Idaho National

Engineering Laboratory

INEL-96/0180

Corrosion in Supercritical Fluids

W. Alan Propp Tom E. Carleson

Kirk W. Daehling

INEL-96/01 80

Corrosion in Supercritical Fluids

W. Alan Propp Tom E. Carleson

Chen M. Wai Pat R. Taylor

Kirk W. Daehling Shaoping Huang

Masud Abdel-Latif

Published May 1996

Idaho National Engineering Laboratory Lockheed Martin Idaho Technologies

Idaho Falls, Idaho 83415

Prepared for the U.S. Department of the Interior

U.S. Bureau of Mines and for the

U.S. Department of Energy Under DOE Idaho Operations Office

Contract DE-AC07-941D13223

DISCLAIMER

Portions of this document may be illegible in electronic image products. Images are produced from the best available original document.

ABSTRACT

Integrated studies were carried out in the areas of corrosion, thermodynamic modeling, and electrochemistry under pressure and temperature conditions appropriate for potential applications of supercritical fluid (SCF) extractive metallurgy. Carbon dioxide and water were the primary fluids studied. Modifiers were used in some tests; these consisted of 1 wt% water and 10 wt% methanol for carbon dioxide and of sulfuric acid, sodium sulfate, ammonium sulfate, and ammonium nitrate at concentrations ranging from 0.00517 to 0.010 M for the aqueous fluids. The materials studied were Types 304 and 316 (UNS S30400 and S31600) stainless steel, iron, and AISI-SAE 1080 (UNS G10S00) carbon steel. The thermodynamic modeling consisted of development of a personal computer-based program for generating Pourbaix diagrams at supercritical conditions in aqueous systems. As part of the model, a general method for extrapolating entropies and related thermodynamic properties from ambient to SCF conditions was developed. The experimental work was used as a tool to evaluate the predictions of the model for these systems. The model predicted a general loss of passivation in iron-based alloys at SCF conditions that was consistent with experimentally measured corrosion rates and open circuit potentials.

For carbon-dioxide-based SCFs, measured corrosion rates were low, indicating that carbon steel would be suitable for use with unmodified carbon dioxide, while Type 304 stainless steel would be suitable for use with water or methanol as modifiers.

... 111

CONTENTS

ABSTRACT .............................................................. lll

INTRODUCTION ........................................................ 1

...

BACKGROUND ......................................................... 1

EXPERIMENTAL PROCEDURE ............................................ 3

Fluids .............................................................. 3

Materials ............................................................ 3

Corrosion Monitoring Apparatus .......................................... 3

Test Loops for Carbon Dioxide-Based Fluids ............................. Aqueous Corrosion Test Apparatus ....................................

3 6

Analysis ............................................................. 7

Carbon Dioxide Tests .............................................. 7 SLRC Corrosion Coupons ........................................... 7 Electrochemistry Experiments ........................................ 7

MODELING ............................................................. 8

RESULTS AND DISCUSSION .............................................. 10

CarbonDioxide ....................................................... 10

Aqueous Tests ........................................................ 12

Modeling ........................................................ 12 Electrochemical ................................................... 15 SLRC CorrosionCoupons ........................................... 16

CONCLUSIONS .......................................................... 17

REFERENCES ........................................................... 19

V

FIGURES

1. Test loop for continuous corrosion monitoring ................................ 4

2. Eh/pH diagrams: (a) Fe - H20 at 25°C and 1 atm., (b) Fe - H20 at 374°C and 220 atm., (c) Fe/Cr - H20 at 25°C and 1 atm., and (d) FelCr - H20 at 374°C and 220 atm . . 13

3. Pourbaix diagrams for the chromium-water system at supercritical conditions (374°C and 220 atm): (a) activity is unit value for all species, (b) activity is 10” for species in solution. (c) activitv is 10 for snecies in solution .............................. -6 14

TABLES

.................................. c 1. Test conditions for the SCF carbon dioxide 10

2. Test conditions for TvDe 304 SS with SCF mixtures containine either methanol or water 11

vi

Corrosion in Supercritical Fluids

INTRODUCTION

The objective of this research was to study corrosion of iron-based alloys in selected supercritical fluids (SCFs) under conditions applicable to extractive metallurgy. This work was in support of a research program in SCF extractive metallurgy conducted at the Salt Lake City Research Center (SLRC) of the Bureau of Mines. Integrated studies were carried out in the areas of corrosion, thermodynamic modeling, and electrochemistry under pressure and temperature conditions appropriate for SCF extractive metallurgy. The research reported here was a collaborative effort between the Idaho National Engineering Laboratory (INEL) and the University of Idaho.

Two primary fluids were studied carbon dioxide and water. The materials chosen for evaluation in the initial studies were Types 304 and 316 (UNS S30400 and S31600) stainless steel (SS). Test results for these materials provided baseline corrosion information to which the corrosion behavior of other materials could be compared. Later tests included iron; AISI-SAE 1080 (UNS G10800) carbon steel (CS) was also used in the water studies.

In parallel with the experimental studies, thermodynamic modeling of metal corrosion in water at SCF conditions was carried out. The modeling effort led to the development of a personal computer (PC) program for generating the high-temperature Pourbaix diagram for a specified system from existing ambient condition thermodynamic data and established procedures for extrapolating to SCF condifions. Variables considered in the model include the properties of the material and composition of the fluid. The results of the experimental corrosion studies have been used to evaluate the predictions of the model.

The following sections present background on SCFs, describe the experiments and modeling, and discuss the results.

BACKGROUND

Supercritical fluids have promise for leaching metal values from ores while generating a minimal amount of hazardous waste. For this research, an SCF is defined as a fluid that has been raised simultaneously above both its critical temperature and critical pressure. In this region, a single homogeneous phase exists-the fluid is essentially a very dense gas. In general, SCFs have physical properties that are intermediate between those of a gas and a typical organic liquid. Therefore, SCFs have potential as solvents in advanced separations and processing operations. As solvents, they have advantages over either gases or liquids. The high diffusivity of SCFs provides high mass transfer rates, in contrast to the limitations imposed by the moderate to low diffusion and mass transfer rates of liquids. Thus, process kinetics are typically faster when SCFs are used rather than liquid solvents. The low viscosity and zero surface tension of SCFs give them superior wetting properties and better small pore penetration, resulting in properties comparable to gases and significantly better than liquids. This capability of SCFs can be used to advantage when solid surface/fluid interactions are important, such as extraction of constituents from solid matrices.

