Article Cite This: J. Phys. Chem. C 2019, 123, 15009−15016
pubs.acs.org/JPCC
Role of Dissolved Oxygen in Iron Oxidation in Supercritical Water: Insights from Reactive Dynamics Simulations Liqiang Ai, Yusi Zhou, and Min Chen*
Downloaded via UNIV AUTONOMA DE COAHUILA on July 31, 2019 at 03:16:19 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Department of Engineering Mechanics, Center for Nano and Micro Mechanics, Tsinghua University, Beijing 100084, P. R. China ABSTRACT: Oxygenated treatment is considered to be an effective chemical water treatment method and is widely used in supercritical power plants. Previous isotope tracer experiments reported that dissolved oxygen and supercritical water (scH2O) react with metals simultaneously, but the atomistic mechanism remains unrevealed. We performed reactive dynamics simulations to investigate the role of dissolved oxygen in iron oxidation in supercritical water. Structural analysis shows that the prior reacted O from O2 (Oxy) will activate the surface Fe atom to react with the O in scH2O (Owa), resulting in a Fe− Oxy−Fe−Owa−H structure. Aside from activating Fe, Oxy has another contribution of absorbing the proton generated after the dissociation of scH2O. These two roles of dissolved O2 result in an O2-boosted scH2O oxidation mechanism. A series of simulations indicate that appropriate O2 concentration will accelerate the formation of the surface protective oxide layer, whereas excessive O2 concentration might cause destructive internal oxidation of Fe. The increase of temperature has little influence on the formation of the surface oxide layer but will increase the internal oxidation rate of O2.
1. INTRODUCTION Supercritical water (scH2O) is widely used as the working fluid in supercritical boilers in fossil power plants due to its high thermodynamic efficiency at elevated temperatures.1,2 However, the corrosion of the structural materials in scH2O becomes a vital issue. To slow down the corrosion, scH2O must be given a strict treatment before injected into the boilers. There are two standard methods of water treatment applied in supercritical boilers. One is the all-volatile treatment (AVT) using ammonia and a reducing agent or an oxygen scavenger such as hydrazine, the other is the oxygenated treatment (OT) using ammonia and oxygen.3 In contrast to AVT, OT has the potential to restrain flow-accelerated corrosion.3 OT is considered to be an effective chemical water treatment method and is widely used in supercritical power plants. So far, a great deal of research on the corrosion of various structural materials of T91, P92, HT9, and others in scH2O with different contents of dissolved O2 has been conducted.4−10 A general conclusion was that for many materials investigated, the oxidation rate increases with the increase in dissolved oxygen contents under the experimental conditions. Among these studies, the experiments performed by Zhu et al.10 was illuminating. They performed an oxygen isotope tracer experiment of ferritic−martensitic steel P92 at 873 K and 25 MPa in scH216O with a dissolved 18O2 content of 2000 ppb, up to an exposure time of 100 h. The oxygen isotope profiles revealed that dissolved O2 and scH2O react with metals simultaneously. They concluded that dissolved O2 mainly changes the oxidation reduction potential of scH2O through the oxide scale, leading to an increase in the oxidation rate. The interpretation is satisfied from the macroscale; however, the detailed atomistic chemical reaction mechanism, © 2019 American Chemical Society
especially how scH2O molecule reacts with the metal surface in the presence of O2, remains unrevealed. On investigating the oxidation mechanism mentioned above, the molecular dynamics (MD) simulation method has its advantage. On the one hand, it could provide an intuitive atomistic picture of the simulated system, which allows the analyzing of the oxidation process. On the other hand, it is convenient to track the oxygen atoms and distinguish between the two kinds of molecules: the H2O molecule and the O2 molecule. In the experiments with isotope tracing techniques, the purity of 18O in labeled O2 could just be above 95%,10 which might introduce errors in the recognition of specific atoms. With MD simulation methods, however, the accuracy of recognizing the “tracer” atoms is 100%. Despite the advantage of MD simulations, it is not an easy job to perform an appropriate simulation for the metal oxidation system. First, the force field