06_Conclusion - Yiwei666/08_computional-chemistry-learning-materials- GitHub Wiki
1. Trace element partitioning between silicate melts – A molecular dynamics approach
5. Conclusions
From the previous discussion we may draw the following conclusions: (1) It is possible to employ a thermodynamic integration scheme to predict trace element partitioning, given that simulation times are long enough to thoroughly sample the phase space. (2) The approach seems sensitive enough (even when simulating small systems sizes) to meaningfully predict partition coefficients with an order of magnitude accuracy. (3) Our simulations confirm that trace elements that require large oxygen coordination shells preferentially partition into depolymerized melts with more loosely bonded oxygen anions enabling them come closer to their ideal coordination environment. The opposite effect seems to hold true for small cations with coordination shells similar to those of Si or Al. (4) To further advance our computational method, systematic studies are needed to evaluate the influence of temperature, pressure as well as finite time (and size) effects on predicted partitioning coefficients and melt structures, to determine realistic trace element activity coefficients and investigate coupled substitutions in particular for the prediction of mineral-melt partitioning.
2. Carbon dioxide in silicate melts: A molecular dynamics simulation study
5. Conclusion
By performing a series of molecular dynamics simulations where a supercritical CO2 phase is in contact with a silicate melt of a given composition (rhyolitic, basaltic or kimberlitic) at different temperatures (1473–2273 K) and pressures (20–150 kbar), we have been able to evaluate the solubility of CO2, the populations of molecular and carbonate species, their diffusivity through the melt and the local structure. The main results can be summarized at it follows. (1) The solubility of CO2 increases markedly with the pressure in the three investigated melts, the behavior being super-Henrian above 20 kbar (∼2 wt% CO2 at 20 kbar and more than 25 wt% at 100 kbar). Surprisingly, the solubility is found to be weakly dependent on the melt composition (as far as the present compositions are concerned) and it is only at very high pressure (above ∼100 kbar) that a clear hierarchy between solubilities occurs, namely rhyolite < MORB < kimberlite. Furthermore at a given pressure the calculated solubility is negatively correlated with the temperature.
(2) In CO2-saturated melts, the proportion of carbonate ions increases under pressure at isothermal condition and decreases when the temperature increases at isobaric condition. Furthermore, at fixed (P, T) conditions the proportion of carbonate ions is higher in CO2-undersaturated melts than in the CO2-saturated melt. Although the proportion of molecular CO2 decreases when the degree of depolymerization of the melt increases, it is still significant in CO2-saturated basic and ultrabasic melts. This finding is at variance with experimental data on CO2-bearing glasses which show no evidence of molecular CO2 as soon as the degree of depolymerization of the melt is high (e.g. basalt). These conflicting results can be reconciled with each other by noticing that a simple low temperature extrapolation of the simulation data for the solubility of CO2 in MORB predicts that the proportion of molecular CO2 might be negligible in the glass at room temperature. A conclusion also reached by several experimental studies (Porbatzki and Nowak, 2001, Morizet et al., 2001, Morizet et al., 2007, Nowak et al., 2003, Spickenbom et al., 2010) which show that an estimation based upon the analysis of quenched glasses tends to underestimate the true contents in molecular CO2 at liquidus temperature.
(3) The carbonate ions are found to be transient species in the liquid phase, with a lifetime increasing exponentially with the inverse of the temperature (from ∼10 ps at 2000 K to ∼100 ps at 1400 K). Therefore carbonate ions are expected to be long-lived species in the glass at room temperature. Contrarily to a usual assumption, the diffusivity of carbonate ions in the liquid silicate is not vanishingly small with respect to that of CO2 molecules: in MORB it is smaller by a factor of ∼6 at 1473 K and by a factor of ∼2 at 2273 K. Although the bulk diffusivity of CO2 is governed primarily by the diffusivity of CO2 molecules, the carbonate ions contribute significantly to it in depolymerized melts because they are preferentially associated with the NBO’s of the melt which are themselves diffusing faster than the BO’s.
