Research on Microstructure and Physical Properties of Molten Carbonate Salt based on Machine Learning
-
摘要: 采用第一性原理计算、机器学习以及经典分子动力学模拟联用的方法对K2CO3和Na2CO3熔融状态下的结构和性质进行计算。计算结果表明K2CO3能量与受力的测试误差分别为8.62×10-4eV/atom和4.67×10-2eV/10-10m；Na2CO3测试误差为1.19×10-3eV/atom和5.31×10-2eV/10-10m。结果表明计算值与实验值吻合较好，K2CO3密度、比热和热导率的计算偏差分别为5.0%、3.3%和8.0%，Na2CO3密度、比热和热导率的计算偏差分别为5.6%、6.5%和3.5%。Abstract: Molten alkali carbonate are widely concerned as potential thermal storage and transfer materials in solar power utilization. As an effective method in molten salt research, computer simulations have been widely used and researchers have been working hard to enhance the accurate of this method. In this paper, local structure and some physical properties of K2CO3 and Na2CO3 at different temperature were calculated by a complex simulation method including First-principle molecular dynamics, Machine learning and Classical molecular dynamics. In this method, First-principle molecular dynamics offered accuracy structure information, Machine learning was used to create deep potential from structure information to describe potential energy of system, and classical molecular dynamics simulation was used to performed large scale simulation. This complex method can reduce calculation errors caused by potential functions and empirical parameters. The calculation results show that energy and force learning test errors of K2CO3 were 8.62×10-4eV/atom and 4.67×10-2eV/10-10m, respectively, test errors of Na2CO3 were 1.19×10-3eV/atom and 5.31×10-2eV/10-10m. In all simulation process, the carbonate ion was a standard equilateral triangle structure in the system, and carbonate clusters were slightly loosened with the increase of temperature; the distance between anions and cations gradually increases with the increase of temperature. Comparing calculated data of the property with experimental value, the result shows that calculated data is in good agreement with the experimental value. The deviations of density, specific heat capacity and thermal conductivity of K2CO3 are 5.0%, 3.3% and 8.0%. The deviations of density, specific heat capacity and thermal conductivity of Na2CO3 are 5.6%, 6.5% and 3.5%.
-
Key words:
- ionic structure /
- properties of molten salt /
- molecular dynamic /
- carbonate salt
-
图 1 K2CO3学习过程中能量和受力的均方根误差 ((a)能量，(b)受力)
Figure 1. The energy and force RMSEs of K2CO3 during learning process ((a)energy, (b)force)
图 2 K2CO3机器学习的预测结果与DFT计算值的对比((a)能量, (b-d)受力)
Figure 2. Comparsion of ML prediction data and DFT calculation of K2CO3 ((a)energy, (b-d)force)
图 3 Na2CO3学习过程中能量和受力的均方根误差 ((a)能量，(b)受力)
Figure 3. The energy and force RMSEs of Na2CO3 during learning process ((a)energy, (b)force)
图 4 Na2CO3机器学习的预测结果与DFT计算值的对比((a)能量, (b-d)受力)
Figure 4. Comparsion of ML prediction data and DFT calculation of Na2CO3 ((a)energy, (b-d)force)
图 5 不同温度下K2CO3径向分布函数的DPMD与FPMD计算结果对比
Figure 5. Comparsion of Radial distribution functions of K2CO3 derived from DPMD and FPMD simulation at different temperature
图 6 DPMD方法计算所得的K2CO3径向分布函数
Figure 6. Radial distribution functions of K2CO3 calculated by DPMD method
图 7 不同温度下Na2CO3径向分布函数的DPMD与FPMD计算结果对比
Figure 7. Comparsion of Radial distribution functions of Na2CO3 derived from DPMD and FPMD simulation at different temperature
图 8 DPMD方法计算所得的Na2CO3径向分布函数
Figure 8. Radial distribution functions of Na2CO3 calculated by DPMD method
图 9 不同温度下K2CO3和Na2CO3的密度实验值与模拟值对比
Figure 9. Comparison of literature and simulated densities of K2CO3 and Na2CO3 at different temperatures
图 10 (a)K2CO3计算中焓随温度的变化，(b) Na2CO3计算中焓随温度的变化
Figure 10. Changes of enthalpy of (a) K2CO3, (b) Na2CO3 with temperature
图 11 K2CO3和Na2CO3的热导率实验值与模拟值的对比
Figure 11. Comparison of literature and simulated thermal conductivity of K2CO3 and Na2CO3
