Removal of Oxygen from Commercially Pure Titanium Melt via a Two-Step Plasma-Arc Melting Process Using Hydrogen

Abstract

Oxygen is one of the principal impurities in off-grade Ti sponges and Ti scraps. To promote the effective use of these low-cost materials, developing reliable technologies for removing oxygen from Ti is essential. This study investigated the removal of oxygen from commercially pure Ti (CP Ti, Gr. 2) with an initial oxygen content of 0.127 mass% using a two-step plasma-arc melting process consisting of hydrogen plasma-arc melting and subsequent Ar plasma-arc melting. In the first step, plasma-arc melting under an Ar-H₂ atmosphere with an H₂ partial pressure of 0.5 atm resulted in dissolved hydrogen content of 2 mass% in the subsurface region of the Ti melt. This high hydrogen content could be attributed to the high hydrogen potential of the atomic hydrogen (H) gas generated in the plasma. Second, Ar plasma-arc melting at a plasma current of 300 A reduced the oxygen content in the subsurface region of the Ti melt to 0.034 mass%. When the plasma current was increased to 600 A, the oxygen content decreased to approximately 0.05 mass%. Although the reduction was less pronounced than that observed at 300 A, the oxygen content decreased more uniformly over a deeper region of the melt. Thermodynamic considerations suggest that the dissolved hydrogen introduced during the first step functioned as a deoxidizer, with the deoxidation driven by the different water vapor partial pressures between the first and second steps. This study clarifies the underlying deoxidation mechanism and the influence of melting conditions on oxygen removal.

1. Introduction

Oxygen is one of the principal impurities in off-grade Ti sponges and Ti scraps. To promote the effective use of these low-cost materials, extensive research focused on removing oxygen from Ti [1–5]. Various methods for solid Ti deoxidation have been reported [6, 7], including those involving Ca, combining Mg and H₂ [8, 9], and rare-earth metals [10–14]. Nevertheless, the deoxidation of solid Ti is not always efficient owing to kinetic limitations. Thus, removing oxygen during melting—an indispensable step in both Ti production and recycling—offers significant advantages. This approach effectively integrates refinement into the melting stage, thereby promoting Ti recycling.

Several research groups have investigated the deoxidation of Ti melt [15–22]. Yahata et al. investigated oxygen removal from Ti-Al-based melts via electron beam melting and achieved oxygen content as low as 0.01–0.05 mass% [15]; however, this process requires a large amount of Al. Bartosinski et al. examined the deoxidation of Ti-6Al-4V melts via pressure electroslag remelting with the addition of CaF₂ and Ca, achieving oxygen content reduction from 0.2 to 0.1–0.15 mass% [16] and from 1 to 0.7 mass% [17]. Nonetheless, the reproducibility of the process remains a significant concern.

Recently, Okabe et al. proposed a deoxidation process using a cold crucible in a high-frequency induction melting furnace in combination with fluxes containing rare-earth metals and their oxyhalides [18]. This method reduced the initial oxygen content from approximately 0.1 mass% to 0.02 mass%. However, rare-earth metals (which are deoxidizers) introduced into the Ti melt tend to form stable oxides with low solubility products in Ti [23, 24], potentially increasing the deformation resistance during subsequent processing and reducing the fatigue properties.

Deoxidation of the Ti melts through hydrogen plasma-arc melting offers several advantages: the deoxidation product (H₂O) does not remain in the Ti ingot; the deoxidizing agent hydrogen can be readily removed from the ingot. Moreover, plasma-arc melting is already employed in current Ti melting processes. Hepworth and Schuhmann [25] showed that the activity coefficient of oxygen decreases with reduced hydrogen content at 1073 K. Although they focused on Ti alloys, their findings indicates that the dissolved hydrogen may increase the activity coefficient of the dissolved oxygen even in Ti melt. Using this technique, Su et al. [19, 20] and Oh et al. [21] reported oxygen reductions from 0.12 to 0.028 mass% in a Ti-6Al-4V melt [19] and from 0.23 to 0.084 mass% in a commercially pure Ti (CP Ti) melt [21], respectively. However, the thermodynamic understanding of the underlying deoxidation mechanisms remains inadequate.

Consequently, we proposed a two-step plasma-arc melting process for deoxidizing a Ti melt [22]. As shown in Fig. 1, hydrogen plasma-arc melting was first used to introduce hydrogen into a Ti melt (first step), followed by Ar plasma-arc melting (second step), during which the oxygen in the melt reacted with the dissolved hydrogen and was removed as H₂O(g) from the system. In contrast to the findings of Su et al. [19, 20] and Oh et al. [21], the second step is required as no evident decrease in oxygen content was observed in the first step. These two steps can be performed continuously without solidifying the melt. Using this process, the oxygen content of the Ti melt was successfully reduced from approximately 1.5 to 0.7 mass% [22].

Fig. 1

Schematic of two-step plasma-arc melting. (online color)

Although the two-step process may appear complex, hydrogen plasma-arc melting inherently requires a subsequent dehydrogenation step. Therefore, both the dehydrogenation and deoxidation processes can be considered the second step. This study investigated deoxidation behavior in a region with a lower oxygen content than that examined in our previous study [22], using a two-step plasma-arc melting process to clarify the influence of melting conditions on oxygen removal, along with the underlying deoxidation mechanism.

