 h (Figure H). These dark areas were likely a consequence of TiH _x formation, which would lead to the localized depletion of Ti relative to the matrix at these locations. These observations suggest that adsorption and absorption of H leading to the formation of TiH _x initiated and proceeded predominantly on the surface around the Ti _x Fe IMPs.

Once the time of polarization was lengthened to 4 h, the FE-SEM image showed that significant corrosion had occurred at the IMP (Figure G). This observation is supported by a large decrease in the Fe signal in the EDX map, which shows Fe only along the edges of the IMP location (Figure H). Additionally, a clear morphological difference was observed between the region around the Ti _x Fe IMP and the rest of the surface (Figure G). The smooth region around the Ti _x Fe IMP is associated with the dark region on the EDX map, while the rest of the surface was rougher and higher in Ti content than the area around the IMP. This dark region, proposed to be TiH _x , grew substantially as the time of polarization increased over the first 4 h.

The FE-SEM images acquired after 10 and 16 h of galvanostatic polarization reveal severe corrosion damage to the Ti _x Fe IMP. The impression that the EDX map for the 10-h specimen shows a light signal for Fe across many regions on the surface is misleading and due to the combination of a low Ti signal and the presence of noise in the Fe EDX map at these locations (Figure J). This can easily be visualized by observing the individual EDX maps for Ti and Fe for this specimen, as found in the Supporting Information section. The only reliable Fe signal appears at the top of the map (Figure J, Location I), indicating a nearly complete dissolution of the Ti _x Fe IMP.

The EDX map for the 16-h specimen also shows a nearly complete dissolution of Fe within the Ti _x Fe IMP; however, there is a distinct location within an IMP that remained concentrated in Fe (Figure L, Location I). To investigate this feature further, elemental analysis (Figure A) and cross-sectional images (Figure B) were obtained. The elemental analysis data presented in Table revealed that a region representing the general surface (Spectrum 1) contained only Ti and small amounts of surface-adsorbed species. Both Fe (1.1 wt %) and Cu (4.4 wt %) were detected in the feature within the IMP (Spectrum 2). Cu is not expected as a meaningful impurity in Ti-2, and Cu-containing IMPs have never been reported before in Ti-2. The cross-sectional image of this location (Figure B) shows that this Cu feature was present within the volume of the predominantly Fe-containing IMP. Despite the highly aggressive conditions, this Cu-containing aspect of the IMP was resistant to dissolution, likely owing to the stability of Cu metal at low potentials in deaerated acidic solutions.

8.

(A) FE-SEM image indicating the locations analyzed for EDX spot analysis (Table ) and (B) cross-section at the relocatable TixFe IMP for the specimen that was subjected to 16 h galvanostatic polarization at −1 mA/cm² in 0.5 M HCl at 60 °C.

2. EDX Spot Analysis Corresponding to Locations in Figure a (wt %) .

Spectrum	Ti	Fe	Cu	C	Cl	Si	K
Spectrum 1	97.2	ND	ND	2.6	0.1	0.1	ND
Spectrum 2	88.6	1.4	4.4	1.6	2.6	0.1	1.4

^aND = Not detected.

Beyond this Cu-containing IMP, Fe was detected in several areas of the map (Figure L). For example, Fe was detected in the center of a Ti _x Fe IMP that had undergone substantial dissolution (Figure L, Location II). This suggests that this IMP extended deep into the material, as some remained even after extensive etching.

Another region of Fe enrichment (Figure L, Location III) was observed at a site where no Fe was detected on the EDX map recorded on the specimen in the as-polished condition (Figure L). The apparent absence of Fe in the pre-experiment map at this location (Figure L) can be attributed to the relative-intensity-based signal displayed by the maps. This Fe signal (Figure L, Location III) is likely to originate from a subsurface Ti _x Fe intermetallic particle that had negligible intensity contribution in the pre-experiment map due to the abundance of surface-exposed Ti _x Fe IMPs. However, following the dissolution of most surface-exposed IMPs within the mapped region during the experiment, this buried location now appears as a region of relatively high intensity. The alternative possibility is that there was a dissolution-reposition event of Fe within the system; however, this seems improbable due to the high solubility of Fe ions under deaerated acidic conditions.