图 12 归一化张量自相关函数((a) K2CO3 (b) Na2CO3)
Figure 12. Normalized stress tensor time auto-correlation function ((a) K2CO3 (b) Na2CO3)
图 13 K2CO3和Na2CO3的黏度实验值与计算值的对比
Figure 13. Comparison of literature and simulated viscosity of K2CO3 and K2CO3
表 1 K2CO3和Na 2CO3的比热的模拟值与实验值对比
Table 1. Comparison of experimental and simulated results of the Cp of K2CO3 and Na 2CO3
Cp/(J·g-1·K-1) K2CO3 Na 2CO3 计算值 1.56 1.72 实验值 1.51 1.83 
下载: 导出CSV
-
[1] BRADSHAW R W, CORDARO J G, SIEGEL N P. Molten nitrate salt development for thermal energy storage in parabolic through solar power systems[R]. San Francisco, California, USA: The American Society of Mechanical Engineers, 2009. [2] SANG L, CAI M, REN N, et al. Improving the thermal properties of ternary carbonates for concentrating solar power through simple chemical modifications by adding sodium hydroxide and nitrate[J]. Sol Energy Mater Sol Cells, 2014, 124: 61-66. doi: 10.1016/j.solmat.2014.01.025 [3] MARK M, CRAIG T, JUDITH V, et al. Concentrating solar power gen3 demonstration roadmap[R]. Golden, Colorado, USA: National Renewable Energy Lab, 2017. [4] KRUIZENGA A M. Corrosion mechanisms in chloride and carbonate salts[R]. Livermore, California, USA: Sandia Narional Laboratories, 2012. [5] BRIK M G. First-principles calculations of structural, electronic, optical and elastic properties of magnesite MgCO3 and calcite CaCO3[J]. Physical B: Condensed Matter, 2011, 406(4): 1004-1012. doi: 10.1016/j.physb.2010.12.049 [6] DUNSTAN M T, GRIFFIN J M, BLANC F, et al. Ion dynamics in Li2CO3 studied by solid-state NMR and first-principles calculations[J]. Journal of Physical Chemistry C, 2015, 119(43): 24255-24264. doi: 10.1021/acs.jpcc.5b06647 [7] GARCIA L J M, MYRDAL J S G, CHRISTENSEN R, et al. DFT+U Study of polaronic conduction in Li2O2 and Li2CO3: Implications for Li-Air batteries[J]. Journal of Physical Chemistry C, 2013, 117(11): 5568-5577. doi: 10.1021/jp3107809 [8] PHAM T A, KWEON K E, SAMANTA A, et al. Solvation and dynamics of sodium and potassium in ethylene carbonate from ab Initio molecular dynamics simulations[J]. Journal of Physical Chemistry C, 2017, 121(40): 21913-21920. doi: 10.1021/acs.jpcc.7b06457 [9] OTTOCHIAN A, RICCA C, LABAT F, et al. Molecular dynamics simulations of a lithium/sodium carbonate mixture[J]. Journal of Molecular Modeling, 2016, 22(3): 61. doi: 10.1007/s00894-016-2921-4 [10] ZAKIRYANOV D O, TKACHEV N K. Thermal conductivity of alkali metal chlorides: Calculation with molecular dynamics method[J]. High Temperature, 2020, 58(1): 54-57. doi: 10.1134/S0018151X20010228 [11] WANG J, WU J, SUN Z, et al. Molecular dynamics study of the transport properties and local structures of molten binary systems (Li, Na)Cl, (Li, K)Cl and (Na, K)Cl[J]. Journal of Molecular Liquids, 2015, 209: 498-507. doi: 10.1016/j.molliq.2015.06.021 [12] JAYARAMAN S, THOMPSON A P, LILIENFELDO A, et al. Molecular simulation of the thermal and transport properties of three alkali nitrate salts[J]. Industrial and Engineering Chemistry Research, 2010, 49(2): 559-571. doi: 10.1021/ie9007216 [13] KOISHI T, KAWASE S I, TAMAKI S, et al. Computer simulation of molten Li2CO3-K2CO3 mixtures[J]. Journal of the Physical Society of Japan, 2000, 96: 3291-3296. [14] COSTA M F. Molecular dynamics of molten Li2CO3-K2CO3[J]. Journal of Molecular Liquids, 2008, 138(1-3): 61-68. doi: 10.1016/j.molliq.2007.08.001 [15] DU L, XIE W, DING J, et al. Molecular dynamics simulations of the thermodynamic properties and local structures on molten alkali carbonate Na2CO3[J]. International Journal of Heat and Mass Transfer, 2019, 131: 41-51. doi: 10.1016/j.ijheatmasstransfer.2018.11.044 [16] DESMAELE E, SATOR N, VUILLEUMIER R, et al. Atomistic simulations of molten carbonates: Thermodynamic and transport properties of the Li2CO3-Na2CO3-K2CO3 system[J]. Journal of Chemical Physical, 2019, 150(9): 094504. doi: 10.1063/1.5082731 [17] KIYOBAYASHIi T, KOJIMA T, OZAKI H, et al. Ionic conductivity of molten alkali-metal carbonates A2CO3 (A = Li, Na, K, Rb, and Cs) and binary mixtures (Li1-xCsx)2CO3 and (Li1-xKx)2CO3: A molecular dynamics simulation[J]. Journal of Chemical Physical, 2019, 151(7): 074503. doi: 10.1063/1.5109912 [18] ROEST D L, BALLONE P, BDEDAUX D, et al. Molecular dynamics simulations of metal/molten alkali carbonate interfaces[J]. Journal of Physical Chemistry C, 2017, 121(33): 17827-17847. doi: 10.1021/acs.jpcc.7b02842 [19] PERDEW J P, BURKE K, ERNZERHOF M. Generalized Gradient Approximation Made Simple[J]. Physical Review Letters, 1996, 77(18): 3865. doi: 10.1103/PhysRevLett.77.3865 [20] BLOCHL P E. Projector augmented-wave method[J]. Physical Review B: Condens Matter Materials Physics, 1994, 50(24): 17953-17979. doi: 10.1103/PhysRevB.50.17953 [21] KRESSE G, JOUBERT D. From ultrasoft pseudopotentials to the projector augmented-wave method[J]. Physical Review B, 1999, 59(3): 1758. doi: 10.1103/PhysRevB.59.1758 [22] TUO P, YE X B, PAN B C. A machine learning based deep potential for seeking the low-lying candidates of Al clusters[J]. Journal of Chemical Physical, 2020, 152(11): 4106. [23] LIANG W S, LU G M, YU J G. Molecular dynamics simulations of molten magnesium chloride using machine‐learning‐based deep potential[J]. Advnced Theory and Simulations, 2020, 3(12): 2000180. doi: 10.1002/adts.202000180 [24] KINGMA D P, BA J L. Adam: A method for stochastic optimization[R]. San Diego, USA: The 3rd International Conference for Learning Representations, 2015. [25] ZHANG L, HAN J, WANG H, et al. Deep potential molecular dynamics: A scalable model with the accuracy of quantum mechanics[J]. Physical Review Letters, 2018, 120(14): 3001. [26] SOMMERS G M, CALEGARI A M F, ZHANG L, et al. Raman spectrum and polarizability of liquid water from deep neural networks[J]. Physical Chemistry Chemical Physical, 2020, 22(19): 10592-10602. doi: 10.1039/D0CP01893G [27] ZENG J, CAO L, XU M, et al. Complex reaction processes in combustion unraveled by neural network-based molecular dynamics simulation[J]. Nature Communications, 2020, 11(1): 5713. doi: 10.1038/s41467-020-19497-z [28] ALLEN M P, TLLDESLEY D J. Computer Simulation of Liquids[M]. New York: Oxford University Press, 1987. [29] MELCHIONNA S, CICCOTTI G, LEE H B. Hoover NPT dynamics for systems varying in shape and size[J]. Molecular Physics, 1993, 78(3): 533-544. doi: 10.1080/00268979300100371 [30] MARTYNA G J, TOBIAS D J, KLEIN M L. Constant pressure molecular dynamics algorithms[J]. Journal of Chemical Physical, 1994, 101(5): 4177-4189. doi: 10.1063/1.467468 [31] WIRNSBERGER P, D FRENKEL, DELLAGO C. An enhanced version of the heat exchange algorithm with excellent energy conservation properties[J]. Journal of Chemical Physical, 2015, 143(12): 124104. doi: 10.1063/1.4931597 [32] ZHANG M, LUSSETTI E, SOUZA L E S D, et al. Thermal conductivities of molecular liquids by reverse nonequilibrium molecular dynamics[J]. Journal of Physical Chemistry B, 2005, 109(31): 15060-15067. doi: 10.1021/jp0512255 [33] JANZ J, ALLEN C, BANSAL N, et al. Physical properties data compilations relevant to energy storage. II. Molten salts: Data on single and multi-component systems: NSRDS-NBS 61[R]. New York: National Bureau of Standards, 1979. [34] OTSUBO S, IBATA T, NAGASHIMA A. Experimental study on the forced Rayleigh scattering method using CO2 laser.[J]. Transactions of the Japan Society of Mechanical Engineers B, 1998, 64(617): 182-187. doi: 10.1299/kikaib.64.182 -
点击查看大图
计量
- 文章访问数: 91
- HTML全文浏览量: 36
- PDF下载量: 15
- 被引次数: 0