2. Experimental Procedures

In addition to the deoxidation of the Ti melt, this study investigated the molten zone formed during plasma-arc melting. Sections 2.1 and 2.2 describe the experimental methods used for the deoxidation process and evaluation of the molten zone, respectively.

2.1 Deoxidation

2.1.1 Plasma-arc melting

A 500-g forged rod of CP Ti (Gr. 2, φ = 30 mm) with an initial oxygen content of 0.127 mass% was used as the raw material for melting. Before melting, the rod surface was wet-polished and subjected to acid pickling to remove the surface oxide film. The pickling solution was prepared by mixing hydrofluoric acid (HF, 46%, Morita Chemical Industries, Co., Ltd.), nitric acid (HNO₃, 28163-70, Kanto Chemical Co., Inc.), and Milli-Q water at a volume ratio of HF:HNO₃:H₂O = 2:13:85.

The plasma-arc melting furnace (AF-102-207CP, Nihon Tokushu Kikai Co., Ltd.)—including the water-cooled Cu hearth configuration—was identical to that used in our previous study [22]. The two-step plasma-arc melting process consisted of hydrogen plasma-arc melting conducted under an Ar-H₂ atmosphere followed by Ar plasma-arc melting under an Ar atmosphere. In the first step, the melting conditions were varied as follows: the melting atmosphere was set to Ar with 0, 0.1, 0.3, or 0.5 atm of H₂; the melting time was 0.3, 0.9, or 2.1 ks; the total gas flow rate (at 298 K and 1 atm) was 10, 20, or 30 L·min⁻¹; the plasma current was 300 A or 500 A. Ultrahigh-purity Ar gas (G-1, >99.9999 vol%, O₂ < 0.1 vol. ppm) and ultrahigh-purity H₂ gas (G-1, >99.99999 vol%, O₂ < 0.02 vol. ppm) were used as the plasma gases. The total gas pressure was set to 1 atm. Following the first-step melting, H₂ gas introduction was terminated, and the process proceeded directly to the second-step melting without solidifying the Ti melt. In the second step, the melting conditions were varied as follows: the melting time was 0.3, 0.9, or 1.8 ks; the total gas flow rate was 10, 20, or 30 L·min⁻¹; the plasma current was 300, 500, or 600 A. After the second-step melting, the plasma power was turned off, and the Ti melt was allowed to cool within the water-cooled Cu hearth. The melting conditions for the first and second steps are presented in Table 1.

Table 1 Melting conditions used in this study.

2.1.2 Direct rapid solidification of the Ti melt

Following the second-step melting, some ingots were obtained via the direct rapid solidification of the Ti melt in addition to conventional cooling within the Cu hearth. As illustrated in Fig. 2, a carbon mold was installed beneath the water-cooled Cu hearth. Immediately after the designated melting time of the second step, the Cu hearth was tilted, allowing the Ti melt to flow into the carbon mold, where it rapidly solidified. After the solidified ingot was collected, the residual Ti remained in the Cu hearth, which was again subjected to two-step plasma-arc melting, followed by direct rapid solidification. The conditions for the first-step melting were as follows: Ar–50%H₂ plasma gas, the melting time was 0.9 ks, the total gas flow rate was 20 L·min⁻¹, and the plasma current was 500 A. The second-step melting was performed with a melting time of 1.8 ks, a total gas flow rate of 30 L·min⁻¹, and a plasma current of 600 A.

Fig. 2

Schematic of direct rapid solidification of Ti after two-step plasma-arc melting.

2.1.3 Oxygen and hydrogen analysis

The hemispherical Ti ingots obtained via cooling in a water-cooled Cu hearth were sectioned, as shown in Fig. 3. The samples were extracted from the Top, Middle, and Inner positions beneath the plasma-arc spot. The extracted samples were approximately 3 × 6 × 1.5 mm in size, with a mass of approximately 0.1 g. The hydrogen and oxygen contents in the extracted samples were measured via an inert gas fusion–thermal conductivity method using a hydrogen analyzer (RHEN602, LECO Corp.) and an inert gas fusion–infrared absorption method using an oxygen/nitrogen analyzer (ONH836, LECO Corp.), respectively. For samples with high hydrogen contents, dehydrogenation heat treatment was conducted before oxygen analysis by holding the samples under vacuum at 1023 K for 3.6 ks. Before the oxygen analysis, the sample surfaces were subjected to wet polishing and acid pickling to remove the surface oxide film using a solution of HF:HNO₃:H₂O = 1:10:10. As described in Section 3.1, the molten zone produced by plasma-arc melting extended >10 mm from the melt surface; therefore, the locations of the three types of samples were confirmed to be in the melted zone. The ingots obtained via the direct rapid solidification method described in Section 2.1.2 were randomly sectioned to a mass of approximately 0.1 g and subjected to oxygen analysis.

Fig. 3

Schematic of the analysis positions of the Ti ingots.