(4) Concerning the structure of the melt network around CO2 species, the carbonate ions are preferentially associated with NBO’s of the melt, with an affinity for NBOs which exceeds that for BOs by almost one order of magnitude. This result explains why the concentration in carbonate ions is positively correlated with the degree of depolymerization of the melt and is hard to detect in fully polymerized melts where the number of NBO’s is close to zero. Furthermore, the network modifier cations are not randomly distributed in the close vicinity of carbonate groups but are ordered in a way which depends at once on the nature of the cation and on the melt composition. But at high temperatures investigated here, there is no evidence of long-lived complexes between carbonate groups and metal cations.
From a geochemical point of view, there is now a body of evidence that incipient melting may occur at depth up to 300 km under oceanic ridges (Dasgupta and Hirschmann, 2006) or in plumes (for a review see Bell and Simonetti (2010)) and carbonate liquids may be at the origin of these melts. In this context it is not immaterial that the present simulation study predicts a very high solubility of CO2 in basaltic and kimberlitic melts at high pressure (as high as 20–25 wt% CO2 at 80 kbar). Although the evaluation of CO2 solubility done here is more like a maximum of what may happen in the source regions, our finding suggests that a great amount of CO2 could be scavenged by interstitial melts (at very low melt fraction) in the upper mantle. So we hope that the present simulation results will promote the development of new degassing scenarios whose the initial step should start in the deep mantle.
3. Partitioning of Si and O between liquid iron and silicate melt: A two-phase ab-initio molecular dynamics study
4. Conclusions
[22] A two-phase ab-initio molecular dynamics simulation method is established to investigate the partitioning of Si and O between liquid iron and silicate melt in the system O-Mg-Fe-Si. The results from the ab-initio calculations are found to be in close agreement with experimental data.
[23] Calculations indicate 2.7 wt% silicon and 0.5 wt% oxygen could be added to the core through the magma ocean process if liquid iron is in equilibrium with silicate melt at the base of a deep magma ocean (39 GPa and 3116 K) and at the current bulk Earth composition. The oxygen content is lower than the current estimate of the core, implying a deeper magma ocean may need to be invoked.
4. Carbon and other light element contents in the Earth’s core based on first-principles molecular dynamics
Conclusions
The partition coefficient of C between liquid iron and silicate melt is determined to be 9 ± 3 based on FPMD. Assuming that C is distributed between the Earth’s core and mantle through magma ocean processes, the core is inferred to contain between 0.1–0.7 wt% C. Carbon thus plays a moderate role in the core density deficit and in the chalcophile and siderophile element distribution during core–mantle segregation processes. We further demonstrate that two-phase FPMD is a viable method by showing that it can provide partition coefficients that are in close agreement with experimental data and geochemical observations. Light element contents of the Earth’s core inferred by applying the partition coefficients agree with those derived from using the cosmochemical volatility curve of elements and geochemical mass balance arguments. Nitrogen deviates the most from this general agreement, suggesting a potential missing N problem for the Earth’s mantle.
5. Magnesium partitioning between silicate melt and liquid iron using first-principles molecular dynamics: Implications for the early thermal history of the Earth's core
5. Conclusion
In the present study, element partitioning data for Mg, Si and O between liquid iron and silicate melt are obtained using first-principles molecular dynamics simulations. These data are combined with previous experimental data to derive the equilibrium constant of magnesium partitioning reaction. It is found that the equilibrium constants are dependent on temperature and the dissolved MgO precipitates from metal as temperature decreases.