describing the atomistic interactions must be able to reproduce the formation and dissociation of chemical bonds between the metal atoms and oxygen atoms. Under such circumstances, the widely applied model of embedded atom method (EAM) or modified embedded atom method (MEAM) fails, although it has great success in simulating the properties of metals and alloys. Moreover, the force field must be computationally efficient; otherwise, it is difficult to reproduce the overall dynamic process of the metal oxidation. Methods based on quantum chemistry calculations are generally more suitable for statics simulations other than dynamic simulations. The reactive force field (ReaxFF) method developed by van Duin and co-workers,11−13 however, Received: February 28, 2019 Revised: May 11, 2019 Published: May 24, 2019 15009
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016
Article
The Journal of Physical Chemistry C
gies. The nonbonded energies of van der Waals and Coulomb are also considered in the total energy. More detailed information about the ReaxFF method is provided in Chenoweth and van Duin’s publications.11,22,23 The ReaxFF parameters used in this simulation were developed by Aryanpour et al.13 The MD simulations were performed in the domain of 28.4 Å × 28.4 Å × 64.2 Å, and the schematic diagram of the simulation system is shown in Figure 1. A 10 × 10 × 5 Fe
provides the possibility for MD simulations of the metal oxidation process in supercritical water with dissolved oxygen. By training the force field parameters out of the train sets calculated by quantum chemistry, ReaxFF is capable of reproducing the formation and dissociation of chemical bonds through its continuous profiles of bond order and interatomic distance.11 ReaxFF is computationally efficient, simulating a system with thousands of atoms for millions of timesteps in several days.14 There are several ReaxFF studies on metal oxidation with water or oxygen at the ambient environment or under the supercritical state.12−21 In 2012, Aryanpour et al.13 developed the first set of ReaxFF parameters for iron (Fe)−oxyhydroxide systems based on density functional theory calculations and the parameter set has been widely applied in subsequent ReaxFF studies. Pan et al.15,16 studied the process of iron oxidation under a typical moist condition and reported that a triplex structure was formed at the end of a three-stage oxidation process to reduce the overall oxidation speed. Aral et al.20,21 investigated the effect of oxidation on the deformation of iron nanowires, and in their ReaxFF MD simulations, the iron nanowires were rapidly surface-oxidized in the atmosphere of either O2 or H2O. DorMohammadi et al.19 studied the initial stages of iron corrosion in pure water and reported the critical stages of the iron corrosion process identified as dissociation of water to OH− and H+, adsorption of OH− on the iron surface, penetration of oxygen into iron to form iron oxides, and dissolution of iron into solution. In 2015, Assowe et al.12 developed a set of ReaxFF parameters for the nickel (Ni)− oxyhydroxide system, simulated the reaction of Ni in aqueous solution at room temperature (300 K) with an external electric field, and reported the generation of Ni composite on the surface. Verners and van Duin17 simulated the Ni nanoplate stress corrosion in water, with the same ReaxFF parameter sets developed by Assowe et al. Zou et al.18 simulated the initial stage of oxidation of Ni in an oxygen atmosphere and proposed a new oxygen diffusion mechanism in which the oxygen atom diffuses via the movement of the oxygen-vacancy pair. Ai et al.14 simulated the oxidation process of Ni in supercritical water and revealed that the dissociation of water is more likely to be a homolytic reaction instead of a heterolytic one under the supercritical state of relatively high temperatures and low densities. The above ReaxFF studies indicated that the ReaxFF method and the existing parameter sets are capable of describing metal reaction with H2O or O2. However, as far as we are concerned, no ReaxFF studies on metal oxidation in scH2O with dissolved O2 have been performed. This study aimed to provide atomistic insight into the role of dissolved O2 in iron oxidation in scH2O, with the method of ReaxFF MD simulations. The concentration of dissolved O2 and temperature are varied for different simulations to investigate their influence on the oxidation rate.
Figure 1. Schematic diagram of the simulation system.