2.2 Molten-zone depth evaluation

The molten-zone depth was evaluated through a melting experiment using a forged α + β type Ti-6Al-4V alloy (590 g) machined into a hemispherical shape identical to that of the water-cooled Cu hearth. Using the α + β type Ti-6Al-4V alloy as a raw material allows the evaluation of the molten zone via microstructural observations after melting. A hemispherical Ti-6Al-4V alloy was used for the melting experiments after cleaning in ethanol. The melting conditions were either Ar plasma-arc melting for 1.2 ks or hydrogen plasma-arc melting in an Ar–30%H₂ atmosphere for 0.9 ks. Under both conditions, the total gas flow rate was maintained at 20 L·min⁻¹, with the plasma current set to 450–500 A. After melting, a semicircular cross-sectional part—including the center of the ingot—was cut to a width of 5 mm, followed by division along its central axis, resulting in two sector-shaped pieces. The sector-shaped pieces were wet-polished using SiC abrasive papers (up to #2400) and buff-polished using Al₂O₃ powder with a diameter of 0.05 µm to obtain a mirror finish. Subsequently, the polished surface was etched via immersion for 10 s in a mixed acid solution (HF:HNO₃:H₂O = 2:13:85). Microstructural observations were performed using a stereomicroscope (Z6APO, Leica Microsystems).

3. Results

3.1 Molten-zone depth

Figures 4(a) and 4(b) show the cross-sectional microstructures of the sector-shaped Ti-6Al-4V alloy ingot after Ar plasma-arc melting and hydrogen plasma-arc melting in an Ar–30%H₂ atmosphere, respectively. As shown in Fig. 4(a), a coarse-grained region was evident, extending 10 mm in depth, starting 12–13 mm beneath the outermost surface of the ingot. This region could be interpreted as the heat-affected zone (HAZ), with the area above it identified as the solidified region. Consequently, the boundary between the solidified region and the HAZ could be considered to delineate the melted zone, indicating that the depth of the molten zone was approximately 12.5 mm. By contrast, for hydrogen plasma-arc melting (Fig. 4(b)), a clear boundary between the solidified region and HAZ could not be identified, making it difficult to assess the melted zone. Generally, hydrogen plasma melting produces a deeper molten zone than Ar plasma-arc melting. Therefore, a molten zone of at least 12 mm could be achieved in both melting processes.

Fig. 4

Microstructures of the cross-section of the Ti-6Al-4V ingot after (a) Ar plasma-arc melting and (b) hydrogen (Ar–30%H₂) plasma-arc melting. (online color)

3.2 Hydrogen content after first-step melting

Figure 5 shows the hydrogen content in the Ti melt after the first-step plasma-arc melting under various H₂ partial pressures in the plasma gas, ranging from 0 to 0.5 atm at a melting time of 0.9 ks, plasma current of 500 A, and total gas flow rate of 20 L·min⁻¹. In a typical plasma reaction environment, hydrogen exists predominantly in its atomic state [26]. Consequently, if molecular hydrogen introduced into the plasma gas completely dissociates into atomic hydrogen, the partial pressure of atomic hydrogen (P_H) can be calculated using eq. (1).

  

\begin{equation} P_{\text{H}} = \frac{2P_{\text{H${_{2}}$}}}{2P_{\text{H${_{2}}$}} + P_{\text{Ar}}} \end{equation} (1)

where $P_{\text{H}_{2}}$ and P_Ar denote the partial pressures of molecular hydrogen (H₂) and Ar in the plasma gas, respectively.

Fig. 5

Effect of the hydrogen partial pressure on the hydrogen content in Ti after the first-step melting for a melting time of 0.9 ks, plasma current of 500 A, and total gas flow rate of 20 L·min⁻¹.

The upper horizontal axis in Fig. 5 represents $P_{\text{H}_{2}}$ introduced into the plasma gas. The dissolved hydrogen content was the highest in the Top position of each ingot for all the investigated hydrogen gas partial pressures. Moreover, the hydrogen content increased with the hydrogen gas partial pressure, reaching a maximum of 2 mass% at an H₂ partial pressure of 0.5 atm.

Figure 6 shows the effects of the melting time, plasma current, and total gas flow rate on the hydrogen content at the Top and Inner positions after the first-step melting at an H₂ partial pressure of 0.5 atm in the plasma gas. For a melting time ≥0.9 ks, the hydrogen content at the Top position stabilized at approximately 2 mass% (Fig. 6(a)). The plasma current and total gas flow rate did not affect the hydrogen content, as shown in Figs. 6(b) and 6(c), respectively. Accordingly, the optimal melting time was set to 0.9 ks to increase the hydrogen content at the Top position. The plasma current was set to 500 A to expand the molten zone, and the total gas flow rate was set to 20 L·min⁻¹. The second-step melting was conducted under the optimized first-step conditions.

Fig. 6

Effects of the (a) melting time, (b) plasma current, and (c) total gas flow rate on the hydrogen content in Ti of the Top and Inner positions after the first-step melting at an H₂ partial pressure of 0.5 atm in plasma gas. The other melting conditions were as follows: (a) plasma current of 500 A and total gas flow rate of 20 L·min⁻¹; (b) melting time of 0.9 ks and total gas flow rate of 20 L·min⁻¹; and (c) plasma current of 500 A and melting time of 0.9 ks.

3.3 Oxygen content after first-step melting

Figure 7 shows the oxygen content in the Ti melt after the first-step plasma-arc melting under the same conditions as Fig. 5. As demonstrated in our previous study [22], no decrease in oxygen content was detected, indicating that a second-step melting is necessary.

Fig. 7

Effect of the hydrogen partial pressure on the oxygen content in Ti after the first-step melting for a melting time of 0.9 ks, plasma current of 500 A, and total gas flow rate of 20 L·min⁻¹.

