A pulse oxidation facility for the study of oxide nucleation behavior Free

The selection of high temperature materials/coatings for gas turbine engine components in aircraft and marine industries is governed by their oxidation performance accompanied by outstanding mechanical properties.^1,2 Usually, an oxide layer between the substrate and the harsh environments prevents further oxidation during service.³ The development of a protective oxide layer depends on the nucleation and lateral growth kinetics prior to establishing full coverage of the substrate.⁴ Generally, oxidation tests are carried out in a furnace without any great difficulties. The schematic of an ideal and actual isothermal exposure as a function of time for long-time exposures is shown in Figs. 1(a) and 1(b), respectively. Figure 1(b) indicates that there is a deviation between the ideal and actual temperature profiles that can be considered as an insignificant error in the measurements when the sample heating time is much lower than the overall exposure time at the target temperature. However, short time furnace exposure oxidation tests cannot be performed isothermally since the heating time and overall exposure time are on the same scale as shown in Figs. 1(c) and 1(d).

For the most part, previous studies of oxidation reactions have followed two approaches involving either bulk samples exposed to oxidation for long times or thin film in situ investigations of initial oxidation on clean surfaces. For example, there are several studies on the furnace oxidation behavior of Ni–Cr/Fe/Co and Ni–Cr–Al alloys in different oxygen atmospheres for 260 h that reported that Cr₂O₃ forms as an external oxide layer and other oxides, such as Fe₂O₃ and Al₂O₃, form as an internal oxide layer, which depends on the composition.^5–10 

There have also been attempts to access short time exposures, such as the work reported by Chattopadhyay and Wood¹¹ on the oxidation behavior of Ni–Cr and Ni–Fe alloys that were exposed at 600 °C under an oxygen pressure of 1 atm for 1–15 min. The resulting oxide layer contained grain and subgrain boundaries and other substructural defects due to either the cation vacancy sinks or more local diffusional paths in oxide. For alloys with compositions of up to 30% Cr, a thin Cr₂O₃ oxide layer was observed to form above the defects due to rapid chromium diffusion to the alloy/oxide interface. However, in this case, the samples were exposed to oxygen during a heating ramp that required more than one minute to reach the reaction temperature followed by a quench after exposure. The non-isothermal exposure complicates the analysis of the short time reaction behavior.

On the other hand, with thin film in situ studies, usually, the native oxide is removed prior to oxidation exposure. These studies have explored the initial nucleation of oxide on different crystallographic surfaces and revealed the dependence of the nucleation behavior and oxide morphology as a function of oxygen partial pressure, Po₂, temperature, and substrate composition.^12–24 In other in situ transmission electron microscopy studies, Yu et al.²⁵ conducted experiments on the early stage oxidation of Ni–Cr and Ni–Cr–Mo alloys and explained that oxide initiation follows the sequence: rock salt and then spinel, followed by corundum oxide in Ni–Cr alloys, whereas the Ni_2−xCr_xO₃ (corundum structure) phase forms after rock salt for Ni–Cr–Mo alloys. In addition, Kirkendall voids are found in Ni–Cr alloy near the metal/oxide interface, while Mo doping inhibits the formation of Kirkendall voids and enhances the corundum oxide nucleation rate in Ni–Cr–Mo alloy. While this approach does give valuable information of the character and morphology of oxidation products on clean surfaces as a function of oxygen pressure, temperature, and crystal orientation, the extension of the results to bulk samples containing a native oxide and different oxidation induced stresses is not clear.

Since the nucleation rate is exponentially related to temperature,²⁶ an accurate analysis of bulk samples in this short time regime is very challenging with current furnace experimental oxidation study techniques. To probe the nucleation and growth behavior of oxides in this high temperature, short time regime, both temperature and gas atmospheres must be precisely controlled. In order to address these requirements, we have developed a novel pulse oxidation facility specialized for short term oxidation tests. The pulse oxidation system enables the analysis of bulk samples with multiple grain orientations over a wide range of exposure temperatures of up to 1200 °C and Po₂ levels from 1.3 × 10⁻⁵ to 0.1 Pa under isothermal conditions for times as short as 10 s. In addition, as a demonstration, Ni-30 wt. % Cr was used to illustrate oxide nucleation at 600 °C after an exposure of 45 s.