super cell is located at the bottom of the domain, with the surface in (100) direction. Fe has the lattice structure of bodycentered cubic and the lattice constant is set to be 2.84 Å, which is determined through ReaxFF MD simulations of pure Fe under the NPT ensembles. A total of 200 water and oxygen molecules are placed above the Fe layer. To investigate the influence of dissolved O2 concentration, a series of different simulations are performed with the numbers of oxygen molecules of 0−80, respectively. The rest of the 200 molecules are water, resulting in the dissolved O2 fraction ranging from 0 to 0.4. To further investigate the influence of temperature on the oxidation rate, ReaxFF MD simulations of Fe oxidation with 180 H2O and 20 O2 are performed at different temperatures, namely 673, 773, 823, 873, 973, and 1073 K, respectively. In Figure 1, the H atom is colored in white and the Fe atom is colored in black. For the sake of clarification, the O atoms are colored in two different colors according to the species of molecules, with red for O in scH2O (noted as Owa) and yellow for O in O2 (noted as Oxy). Periodical boundary conditions are applied in both x and y directions. For z direction, the fixed boundary condition is applied, and a reflecting wall is set at the upper boundary to maintain the scH2O and O2 molecules in the simulation domain. The simulations were carried out with NVT ensembles at the constant temperature of 873 K for 1 000 000 steps, and the temperature was controlled using a Nose−Hoover thermostat with a damp parameter of 25 fs. The numbers of the scH2O and O2 molecules are calculated in advance so that the final equilibrium pressure is around 25 MPa. The above temper-
2. COMPUTATIONAL DETAILS The Fe−scH2O−O2 system is modeled using ReaxFF. The total energy is expressed as follows Etotal = E bond + Eval + Etors + Eover + Eunder + EvdW + ECoul
(1)
The energy of the system consists of geometry-related terms, such as bond, three-body angle, four-body torsional angle energies, and overcoordination and undercoordination ener15010
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016
Article
The Journal of Physical Chemistry C
the temperature is stable around 873 K. The pressure fluctuates around 25 MPa after a short period of the initial adsorption stage. During the initial period, there is a decrease in the pressure that is due to the surface adsorption of scH2O and O2 molecules. During the adsorption of these molecules, the number density of the molecules in the calculation box decreases, resulting in a decrease of pressure. After the Fe surface reaches saturation of adsorption, the pressure reaches equilibrium around 25 MPa. It is also noticed that the pressure reached a minimum at approximate 20 ps, before finally fluctuating around 25 MPa. This is because, at the initial adsorption period, many molecules rapidly move toward the surface layer, resulting in a local high-density surface layer and leaving behind a temporary low-density region. After the saturation of adsorption, the surplus molecules move away from the surface layer, and the pressure of the calculation box returns to normal. Figure 2b shows the evolution of system potential energy. The potential energy reflects the exothermicity of the iron surface oxidation. During the adsorption and reaction of scH2O and O2 on the Fe surface, the potential energy gradually decreases. The heat generated is removed out of the simulation system by the Nose−Hoover thermostat. The damping constant of the Nose−Hoover thermostat decides the frequency to remove the energy. Three different damping constants of 25, 100, and 250 fs are tested. Figure 2c shows the influence of damping constant of the Nose−Hoover thermostat on the evolution of system temperature. The difference only exists during the initial 0.5 ps. With a smaller damping constant, the temperature reaches equilibrium more rapidly. After the initial short period, the temperature keeps stable and there is little influence from the damping constant. In our ReaxFF MD simulations, the damping constant of 25 ps is applied. 3.2. Atomistic Oxidation Mechanism of Fe in scH2O and O2. ReaxFF MD simulations indicate that the oxidation of Fe in scH2O and O2 is a continuous process. Both scH2O and O2 react with Fe, and different species of products, including Fe hydrates, hydroxides, and oxides, are generated during the reaction. To investigate the atomistic oxidation mechanism of Fe in scH2O and O2 from the ReaxFF MD simulations, different molecular species are recognized according to the types and amounts of the chemical bonds formed. For the simulation of Fe oxidation with 180 H2O and 20 O2 at 873 K, several snapshots of simulated atoms are plotted with the balland-stick model at different time steps along the simulation, as shown in Figure 3. The dynamic bonds between Fe−O and O−H are determined through the distances of atoms. The criteria of distances are determined according to Fe−O and O−H bond order, resulting as rcut‑FeO = 2.1 Å and rcut‑OH = 1.2 Å. As shown in Figure 3, during the initial stage, O2 reacts with Fe rapidly. Some of the Oxy atoms remain at the Fe/scH2O surface, whereas the other Oxy atoms migrate inside the bulk metal Fe. Meanwhile, the scH2O molecules are adsorbed on the Fe free surface and do not dissociate immediately after the adsorption. After approximate 5 ps, the scH2O molecules begin to dissociate gradually, and Fe hydroxides are formed at the Fe surface. It is worth noticing that the Fe hydroxides contain both Owa and Oxy, suggesting that the H+ ions formed after dissociation of scH2O are rapidly chemical associated with the Oxy atoms generated from predissociated O2. The oxidation
ature and pressure of 873 K and 25 MPa are the typical temperature and pressure in the superheaters of scH2O boilers and are consistent with the parameters applied in the experiments by Zhu et al.10 The simulations were performed with the open-source LAMMPS package24,25 at a timestep of 0.25 fs. The calculations were completed on the “Explorer 100” cluster system of Tsinghua National Laboratory for Information Science and Technology.