3.4 Oxygen content after second-step melting

Figure 8 shows the effects of melting time, plasma current, and total gas flow rate on the oxygen content after the second-step melting. The oxygen content decreased as the melting time increased beyond 0.9 ks (Fig. 8(a)) for the Top and Middle positions, reaching approximately 0.08 mass% after melting for 1.8 ks. A decline in oxygen content was evident at low plasma currents (Fig. 8(b)). When the current was 300 A, insufficient melting was observed on the Ti surface near the water-cooled Cu hearth during Ar plasma-arc melting, which could be attributed to the reduced plasma power. An increase in total gas flow rate reduced the oxygen content (Fig. 8(c)) at the Top and Middle positions, and the lowest oxygen content was found at a gas flow rate of 30 L·min⁻¹.

Fig. 8

Effects of the (a) melting time, (b) plasma current, and (c) total gas flow rate on the oxygen content in Ti after the second-step melting. The other melting conditions were as follows: (a) plasma current of 500 A and total gas flow rate of 20 L·min⁻¹; (b) melting time of 0.9 ks and total gas flow rate of 20 L·min⁻¹; and (c) plasma current of 500 A and melting time of 0.9 ks.

Based on the above results, two conditions—low-current and high-current conditions—were set for the second-step melting process. The low-current condition was defined as the combination of melting parameters that resulted in the lowest oxygen content, which constituted a melting time of 1.8 ks, plasma current of 300 A, and total gas flow rate of 30 L·min⁻¹. The high-current condition retained the same melting time and total gas flow rate as the low-current condition, but the plasma current was increased to 600 A to expand the molten zone.

Figure 9 shows the oxygen content of the Ti after melting under the low-current and high-current conditions. Under the low-current condition, the oxygen content at the Top position was 0.034 mass%, the lowest value observed in this study. However, the oxygen content tended to increase with depth. By contrast, under the high-current condition, the oxygen content remained constant at approximately 0.05 mass% across all positions (Top, Middle, and Inner). This uniformity could be attributed to the expanded molten zone resulting from the increased plasma power.

Fig. 9

Oxygen content in the Ti melt under low-current and high-current conditions.

The appearance of the specimen obtained from the direct rapid solidification experiment is shown in Fig. 10(a). The masses of the ingots obtained from the first and second solidification experiments were 38 and 48 g, respectively. The results of oxygen content analysis for each specimen are shown in Fig. 10(b). In both the first and second experiments, the oxygen content of the ingots was lower than that of the raw CP Ti (0.127 mass%), confirming that the deoxidation reactions during the two-step plasma-arc melting process proceeded effectively within the molten zone.

Fig. 10

(a) Appearance and (b) oxygen content of the cast Ti produced by the direct rapid solidification experiments. (online color)

4. Discussion

The corresponding chemical reactions at each step can be expressed as follows:

  

\begin{equation} \text{First step}\quad \text{H$_{2}$(g)} = \text{2H(g)} \end{equation} (2)

  

\begin{equation} \phantom{\text{First step}}\quad\text{H(g)} = \underline{\text{H}}\text{(mass%, in Ti(l))} \end{equation} (3)

  

\begin{align} \text{Second step}\quad &2\underline{\text{H}}\text{(mass%, in Ti(l))} \\ &+ \underline{\text{O}}\text{(mass%, in Ti(l))} = \text{H$_{2}$O(g)} \end{align} (4)

In the following discussion, the gas constant is denoted as R (8.3145 J·mol⁻¹·K⁻¹); the temperature is denoted as T (K); the standard Gibbs free energy change of equation (i) is denoted as $\Delta G_{(i)}^{ \circ }$ (J); the partial pressures of molecular and atomic hydrogen gas are denoted as $P_{\text{H}_{2}}$ (atm) and P_H (atm), respectively; the partial pressure of water vapor is denoted as $P_{\text{H}_{2}\text{O}}$ (atm); the contents of dissolved hydrogen and oxygen in the Ti melt are denoted as C_H (mass%) and C_O (mass%), respectively.

4.1 Hydrogen content

The hydrogen dissolution reaction of molecular hydrogen gas into the Ti melt, previously investigated by Lakomskiy and Kalinyuk [27], is represented by eq. (5). The hydrogen content of the Ti melt follows Sieverts’ law. According to their study, the standard Gibbs free energy change in eq. (5) can be expressed by eq. (6):

  

\begin{equation} \text{H$_{2}$(g)} = 2\underline{\text{H}}\text{(mass%, in Ti(l))} \end{equation} (5)

  

\begin{equation} \Delta G_{(5)}^{\circ}/\text{J} = - 97396 + 79.30T \end{equation} (6)

As mentioned in Section 3.2, hydrogen in the plasma can be assumed to exist in an atomic state (H(g)). The standard Gibbs free energy change for the dissociation of molecular hydrogen into atomic hydrogen given by eq. (2) can be expressed by eq. (7) using the NIST-JANAF Thermochemical Tables [28]:

  

\begin{equation} \Delta G_{(2)}^{\circ}/\text{J} = 461078 - 123.19T \end{equation} (7)

The standard Gibbs free energy change of eq. (3) can be calculated using eqs. (6) and (7) and expressed by eq. (8):

  

\begin{equation} \Delta G_{(3)}^{\circ}/\text{J} = - 279237 + 101.245T \end{equation} (8)