3.2.4.2. FIB Cross-Sections and TiH _x Growth Mechanism

Cross-sections were milled across the relocatable Ti _x Fe IMPs by using a focused ion beam (FIB) to determine the extent of IMP dissolution and to reveal the subsurface hydride growth process (Figure ). The exact location and orientation of the cross-section relative to the FE-SEM image analyzed in Figure can be found in the Supporting Information section.

9.

FIB cross-sections across the relocatable TixFe IMP in the (A) control specimen and (B–F) specimens that were subjected to varying periods of galvanostatic polarization at −1 mA/cm² in 0.5 M HCl at 60 °C.

The Ti _x Fe IMP control specimen experienced minimal corrosion damage (Figure A). No dark contrast regions were observed, except possibly for a small region near the surface on the right side of the cross-section. We propose that the dark contrast regions, which developed as the duration of polarization was extended, represented TiH _x . Given the minimal corrosion damage and hydride growth observed in the cross-section of the control specimen, any notable corrosion damage or hydride growth on the remaining specimens can be attributed to the galvanostatic polarization phase of the experiment.

For all specimens subjected to galvanostatic polarization, a noticeable dark contrast region developed around the IMP (Figure B–F). The images presented in Figure were collected in the backscattered electron (BSE) mode. In this mode, areas with lower average atomic numbers appear darker than areas with higher average atomic numbers. Therefore, it is plausible that this dark area corresponds to TiH _x , which has a lower average atomic number than the Ti matrix. This interpretation is supported by the topographic images (Figure ) and the development of dark regions in the EDX maps (Figure ), both of which suggest TiH _x formation around the IMPs after galvanostatic polarization. This is consistent with the XRD analysis, which confirmed the presence of substantial amounts of hydride on the surface of the polarized specimens (Figure ), further supporting the proposal that these large dark contrast regions are TiH _x . Below, the dark contrast regions will be referred to as TiH_x.

After 20 min of galvanostatic polarization, only minimal corrosion penetration was seen at the Ti _x Fe IMP (Figure B), and TiH _x had only formed around the IMP and not on the surrounding Ti matrix, indicating that TiH _x formation initiates at the locations of Ti _x Fe IMPs. After 20 min, the hydride had grown ∼1–2 μm both laterally and vertically into the material. More noticeable signs of IMP corrosion were observed after 1 h of galvanostatic polarization (Figure C). After 1 h, the TiH _x had grown ∼7 μm laterally and ∼4–5 μm vertically, much more than after 20 min of polarization. The development of a protrusion from the surface over each of the Ti _x Fe IMPs shown in Figure C is apparent in this cross-section (Figure C) and can be attributed to the expansion of the lattice as TiH _x precipitated. This image shows that while TiH _x formation initiates around a Ti _x Fe IMP, its subsequent growth spreads laterally and vertically into the Ti matrix, yielding an arc profile. The arc is a consequence of hydrogen ingress being rate-limited by diffusion. This preference for lateral hydride expansion is illustrated by comparing the vertical (∼4–5 μm) to the lateral growth (∼7 μm) of hydride after 1 h of polarization (Figure C). Additionally, the ability of TiH _x to catalyze proton reduction would lead to high surface coverages at the Ti/TiH₂/solution triple interface and faster lateral than vertical growth because lateral expansion would not require slow diffusion through the solid.

The evidence discussed in the topographic images (Figure ), EDX maps (Figure ), and cross-sectional images (Figure B,C) clearly indicates that hydride formation around Ti _x Fe IMPs is much more rapid than on the surface of the Ti matrix away from the IMPs. This microstructural preference can be attributed to a combination of factors, particularly the significantly higher exchange current density for proton reduction on Fe (i_o = 2.0–3.0 × 10^–6 A/cm²) , vs that on Ti (i_o = 5.0–6.3 × 10^–9 A/cm²). , Zhu et al., using scanning electrochemical microscopy (SECM), determined that specific locations on the surface of Ti-2 preferentially supported the reduction of a redox mediator. Grain boundary maps derived from SECM images revealed that these reactive locations are situated along grain boundaries and at triple points. It was proposed that these locations were Fe-containing IMPs, but no direct analytical proof was provided. Additionally, during crevice corrosion, these authors claimed that Ti _x Fe IMPs catalyzed proton reduction inside the crevice, leading to a decreased penetration depth and extent of intragranular corrosion. , The Ti _x Fe IMPs not only support a greater proportion of the cathodic current but also serve as preferential sites for H absorption due to the higher diffusion coefficient of H in the β-phase (9.1 × 10^–5 cm²/s) IMP than in the α-phase matrix (2.6 × 10^–10 cm²/s). Moreover, it is likely that there is less residual oxide on the Ti _x Fe IMPs than on the Ti matrix, which would act as an inhibitor of H⁺ reduction and H absorption. Finally, once hydride formation is initiated on the surface, it likely catalyzes the HER and increases the rate of hydride formation in an autocatalytic cycle.