1

The densities of SCFs are strongly influenced by temperature and pressure, particularly in the region close to the critical point. To a first approximation, solubility is a direct function of density. This property allows solubility to be controlled by relatively small changes in temperature or pressure. Similar control of solubility is not possible with either liquids or gases.

Much of the initial effort to commercialize SCF technology occurred in Europe in the 1970s.' The fluid used in the majority of the early applications was carbon dioxide because it reaches supercritical conditions at moderate temperature and pressure (31°C and 1,072 psia), has negligible environmental impact, and is nontoxic, readily available, easy to handle, and inexpensive. These attributes make carbon dioxide especially attractive as an SCF solvent. Low- molecular-weight organics, water, and other inorganics, such as ammonia and nitrous oxide, have also been studied. Vessel design restrictions have confined initial uses to batch operations, which limits process throughput. This throughput constraint, coupled with high capital equipment costs, has limited successful commercial applications of SCFs to date. Potential SCF processes must meet most of the following five criteria:

1. A relatively low volume feedstock.

2. A high unit value product.

3. Significant health or environmental issues associated with the use of alternative solvents.

4. Use of SCFs, which significantly improves process efficiency or reduces overall costs.

5. No viable alternative process.

As can be expected given the above criteria, most of the significant initial commercial applications of SCF technology have been in food processing, pharmaceutical, and related industries. Recently, there has been a widespread effort throughout various segments of industry, as well as governmental agencies, to expand the role of SCF technology in diverse areas including processing, environmental remedial action, parts cleaning, and as a reaction medium. Typical of these activities were the exploratory studies conducted by the Bureau of Mines in the areas of extractive metallurgy and recovery of strategic materials. In these studies, water-based fluids were used in extractive metallurgy and carbon dioxide with the strategic materials. Related studies investigating the scientific feasibility of using carbon dioxide-based fluids in extractive processes were conducted at the INEL and are reported elsewhere? The materials performance studies, both experimental and theoretical, summarized in this report were conducted in support of these extractive studies, thus the emphasis on aqueous and carbon dioxide-based fluids.

Information on performance of materials in environments containing carbon dioxide is available in the literat~re.3-l~ However, most of this information relates to the petroleum and natural gas industries where the environment is a complex mixture containing not only carbon dioxide but also water, hydrogen sulfide, chlorides, and other components. There is also information available on the performance of stainless steels in dry carbon dioxide, but the temperature/pressure conditions investigated were beyond the range of the present study, which is 150 to 250°C at 1,OOO to 3,000 psig. For aqueous fluids, most of the data reported in the

literature on corrosion of iron alloys were taken under subcritical conditions. Electrochemical studies in supercritical water are rare. Some of the data available are from the power generation industry, 'geochemistry, and related fields;16-18 other data have been reported elsewhere in the

Thus, the literature results were not directly applicable to the research program at SLRC or to the supporting research conducted at the INEL.

EXPERIMENTAL PROCEDURE

Fluids

The two fluids of primary interest to the SLRC, carbon dioxide and water, were studied. Initial work used straight carbon dioxide as the fluid. In later studies, either 1 wt% water or 10 wt% methanol was added as an entrainer. Methanol was selected because SLRC was interested in using it or similar fluids to remove machining oil from alloy scrap to facilitate recovery of strategic constituents by recycling. Water was selected both because of its potential for greatly enhancing the solvent power of the fluid and because it is a constituent in many systems.

In addition to pure water, aqueous fluids containing low levels of various inorganic constituents were studied. These solutions included sulfuric acid, sodium sulfate, ammonium sulfate, and ammonium nitrate in concentrations ranging from 0.005M to 0.01OM. The aqueous studies provided experimental data for evaluating the predictions of high-temperaturehigh- pressure Pourbaix diagrams developed in the thermodynamic modeling by the University of Idaho.

Materials

Types 304 and 316 SS, commonly used in applications requiring corrosion resistance, were chosen for initial studies because they were the alloys employed by the SLRC. Test results for these materials provided baseline corrosion information against which the corrosion behavior of other materials could be compared. Iron was selected as the other material for study with carbon dioxide-based fluids. For water-based fluids, the two stainless steels, 1080 CS and pure iron, were studied.

Corrosion Monitoring Apparatus

For the studies in carbon dioxide-based fluids, corrosion was monitored by a combination of surface analysis techniques and an electrical technique, which will be discussed below. For water- based fluids, electrochemical methods were the primary technique for assessing corrosion. The latter studies were supplemented by selected standard coupon studies conducted at the SLRC. The various studies will be discussed separately.

Test Loops for Carbon Dioxide-Based Fluids

Loop Design. The experimental system used for the initial studies was a closed, small- volume, thermal loop developed specifically for these experiments (Figure 1). The goal was to use off-the-shelf components that could be readily assembled. For safety reasons, cell volume was small, on the order of 100 mL or less. For simplicity and minimum cost, the cell operated as a

3

3000 psi 230 O C valve

Overall length 14-16 in.

Overall width 10-12 in.

- Conducting test wire

l--Conax connector

1-4 in. radius of curvature

- 0.5 in., 0.065 in. wall SS tubing with Swagelok fittings

6 13 471

Figure 1. Test loop for continuous corrosion monitoring.

4

closed system after initial filling. The test loop was fabricated from Type 316 SS double ferrule fittings and Type 304 SS tubing. It was found that use of heavy-wall tubing was required to consistently achieve a leak-tight system. The loop was capable of operating at pressures in excess of 3,000 psig at temperatures over 200°C. The loop volumes ranged from 30 to 90 mL depending on loop dimensions. To provide real-time corrosion information, the test loop was designed to permit continuous monitoring of corrosion. Initial tests did not incorporate this feature and relied only on physical examination of specimens upon completion of the experiment.