3. RESULTS AND DISCUSSION 3.1. Temperature and Pressure Evolution of the System. Figure 2a shows the evolution of temperature and
Figure 2. (a) Evolution of temperature and pressure for the simulation of Fe oxidation with 180 H2O and 20 O2. (b) Evolution of system potential energy for the simulation of Fe oxidation with 180 H2O and 20 O2. (c) Influence of damping constant of the Nose− Hoover thermostat on the evolution of system temperature.
pressure during the simulation of Fe oxidation with 180 H2O and 20 O2 at 873 K. To exclude the influence of surface adsorption, the temperature and pressure are calculated for scH2O and O2 molecules within a calculation box, which is 10 Å above the surface (24.2−64.2 Å in z direction). The temperature and pressure are calculated every 25 fs and averaged over the time span of 2.5 ps. As shown in Figure 2a, 15011
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016
Article
The Journal of Physical Chemistry C
consistent with that of the interface oxygen, suggesting that the mechanism of dissociation of scH2O is associated with the oxygen in the surface layer. To further investigate the atomistic oxidation mechanism of Fe in scH2O and O2, we analyze the chemical reaction process by tracking several specific atoms through frames of the MDcalculated trajectories. Figure 5 illustrates the three-step
Figure 3. Snapshots at different timesteps for the simulation of Fe oxidation with 180 H2O and 20 O2 at 873 K.
process goes on until most of the O2 are consumed and a surface oxide/hydroxide layer is formed. To quantitatively analyze the oxidation process, the evolution of numbers of different species of scH2O and O2 is shown in the stacked bar graph of Figure 4. The color bars
Figure 4. Evolution of numbers of water and oxygen for the simulation of Fe oxidation with 180 H2O and 20 O2 at 873 K.
Figure 5. (a) Microstructure of Oxy activated Fe. (b) Microstructure of Fe−Oxy−Fe−Owa−H. (c) Microstructure of Oxy absorbing the proton generated after the dissociation of scH2O.
from bottom to top represent consumed oxygen, inner oxygen, interface oxygen, reacted water, adsorbed water, and consumed water, respectively. The numbers of the species are counted every 2.5 ps and based on recognizing the different molecules according to the bond order. The oxygen here refers to Oxy and is counted as atoms, so at the beginning, the number of the unconsumed oxygen is 40 corresponding to a total of 20 O2 molecules. The consumed oxygen contains two portions, at the Fe/scH2O interface (noted as interface oxygen) or migrated into the bulk metal Fe (noted as inner oxygen). The consumed scH2O also contains two portions, the adsorbed scH2O and the reacted scH2O. As shown in Figure 4, O2 is consistently consumed, the inner oxygen number almost remains unchanged, whereas the interface oxygen atoms are gradually accumulated. For scH2O, after a 5 ps initial adsorption stage, the number of consumed scH2O reaches an equilibrium; this might result from the saturation of the Fe surface. The adsorbed scH2O is gradually converted into the reacted scH2O. The increasing trend of the number of the reacted scH2O is
mechanism of the O2-boosted scH2O oxidation of Fe. The first step is the reaction of dissolved O2 with Fe. Because of the extremely high reactivity, O2 reacts easily on contacting the Fe surface. Each O2 dissociates into two Oxy, and although several Oxy atoms have the potential to migrate inside the bulk Fe, most of the Oxy stay around the Fe surface. The generated Fe oxide exhibits a structure where Oxy lying between the first and second layer of Fe is chemically bonded with two upper layer Fe atoms and two lower layer Fe atoms, as shown in Figure 5a. The upper layer Fe atoms are activated and have the potential to break the O−H bond in scH2O. After the dissociation of scH2O, the −OH radical is chemically bonded with the activated Fe to form a Fe hydroxide. The reaction process follows a Fe−Oxy−Fe−Owa−H picture, as shown in Figure 5b. Even though attached to an Oxy activated Fe, scH2O does not dissociate automatically, and the outlet of the generated proton (H+) is another issue. By analyzing the microstructure of the MD-calculated trajectories, it is revealed that when a scH2O is 15012
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016
Article
The Journal of Physical Chemistry C