The newly derived equilibrium constants are used to calculate the composition of precipitates, and it is found that the MgO is the dominant ingredient compared to FeO and SiO2, taking up to ∼90 wt% of the precipitates. In combination with the thermal evolution of the Earth's core, the equilibrium constants are further applied to calculate the exsolution time and the Ohmic dissipation entropy for different initial core compositions, which give the information on whether the oxide melt can precipitate in time and provide enough power to sustain a geodynamo in the very early history of the Earth before the inner core nucleation. It is found that for the high silicon composition generated by the Grand Tack paradigm and for a reasonable Mg content (∼2 wt%) added into the core by the moon-forming Giant Impact with the impactor's size equal or larger than that of the Mars, oxide melt could start to precipitate at 3.5 Ga and provide enough power to drive the geodynamo before the inner core nucleation. By contrast, only when a pure compositional convection is taken into account, can the low silicon composition generated by EJS/CJS model and the O-rich composition generate magnetic field. In addition, the precipitation-driven geodynamo explains the gradual decrease of palaeomagnetic field intensity from 3.5 to 1.3 Ga.
Variation of the Earth's magnetic field intensity may now be understood by the switch of mechanisms for driving the geodynamo, with the fast cooling from very high temperature at first, followed by the light-element precipitation, in particular magnesium exsolution, and then the inner core nucleation to power the geodynamo in sequence. The Earth's accretion processes, the Moon-forming Giant Impact, the resulting magma ocean all play important and indispensable roles in concert in the early history of the Earth and in its subsequent evolution.
6. Molecular dynamics simulations of Y in silicate melts and implications for trace element partitioning
4. Conclusions
We combined MD simulations and EXAFS spectroscopy to investigate the structural environment of Y as a trace element in silicate melts of varying composition. For the MD, a new interaction potential including polarization was constructed for the system Y―Ca―Al―Si―O, which proved to be accurate, transferable and computationally efficient. The simulations revealed two structural trends: First, the average coordination number of Y decreases when the melt polymerization decreases (i.e. when the Ca content increases). This change is accompanied by a decrease of the average Y―O distance by about 4%, and at the same time, oxygen disorder around Y is reduced. A very similar variation is also seen in EXAFS experiments on glasses, which corroborates the reliability of the simulation results.
Second, the MD simulations for the three Ca-bearing melts indicate that the second (cationic) coordination shell around Y exhibits a larger Ca/(Si + Al) ratio than the bulk composition. In other words, Y tends to form clusters with the network modifier Ca, which implies that for a given melt, it is energetically more favorable for Y to share oxygen with Ca than with the network formers Si and Al. This, in turn, suggests that, given two melts of different composition, Y should partition preferentially into the one with larger Ca/(Si + Al) ratio, i.e. into the less polymerized melt. Indeed, modeling the exchange reaction of Y and Al between a Ca-free and a Ca-bearing melt by means of thermodynamic integration, we confirmed that minimization of Gibbs free energy drives Y into the Ca-bearing melt.
In summary, using simple systems, we presented computational and experimental evidence on how the influence of melt composition on trace element partitioning can be rationalized in terms of atomic-scale processes. We found a systematic influence of melt composition on the microscopic melt structure around Y and investigated the energetic implications of structural changes. The exemplary result that Y incorporation into melts is facilitated by the presence of network modifiers is consistent with the general observation that REE prefer depolymerized melts to polymerized ones. Although most systems which are studied experimentally, and Nature itself, are more complex than the melts investigated in this study, we still hold that the underlying mechanisms are the same in both cases.
7. Partitioning of sulfur between solid and liquid iron under Earth’s core conditions: Constraints from atomistic simulations with machine learning potentials
6. Concluding remarks
In this study, we derive new generation nonparametric interaction potentials for Fe-S systems applicable under Earth’s core conditions. Based on machine learning techniques, these Gaussian Approximation Potentials are shown to reproduce the first principles simulation results with unprecedented accuracies, including the interatomic forces, local structures and, most importantly, the free energies that fundamentally govern all thermodynamic properties. With a similar approach, we will be able to derive accurate potentials for more complex systems (e.g., multicomponent systems including elements of Ni, O, S, Si, C, H), which are very difficult to investigate solely with first principles techniques due to the increase in the size of the phase space.