Corrosive attack on a wire specimen causes an increase in resistance as the specimen diameter decreases. Therefore, monitoring the resistance of the wire also monitors corrosion. The technique for continuous corrosion monitoring developed for these studies was based on this principle. An overview of the general approach and underlying principles has been presented by the National Association of Corrosion Engineers.23 The resistance of a 0.020-in.-diameter wire specimen was monitored as a function of time under test conditions. Because resistance is temperature dependent, a reference specimen of the same wire, mounted in a reference test loop filled with argon, was used to compensate for temperature fluctuations. After being filled, the two loops were clamped together so that they operated as a single unit. The ends of the specimens extended out of the loops through Conax connectors fitted with Teflon seals. The specimens were silver-soldered to short lengths of heavy-gauge copper bus wire to facilitate making low-resistance, mechanically robust electrical connections to the test specimens. At the bottom ends of the reference and test loops, the two bus wires were silver-soldered together, forming an interconnection between the two specimens that provided a common point for connecting the measuring instrument to both specimens. The specimens were cleaned with nonchlorinated organic solvents to remove all traces of oil and grease, air-dried, and then mounted in the loops prior to filling.

Pressure Testing and Loop Filling. Prior to use, each loop was pressure-tested with argon to ensure that the system was leak tight; if not, the loop was rebuilt and retested. The volume of each loop was calculated from the mass of argon required to produce a known pressure (around 1,OOO psig). Prior to filling, the loop was evacuated and flushed repeatedly with carbon dioxide. The loop was filled by transferring carbon dioxide from a 1-L reservoir to the loop. The transfer was achieved by immersing the cold leg in liquid nitrogen and condensing the carbon dioxide into the loop.

The reference loops were filled with high-pressure argon instead of carbon dioxide. Fill pressure was determined by the anticipated loop operating conditions. The objective was to have similar temperatures and pressures in both loops of a duplex unit during testing.

Temperature Control. One leg of the test loop was heated to induce fluid flow and prevent stagnation. Several turns of a heating tape were close-wrapped around the bottom of the hot leg to produce a localized source of heat, and the remainder of the tape was open-wrapped up the hot leg to maintain temperature in the leg. Aluminum spacer blocks mounted between the test and reference loops on both the hot and cold legs provided mechanical support and thermal links between the loops to minimize temperature differences. Each pair of loops was enclosed in insulation to reduce heat loss and minimize temperature fluctuations within the system.

5

Power to the heating tape was controlled by a variable autotransformer. For each loop, skin temperatures were monitored by strategically placed thermocouples attached directly to the outer surface. The temperature at the hottest point of the loop ranged from 170 to 250°C. The temperature gradient around each loop depended upon the manner in which the loop was insulated and the heat transfer properties of the fluid; neither was quantified.

Corrosion Monitoring. Specimen resistance was measured using the four-terminal technique. This technique, commonly used to accurately measure low-valued resistances, eliminates the errors introduced by meter lead resistance. With one set of leads, a current source is used to force a current through the unknown resistance, developing a voltage across it. The resulting voltage is measured with a high-input impedance meter using the second set of leads. The resistances of the reference specimen and the test specimen were each measured independently, but the ratio of their resistances was followed as a function of time. A variety of instruments were evaluated during the course of these studies. The best type of instrument for this application proved to be one that provides internal compensation for thermally generated contact voltages, eliminating them as an additional source of error in specimen resistance measurements. However, these errors should be small and essentially constant even if not compensated for since temperature variations at the contacts were small. Therefore, these errors should have been essentially compensated for by use of the resistance ratio technique.

Aqueous Corrosion Test Apparatus

Coupon tests were performed in the SCF extraction system at the SLRC. Electrochemical tests were performed using an experimental system designed and fabricated at the University of

The latter system is a computer-controlled, flow-through system consisting of a high- pressure pump for fluid delivery and system pressurization and a high-temperaturehigh-pressure potentiostat for electrochemical measurements. A heating tape was used as the heat source. The vessel was fabricated from Type 316 SS. Two Conax fittings were used for the connection between the potentiostat and the electrodes in the reactor. The fittings, using Lava sealants, are capable of operation at temperatures up to 980°C and pressures up to 10,OOO psig without loss of seal.

The pump, a reciprocating piston type, can handle a wide variety of solvents, generate pressures up to 6,000 psi (about 400 atm), and be accurately controlled at flow rates between 46 and 460 mL/h. The pump has a sapphire plunger and ruby ball check valves. The fittings used employ a reverse-threaded ferrule instead of a compression-type ferrule for sealing. The threaded ferrule permits operation at temperatures over 540°C (high temperatures do not affect the positioning of the ferrule on the tubing). In addition, these fittings provide a more reliable seal when repeatedly assembled and disassembled. A pressure transducer and a thermocouple probe were incorporated. A PC program was developed to control the system using the signals from the transducer and the thermocouple to control the pump and the heating tape.

The reference electrode used was an internal, silver-silver chloride type, based on a design by A. K Agrawal et al.” The concentration of the KCl solution in the electrode for this study was 0.1M. Additional details of electrode design, operation, and calibration are discussed elsewhere. 24-26

6

Analysis

Carbon Dioxide Tests

On selected tests, fluid samples were analyzed by mass spectrometry for carbon monoxide that might have formed as a byproduct of carbon dioxide decomposition during corrosive attack The results obtained were compared to those from the carbon dioxide starting fluid.

In addition, selected samples of the various materials under study were submitted for scanning electron microscopy (SEM) and energy dispersive x-ray spectroscopy (EDS) analysis. These samples included not only sections of the wire specimens, but also sections of the loops. Separate samples were submitted for Auger electron spectroscopy after ultrasonic cleaning in methanol. Depth profiles were performed under the same sputtering conditions to provide consistent results for comparison. Extended sputtering, to reach substrate composition, was performed on samples immediately above the hottest zone to obtain an assessment of the relative total thickness of the surface oxide layer.

SLRC Corrosion Coupons

For the tests performed at the SLRC, the coupons were allowed to air dry after each test, but were not cleaned before being transported back to the INEL. Samples of the feed and effluent solutions were collected during each run for analysis by inductively coupled plasma spectroscopy. The wash solution from each test, i.e., the aqueous liquid collected after each test when the lines and reactor vessel were cleaned with 10% (vh) nitric acid, was also collected. All coupons were reweighed before cleaning. Selected coupons were submitted for S E W D S analysis before cleaning.