above zero. For those O2-activated Fe atoms, the increase of atom charge will enhance the electrostatic interaction between Fe and adsorbed Owa in scH2O and increase the probability of the dissociation of scH2O. The variation of the atom charge of Fe is a direct reflection of the O2 activating process. Figure 8 shows the number density of different atom species along z direction. The red line represents Owa, the yellow line
dissociated, there is always a nearby Oxy on the Fe surface. The surface Oxy is usually located at the bridge site on the Fe surface and it is exposed to the other atoms attached to the Fe surface. The Oxy is negatively charged and has notable electrostatic interaction with neighboring H atoms. During the vibration of a nearby scH2O adsorbed on the Fe surface, when one of the H atoms in scH2O is close enough to the Oxy, it might be absorbed by Oxy and forms an −OH radical. The scH2O molecule loses a proton and forms another −OH radical, as shown in Figure 5c. The above pathway for the oxidation process is illustrated in the schematic diagram of Figure 6. The arrow represents the path for proton. As a result,
Figure 8. Number density of different atom species along z axis. The red line represents Owa, the yellow line represents Oxy, and the black line represents Fe. Figure 6. Schematic representation of pathway for the oxidation process. The arrow represents the path for proton.
represents Oxy, and the black line represents Fe. The number density is calculated every 1000 timesteps and averaged from 125 to 250 ps. The profile of Fe reveals a layered structure. The profile of Oxy has a peak around 12 Å, right between the first and second layer of Fe. Another peak is around 14 Å, resulting from the surface Fe hydroxide. The profile of Owa has three peaks. The peak around 17 Å is due to the layered structure of scH2O near Fe surface. The highest peak is around 15 Å, resulting from the adsorption of scH2O on Fe surface. The third peak around 14 Å is right at the position of Oxy over Fe surface, indicating the existence of Owa involved Fe hydroxide. The profiles of Owa and Oxy support the O2boosted scH2O oxidation mechanism. 3.3. Oxidation Rates at Varied Dissolved Oxygen Concentrations and Temperatures. To investigate the influence of the concentration of the dissolved O2 on the oxidation of Fe in scH2O and O2, we compare the results from a series of different simulations with the numbers of oxygen molecules of 0−80, respectively. And the O2 fraction ranges from 0 to 0.4. Figure 9 shows the evolution of numbers of reacted water for simulations with different O2 concentrations. For the simulations without dissolved O2, the water molecules seldom react. For other simulations with dissolved O2, the
the generated hydroxide layer on the Fe surface contains both Owa and Oxy, which is consistent with the experimental results by Zhu et al. The absorption by Oxy is a promotion of the dissociation of scH2O. Our previous studies on Ni oxidation in pure scH2O showed that without O2, scH2O reacts with Ni at a much lower rate, under that circumstance, the generated proton is absorbed by isolated scH2O to form an H3O+. In the simulations with dissolved O2, no H3O+ is observed. To provide more detailed information supporting the above O2-boosted oxidation mechanism for Fe in scH2O, we performed charge analysis for surface Fe atoms. Figure 7
Figure 7. Evolution of atom charge for O2-activated Fe atom A (red) and B (magenta) and nonactivated Fe atom C (blue) and D (black).
shows the evolution of atom charge for O2-activated Fe and nonactivated Fe. The surface layer of Fe atoms is shown in the inserted figure of Figure 7. At 25 ps, Fe atoms A and B are already bonded with Oxy, whereas C and D remain unbonded within the time span of 0−25 ps. Charge analysis reveals that the atom charge of A increases to around 0.35 e after bonded with an Oxy at about 5 ps, and the atom charge of B increases to around 0.35 e after bonded with an Oxy at about 10 ps. The charges of nonactivated Fe atoms C and D remain a little
Figure 9. Evolution of numbers of reacted water for simulations with different O2 concentrations. 15013
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016
Article
The Journal of Physical Chemistry C number of reacted water increases, with a decreasing growth rate, and reaches an equilibrium after about 125 ps. The numbers of consumed water, absorbed water, and reacted water at different O2 fractions are calculated for comparison, as shown in Figure 10. All the numbers are
Figure 10. Numbers of consumed water, absorbed water, and reacted water at different O2 fractions.