If the dissolved hydrogen in the Ti melt obeys Henry’s law, the hydrogen content introduced into the melt by molecular and atomic hydrogen gases can be expressed as eqs. (9) and (10), using the standard Gibbs free energy changes shown in eqs. (6) and (8), respectively:

  

\begin{equation} \log (C_{\underline{\text{H}}}) = \frac{1}{2}\left(\log P_{\text{H${_{2}}$}} - \frac{\Delta G_{(5)}^{\circ}}{2.303RT} \right) \end{equation} (9)

  

\begin{equation} \log (C_{\underline{\text{H}}}) = \log P_{\text{H}} - \frac{\Delta G_{(3)}^{\circ}}{2.303RT} \end{equation} (10)

The content of dissolved hydrogen in the Ti melt (log(C_H)) calculated using eqs. (9) and (10) is depicted in Fig. 11 with respect to the inverse temperature of the melt (1/T), where the partial pressure of the molecular hydrogen ($P_{\text{H}_{2}}$) was 0.5 atm and the partial pressure of the atomic hydrogen (P_H) was 0.67 atm, calculated using eq. (1).

Fig. 11

Thermodynamic calculation of the hydrogen content in the Ti melt.

The dotted line in Fig. 11 represents the calculated values considering the equilibrium of eq. (2), i.e., the dissociation equilibrium between molecular and atomic hydrogen. At lower temperatures (where the degree of dissociation was low), the dotted line approached the behavior corresponding to hydrogen introduction via molecular hydrogen. By contrast, at higher temperatures (where the degree of dissociation approaches unity), it asymptotically followed the behavior associated with atomic hydrogen introduction. The hydrogen content after the first-step melting under the molecular hydrogen partial pressure of 0.5 atm was approximately 2 mass% at the Top position (Fig. 6), considerably higher than the values estimated assuming hydrogen introduction via molecular hydrogen or dissociation equilibrium between molecular and atomic hydrogen, as shown in Fig. 11. This result suggests that the atomic hydrogen generated in the plasma remained present even on the surface of the Ti melt, where the temperature was lower than that of the plasma, and that hydrogen was introduced into the melt in accordance with the high hydrogen potential of atomic hydrogen. Consequently, compared with conventional hydrogen gas dissolution, hydrogen plasma-arc melting increased the dissolved hydrogen content, which contributed to the progression of the deoxidation reaction during the second-step melting.

4.2 Deoxidation mechanism

If the dissolved oxygen and hydrogen in the Ti melt obey Henry’s law, the oxygen content in the Ti melt (log(C_O)) in eq. (4) can be expressed by eq. (11) at equilibrium.

  

\begin{equation} \log (C_{\underline{\text{O}}}) = - 2\log (C_{\underline{\text{H}}}) + \log P_{\text{H${_{2}}$O}} + \frac{\Delta G_{(4)}^{\circ}}{2.303RT} \end{equation} (11)

The standard Gibbs free energy changes shown in eqs. (12) and (13) ($\Delta G_{(12)}^{ \circ }$ and $\Delta G_{(13)}^{ \circ }$, respectively) are required to calculate the standard Gibbs free energy change in eq. (4)—that is, $\Delta G_{(4)}^{ \circ }$.

  

\begin{equation} \text{H$_{2}$(g)} + \frac{1}{2}\text{O$_{2}$(g)} = \text{H$_{2}$O(g)} \end{equation} (12)

  

\begin{equation} \text{O$_{2}$(g)} = 2\underline{\text{O}}\ \text{(}\textit{mass}\text{%, in Ti(l))} \end{equation} (13)

$\Delta G_{(12)}^{ \circ }$ and $\Delta G_{(13)}^{ \circ }$ can be derived from eqs. (14) and (15), respectively, using the NIST-JANAF Thermochemical Tables [28] and the data reported by Fitzner [29]:

  

\begin{equation} \Delta G_{(12)}^{\circ}/\text{J} = - 252610 + 58.49T \end{equation} (14)

  

\begin{equation} \Delta G_{(13)}^{\circ}/\text{J} = - 980730 + 147.58T \end{equation} (15)

According to eqs. (6), (14), and (15), $\Delta G_{(4)}^{ \circ }$ can be expressed as

  

\begin{equation} \Delta G_{(4)}^{\circ}/\text{J} = 335151 - 94.60T \end{equation} (16)

Using eq. (16), eq. (11) can be rewritten as

  

\begin{equation} \log (C_{\underline{\text{O}}}) = - 2\log (C_{\underline{\text{H}}}) + \log P_{\text{H${_{2}}$O}} + \frac{17503}{T} - 4.940 \end{equation} (17)

Clarifying the relationship between the dissolved hydrogen and dissolved oxygen shown in eq. (17) requires determining the melt temperature (T) and partial pressure of water vapor ($P_{\text{H}_{2}\text{O}}$). The melt temperature can be estimated based on the dissolved hydrogen content observed in the first step. Specifically, the melt temperature was defined as the temperature at which the measured hydrogen content of C_H = 2 mass% was achieved via the reaction described by eq. (3). In other words, the temperature at which the C_H = 2 mass% line (dashed line in Fig. 11) intersects the calculated C_H values based on eq. (3) in Fig. 11—that is, approximately 2500 K—was used as the melt temperature. At this temperature, eq. (17) can be written as eq. (18).