Electrochemistry Experiments

Polarization curves (current density versus applied potential) were obtained by scanning the applied voltage while operating the computer and the potentiostat remotely. The effect of the internal resistance drop in pure water was determined by comparing the results obtained with pure water to those with a 0.005M Na2S04 solution at the same supercritical conditions. Polarization curves for both fluids were similar in shape except that the current density for pure water was less than that for the sodium sulfate solution, and the curve for pure water had a larger potential range. Nevertheless, the effect of the internal resistance drop should be quite small when the current density is close to zero, assuming the dilute sodium sulfate solution is inert to the redox reaction on the working electrode surface. Within the overpotential range of 20.2 mV, the measured current density was low, justifying this assumption. The two polarization curves coincided in this region.

The polarization curves indicated that a high scanning rate affects the current density. This phenomenon can be related to the capacitance effect of the electric double layer on the working electrode surface. It can be shown that if the scanning rate is low enough, the time increment between each deviation of the applied potential is much larger than the characteristic relaxation time, so that the output current density of the polarization curves will not be affected by the scanning rate. The experimental results showed that the appropriate scanning rates are those

7

smaller than 10 mV/s; therefore, the scanning rate applied for this study was 5 mV/s. Further discussion of these studies has been presented elsewhere. 24-26

MODELING

Thermodynamic modeling of the corrosion process in supercritical aqueous systems was selected as the focus of the modeling studies because of the potential for applying the results to a broad range of high-temperaturehigh-pressure processes. Such processes include power generation, geochemistry, hydrothermal crystal growth, desalination, geothermal utilization, extractive hydrothermal metallurgy, and conversion of fossil fuels. The primary objective was development of a PC-based thermodynamic model for predicting behavior of materials under SCF conditions in aqueous systems. The thermodynamic analysis should provide information on the stable species, such as corrosion products, at given conditions of temperature, pressure, and system composition. The model is based on Pourbaix diagrams, potential versus pH diagrams, extrapolated to SCF conditions. The results of the experimental studies in aqueous systems, both electrochemical and coupon tests, were used to evaluate the model.

Thermodynamic property data for chemical species at high temperatures and high pressures are not extensive. Fortunately, extrapolation of thermodynamic properties from ambient to elevated conditions provides a method of conducting the thermodynamic analysis, though the reliability of the analysis is dependent upon the accuracy and reliability of the extrapolation techniques employed. A detailed review of these methods as applied to the prediction of Pourbaix diagrams at elevated conditions is available.28

The Pourbaix diagram, a graphical presentation of stable species within a pH-potential region, is a valuable tool for analyzing the electrochemical equilibria in aqueous systems. With this diagram, the reaction products under given conditions can be determined. For example, if a redox-reaction product of a metal is a soluble ionic species, its solid oxide would not be expected to be a stable species; therefore, passivation of the metal in such conditions is impossible. The diagrams for various metals and their oxides at room temperature are widely available. A PC program was developed to calculate and plot pH-potential diagrams for systems consisting of two elements, one metal and one nonmetal, in the presence of water at 25°C and at specific activities of the species. The diagram is in the form of an array of symbols representing the fields of stable species. For thermodynamic analysis of the corrosion of iron alloys in supercritical water, the program was modified based on standard thermodynamic property extrapolation methods to produce the corresponding Pourbaix diagram at the required temperature and pressure.

The development of the calculations is based upon the following general chemical reaction at equilibrium:

nIM(I) + nIIM(II) + n002 + nHH+ + nwH20 + nLL = 0

where M(1) and M(I1) represent species containing the given metallic element M in different oxidation states, I and II; ni is the stoichiometric coefficient of species i; and L represents the general nonmetallic species with a charge ZI At constant temperature and pressure, the equilibrium condition is expressed as

8

"H "W "L M; = AGO + RT In($;: fff" aH aw aL = o

where R is the ideal gas constant, T is the absolute temperature, ai is the activity of species i, and f,-, is the fugacity of oxygen above the solution. In this expression, AGO is the standard Gibbs free energy change for the reaction, which is given by the following equation:

A G O = nIpoI + nTIpoII + nopoo + nHpoH + nWpow + nLpoL (3)

where poi is the standard chemical potential of the corresponding species at the given temperature and pressure.

. The method for extrapolating the chemical potential to elevated temperatures and pressures is based on the following equation, which expresses the temperature and pressure dependence of the chemical potential of species k2'

where pi is the chemical potential, T is the system temperature, P is the system pressure, the superscript denotes the thermodynamic properties at the reference state (e.g., at To and Po), si is the specific entropy, vi is the specific volume, (aSi(T,P0)/ar), is cp(T,Po)/r with cp being the specific heat at constant pressure, api is the coefficient of thermal expansion, and pTi is the isothermal compressibility.

Equations (2) through (4) can be combined to obtain the value of the reaction equilibrium constant at any specified conditions. For Pourbaix diagrams, the boundary between two regions with different predominant species is taken as the straight line resulting from setting the activity ratio of the two species equal to 1 in the equilibrium constant. A detailed discussion of the calculation procedures and general information on preparing Pourbaix diagrams can be found in various source^.^^^^ Additional discussion and details of the present calculations, including

9

assumptions used in estimating entropies and the specific volume of water, have been presented elsewhere. 31-32 The results obtained with the model will be discussed below.

In the calculations, the fourth term of Equation (4) was neglected, since the contribution of this term to the total value of chemical potential is only about 10% for anionic species containing oxygen, and even less for other anionic species. The temperature and pressure dependency of the coefficient of thermal expansion, etPL and the isothermal compressibility, PTi, up to elevated conditions are seldom available. Their contribution to the chemical potential value is usually small, except for water near its critical point. Therefore, it was assumed that the terms of Equation (4) involving these parameters for all species other than water are negligible. In other words, the volumes of solid species were regarded as constant, and the volumes of species in aqueous solutions were considered to be the same as water. This assumption is adequate because the concentrations of corrosion products in solution are usually very low.