averaged from 125 to 250 ps. The total consumed water remains almost unchanged with the increase of O2 fraction. The number of reacted water sharply increased with the increase of O2 fraction, whereas the number of adsorbed water sharply decreased with the increase of O2 fraction. These results indicate that the total number of the total scH2O, including the adsorbed scH2O and the eventually reacted scH2O, has a limitation due to the saturation of Fe surface. With the increase of O2 fraction, the adsorbed scH2O on the Fe surface is more likely to dissociate and react with Fe. The result is a direct reflection of the O2-boosted scH2O oxidation. Aside from the calculation of consumed and reacted water molecules, bond order analysis is also applied to investigate the influence of O2 concentration on the oxidation of Fe. The bond orders for Fe−Owa and Fe−Oxy chemical bonds are calculated every 2.5 ps and averaged over the time span of 100−150 ps. Figure 11 shows the distribution of bond orders for the simulations with two different O2 concentrations. The total numbers of Fe−Oxy bonds are larger than those of Fe− Owa bonds for both cases. The numbers of Fe−Oxy bonds are significantly larger for higher O2 concentration whereas the numbers of Fe−Owa bonds are almost the same. There are two peaks in the distributions of bond orders for Fe−Oxy at approximate 0.42 and 0.51, respectively. There is a peak in the distributions of bond orders for Fe−Owa at approximate 0.68. The positions of the peaks are consistent for different O2 concentrations. A larger bond order means a stronger bond between Fe and O atoms. The stronger Fe−Owa bonds come from those surface −OH radicals and the weaker Fe−Oxy bonds are the bonds formed inside bulk Fe. The bond order analysis also indicates that the attached scH2O on the Fe surface has a limitation, the increase in O2 fraction will increase the dissociation rate of attached scH2O. Meanwhile, the surplus O2 will react with inner layer Fe which is usually harmful to the strength of the structure in practice. These results might inspire the application of OT in supercritical power plants that appropriate O2 concentration will accelerate the formation of the surface protective oxide layer, whereas excessive O2 concentration might cause destructive internal oxidation of Fe.
Figure 11. Distribution of bond orders for Fe−Owa and Fe−Oxy chemical bonds for the simulations of Fe oxidation with (a) 160 H2O and 40 O2 and (b) 180 H2O and 20 O2.
To further investigate the influence of temperature on the oxidation rate, ReaxFF MD simulations of Fe oxidation with 180 H2O and 20 O2 are performed at different temperatures, namely 673, 773, 823, 873, 973, and 1073 K, respectively. Bond order analysis is applied. The total number of Fe−Owa and Fe−Oxy bonds with a bond order greater than 0.3 is calculated every 2.5 ps. Figure 12 shows the evolution of
Figure 12. Evolution of numbers of Fe−Owa and Fe−Oxy chemical bonds at different temperatures for the simulations of Fe oxidation with 180 H2O and 20 O2.
numbers of Fe−Owa and Fe−Oxy chemical bonds from 0 to 250 ps at different temperatures of 673, 773, and 873 K. It is shown that the number of Fe−Oxy bonds increases with time and the increasing rate gradually decreases for all temperatures. The increasing rate is larger at higher temperatures. The number of Fe−Owa bonds increases during the initial 20 ps and becomes stable later. These results also indicate a surface saturation for Owa, and the increase of Fe−Oxy bonds is due to migration of Oxy inside bulk Fe. The migration of Oxy has to cross the surface oxidation layer, resulting in a significant 15014
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016
Article
The Journal of Physical Chemistry C
The evolution of the number of Fe−Oxy bonds satisfies the logarithmic rate relation, and the activation energy of internal oxidation of Fe is calculated to be 3.04 kJ/mol according to the Arrhenius formula. These result might raise some inspiration for the development of OT used in supercritical power plants.
decreasing rate. The evolution of the number of Fe−Oxy bonds satisfies the logarithmic rate relation as the expression below. NFe ‐ Oxy = k ln t + C
(2)
The logarithmic rate constant k at different temperatures is calculated through the linear fit of NFe−Oxy and ln t. The logarithmic rate relation, other than parabolic rate relation, indicates a greater resistance inside the surface oxide layer than the pure diffusion process. According to Arrhenius formula, rate constant k at different temperatures satisfies the following equation.