  

\begin{equation} \log (C_{\underline{\text{O}}}) = - 2\log (C_{\underline{\text{H}}}) + \log P_{\text{H${_{2}}$O}} + 2.061 \end{equation} (18)

The partial pressure of the water vapor was expected to differ considerably between the first and second steps. For the second step, which involved Ar plasma-arc melting, the water vapor partial pressure ($P_{\text{H}_{2}\text{O}}^{\text{Ar}}$) was assumed to be 5.4 × 10⁻⁷ atm, based on the dew point of the supplied Ar gas (193 K). By contrast, for the first step involving hydrogen plasma-arc melting in the Ar-H₂ atmosphere, the water vapor partial pressure ($P_{\text{H}_{2}\text{O}}^{\text{Ar–H}_{2}}$) was estimated using the equilibrium shown in eq. (4), which required the content of dissolved hydrogen and oxygen in the first step. The measured content of dissolved hydrogen at the Top position (2 mass%) and an initial oxygen content (0.127 mass%) were adopted because deoxidation did not proceed during the first step. By substituting these values into eq. (18), the partial pressure of water vapor was $P_{\text{H}_{2}\text{O}}^{\text{Ar–H}_{2}} = 4.4 \times 10^{ - 3}$ atm. The water vapor pressure calculated using eq. (4) indicates the pressure at the interface between the Ti melt and gas phase.

Figure 12 shows the relationship between the dissolved oxygen and hydrogen content in the Ti melt obtained at the estimated melting temperatures and the water vapor partial pressures for the first and second steps. Lines A and B in the figure represent the relationships for the first and second steps, respectively. As reported by Okabe et al. [18], $\Delta G_{(13)}^{ \circ }$ values exhibited significant uncertainty. For eq. (18), the $\Delta G_{(13)}^{ \circ }$ value reported by Fitzner [29] was used. Several other research groups [30–33] also reported $\Delta G_{(13)}^{ \circ }$ values, with some of them [30, 31, 33] appearing to be lower than Fitzner’s value. Therefore, Fig. 12 contains a solid Line B calculated using eq. (18) based on Fitzner’s $\Delta G_{(13)}^{ \circ }$ value [29], as well as a dotted Line B calculated using a $\Delta G_{(13)}^{ \circ }$ value of −700 kJ at 2500 K. The −700 kJ value closely approximates the average of two values; one is −612 kJ reported by Fitzner [29] at 2500 K, and the other is −763 kJ extrapolated to 2500 K from the data provided by Kobayashi and Tsukihashi [30]. The dotted Line B is described by eq. (19).

  

\begin{equation} \log (C_{\underline{\text{O}}}) = - 2\log (C_{\underline{\text{H}}}) + \log P_{\text{H${_{2}}$O}} + 2.982 \end{equation} (19)

Notably, Line A is independent of the $\Delta G_{(13)}^{ \circ }$ value because it is based on the experimentally measured hydrogen and oxygen contents obtained in the first-step melting. In the figure, the changes in the dissolved oxygen and hydrogen contents are indicated by dashed arrows. The initial state—i.e., the condition before melting—is represented by a filled circle. Considering the hydrogen content was close to zero before the first step, it was plotted at log(C_H) = −2 for convenience. During the first step, the oxygen content remained practically unchanged, whereas the hydrogen content increased. Consequently, the state shifted from the filled circle to the open circle on Line A, corresponding to the value calculated using $P_{\text{H}_{2}\text{O}}^{\text{Ar–H}_{2}}$. In the second step, the state moved further from the open circle on Line A toward the open squares on Line B. With the decrease in hydrogen content, the deoxidation described by eq. (3) proceeded. In other words, a decrease in the chemical potential of H₂O—i.e., the deoxidation product—was essential for deoxidation. The position of the open square was set using an oxygen content of 0.034 mass%—obtained via deoxidation in this study (Fig. 9). If the deoxidation reaction expressed by eq. (4) was the sole mechanism consuming dissolved hydrogen in the second step, the oxygen content would have decreased significantly compared to that observed in this study. This discrepancy detected in the second step was presumed to arise from the competition between the hydrogen molecule formation reaction (eq. (20)) and the deoxidation reaction (eq. (4)).

  

\begin{equation} 2\underline{\text{H}} \text{(}\textit{mass}\text{%, in Ti(l))} = \text{H$_{2}$(g)} \end{equation} (20)

As the hydrogen potential in the Ar plasma-arc melting (second step) was low, only dehydrogenation was expected to occur in the later stage of the second step, resulting in a transition from the open square to the open triangle, as shown in Fig. 12.

Fig. 12

Schematic of the changes in the hydrogen and oxygen content in the Ti melt during two-step plasma-arc melting.

In this study, the deoxidation reaction did not proceed during the first step of the two-step plasma-arc melting process. In principle, the one-step deoxidation in the first step, i.e., a shift from the filled circle toward the lower-oxygen-content region on Line A, is thermodynamically feasible. However, this shift has not yet been achieved, and the underlying reason remains unclear. Some factors are presumed to inhibit the removal of the deoxidation product (H₂O) from the Ti melt surface during hydrogen plasma-arc melting.