The substantial initial efforts of training the machine learning potentials provide a return in the remarkable efficiency in sampling the phase spaces of iron and its alloys under various temperature and pressure conditions. It is then possible for us to simulate free energies and predict phase behaviors with fundamentally rigorous thermodynamic integration method within affordable computational cost. In fact, to thoroughly sample the phase space around the liquidus and solidus of Fe-S solutions, we have carried out over 500 independent atomistic simulations in this study, each with 180 atoms and at least 20,000 steps. The current implementation of the GAP models takes about 12 CPU seconds for each step in each run. With parallel acceleration of 24 CPU cores for each run, we have managed to accomplish the simulations within two months. As a comparison, the direct high precision DFT simulations are about three orders of magnitudes slower than the GAP simulations.
As a first application of the framework mentioned above, we focus in this paper on the partition coefficients of sulfur between the solid and liquid iron under Earth’s core conditions. While the results at ICB are in good agreements with early DFT simulations, we obtained the melting and partitioning behaviors over the entire relevant T-P regime of the Earth’s core. In particular, the invariance of partition coefficients from 250 GPa to higher pressures found in this study provides new constraint on the compositions and dynamics of Earth’s inner and outer core.
Finally, since the phase behaviors of iron alloys are comprehensively complex, it should be noted that much more endeavors are needed beyond our current efforts of predicting sulfur partitioning in Fe-S binary systems under core conditions. The interplays of different impurities, the possible stabilization of face-cubic-centered (fcc) or even body-cubic-centered (bcc) structures, the immiscibility of liquid iron-alloying systems, the heterogeneities of the Earth’s inner core, etc., can all be important to estimate the roles of light elements in real Earth’s core. Interests in these issues would imply quickly growing demands of computations over broader phase spaces. The high accuracy and efficiency gained by the framework proposed in this study would benefit providing new constraints over all these issues.
8. Coordination of Zr4+/Hf4+/Nb5+/Ta5+ in silicate melts: insight from first principles molecular dynamics simulations
5. Conclusions
In this study, by using first principles molecular dynamics (FPMD) technique, we investigated the coordination chemistry of Zr4+, Hf4+, Nb5+, Ta5+ and F− in anhydrous and hydrous F-free/F-bearing albite melts. We found that F− bound with Si and Al in silicate melts, but could not form stable complexes with Zr4+/Hf4+/Nb5+/Ta5+ in the melts. Zr4+, Hf4+, Nb5+ and Ta5+ formed 6-fold coordinations in silicate melts with the average Zr4+/Hf4+—O2− and Nb5+/Ta5+—O2− bond lengths of about 2.0 Å and 1.9 Å respectively. Our simulation results, together with previous solubility experiments by Fiege et al. (2011) and Aseri et al. (2015), indicate that F− does not have direct chemical interaction with Zr4+/Hf4+/Nb5+/Ta5+. These findings improve our understanding of the correlation between F− and Zr4+/Hf4+/Nb5+/Ta5+, and suggest that more attention should be paid to the indirect effect of fluorine on the magmatic ore-forming processes of HFSE.
9. The effect of CaO(MgO) on the structure and properties of aluminosilicate system by molecular dynamics simulation
4. Conclusions
The molecular dynamics simulation method was used to explain the similarities and differences between CaO and MgO in structure and properties of aluminosilicate at 1773 K. And these simulations reproduce well the changes in aluminosilicate structure. By comparison, it was found that the two metal oxides have substantially no significant effect on the short-range ordering of the aluminosilicate structure. At the same mass ratio, the degree of polymerization of the aluminosilicate containing MgO is higher than that of the aluminosilicate containing CaO. And when the two metal oxides increase in the same quality, MgO has a greater degree of damage to the network structure. Compared with the effect of CaO, more bridge oxygen is converted to free oxygen, resulting in a lower degree of polymerization of the system. Similarly, the diffusion coefficient of every atom in the magnesium aluminosilicate system is higher than that of the calcium aluminosilicate, and the total diffusion coefficient of the system is also higher, indicating that under the same mass conditions, the damage of the network structure to MgO is greater. As expected, the viscosity of both systems shows a decreasing trend. With the increase of metal oxides, the viscosity of magnesium aluminosilicate system is always lower than that of calcium aluminosilicates. The free running temperature of both systems decreases with the increase of metal oxides, and the potential energy of the two systems shows the opposite trend. From the perspective of energy, it could also be seen that the stability of the magnesium aluminosilicate system is lower than that of the calcium aluminosilicate system. Therefore, under the same mass fraction, the effect of MgO on aluminosilicate is the same as that of CaO, but it is stronger than CaO.