There is substantial evidence in the literature to suggest that TiH _x formation initiates at IMPs, and the results described herein serve as the clearest evidence for this phenomenon. Noël observed frequent repassivation events on a freshly polished Ti-2 electrode in concentrated HCl solutions at 25 °C. It was proposed that during the active corrosion of Ti, Fe IMPs provide a site for the initiation of hydride formation, with the hydride then growing on the surface until it becomes sufficiently catalytic for proton reduction to enforce repassivation. That such a process can occur is supported by Liu et al., who demonstrated that a Ti-2 electrode that achieved an active corrosion condition subsequently established a passive open-circuit potential after the galvanostatic formation of TiH _x . In this study, time-of-flight secondary ion mass spectrometry (ToF-SIMS) analysis of a specimen that had a galvanostatically grown hydride displayed a nonuniform distribution of H, potentially due to the preferential formation of TiH _x at Fe-containing IMPs. Additionally, Yan et al. cathodically charged Ti-2 specimens and found evidence to suggest that Ti _x Fe IMPs are preferential locations for hydride formation using dynamic secondary ion mass spectrometry (D-SIMS). Finally, Tan et al. showed hydride growth initiation at Ti _x Fe IMPs due to manual polishing. The authors proposed that H pick-up on Ti-2 first occurs primarily at IMPs, leading to hydride formation once the solubility limit of the β-phase or the abutting grain boundary is reached.

The cross-section of the specimen after 4 h of polarization shows that the majority of the IMP dissolved, and the remaining portion resided deep within the original IMP location (Figure D). The EDX map of this specimen indicated that Fe was present only along the edge of the IMP (Figure H), with the cross-sectional view showing that the remaining Fe, detected with EDX, was from a portion of the IMP that was buried beneath the surface. The depth of this IMP was shallower than the IMP studied in the 1-h experiment; therefore, although the majority had dissolved after 4 h, it is difficult to say whether there was a significant increase in total dissolution from 1 to 4 h. Based on the subsurface dark area visible in the cross-section after 4 h (Figure D), the hydride had grown around the IMP to a depth of ∼5–6 μm below the surface and ∼7 μm laterally, which is similar to the size of the hydride region around the IMP after 1 h (Figure C). In contrast to the 1-h specimen, on the 4-h specimen, the hydride formed at locations on the Ti matrix and penetrated the metal along preferred pathways after 4 h (e.g., Figure D, Location I). Additionally, hydride formed ∼5 μm below the surface (Figure D, Location II) around a light-contrast feature that could have been a Ti _x Fe IMP. Since Fe IMPs form along the grain boundaries due to the preference of Fe for the β-phase, it is plausible that there exists a grain boundary linking the relocatable surface IMP to this subsurface IMP. The presence of a β-phase grain boundary would serve as a diffusion pathway for H, leading to hydride formation once the H concentration limit was exceeded.

After 10 h, the IMP appeared to be fully dissolved (Figure E); however, the EDX analysis (Figure J) indicates that a small amount of Fe was present at this location. Hydride grew to ∼4 μm below the surface of the IMP as well as at various locations on the Ti matrix. The depth of ∼4 μm after 10 h is less than the ∼5 μm found after 4 h (Figure D), which is likely due to differences in microstructural features at the different locations, such as in the number and nature of grain boundaries or differences in the composition of the Ti _x Fe IMP, which would lead to variations in HER kinetics. Furthermore, the Ti _x Fe IMP selected on the 10-h specimen was linear (Figure I) and, therefore, likely positioned along a single grain boundary. In contrast, the IMP selected for the 4-h specimen spread had three limbs and was likely situated across a triple point. Additionally, the difference in hydride depth could have been due to differences in the size of the Ti _x Fe IMPs. The IMP chosen for the 4-h experiments has a larger surface area and, therefore, presumably can support a larger amount of cathodic current, resulting in more hydride formation despite a shorter time of polarization.