■
Corresponding Author
*E-mail:
[email protected]. Phone: +86 10 62797062.
ji E zy k = k 0 expjjj− a zzz j kBT z (3) k { where kB is the Boltzmann constant and Ea is the activation energy. The Arrhenius equation implies the linear relation between ln k and 1/T as the expression below. ln k = ln k 0 −
Ea kBT
AUTHOR INFORMATION
ORCID
Min Chen: 0000-0002-8900-8855 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the National Key Basic Research Program of China under Grant number 2015CB251502 and the National Natural Science Foundation of China under grant numbers 51776104 and 51621062. The computations were performed at the National Laboratory for Information Science and Technology of Tsinghua University in China.
(4)
Figure 13 presents the relationship between ln k and 1/T. According to the linear fit, the activation energy is calculated to be 3.04 kJ/mol.
■
REFERENCES
(1) Sarrade, S.; Féron, D.; Rouillard, F.; Perrin, S.; Robin, R.; Ruiz, J. C.; Turc, H. A. Overview on Corrosion in Supercritical Fluids. J. Supercrit. Fluids 2016, 120, 335−344. (2) Betova, I.; Bojinov, M.; Kinnunen, P.; Penttilä, S.; Saario, T. Surface Film Electrochemistry of Austenitic Stainless Steel and Its Main Constituents in Supercritical Water. J. Supercrit. Fluids 2007, 43, 333−340. (3) Dooley, R. B.; Chexal, V. K. Flow-Accelerated Corrosion of Pressure Vessels in Fossil Plants. Int. J. Pressure Vessels Piping 2000, 77, 85−90. (4) Zhang, N.-q.; Xu, H.; Li, B.-r.; Bai, Y.; Liu, D.-y. Influence of the dissolved oxygen content on corrosion of the ferritic-martensitic steel P92 in supercritical water. Corros. Sci. 2012, 56, 123−128. (5) Was, G. S.; et al. Corrosion and Stress Corrosion Cracking in Supercritical Water. J. Nucl. Mater. 2007, 371, 176−201. (6) Bischoff, J.; Motta, A. T. Oxidation behavior of ferriticmartensitic and ODS steels in supercritical water. J. Nucl. Mater. 2012, 424, 261−276. (7) Ampornrat, P.; Was, G. S. Oxidation of Ferritic−Martensitic Alloys T91, Hcm12a and Ht-9 in Supercritical Water. J. Nucl. Mater. 2007, 371, 1−17. (8) Ren, X.; Sridharan, K.; Allen, T. R. Corrosion of Ferritic− Martensitic Steel Ht9 in Supercritical Water. J. Nucl. Mater. 2006, 358, 227−234. (9) Chen, Y.; Sridharan, K.; Allen, T. Corrosion behavior of ferriticmartensitic steel T91 in supercritical water. Corros. Sci. 2006, 48, 2843−2854. (10) Zhu, Z.; Xu, H.; Jiang, D.; Yue, G.; Li, B.; Zhang, N. The role of dissolved oxygen in supercritical water in the oxidation of ferriticmartensitic steel. J. Supercrit. Fluids 2016, 108, 56−60. (11) van Duin, A. C. T.; Dasgupta, S.; Lorant, F.; Goddard, W. A. Reaxff: A Reactive Force Field for Hydrocarbons. J. Phys. Chem. A 2001, 105, 9396−9409. (12) Assowe, O.; Politano, O.; Vignal, V.; Arnoux, P.; Diawara, B.; Verners, O.; van Duin, A. C. T. Reactive Molecular Dynamics of the Initial Oxidation Stages of Ni(111) in Pure Water: Effect of an Applied Electric Field. J. Phys. Chem. A 2012, 116, 11796−11805. (13) Aryanpour, M.; van Duin, A. C. T.; Kubicki, J. D. Development of a Reactive Force Field for Iron-Oxyhydroxide Systems. J. Phys. Chem. A 2010, 114, 6298.
Figure 13. Relationship between ln k and 1/T.