RESULTS AND DISCUSSION

Carbon Dioxide . Overall, the results indicate that lower-grade alloys, such as carbon steels, would be

candidate materials of construction for use with straight carbon dioxide as a fluid, and Type 304 SS would be a suitable candidate for use with carbon dioxide mixtures containing either water or methanol.

Table 1 summarizes the conditions of the corrosion tests conducted with straight carbon dioxide as the fluid. Only limited corrosion of Type 304 SS occurred in carbon dioxide. Limited general surface oxidation was observed for both Types 304 and 316 SS, and some pitting was observed in Type 316 SS at sharp corners where fluid flow changed direction. The implied corrosion rates were low, precluding unambiguous assessment of corrosive attack using the fairly gross techniques afforded by surface examination. From the continuous corrosion monitoring tests, the corrosion rate for Type 304 SS was calculated at 0.05 mil per year. For iron, corrosion rates ranging from 0.1 to 0.4 mil per year were obtained; although low, these corrosion rates are approximately two to ten times higher than those for stainless steel.

Table 1. Test conditions for the SCF carbon dioxide.

Test materials

Time at test conditions

(h)

Temperature at hottest point

(“C) 304,316 SS 742 to 9,177 149 to 238 1,420 to 2,284

Iron 4,964 162 to 183 1,264 to 1,777

10

Tests with Type 304 SS and mixtures of carbon dioxide containing either 1 wt% water or 10 wt% methanol had corrosion rates of 0.02 and 0.001 mil per year, respectively. That the corrosion rates observed in mixtures were apparently lower than in unmodified carbon dioxide can be ascribed to the limited number of tests conducted with mixtures; statistically, these results are equivalent to those for unmodified carbon dioxide. The test conditions are summarized in Table 2.

EDS analyses of samples sectioned from the loops indicated minor amounts of sulfur, chlorine, and calcium in the Type 304 SS samples that were not present in the reference material. However, EDS did not indicate corrosive attack of Type 304 or 316 SS by the SCF carbon dioxide environment; the inherent surface roughness of the samples completely masked the limited corrosion that occurred. Auger analyses showed that the surface oxide films had significant concentrations of carbon as well as trace amounts of chlorine and sulfur. The chlorine was most probably from contamination. The sulfur may have been either surface contamination or a minor constituent of the alloy. Three primary sources for the carbon were identified surface contamination from the environment, minor alloy constituent, or product of the decomposition of the carbon dioxide fluid during corrosion. The carbon concentrations were highest in the hottest regions of the loops, whereas the concentrations of chlorine and sulfur were higher in the cooler portions of the loops.

Auger depth profiles for all the samples were similar. The oxygen sputtered off more rapidly from the reference sample than the other samples, showing that some corrosion occurred during the test. The oxide layer on the reference samples was ascribed to exposure to the atmosphere. Differences in the thicknesses of the oxide films among test loops were ascribed to differences in the surface roughnesses of the tubing used to fabricate the loops. The oxide film thickness was not related to alloy or temperature.

Carbon monoxide was not detected in the fluid either before or after testing, precluding decomposition of CO, with the intermediate formation of CO as a significant corrosion mechanism.

Table 2. Test conditions for Type 304 SS with SCF mixtures containing either methanol or water.

Secondary fluid

Time at test Temperature at hottest conditions point Pressure

(h) (“C) (psig) 10 wt% Methanol 2,300 265 1,350 to 2,400

1 wt% Water 3,900 185-200 2,000 to 2,400

11

I

Aqueous Tests

Modeling

Since chromium is a major component in iron alloys, the pH-potential diagrams for both iron and iron-chromium were generated. The diagrams for iron and iron-chromium in water at its critical point and at ambient conditions are shown in Figure 2. The Y-axis values of potential are relative to the standard hydrogen electrode. All dissolved species have been taken at unit activity.

In the diagrams, a solid line indicates the border between two species. The area between the two diagonal dashed lines is the stable region of water. Above the upper line, water is oxidized to oxygen; below the lower one, it is reduced to hydrogen. Figure Za, the Pourbaix diagram for iron/water at atmospheric conditions, indicates that (a) for pH 7 water, iron metal is the stable species for electrochemical potentials below about -0.6 V, (b) passivation (formation of a stable oxide film) would occur between -0.6 and +1 V, and (c) above -1 V, active corrosion (formation of Fe042-) would occur. Consequently, moderate corrosion rates (in the absence of oxygen) are anticipated. Figure 2b is the Pourbaix diagram for iron at supercritical water conditions. Because both the noble (iron metal) and active corrosion (Fe02.) regions have shifted to lower potentials, one would expect accelerated corrosion under supercritical water conditions. The effect of chromium on the corrosion behavior is seen in Figures 2c and d. At atmospheric conditions, passivation (formation of an insoluble chromium oxide) occurs over much of the intermediate H region (pH 6 to 11). At supercritical fluid conditions, active corrosion

chromium is apparently lost at supercritical water conditions due to the formation of soluble species.

[formation of (CrO, P )] occurs in most of this intermediate region. Thus, the passivation due to

Pourbaix diagrams, presented in Figure 3, were also generated to depict the effect of increasing concentrations of ionic chromium species (for activities ranging from 1 to 1 x lod). Near pH 7 and for potentials between the breakdown potentials of water, the region for solid chromium oxide decreases with decreasing activity and the stability region for ionic chromium species increase, again indicating loss of passivation. For the solid oxides, the region at an activity of 1 is smaller for supercritical conditions than ambient, and an activity of 1 x lo4 no longer exists. As the concentration decreases, the insoluble chromium oxide (Cr203) gives way to a soluble ionic chromite (CrOi) species, which is a further indication that loss of passivation is probable at SCF conditions because the behavior is exhibited over such a wide range of concentrations. Thus, enhanced corrosion of stainless steel alloys is expected in the supercritical region in spite of their chromium content.

Finally, Pourbaix diagrams were prepared for a variety of sulfur species and different concentrations of ionic species. The diagrams for the sulfide species (reducing environment) indicate that iron sulfide is a stable species at ambient conditions, whereas iron oxide and ferrite ion are stable at supercritical conditions. This indicates that pyritic materials may be decomposed at supercritical water conditions and that iron passivation may also be decreased (much more of the region is occupied by the ferrite ion). (The decomposition of pyritic materials has hydrometallurgical implications but is not directly related to corrosion.) For oxidizing conditions

12

Figure 2. E,,/pH diagrams: (a) Fe - H20 at 25°C and 1 atm., (b) Fe - H20 at 374°C and 220 atm., (c) Fe/Cr - H20 at 25°C and 1 atm., and (d) Fe/Cr - q 0 at 374°C and 220 atm.