10. About hopping mechanism for sodium diffusion in silicate liquids with low sodium concentrations: Molecular dynamics simulation
4. Conclusion MD simulation has been carried out for Na2O.4SiO2 melt at 1873 K and ambient pressure. The structure and dynamics is studied in terms of Voronoi polyhedron, Sisingle bondO subnet and Osingle bondO cluster. The simulation shows that Na atoms are placed in O polyhedrons and not in Si ones. Average volume of Si polyhedron is equal to 8.13 Å3 that significantly smaller than BO, NBO and FO polyhedron (20.66, 33.11 and 42.4 Å3, respectively). The volume of O polyhedron is almost independent with number of Na located inside the polyhedron, indicating that the interaction between Na and O weakly affects the local arrangement of network formers. The simulation also reveals that Na atoms are not uniformly distributed through O polyhedrons, but they mainly gather in NBO and FO polyhedrons. In particular, 75.86% of total Na are placed in NBO and FO polyhedrons, the total volume of which is only 27.36% of the simulation box.
We propose a new mechanism for sodium diffusion in alkali-silicate melt with low alkali concentration. Accordingly, Na moves through the network structure by two ways: they move collectively with network formers staying in the same polyhedron, and frequently displace from sites to sites located in different O polyhedrons. The second movement type gives the essential contribution to sodium diffusivity, and it is the origin of very fast mobility of sodium. As NBO ↔ BO happen, Na atoms redistributed between O polyhedrons make the collective motion with network formers. Each BO, NBO has 1 and 2 sites, respectively. The site may be empty or occupied by one sodium. The site energy for Na located in NBO site is significantly smaller than BO site, and the transition energy for Na moving from present site to neighboring BO site is larger than the one to NBO site. The hops of Na between polyhedrons are strongly correlated with each other at low temperature. The movement of Na through BO sites resembles the interstitial mechanism, while Na atoms diffuse through NBO sites by mobile-vacancy mechanism.
The dynamical structure is found to be strongly heterogeneous. During 150 ps we observe two distinct regions occupied by Sisingle bondO subnet and Osingle bondO cluster, the O atoms of which do not make any NBO ↔ BO. Here Si and BO are placed in the first region, while NBO and FO reside in the second region. The sodium number density in the second region is by 10 times larger than the first one. In fact, during 150 ps the second region resembles a preferential diffusion pathway for sodium and possesses a volume of 20.5% of simulation box.
11. Molecular Dynamics Simulation of the Structure and Properties of CaO-SiO2-CaF2 Slag Systems
Conclusions
In this study, we investigated the microstructure of CaO-SiO2-CaF2 (CSF) molten slag at different basicities and with different CaF2 contents using MD simulations and FT-IR measurements. The conclusions are as follows:
(1) MD simulations show that in the CSF system, the average bond length of Si-O, Ca-O, and O-O are 1.61, 2.31, and 2.61 Å, respectively, whereas the coordinate number of Si does not change. The average angle of O-Si-O is unaffected by the basicity and remains at 109.2.
(2) An increase in the basicity increases the values of Q0, Q1, and Q2 and decreases the values of the Q3 and Q4 species. Based on these calculations, the addition of CaF2 in the system has no significant effect on the depolymerization of the silicate network at a low or high basicity.
(3) The FT-IR measurements demonstrated that the CaF2 has no obvious effect on the depolymerization of the network structure, and it acts mainly as a diluent. With increasing basicity, the corresponding peaks of the silicon tetrahedron structure shift to lower wave number regions, indicating a change in the melt structure from complex to simpler.