After 16 h of galvanostatic polarization, the IMP was completely dissolved (Figures B and F), except for the Cu-rich region discussed above (Figure ). Due to the narrow width of the cross-section, approximately 12 μm, lateral information regarding the dimensions of hydride growth is limited (Figure F). The extent of hydride growth was similar to the results obtained for 4 and 10 h of polarization: the hydride had grown to a depth of ∼4–5 μm, formed in some areas on the Ti matrix, and grew into the metal along preferred directions. The cross-sections showed no significant visual differences between the 4-h, 10-h, and 16-h specimens; however, it is expected that overall surface coverage increases between 4 and 16 h due to the slowly increasing potential and the increasing TiH _x /Ti ratio seen by XRD.

Under the conditions tested in this study, the TiH _x layer growing at the IMPs’ locations reaches a thickness of ∼5 μm after 1 h, and further increases in polarization time do not increase the TiH _x layer thickness. For the TiH _x layer to grow, there must be diffusion of H atoms from the surface through the existing hydride layer to the metal-hydride interface. Due to the diffusion coefficient of H in TiH _x being lower than that in Ti metal, the layer acts as a H diffusion barrier. Phillips et al. determined a H diffusion coefficient of 4.3 × 10^–12 cm²/s for a hydride layer grown at −1 mA/cm² in 0.05 M H₂SO₄ at 25 °C on Ti-1. This is 2 orders of magnitude lower than the H diffusion coefficient in the α-Ti matrix (2.6 × 10^–10 cm²/s). Therefore, at a thickness of ∼5 μm, a steady-state condition between TiH _x growth and dissolution is established due to the extremely low rate of H diffusion from the surface to the metal-hydride interface that is required to grow the TiH _x layer (Figure ). At this steady-state condition, it is likely that the surface coverage of adsorbed H is at saturation, and since the hydride layer is limiting diffusion of H, a larger fraction of the adsorbed H evolves as H₂ rather than participating in TiH _x formation.

It is evident that at longer polarization times, TiH _x begins to form on the surface of the Ti matrix. Although the composition of the grain is uniform, the hydriding of the surface on the grain appears to be nonuniform (e.g., Figure D, Location I). These preferential matrix initiation locations could be due to features such as roughness or stress, and once the hydride has been initiated, further hydriding occurs at this location (i.e., H⁺ reduction/H absorption is catalyzed at this location rather than on the adjacent unhydrided matrix). Furthermore, at longer polarization times (Figure D–F), hydride growth into the metal occurs in preferential directions. This directional growth indicates preferential H diffusion pathways, possibly along grain boundaries or specific lattice planes.

After 10 and 16 h of galvanostatic polarization, the IMPs were essentially fully dissolved, yet the potential readings and Tafel slope analysis presented in Figure indicate that the surface of Ti-2 remained catalytic for the HER. When the cathodic current was first applied, Ti _x Fe IMPs supported the rapid reduction of H⁺ and H absorption, leading to local TiH _x formation and a rapid increase in potential. As the IMPs dissolved, this cathodic activity of the surface should have ceased if only the IMPs provided sites to catalyze H⁺ reduction. Therefore, the maintenance of catalysis appears to be attributable to the growing hydride layer that either originated at an IMP or formed more generally on the surface of the Ti matrix. This led to a slowly increasing potential that was typically observed after 4 h due to the increasing TiH _x surface coverage (Figure ).