Cr042-

-

-

-

-2 2 6 10 14

8.1 741 PH

a

2.20 I I I l l I w r O 4

HCrO, - - Cr042- -

h Cr+3 $ -0.05 - - W

-0.80 -

-1.55 - Cr

i I -2 2 6 10 14

8-1 739 b PH

Cr042-

I I l l l

-

w

-2 2 6 10 14

C 8-1 740 PH

Figure 3. Pourbaix diagrams for the chromium-water system at supercritical conditions (374°C and 220 atm): (a) activity is unit value for all species, @) activity is 10” for species in solution, (c) activity is lo4 for species in solution.

14

in the iro&ater/sulfate and ironhater systems, sulfates do not appreciably affect corrosion for either ambient or supercritical conditions.

Electrochemical

Experimental measurements are consistent with the theoretical predictions, indicating increased corrosion at supercritical temperatures and pressures relative to atmospheric conditions. Polarization curves were obtained for pure iron, 1080 CS, and Types 304 and 316 SS near the critical point of water. Loss of passivation was indicated by the curves for 304 and 316 SS. Based on the discussion of the internal resistance drop in the analysis section, the potential range used for fitting the model was selected as Eo k 0.2 V.

Although the exchange current density of pure iron is three times larger than that of stainless steel at ambient conditions, they are comparable at supercritical conditions. There appears to be a maximum value of the exchange current density around the critical temperature of water for each of the materials investigated. This maximum may be due to the change in dielectric constant and conductivity as the critical point of water is passed. Polarization curves also showed that the passivation-voltage ranges for the stainless steels near the critical point are much smaller than those at ambient conditions. This observation indicates that passivation may be lost around the critical point. A detailed study of 304 SS was conducted from ambient to supercritical conditions (temperature up to 530°C and pressure up to 300 atm). A plot of exchange current versus temperature showed that at a fixed pressure, the exchange current density increases exponentially with temperature up to the critical point, reaches a maximum, and then decreases. Comparison of calculated exchange current densities at 230 atm compared with the experimental values shows a deviation between the experimental result and the model value of less than k0.002 Man2 under supercritical conditions.

Based upon the electrodics, the exchange current density is related to the activation energy?3 Assuming an Arrhenius-type relation for the effect of temperature upon corrosion rate, the quantities analogous to activation energy range between 4.30 and 6.44 kcal/mol. The literature indicates that activation energies for mass-transfer-limited processes range between 1 and 3 kcaVmo1 and for reaction-limited processes between 10 and 20 kcal/mol.34 Based upon this criterion, the observed corrosion of 304 SS in pure water lies between the mass transfer and reaction rate limited cases.

In addition to plotting the polarization curves, the open circuit electrochemical potentials of various iron alloys in water at ambient and supercritical conditions were measured. For iron and its alloys (1080 CS, 304, and 316 SS), the open circuit potential values vary from -0.112 to +0.055 V in water at its critical point, and from -0.138 to -0.060 V in water at ambient conditions. The values were measured against a silver-silver chloride/O.lM KCl reference electrode. The potentials versus the standard hydrogen electrode were calculated from those measured with this reference electrode; calculated potentials were 0.0 and -0.1 V at critical and ambient conditions, respecti~ely.3~ Locating the pH and open circuit potential values in the Pourbaix diagrams shown in Figure 2, Cr203, Fe203 (hematite) and Fe304 (magnetite) are the oxidation products at ambient conditions, while CrOZ- and Fe203 are predicted to be the primary corrosion products at critical conditions. Thus, chromium may join in the formation of the passivation film for iron alloys at ambient conditions, but at supercritical conditions, this passivity is lost due to the

15

formation of a soluble oxidation product of chromium. This result is consistent with the phenomenon observed in the polarization analysis of pure iron and 304 SS as discussed above, where the exchange or corrosion current density of 304 SS was found to be one-third that of pure iron at ambient conditions and comparable to it at supercritical conditions. In addition, the measured open circuit potential falls inside the calculated stable region of water, providing further indication that the electrochemical measurements are consistent with the thermodynamic predictions. Further details of these experimental studies are available e l s e ~ h e r e . ~ ~ ~ ~ ~ ~ ~ ~

SLRC Corrosion Coupons

A single coupon, in the form of a long strip of mill-finish Type 316 SS, was used in each of the first series of tests performed by SLRC personnel. Corrosion occurred during the tests as evidenced by a substantial oxide layer that did not readily sputter off. Samples were also submitted for SEM/EDS analyses. Evidence of intergranular corrosion was noted for all samples. The analyses obtained were in good agreement with the nominal composition for all constituents with the exception of molybdenum; the results for this element were consistently higher than the nominal value. This fact is significant in light of the results obtained from the later testing discussed below.

The coupons from the second series of tests were all discolored and darkened by exposure to the test environments regardless of the feed solution composition. All coupons showed small weight gains except the Type 304 SS coupons from the run that used a dilute sulfuric acid fluid; these samples showed a net weight loss. Typically, weight changes were very small and approached the limit of precision of the balance used.

Selected samples were submitted for SEMEDS analysis before cleaning. Some intergranular corrosion was evident. This type of corrosive attack may be inherent to exposure of the stainless alloys to the supercritical media or it may be related to sensitization of the materials. Varying amounts of surface deposition or scaling were noted on all coupons. In most cases, the weight loss attributable to loss of material by intergranular corrosion was more than offset by the weight gain from the surface deposition, and a net weight gain resulted. However, the Type 304 SS coupons from the sulfuric acid run experienced a net weight loss.

The EDS analyses were in agreement with the nominal alloy composition for all constituents reported except for molybdenum, which was found at two to three times higher levels than expected for all samples. To determine if these high molybdenum levels were a surface phenomenon or were related to the bulk composition, the edge of one sample was examined by EDS at four locations from the surface toward the center. The values ranged from 0.3 wt% molybdenum in the bulk of the sample to 23 wt% at the surface; the surface value is significant. Possible sources of this contamination are discussed below.