(4) The viscosity values calculated via the MD simulation and FactSage software were essentially the same. A good linear relationship was observed between the viscosity of the molten slag and the microstructure parameters from NBO/T. Therefore, with increasing basicity or increasing CaF2 contents, the viscosity of the molten slag decreased. CaF2 may reduce the viscosity value but mainly acts as a diluent.
12. Stabilization mechanism of arsenic-sulfide slag by density functional theory calculation of arsenic-sulfide clusters
4. Conclusion
In summary, DFT calculations were applied to investigate the structural stability of various As‒S clusters. The analysis of structural character suggest the S multimers-covering-As2S3 configuration possessed the highest stability amongst the candidates. The binding energy and formation energy of S multimers-covering-As2S3 is 3.74 and 26.0 eV, respectively. In typical, the electronic information confirms the stabilized mechanism is contributed to the 4p-orbital (As) binding with 3p-orbital (S) decreases energy level of HOMO. Motivated by the calculation results, a rational design was proposed by adding additional S into the As2S3 powders in the hydrothermal reaction to produce a chemically stable compound. Based on the standard toxicity leaching experiments, the As concentration in the leachate is only 0.8 mg/L, which is far lower than the As‒S compounds without interaction with S. The theoretical understanding on the structure-stability relationship of As‒S clusters and the inspired treatment method open a hopeful window for large-scale stabilization treatment of As‒S slag.
13. Ab Initio Modeling of Structure and Properties of Single and Mixed Alkali Silicate Glasses
In this paper, we presented the results of simulation on three single and two mixed alkali silicate glasses containing 20 and 30 mol % of alkali oxide using an accurate AIMD technique. The resulting structures in the form of BL and BA distributions are analyzed in detail showing the difference in the case of Li doping. The main focus is on the electronic structure, interatomic bonding, and their compositional dependence using the ab initio OLCAO method. The results are generally in good agreement with experiments. The doping of alkali oxide in silica network lead to the depolymerization of the network structure with different effects for different alkali oxides. This effect is more prominent with increasing size of the ion and with increasing alkali oxide content. The electronic structure calculation shows that the TBOD decreases with increased alkali content or a reduction in the internal cohesion in alkali silicate glass. The Li-doped silicate shows more difference in the calculated mechanical and optical properties than Na and K doped glass. The main conclusion is that MAE is only observed in the elastic properties showing deviation of mechanical parameters from the simple additive rule from single ion glasses. However, the present study did not address the possibility of phases separations in discussing the MAE, in which the two different alkali atoms in the silicate may tend to cluster in regions dominated by type. Such scenarios cannot be ruled out and can be investigated using the same methodology adapted in the paper. It will require separated modeling with a segregated region with the same number of alkali ions very similar to the work we published in ref 49 for mixed a-Si1–xGexO2 (x = 0 to 1) glass with a much larger model. That being said, we believe that the segregated model is unlikely because the simulation we did is with AIMD with quenching from high temperature down to low temperature. Any segregation that may exist will be fully accounted for. The implications of the present work can exemplify that AIMD in conjunction with the rigorous quantum mechanical evaluation of the physical properties will be the preferred venue for studying complex multicomponent glasses over the classical approach. These include the properties related to ionic motion in order to explore the ways and means to increase the durability of the glass and the origin of MAE in alkali silicate glass. The subject of doping and codoping with multiple cations in bulk silica glass and on surfaces with the presence of water molecules are viable possibilities using the methodologies demonstrated in this paper, assuming sufficient computational resources are available in using larger models. Such work, in conjunction with parallel experimental investigation can accelerate the speed of development for better products in commercial markets. Some of these projects are currently in progress and will be reported in the near future.
14. Effect of alkaline oxides on aluminate slag structure by first principles calculation
4. Conclusions Through the first-principles molecular dynamics simulation, the influence mechanism of 8 mol% alkaline oxides (Na2O, Li2O, BaO, MgO) on the molecular-atom-electron multi-scale structure characteristics of aluminate slag (CaO/Al2O3 = 1.3) at 1873 K was analyzed. The following conclusions were obtained.