To approach as closely as possible to the idealized experimental behavior, as illustrated in Fig. 1(d), a pulsed oxidation facility was developed and established. The pulsed oxidation experimental facility is shown as a schematic diagram in Figs. 2 and 3, respectively. The experimental facility is comprised of a test sample (A), an alumina support with a recessed sample holder (B), a stainless steel sample support (C), a fiber-optic sensor for temperature measurement (E), a feedthrough for a vacuum pump (F), a bolt (G), a base (H), an exhaust valve (I), an exhaust feedthrough (J), a fiber-optic feedthrough (K), a fiber-optic cable (L), a quartz dome (M), induction coils (N), a quartz gas inlet rod (O), an interface between the base and the dome (P), a residual gas analyzer (Q), a gas inlet valve (R), gas vessels (S), and gas vessel shut-off valves.

The vacuum chamber was constructed of 304L stainless steel. The chamber volume was deliberately minimized to 0.0052 m³ to maintain short backfill times and minimize Ar gas usage. The sample chamber was constructed of a graded glass dome with the cylindrical dimensions of 0.145 m length and 0.046 m diameter, terminating in quartz glass. The quartz was selected due to a combination of high temperature stability and low conductivity. This minimizes the amount of heat transfer to the dome during heating of the sample; above 1000 °C, the dome is merely warm to the touch. A cylindrical alumina pedestal with the dimensions of 0.063 m length and 0.025 m diameter was selected as a sample holder since it does not react with the magnetic field produced by using the electromagnetic induction heater (Magneforce heating system 2500 R2) during experimentation.

A stepper motor actuated gate valve (VAT, adaptive pressure controller PM-4) was used to separate the sample holder chamber and the vacuum chamber. This gate can isolate the roughing and turbopumps (the nXDS 6i scroll pump with a maximum pump down pressure of 2 × 10³ Pa and the Hi pace 300 Pfeiffer vacuum turbopump with a maximum pump down pressure of 1 × 10⁻⁸ Pa) and act as a controllable throttle to the gas outflow from the sample holder chamber. An ion gauge (Kurt J. Lesker Company, 354 Series Ionization Vacuum Gauge Module) was used to read the pressure in the chamber. A proportional integral differential (PID) control loop was used to connect and modulate the mass flow controller (MFC) outputs (Alicat scientific MC-series) to the chamber to maintain the required pressure. These MFCs can control the gas outflow from 0 to 5 SCCM (standard cubic centimeters) with a high degree of granularity and short response times. With a combination of altering the mass flow controller output and adjusting the gate position, the chamber can maintain a wide range of chamber pressures and compositions to enable tight control of oxidation and reduction. In addition, this configuration enables experiments to be carried out in flowing gas so that reaction products are swept away from the sample and ensuring the programmed nominal gas composition is what the sample surface experiences.

During the experiment, the gate valve is set to a user-configurable intermediate position. This is done to maintain an active flow through the chamber. The pressure input from the ion gauge is used to modulate the gas flow to maintain a steady-state pressure and the corresponding flow rate. In this steady-state, the gas outflow through the gate to the pumps precisely matches the inflow from the active MFC(s), which is specified in SCCM. A typical steady-state flow rate for an oxidation experiment with a 0.06 Pa setpoint is 0.5 SCCM. Due to the small overall chamber volume (0.0052 m³ or 5.2 l), a simple estimation of the time required for the replacement of the entire gas atmosphere within the chamber can be made by comparing the total number of mol in the chamber, n_chamber, during the experiment to $ṅ$⁠, the number of mol introduced in 1 s by a 0.5 SCCM inflow,

The small chamber size, adequate pumping capability, and comparatively large gas flows enable a rapid turnover of the gas environment. By controlling the total chamber pressure, available Po₂ can be calculated based on the gas mixture introduced during the oxidation phase of the experiment. This pressure control is illustrated in Fig. 4. During the “heat-up,” “stabilization,” and “oxidation” phases, the pressure was actively controlled by the MFC output, which is, in turn, modulated by the regulating PID loop.

A Hiden Analytical HAL 200 RC residual gas analyzer (RGA) detector was incorporated into the pulsed oxidation chamber to further expand the instrument capabilities. The detector can resolve ions with mass, m, to charge, z, ratios (m/z) of 1–200 that allows for the detection of any gaseous products and other unwanted components (H₂O, N₂, and O₂) prior to experimentation. This allows for the direct measurement of the partial pressures of all the gasses in the chamber, including P_O2. Additionally, the ability to detect unwanted gases is useful for repairs and maintenance. An example output from the RGA is presented in Fig. 5. The system was scanned from a mass-to-charge ratio of 0.4–50, while the pulsed oxidation chamber was in the pumped down state, revealing N₂, H₂O, and O₂ as the principal constituents.