The fluid pH was measured at the conclusion of each test, at ambient temperature and after the fluid had been exposed to air. The inductively coupled plasma spectroscopy analyses were limited to the major alloy constituents (iron, chromium, nickel, manganese) and molybdenum and sulfur. Contamination in the feed solutions was undetectable. As anticipated, all effluent and wash fluids contained major alloy constituents at varying levels, picked up as corrosion products during testing. In addition, all the samples had detectable levels of molybdenum, the presence of

16

which is consistent with the results of the coupon analyses. The concentration of molybdenum ranged from 4 to 20 ppb, significantly above the undetectable levels of the feed solution. The source(s) of the molybdenum contamination observed during the tests were not isolated. A possible source of molybdenum could be selective leaching from the Type 316 SS components of the experimental system. The analytical results for sulfur, expressed as an element, showed a significant decrease in levels between the feed and the effluent, which is consistent with the observation of precipitate formation and the semiquantitative nature of the sulfur analyses. (All effluent and wash samples collected, including those from the run with ammonium nitrate feed, contained yellowish turbidity and larger yellowish solids that settled out during each test. The sulfur values are only semiquantitative because of the presence of these solid sulfur precipitates.)

The results of these limited coupon studies are consistent with both the thermodynamic predictions and the electrochemical studies discussed previously.

CONCLUSIONS

A PC program was developed for electrochemical equilibria calculations and graphical pH- potential diagram presentation of a one-metal/one-nonmetalhvater system. The program can be used for temperatures and pressures exceeding the supercritical point of water. The calculations show that hematite (Fe203) is the oxidation product of iron in supercritical water, and the oxidation product of chromium in supercritical water is an ionic, water-soluble species, either CrO2- or CrO;. Thus, the passivation effect of chromium is lost in supercritical water. A general method for extrapolating entropies and other related thermodynamic properties to the evaluated temperatures was developed. The model is potentially applicable to a wide variety of aqueous processes under extreme conditions, such as in the fields of power generation, geochemistry, hydrothermal metallurgy, hydrothermal crystal growth, corrosion, and conversion of fossil fuels. Currently, the application of the model is limited by lack of thermodynamic data on species of interest.

Thermodynamic predictions were consistent with experimentally measured corrosion rates and open circuit potentials. The results indicate that enhanced corrosion of stainless alloys containing chromium may be expected in supercritical water. These corrosion rates appear comparable to those for mild steel or iron. The agreement between the predictions of the model and the experimental results indicates the success of the model in its current state of development as well as its potential.

The exchange current density, open circuit potential, and transfer coefficients were calculated for different alloys around the supercritical point of water. Passivation of Types 304 and 316 (UNS S30400 and S31600) SS appears to be lost around the critical point. Extensive experiments were conducted on 304 SS in the supercritical region. Based upon the Arrhenius relation for the corrosion rate, the activation energy for corrosion of 304 SS in supercritical pure water fell between values corresponding to mass transfer limited and reaction rate limited processes.

As part of the experimental system for electrochemical measurement, a silver/silver chloride reference electrode was developed and successfully used at supercritical water conditions. This

17

electrode potentially provides an additional means of monitoring aqueous processes in similar severe service conditions.

For carbon dioxide-based SCFs, measured corrosion rates were low, indicating that carbon steel would be suitable for use with straight carbon dioxide, while Type 304 SS would be suitable for carbon dioxide modified with either water or methanol.

18

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

REFERENCES

G. M. Schneider, E. Stahl, and G. Wilke (eds.), Extraction mth Supercrirical Gases, Verlag Chemie, Deerfield Beach, Florida, 1980, 189 pp.

W. A. Propp, T. E. Carleson, C. M. Wai, and S. Huang, TranspoH of Metal Sulfides in Supercritical Carbon Dioxide, INEL96/0179, May 1996.

D. R. Holmes, R. B. Hill, and L. M. Wyatt (eds.), "Corrosion of Steels in CO;, Proceedings of the British Nuclear Energy Society International Conference at Reading University, September 23-24, 1974.

R. J. Taunt and R. E. Pendlebury, "Oxidation of Mild Steel Crevices in High Pressure Carbon Dioxide," Werhtofle und Korrosion, 30, 1979, pp. 246-254.

J. C. P. Garrett, J. T. Crook, S. K. Lister, P. J. Nolan, and J. k Twelves, "Factors in the Oxidation Assessment of AIS1 Type 310 Steels in High Pressure Carbon Dioxide," Corrosion Science, 20, 1, 1982, pp. 37-50.

I. Matsushima, K. Masamuro, and S. Hashizume, "Corrosion Resistance of High-Alloy Oil Country Tubular Steels as Determined by Polarization Behavior in High Temperature High Pressure Environments," International Congress on Metallic Corrosion, Toronto, Canada, June 3-7, 1984, Vol. 2, National Research Council of Canada, Ottawa, Canada, pp. 609-614.

M. E Taylor, K. F. Hampson, and A. Y. Nehra, "Preliminary Results of the Oxidation Behavior of Contaminated Crevices between Mild Steel Components Exposed to C02-based Atmospheres," British Corrosion Journal, 13, 1978, pp. 79-81.

H. Kurahashi, Y. Sone, H. Otsubo, and Y. Nkai, "Corrosion and Stress Corrosion Behavior of Duplex Stainless Steels in H2S-C02-CI- Environment," Transactions of the Iron and Steel Imtitute OfJapan, 24, 4, 1984, p. B-129.

J. D. Garber, R. Perkins, and H. Y. Su, "High Temperature Wheel Test Simulates CO, Corrosion," Oil and Gas Joumal, 84, 17, April 28, 1986, pp. 62-64.

K. Mudge and C. J. Levesque, "Materials Choice Can Cut Downhole Corrosion," Oil and Gas Journal, 80, 39, September 27, 1985 pp. 245-248.