5. Conclusions

The deoxidation behavior of a Ti melt with an initial oxygen content of 0.127 mass% during two-step plasma-arc melting was investigated, with a particular focus on the melting conditions and deoxidation mechanisms. The following conclusions were drawn:

1. (1)    The hydrogen content in the subsurface region of the Ti melt reached 2 mass% under the first-step condition of an H₂ partial pressure of 0.5 atm in plasma gas, melting time of 0.9 ks, total gas flow rate of 20 L·min⁻¹, and plasma current of 500 A. This high hydrogen content could be attributed to the high hydrogen potential of the atomic hydrogen generated in the plasma.
2. (2)    The oxygen content in the subsurface region of the Ti melt was reduced from 0.127 to 0.034 mass% under the second-step condition of a melting time of 1.8 ks, total gas flow rate of 30 L·min⁻¹, and plasma current of 300 A. When the plasma current in the second step was increased to 600 A, the oxygen content decreased to approximately 0.05 mass%. This condition resulted in a more uniform reduction in the oxygen content throughout the deeper region of the melt compared with the 300-A case.
3. (3)    A reduction in oxygen content was also confirmed in the cast Ti produced by direct rapid solidification after the second step, indicating that the deoxidation reaction proceeded within the molten zone.
4. (4)    The hydrogen introduced during the first step acted as a deoxidizer in the second step, wherein the deoxidation reaction was driven by the different water vapor partial pressures between the two steps. No deoxidation reaction occurred during the first step alone.

Acknowledgments

This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan (grant number: 22K18305) and The Light Metal Educational Foundation, Inc.