The data produced by the RGA were further processed by selecting the peak pressure value in each region of interest, corresponding to each gas species resolved in the plot, presented in Table I. In this way, the partial pressure of each gas species can be quantified for different stages of the pulsed oxidation experimental procedure. Scans were collected after pumpdown, after heat-up immediately prior to the oxidation exposure, during the exposure, and 3 min into cool down.

Table I indicates that the partial pressure of oxygen was maintained at 4.10 * 10⁻⁴ Pa immediately prior to the oxygen exposure. During the experiment, the total pressure rose a setpoint of 6.7 * 10⁻³ Pa, corresponding to an increase in the oxygen partial pressure to 1.03 * 10⁻³ Pa, before dropping to 5.16 * 10⁻⁵ Pa only 3 min after the experiment. Note that the RGA has an interlocked upper pressure limit of 1.33 * 10⁻² Pa for safe operation. To avoid triggering the interlock, a lower experimental pressure was employed for typical oxidation exposures, as indicated in Sec. II B, but the gas composition will be maintained.

Sample temperature is measured by using an optical pyrometer coupled to a sapphire light pipe (Photrix OEM optical thermometer) fixed at the bottom of the sample holder. This provides temperature data with low latency (1 µs response time). Control of the sample temperature has been accomplished by establishing a PID loop between the induction heater and the output from the optical pyrometer. An example of the temperature control capability is demonstrated in Fig. 6. A cylindrical specimen of Ni-30% Cr was heated to a target setpoint of 600 °C for an intended oxidation duration of 45 s.

During the “heat-up” phase, the heater is set to the maximum output; the sample heats at 10.0 °C/s over the 300–500 °C range. Near the temperature setpoint, the associated PID loop takes over control and throttles the heater to maintain the target temperature, called the “stabilization” phase. During both the heating and stabilization phases, the chamber pressure is maintained at the experimental pressure setpoint, but with an Ar flow, to prevent unintended oxidation.

Once the user has deemed both the sample temperature and the chamber pressure to be adequately stabilized, the experiment proceeds to the “oxidation” phase. At this time, the Ar flow is replaced with the oxidizing gas mixture (typically, O₂ or 20% O₂ 80% Ar), modulated by the same pressure PID loop. As illustrated in Fig. 4, this change-over results in brief pressure oscillations, but within 15 s, the PID loop can reach a steady-state. This rapid decrease in pressure illustrates the pumping efficacy of the system; the setpoint pressure is only maintained with adequate flow from the MFCs. The pressure measured by the ion gauge is reflective of the process gas mixture.

After the experimental duration has elapsed, the system enters the “cool down” phase. Several actions occur simultaneously, all intended to slow the oxidation rate as much as possible, while the sample still retains heat. The heater is shut off, the oxidizing gas flow is terminated, the Ar flow is set to the maximum output (5 SCCM), and the gate valve is moved from the intermediate position to the fully open position to maximize the effective pumping rate. These actions in concert act to minimize the amount of oxidation, which occurs during the final cooldown. This is of great importance due to the relatively modest cooling rates (<2 °C/s) observed in samples cooling under these conditions.

To coordinate the required set of simultaneous actions, all experimental functions are controlled by a master LabVIEW program. It allows for many user-configurable experimental parameters and runs the exposure automatically after the user loads the sample on the holder. Experimental repeatability is improved by using this program to regulate all key exposure parameters: total pressure, oxygen partial pressure, temperature, and time.

Prior to the commencement of the pulsed oxidation experiment, Ni-30 wt. % Cr was arc melted followed by cold rolling and then annealing at 1250 °C for 120 h to obtain a microstructure with a large grain size (1–3 mm). Subsequently, samples were cut with dimensions of 12 mm diameter and 3 mm thickness. The samples were polished up to 4000 grit grinding and followed by diamond paste and then colloidal silica polishing. Successively, samples were rinsed with water followed by ultrasonication in methanol for 15 min. With the large grain size, specific grain orientations can be identified by electron back scattered diffraction (EBSD) and marked for subsequent examination after oxidation exposure. This allows for the observation of orientation dependence of oxide nucleation behavior in bulk samples under identical exposure conditions.