The observation of IMP dissolution as a function of polarization time would not typically be expected under cathodic polarization, since the loss of metal inevitably involves anodic rather than cathodic processes, but Ti is extremely reactive, with E° = −1.63 V. Despite the passage of a higher cathodic current at the IMPs than on the Ti matrix, the greater depth of corrosion damage at the Ti _x Fe IMPs compared to that on the Ti matrix (e.g., Figure B) indicates that the IMPs are more anodically active than the Ti matrix, likely because the oxide over the IMPs is not as passivating as that over the Ti matrix. The control specimen, subject to only 15 min of active corrosion on an open circuit, showed small yet noticeable signs of preferential corrosion on the IMPs (Figures A and A), which supports the hypothesis that the Ti _x Fe IMPs are less stable than the Ti matrix in acidic conditions at 60 °C. In a study on crevice corrosion, Standish et al. claimed, but did not unequivocally demonstrate, that Ti _x Fe IMPs on Ti-2 would be more active anodically than the Ti matrix. When coupled with the catalysis of H⁺, this would lead to anodic dissolution of the Ti _x Fe IMP, releasing Ti³⁺ and Fe²⁺ to solution.

A possible further explanation for the enhanced anodic activity on cathodically polarized IMPs is that the Ti _x Fe IMPs undergo chemical dissolution after becoming hydrided. The cross-sectional images indicate that the Ti _x Fe IMP did not appear to become hydrided (i.e., did not develop a dark contrast region) but rather facilitated hydride growth in the surrounding matrix. It is therefore possible that the cathodically formed TiH _x produced directly from Ti in the Ti _x Fe IMP was unstable and dissolved rapidly after formation. This mechanism has been proposed for Mg dissolution upon forming Mg hydride, where the hydride first forms via a reaction between metal and adsorbed H (Reaction 5), followed by hydride dissolution in H₂O (Reaction 6). Here, the reactions were balanced to consider the formation of TiH₂.

$$\mathrm{Ti}+{\mathrm{2H}}_{\mathrm{Ads}}\to {\mathrm{TiH}}_2$$ 5

$${\mathrm{4TiH}}_2+{\mathrm{4H}}_2\mathrm{O}\to {\mathrm{4Ti}}^{\mathrm{3+}}+{\mathrm{4OH}}^{-}+{\mathrm{6H}}_2$$ 6

Furthermore, the formation of hydride from the Ti in the Ti _x Fe IMP would lead to further enrichment of Fe at these locations, rendering the IMP more unstable and anodically reactive.

In addition to Ti _x Fe IMP dissolution, the Ti matrix also began to dissolve, as indicated by the roughening of the surface as the polarization time increased (Figures and ). The roughening appears to have been on a subgrain-size scale and unassociated with grain boundaries. The anodic current was initially concentrated at the IMP, but once the IMP dissolved, the anodic current would have been supported by the Ti matrix. The roughening of the Ti matrix could also have been caused by the chemical dissolution of TiH _x via Reaction 6. Overall, the rate of TiH _x dissolution was likely less than the rate of hydride formation, leading to a surface coverage increase over time. However, surface roughening may have arisen from varying initiation times and the dissolution and formation rates of TiH _x across the surface.

4. Conclusions

We investigated the mechanism of TiH _x growth on Ti-2 under simulated crevice corrosion conditions and determined that TiH _x formation initiated at Ti _x Fe IMPs and caused the development of a protruding feature due to the volume increase associated with TiH _x precipitation. After initiation, TiH _x continued to grow both laterally and vertically in an arc shape around the Ti _x Fe IMP. After 1 h, the TiH _x layer reached a thickness of ∼5 μm, representing a steady-state thickness that did not change as the polarization time increased. After 4 h, the TiH _x formed at areas on the Ti matrix that were not associated with Ti _x Fe IMPs. Titanium hydride was observed to grow into the metal in preferential directions, possibly related to preferential H diffusion pathways such as grain boundaries and specific favorable lattice planes. The XRD data, electron microscopy observations, and the potential of the electrodes during galvanostatic polarization all support the observation of a rapidly growing hydride layer within the first 4 h, followed by slower growth over the remainder of the experiment.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcc.5c03765.

• Tafel slope analysis for the time points indicated in Figure (Figures S1–S7); electrochemical data for the specimen with relocatable Ti _x Fe IMPs (Figure S8); XRD reference data used to identify the peaks in the experimental XRD data (Figure S9); individual Ti and Fe EDX maps for the relocatable Ti _x Fe IMPs (Figures S10–S15); FE-SEM images orientated and marked to show the location where a FIB cross-section was milled (Figures S16–S22) (PDF)

The authors declare no competing financial interest.

Published as part of The Journal of Physical Chemistry C special issue “Jacek Lipkowski Festschrift”.