REFERENCES

• 1)   O.  Takeda and  T.H.  Okabe: Current status of titanium recycling and related technologies, JOM 71 (2019) 1981–1990. doi:10.1007/s11837-018-3278-1
• 2)   R.O.  Suzuki,  S.  Natsui and  T.  Kikuchi: Recent studies on titanium refining: 2017–2020, Mater. Trans. 62 (2021) 905–913. doi:10.2320/matertrans.MT-M2020373
• 3)   Y.  Xia,  J.  Zhao,  Q.  Tian and  X.  Guo: Review of the effect of oxygen on titanium and deoxygenation technologies for recycling of titanium metal, JOM 71 (2019) 3209–3220. doi:10.1007/s11837-019-03649-8
• 4)   Y.  Zhang,  W.  Lu,  P.  Sun,  Z.Z.  Fang,  S.  Qiao,  Y.  Zhang and  S.  Zheng: Deoxygenation of Ti metal: A review of processes in literature, Int. J. Refract. Met. Hard Mater. 91 (2020) 105270. doi:10.1016/j.ijrmhm.2020.105270
• 5)   M.  Niinomi,  T.  Narushima and  T.  Nakano: Recent research and development in the processing, microstructure, and properties of titanium and its alloy, Mater. Trans. 65 (2024) 1600–1611. doi:10.2320/matertrans.MT-M2024082
• 6)   T.H.  Okabe,  R.O.  Suzuki,  T.  Oishi and  K.  Ono: Thermodynamic properties of dilute titanium-oxygen solid solution in beta phase, Mater. Trans. JIM 32 (1991) 485–488. doi:10.2320/matertrans1989.32.485
• 7)   Y.  Zhang,  Z.Z.  Fang,  Y.  Xia,  Z.  Huang,  H.  Lefler,  T.  Zhang,  P.  Sun,  M.L.  Free and  J.  Guo: A novel chemical pathway for energy efficient production of Ti metal from upgraded titanium slag, Chem. Eng. J. 286 (2016) 517–527. doi:10.1016/j.cej.2015.10.090
• 8)   Y.  Zhang,  Z.Z.  Fang,  P.  Sun,  T.  Zhang,  Y.  Xia,  C.  Zhou and  Z.  Huang: Thermodynamic destabilization of Ti-O solid solution by H₂ and deoxygenation of Ti using Mg, J. Am. Chem. Soc. 138 (2016) 6916–6919. doi:10.1021/jacs.6b00845
• 9)   Y.  Xia,  Z.Z.  Fang,  Y.  Zhang,  H.  Lefler,  T.  Zhang,  P.  Sun and  Z.  Huang: Hydrogen assisted magnesiothermic reduction (HAMR) of commercial TiO₂ to produce titanium powder with controlled morphology and particle size, Mater. Trans. 58 (2017) 355–360. doi:10.2320/matertrans.MK201628
• 10)   C.  Zheng,  T.  Ouchi,  A.  Iizuka,  Y.  Taninouchi and  T.H.  Okabe: Deoxidation of titanium using Mg as deoxidant in MgCl₂-YCl₃ flux, Metall. Mater. Trans. B 50 (2019) 622–631. doi:10.1007/s11663-018-1494-2
• 11)   A.  Iizuka,  T.  Ouchi and  T.H.  Okabe: Development of a new titanium powder sintering process with deoxidation reaction using yttrium metal, Mater. Trans. 61 (2020) 758–765. doi:10.2320/matertrans.MT-M2019340
• 12)   A.  Iizuka,  T.  Ouchi and  T.H.  Okabe: Ultimate deoxidation method of titanium utilizing Y/YOCl/YCl₃ equilibrium, Metall. Mater. Trans. B 51 (2020) 433–442. doi:10.1007/s11663-019-01742-6
• 13)   L.  Kong,  T.  Ouchi and  T.H.  Okabe: Deoxidation of Ti using Ho in HoCl₃ flux and determination of thermodynamic data of HoOCl, J. Alloy. Compd. 863 (2021) 156047. doi:10.1016/j.jallcom.2020.156047
• 14)   G.  Kamimura,  T.  Ouchi and  T.H.  Okabe: Deoxidation of titanium using cerium chloride flux for upgrade recycling of titanium scraps, Mater. Trans. 63 (2022) 893–902. doi:10.2320/matertrans.M-M2022805
• 15)   T.  Yahata,  T.  Ikeda and  M.  Maeda: Deoxidation of molten titanium by electron-beam remelting technique, Metall. Trans. B 24 (1993) 599–604. doi:10.1007/BF02673175
• 16)  M. Bartosinski, J. Reitz and B. Friedrich: Modelling the oxygen content of titanium during deoxidation in the pressure electroslag remelting (PESR) process, (European Metallurgical Conference, EMC 2013, 2013) pp. 1085–1092.
• 17)   M.  Bartosinski,  S.  Hassan-Pour,  B.  Friedrich,  S.  Ratiev and  A.  Ryabtsev: Deoxidation limits of titanium alloys during pressure electro slag remelting, IOP Conf. Series Mater. Sci. Eng. 143 (2016) 012009. doi:10.1088/1757-899X/143/1/012009
• 18)   T.H.  Okabe,  G.  Kamimura,  T.  Ikeda and  T.  Ouchi: Direct production of low-oxygen concentration titanium from molten titanium, Nat. Commun. 15 (2024) 5015. doi:10.1038/s41467-024-49085-4
• 19)   Y.  Su,  L.  Wang,  L.  Luo,  X.  Jiang,  J.  Guo and  H.  Fu: Deoxidation of titanium alloy using hydrogen, Int. J. Hydrogen Energ. 34 (2009) 8958–8963. doi:10.1016/j.ijhydene.2009.08.053
• 20)   B.  Jiang,  L.  Wang,  G.  Zhou,  X.  Liu,  H.  Lu,  G.  Zhu,  J.  Du,  C.  Zhao,  B.  Su,  B.  Wang,  R.  Chen and  Y.  Su: Environmentally friendly and simple recycling of titanium alloy scrap via deoxygenation with hybrid hydrogen plasma arc, Adv. Sci. 12 (2025) 2414747. doi:10.1002/advs.202414747
• 21)   J.-M.  Oh,  K.-M.  Roh and  J.-W.  Lim: Brief review of removal effect of hydrogen-plasma arc melting on refining of pure titanium and titanium alloys, Int. J. Hydrogen Energ. 41 (2016) 23033–23041. doi:10.1016/j.ijhydene.2016.09.082
• 22)   M.  Watanabe,  F.  Sato,  R.  Abe,  K.  Ueda,  D.  Matsuwaka,  F.  Kudo and  T.  Narushima: Deoxidation of Ti melt by newly developed two-step plasma arc melting process using hydrogen, Metall. Mater. Trans. B 50 (2019) 1553–1558. doi:10.1007/s11663-019-01608-x
• 23)   S.  Hotta,  K.  Yamada,  T.  Murakami,  T.  Narushima,  Y.  Iguchi and  C.  Ouchi: β grain refinement due to small amounts of yttrium addition in α+β type titanium alloy, SP-700, ISIJ Int. 46 (2006) 129–137. doi:10.2355/isijinternational.46.129
• 24)   K.  Ueda,  S.  Nakaoka and  T.  Narushima: β grain refinement of α+β type Ti-4.5Al-6Nb-2Fe-2Mo alloy using rare earth oxide precipitates, Mater. Trans. 54 (2013) 161–168. doi:10.2320/matertrans.MC201207
• 25)   M.T.  Hepworth and  R.  Schuhmann, Jr.: Thermodynamic properties of titanium-oxygen-hydrogen alloys, Trans. Metall. AIME 224 (1962) 928–935.
• 26)  V. Dembovský: Plasma Metallurgy: The principals, (Elsevier, 1985) pp. 71–72.
• 27)   V.I.  Lakomskiy and  N.N.  Kalinyuk: Solubility of hydrogen in liquid titanium and nickel, Russ. Metall. 2 (1966) 80–85.
• 28)  M.W. Chase, Jr.: NIST-JANAF Thermochemical Tables Fourth Edition Part II Cr–Zr, (American Institute of Physics, 1998) p. 1261 and 1324.
• 29)   K.  Fitzner: Prediction of thermodynamic properties of oxygen in liquid transition metals, Thermochim. Acta 52 (1982) 103–111. doi:10.1016/0040-6031(82)85188-5
• 30)   Y.  Kobayashi and  F.  Tsukihashi: Thermodynamics of oxygen in molten Ti-Al and Zr-Al alloys, High Temp. Mater. Process 19 (2000) 211–218. doi:10.1515/HTMP.2000.19.3-4.211
• 31)   M.  Maeda,  T.  Yahata,  K.  Mitugi and  T.  Ikeda: Aluminothermic reduction of titanium oxide, Mater. Trans. JIM 34 (1993) 599–603. doi:10.2320/matertrans1989.34.599
• 32)   Y.A.  Chang and  D.C.  Hu: On the Gibbs energy interaction parameters of oxygen and nitrogen in liquid alloys, Metall. Trans. B 10 (1979) 43–48. doi:10.1007/BF02653970
• 33)  I.S. Klykov: Deoxidation of Metals, Translated by K. Endo and Edited by K. Gunji, (Nichi-So Tsushinsha, 1977) p. 27.