(1) The stability between each atom and oxygen was in the order of Al-O > Mg-O > Ca-O > Li-O > Ba-O > Na-O. With the addition of alkaline oxides, the stability of Al-O and Ca-O bonds in each slag system was NAC ≈ BAC > LAC ≈ AC > MAC, indicating that BaO and Na2O increase the stability of Al-O and Ca-O bonds, while MgO was contrary action. Furthermore, due to the amphoteric characteristics of Al2O3, the proportion of AlV and the CN of Al were AC > MAC > LAC > BAC > NAC, describing that the addition of alkaline oxides in CaO-Al2O3 slag was beneficial to promote the stability of [AlO4]5- tetrahedron, and the strength was Na2O > BaO > Li2O > MgO.
(2) With adding 8 mol% alkaline oxide in CaO-Al2O3 slag, the alkaline cations partially replace the original Ca2+ combined with Onb to promote the structure depolymerization, and the ability was ranked as NAC > BAC > MAC > LAC, which were consistent with the results of O1s XPS and FTIR spectra. Moreover, alkaline cations also combined with Ob and Ot to play a charge compensation role with the ability sequenced as Na+ > Li+ > Ba2+ > Mg2+ because of the different polarizability. In view of the excellent characteristics of in Al-O structure stabilization, charge compensation and structure depolymerization, adding appropriate amount of Na2O/BaO was necessary to effectively regulate the performances of low/non-reactive mold flux.
(3) It can be concluded from PDOS and ELF that ionic bonds were formed between alkaline cations and O. In the Al-O tetrahedral structure, the localized electrons around O formed small synapses pointing to Al, forming a charge transfer bond, and a small number of electrons were localized between O and O, forming a chemical bond, which surrounds Al to form a closed structure. Therefore, alkaline oxides played two roles of depolymerization and charge compensation. It is through the interaction of alkaline cations with the closed structure of Al-O tetrahedron to maintain the balance of force and charge, which together constituted a relatively stable slag system.
15. First-principles study on microstructure of CaO-Al2O3-B2O3 slag
4. Conclusions The electronic structure and evolution law of CaO-Al2O3-B2O3 system slag with various C/A ratios were calculated by first-principles molecular dynamics. The following conclusions were obtained:
(1) With the increase of C/A, the bond length of B-O bond remained unchanged, while the bond length of Al-O bond decreased. The stability of the corresponding bonds of each atom was in the order of B-O > Al-O > Ca-O > O-O. The CNs of B-O and Al-O increased first and then decreased.
(2) The proportion of Of and Onb augmented, while the proportion of Ob and Ot decreased monotonously with the increase of C/A. In addition, XPS and FTIR results were consistent with the simulation results of that Al-O and B-O network depolymerized.
(3) From the ELF, Ca and O existed in the form of ionic bond, and B and Al connected with O formed charge-transfer bond. From the Bader charge, the oxygen ion charge from low to high was Onb < Ob < Ot.
(4) The structural evolution from BIII to BIV was accompanied by the charge transfer from the new bonded O to BIII, and the near-plane B-O triangular structure became tetrahedral. A stable electronic cloud connection was formed between each O of the B bond, which was beneficial to the stability of the B-O network structure.
(5) The structural evolution process between AlIV, AlV and AlVI was also accompanied by charge transfer. When AlIV transformed to AlV, the newly bonded O atoms gained charge and Al was pulled from the middle of the four O atoms to the edge of the tetrahedron, while when AlV changed to AlVI, the newly bonded O atoms lost charge and formed an octahedron structure. There were few stable electron cloud connections between O bonded with Al, which may be the essential reason for Al2O3 to be amphoteric oxide.