C. Cleave, J. M. Calvert, D. G. Lees, and C. P. Rowlands, "A Study of the Mechanism of Corrosion of Some Ferritic Steels in High-pressure Carbon Dioxide with the Aid of Oxygen-18 as a Tracer; Part I Low-Silicon Mild Steel," Roceedings of the Royal Society of London A, 379, 1982, pp. 404-427.

12. C. Cleave, J. M. Calvert, D. G. Lees, and C. P. Rowlands, "A Study of the Mechanism of Corrosion of Some Ferritic Steels in High-pressure Carbon Dioxide with the Aid of

19

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Oxygen-18 as a Tracer; Part I1 High-Silicon Mild Steel," Proceedings of the Royal Society of London A, 379, 1982, pp. 429-438.

NACE Task Group T-1-3, L. E. Newton, Jr., Chairman, and R. H. Hauser, Vice Chairman, CO, Corrosion in Oil and Gas Production: Selected Papers, Abstracts, and References, NACE, Houston, TX, 1981,687 pp.

NACE, "High Temperature Corrosion Data: A Compilation by Technical Unit Committee T-5B on High Temperature Corrosion," Corrosion, 11, 5, 1955, pp. 59-63.

L. k Morris, "Resistance to Corrosion in Gaseous Atmospheres," Handbook of Stuinless Steels, D. Peckner and I. M. Bernstein (eds.), McGraw-Hill Book Company, New York, NY, 1977, pp. 17-1 through 17-33.

.

J. R. Weeks, "Corrosion of Steam Generator Turbine in Operating Pressurized Water Reactors," Corrosion Problems in Energy Conversion and Generation, Electrochemical Science, 1975.

W. E Bogaerts and C. Bettendorf, Electrochemistry and Corrosion of Alloys in High- Temperature Water, EPRI NP-4705, July 1986.

R. W. Staehle, D. de G. Jones, and J. E. Slater (eds.), "High Temperature High Pressure Electrochemistry in Aqueous Solutions," Proceedings of a conference held at the University of Surrey, England, January 7-12, 1973, NACE-4, Houston, Texas; National Association of Corrosion Engineers, 1976, 570 pp.

L. Tomlinson, M. H. Hurdus, C. B. Ashmorc, and P. J. B. Silver, "Sodium Heated Steam Generator Tubes: Effect of Heat Flux on the Deposition of Magnetite from Solution and Corrosion of the Underlying Steel," Corrosion, 41,5, Max 1985, pp. 257-264.

M. Hishida, J. Takabayashi, T. Kawakubo, and Y. Yamashina, "Polarization Curve Measurement in High Purity Water at Elevated Temperatures," Corrosiun, 41, 10, 1985, pp. 570-574.

M. E. Indig and D. A. Vermilyea, "Corrosion of Sensitized Austenitic Stainless Steel in Hot Aqueous Solutions Under Natural and Electrochemical Control," Corrosion, 31, 2,1975, pp. 51-59.

J. P. Carter, B. S. Cavino Jr., T. J. Driscoll, W. D. Riley, and M. Rosen, "Gaseous Corrosion of Metals in a 60% H20, 40% HC1 Environment," Comosion, 40,5, 1984, pp. 205-214.

NACE, Electrical and Electrochemical Methods for Determining Corrosion Rates, NACE Publication 3D170, 1984 Revision, 10 pp.

K. W. Daehling, Design and Construction of a Supercritical Fluid Reactor System for a Corrosion Study of Metals in Supercritical Fluids, M. S. thesis, University of Idaho, Moscow, ID, January 1988, 135 pp.

20

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

S. Huang, Studies of Corrosion and Extraction in Supercritical Fluids, Ph.D. dissertation, University of Idaho, Moscow, .ID, October 1990, pp. 37-57.

S. Huang, K Daehling, T. E. Carleson, M. Abdel-Latif, P. Taylor, C. Wai, and A. Propp, "Electrochemical Measurements of Corrosion of Iron Alloys in Supercritical Water," in Supercritical Fluid Science and Technology, K. P. Johnston and J. M. L. Penninger (eds.), ACS Symposium Series #406, Washington, D C American Chemical Society, 1989, pp. 287-300.

A. K Agrawal and R. W. Staehle, "Technical Note: A Silver-Silver Chloride Reference Electrode for the High Temperature and High Pressure Electrochemistry," Corrosion, 33, 11, 1977, pp. 418-419.

P. Duby, "Graphical Representation of Equilibrium in Aqueous Systems at Elevated Temperatures," in High Temperahsre High Pressure Electrochemistry in Aqueous Solutions, R. W. Staehle, D. de G. Jones, and J. E. Slater (eds.), Proceedings of a conference held at the University of Surrey, England, January 7-12, 1973, NACE-4, Houston, l X National Association of Corrosion Engineers, 1976, pp. 353-374.

M. Model1 and R. C. Reid, Thermodynamics and Its Applications, 2nd edition, Prentice-Hall, New Jersey, 1983, p. 144.

M. Pourbaix, Lecfures on Electrochemical Corrosion, translated by J. A. S. Green, Plenum Press, New York, 1973,336 pp.

S . Huang, Studies of Corrosion and Extraction in Supercritical Fluids, PbD. dissertation, University of Idaho, Moscow, ID, October 1990, pp. 16-36.

S. Huang, K. Daehling, T. E. Carleson, P. Taylor, C. Wai, and A. Propp, "Thermodynamics Analysis of Corrosion of Iron Alloys in Supercritical Water," in Supercritical Fluid Science and Technology, K. P. Johnston and J. M. L. Penninger (eds.), ACS Symposium Series #406, Washington D C American Chemical Society, 1989, pp. 276-286.

J. O'M. Bockris and A. K. N. Reddy, Modem Electrochemistry, 2, Plenum Press, New York, 1970, p. 1152.

0. Levenspiel, Chemical Reaction Engineering, John Wiley and Sons, New York, 1972.

S. Huang, K. Daehling, T. E. Carleson, and A. Propp, "Construction and Calibration of an Internal Silver/Silver Chloride Reference Electrode for Supercritical Fluid Studies," Corrosion, 47, 3, 1991, pp. 185-188.

S. Huang, Studies of Corrosion and Extraction in Supercritical Fluids, Ph.D. dissertation, University of Idaho, Moscow, ID, October 1990, pp. 58-93.

21