The polished Ni-30 wt. % Cr sample was placed on the alumina sample holder and then covered with the quartz dome. Both the turbopump and the roughing pump were operated continuously during the exposure. Subsequently, when the chamber reached the pressure of 1.3 × 10⁻³ Pa, the chamber was backfilled with Ar gas to reach the pressure of 6.7 × 10⁻¹ Pa and then reduced again to 1.3 × 10⁻³ Pa, and it was repeated the same way for three times to flush out residual gases in the chamber prior to the start of the experiment. Next, the pressure in the chamber was fixed at 6.7 × 10⁻² Pa and then the sample was heated by induction to the target temperature of 600 °C. After reaching the target temperature, a gas mixture of 80% Ar and 20% O₂ was introduced at the same pressure setpoint of 6.7 × 10⁻² Pa to begin oxidation for an exposure of 45 s. After finishing the exposure time, the 80% Ar and 20% O₂ gas flow was halted along with a termination of heating, and 5 SCCM of Ar gas was introduced at the chamber to inhibit the oxidation during the cooling of the sample. Figure 7 shows the scanning electron microscopy (SEM) micrographs and probability distribution of the Ni-30 wt. % Cr oxidized sample at 600 °C for 45 s. Interestingly, it is observed that inward and outward oxide islands are forming on (111) and (100) oriented grains, respectively. The inward oxide islands [dark islands in Fig. 7(a)] are in the range of 50–300 nm, whereas outward oxide islands [bright islands in Fig. 7(b)] are in the range of 50–600 nm and are well separated. The oxide island nucleation density was measured on both the planes; their densities are 2.9 × 10¹¹ and 7 × 10¹² m⁻² on (111) and (100) planes, respectively. Similar inward and outward growing oxide islands were also observed in the former study on Ni-30 wt. % Cr at 600 °C for 60 s by Yu et al.²⁷ using the same experimental facility. In addition, in the previous study, an energy dispersive spectroscopy (EDS) analysis on cross-sectional micrographs by Scanning Transmission Electron Microscopy-High Angle Annular Dark-Field imaging (STEM-HAADF) of (111) and (100) oriented grains indicated that the inward oxide islands that have composition consistent with corundum and outward oxide islands are rock salt, respectively. The analysis indicates that cation vacancies dominate the diffusion in the rock salt structure, whereas oxygen vacancies dominate diffusion in the corundum structure. The evolution of the corundum and rock salt oxide size distribution after different exposure times at different temperatures provides valuable information on the nucleation and growth behavior during the initial stage of oxidation. In addition, in another pulse oxidation experiment, Yu et al.²⁸ studied the oxidation behavior of Ni–Cr–Mo alloys in the solution of K₂S₂O₈–Na₂SO₄ and at Po₂ of 1.3 × 10⁻² Pa in the temperature range of 500–800 °C and demonstrated that the oxide composition exceeded the thermodynamic solubility limits. This non-equilibrium solute capture in oxides is controlled by the rapidly moving interface.

A pulse oxidation facility was developed for oxide nucleation behavior of high temperature materials. The pulse oxidation system enables the analysis of bulk samples with a native oxide and multiple grain orientations over a wide range of exposure temperatures of up to 1200 °C and Po₂ levels from 1.3 × 10⁻⁵ to 0.1 Pa under isothermal conditions for times as short as 10 s. This device enables the effects of grain orientation on short time oxidation behavior to be demonstrated by exposing differently oriented pre-characterized grains to identical oxidation conditions. For example, oxide nucleation behavior of the pure Ni-30 wt. % Cr substrate at 600 °C for 45 s is illustrated by using this facility. Based on the previous study²⁷ and the present study, corundum and rock salt oxide islands are forming on both (111) and (100) oriented grains, respectively. The corundum oxide islands are in the range of 50–600 nm, whereas rock salt oxide lands are in the range of 50–600 nm and are well separated. The oxide nucleation densities are 2.9 × 10¹¹ and 7 × 10¹² islands/m² on (111) and (100) planes, respectively. This facility can also be used up to 1200 °C and provides new opportunities for early stage oxidation studies.

This work was supported by ONR MURI “Understanding Atomic Scale Structure in Four Dimensions to Design and Control Corrosion Resistant Alloys” under Grant No. N00014-16-1-2280. We are most grateful to Professor L. D. Marks of Northwestern University for his initial proposal for the need for a pulse oxidation facility.

The data that support the findings of this study are available from the corresponding author upon reasonable request.