4. CONCLUSIONS Reactive dynamics simulations are performed to investigate the role of dissolved oxygen in iron oxidation in supercritical water. The evolution of numbers of water and oxygen indicates that the oxidation of Fe in scH2O and O2 is a continuous process. Both scH2O and O2 react with Fe, and different species of products, including Fe hydrates, hydroxides, and oxides, are generated during the reaction. Structural analysis reveals an O2-boosted scH2O oxidation mechanism. The prior reacted Oxy will activate the surface Fe atom to react with Owa, resulting in a Fe−Oxy−Fe−Owa−H picture. Aside from activating Fe, Oxy has another contribution of absorbing the proton generated after the dissociation of scH2O. Charge analysis and number density profiles support the O2-boosted scH2O oxidation mechanism. A series of simulations with different O2 fractions indicates that appropriate O2 concentration will accelerate the formation of the surface protective oxide layer, whereas excessive O2 concentration might cause destructive internal oxidation of Fe. A series of simulations with different temperatures indicates that the increase of temperature has little influence on the formation of the surface oxide layer but will increase the internal oxidation rate of O2. 15015
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016
Article
The Journal of Physical Chemistry C (14) Ai, L.; Zhou, Y.; Huang, H.; Lv, Y.; Chen, M. A Reactive Force Field Molecular Dynamics Simulation of Nickel Oxidation in Supercritical Water. J. Supercrit. Fluids 2017, 133, 421−428. (15) Pan, T.; van Duin, A. C. T. Steel Surface Passivation at a Typical Ambient Condition: Atomistic Modeling and X-Ray Diffraction/Reflectivity Analyses. Electrocatalysis 2011, 2, 307−316. (16) Pan, T. Y.; Xi, Y. P. Physicochemical Nature of Iron Oxidation in a Damp Atmospheric Condition. Acta Metall. Sin. (Engl. Lett.) 2011, 24, 415−422. (17) Verners, O.; van Duin, A. C. T. Comparative Molecular Dynamics Study of Fcc-Ni Nanoplate Stress Corrosion in Water. Surf. Sci. 2015, 633, 94−101. (18) Zou, C.; Shin, Y. K.; van Duin, A. C. T.; Fang, H.; Liu, Z.-K. Molecular Dynamics Simulations of the Effects of Vacancies on Nickel Self-Diffusion, Oxygen Diffusion and Oxidation Initiation in Nickel, Using the Reaxff Reactive Force Field. Acta Mater. 2015, 83, 102−112. (19) DorMohammadi, H.; Pang, Q.; Á rnadóttir, L.; Isgor, O. B. Atomistic Simulation of Initial Stages of Iron Corrosion in Pure Water Using Reactive Molecular Dynamics. Comput. Mater. Sci. 2018, 145, 126−133. (20) Aral, G.; Islam, M. M.; Wang, Y.-J.; Ogata, S.; van Duin, A. C. T. Oxyhydroxide of Metallic Nanowires in a Molecular H2o and H2o2 Environment and Their Effects on Mechanical Properties. Phys. Chem. Chem. Phys. 2018, 20, 17289−17303. (21) Aral, G.; Wang, Y. J.; Ogata, S.; van Duin, A. C. T. Effects of Oxidation on Tensile Deformation of Iron Nanowires: Insights from Reactive Molecular Dynamics Simulations. J. Appl. Phys. 2016, 120, 5195. (22) Senftle, T. P.; Hong, S.; Islam, M. M.; Kylasa, S. B.; Zheng, Y.; Yun, K. S.; Junkermeier, C.; Engelherbert, R.; Janik, M. J.; Aktulga, H. M. The Reaxff Reactive Force-Field: Development, Applications and Future Directions. npj Comput. Mater. 2016, 2, 15011. (23) Chenoweth, K.; van Duin, A. C. T.; Goddard, W. A. Reaxff Reactive Force Field for Molecular Dynamics Simulations of Hydrocarbon Oxidation. J. Phys. Chem. A 2008, 112, 1040−1053. (24) Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117, 1−19. (25) Aktulga, H. M.; Fogarty, J. C.; Pandit, S. A.; Grama, A. Y. Parallel Reactive Molecular Dynamics: Numerical Methods and Algorithmic Techniques. Parallel Comput. 2012, 38, 245−259.
15016
DOI: 10.1021/acs.jpcc.9b01936 J. Phys. Chem. C 2019, 123, 15009−15016