16. Ab initio molecular dynamics assessment of thermodynamic and transport properties in (K,Li)Cl and (K, Na)Cl molten salt mixtures
4. Concluding remarks
We have investigated (K,Li)Cl and (K,Na)Cl mixtures with five different compositions and at three different temperatures and compared our computational results with experimental data where available. In most cases, our computational results, from ab initio molecular dynamics, are consistently in better agreement with experimental measurements than those classical molecular dynamics. Most of the physicochemical properties of the mixtures depend on composition and temperature. Properties, such as the atomic charges, show some degree of additive behavior, while others, including ionic conductivity, show negative deviations from additivity. The primary reasons for this difference can be traced to the ability of ab initio molecular dynamics to account for both polarizability of ions as well as the charge fluctuations that accompany change in coordination environments. In studying the mixing properties of the salts, we found that it is mainly driven by entropy and that, KCl and LiCl mix better than KCl and NaCl. This study provides the first systematic ab initio insights into molten salt mixture systems. It also confirms the power of the theoretical approaches adopted to provide insights and enrich missing data for the predictions of thermal properties that are difficult to measure.
17. Comparison of desulfurization mechanism in liquid CaO-SiO2 and MnO-SiO2: An ab initio molecular dynamics simulation
- Conclusions
In this work, we explored the application of AIMD calculations on predicting slags’ structure, charge distribution and the sulfur dissolution mechanism. A comparative study of liquid CaO-SiO2 and MnO-SiO2. was conducted, which would be ready to be extended to other systems. After melting the solid CaO-SiO2 and MnO-SiO2., the structural features and charge distribution of the two systems are calculated under 2000 K. Subsequently, further calculations were made for the sulfur-doped system. The main conclusions are as follows:
1.Both liquid CaO-SiO2 and MnO-SiO2 are composed of Si-O and Ca-O/Mn-O units. Among them, Si-O has a strong polar bond while Ca-O and Mn-O show ionicity, and Mn-O has weaker ionicity than Ca-O. A small amount of Mn-Mn clusters are found in liquid MnO-SiO2.
2.Charge analysis depicts that there is less charge around Ca, while there are relatively more charges around Mn. Bader charge analysis shows that the valence of oxygen in the MnO-SiO2 slag is widely distributed within the range from − 1.7 e− 0.8 e, and that of oxygen in CaO-SiO2 is − 1.7 e− 1.4 e. The Bader charge of Ca and Mn atoms are distributed between 1.5 and 1.7e and 0.6–1.6e, while Si is mainly concentrated around 3e.
3.Sulfur doping calculations show that S forms a stable bonding structure with Mn atoms when incorporated into the liquid MnO-SiO2, and the Si–S bond is hardly found in the S-doped MnO-SiO2 silicate. However, in the CaO-SiO2 system, the S atom does not undergo rapid bond transitions, tends to form Si-S-Ca bonding structures.
4.The uneven distribution of charge in MnO-SiO2 system can affect the transformation of oxygen types, resulting in the decrease of bridged oxygen and the increase of non-bridged oxygen during desulphurization. However, there is no charge effect in CaO-SiO2 system, and the non-bridged oxygen will be consumed in the desulfurization process, resulting in the decrease of non-bridged oxygen and the increase of bridged oxygen.
18. An Ab Initio Molecular Dynamics Simulation of Liquid FeO–SiO2 Silicate System with Sulfur Dissolving
Conclusions
In this work, we explored the application of AIMD calculations on predicting silicates’ chemical properties and explaining the desulfurization mechanism of them. Liquid FeO·SiO2 was taken as an example to conduct the simulation, which would be ready to be extended to other systems. After melting the solid FeO·SiO2, we calculated the structural features, chemical and dynamical properties of the liquid FeO·SiO2 under 2000 K. The liquid FeO·SiO2 is primarily constituted by Si–O and Fe–O bonds, with the former being strong covalent bonds, while the latter showing the feature of ionic bonding. Charge analysis shows that Fe and O have a wide range of charge states, while the Bader charge of Si is relatively concentrated. At the same time, there are Fe-Fe clusters in the structure. The sulfur atom that is incorporated into the liquid FeO·SiO2 tends to form a stable bonding structure with three iron atoms, and the Si–S bond seems to be unstable thus unable to exist in the S-doped FeO·SiO2 silicate.