节点文献

反渗透膜及水溶液内扩散过程的分子模拟研究

Studies on the Molecular Simulation of Diffusion Process in Reverse Osmosis Membrane and Water Solution

【作者】 刘清芝

【导师】 胡仰栋;

【作者基本信息】 中国海洋大学 , 海洋化学工程与技术, 2007, 博士

【摘要】 本论文针对目前传统试验手段用在研究分子扩散过程以及扩散系数方面的局限性,通过分子动力学模拟(Molecular Dynamic Simulation)的手段,深入详细地研究了小分子在反渗透复合膜以及水溶液内的扩散过程。主要进行了四方面的研究工作:分子在体系内扩散系数的分子模拟方法研究;甘氨酸等12种氨基酸分子在水溶液内扩散系数的模拟;氧气等9种气体分子在水溶液内扩散系数的模拟;水和盐分子(NaCl, MgCl2,CaSO4,K2SO4)在8种反渗透复合膜中扩散状态及扩散系数的研究,膜的结构单体与水和盐分子在膜内扩散系数关系的分析与讨论。1)论文首次从模型的分子数样本以及模拟步长出发,对影响分子在体系内扩散系数的主要因素进行了研究。并针对动力学模拟数据提出了新的分析方法。①对浓度分别为0.7mol/L和0.9mol/L的甘氨酸水体系进行了动力学模拟。建立模型时分别加入了不同数目的分子:对0.7mol/L的甘氨酸模型的构建,分别在立方晶格中加入了3、5、7、10、12个甘氨酸分子;对于0.9mol/L的甘氨酸则分别在立方晶格中加入了4、7、9、13、15个甘氨酸分子。扩散系数模拟值与实验值的对比表明分子数样本对于扩散系数的模拟结果有较大的影响。分子数越多,则模拟结果与实验值越接近,当分子数达到一定值时,扩散系数趋于一个稳定的值。论文还针对不同温度下氧气在水中的扩散过程作了关于样本数的模拟。分别考察了温度为303K和313K条件下的气体扩散。模拟体系中氧气分子数分别采用1,2,3个。结果表明随着模型中分子数的增多,得到的扩散系数的值也逐渐增大,并逐渐恒定于正确值。②通过对比传统分析方法以及微分-区限变分法用在均方位移(MSD)曲线上的优缺点,本论文提出了新的数据分析方法-聚类分析方法。该方法克服了传统方法用在MSD线性不好时给分析结果带来误差的缺点,对微分-区限变分法进行了改进。将聚类方法和微分-区限变分法应用在氧气在水中的扩散系数的分析中,实验值,微分-区限变分法值和聚类方法值分别为2.10×10-9m2s-1,1.80×10-9m2s-1,2.00×10-9m2s-1,聚类分析方法的误差要小于微分-区限变分法,充分证明聚类方法是一个有效的分析方法。③模拟甲烷和氨气在水中的扩散过程时,对0.1fs的步长值延长了2-3倍的模拟时间,结果表明模拟时间的延长对最后的模拟结果没有明显的影响,从而排除了模拟时间对扩散系数模拟结果的影响。另外,论文对甲烷,氨气和二氧化碳气体在水中的扩散过程采用了0.1fs-1.5fs范围内几个不同的步长值,模拟结果表明,步长值对扩散系数的模拟结果有较大影响。随着步长的增加,扩散系数有一个极值,与该极值相邻的两个步长对应的扩散系数与实验值最为接近。这两个步长值为极值对应的步长值的70%和150%。结合步长在模拟计算中的作用,本论文认为对于一个特定体系,应在0.1fs到1.5fs之间取多个值进行模拟,取模拟值最大的步长的70%即为模拟的合理值。2)本论文首次通过分子模拟的手段研究了氨基酸在水溶液内的扩散过程,并利用微分-区限变分法计算了氨基酸在水中的扩散系数。目前氨基酸在水中扩散系数主要通过实验手段来获得,对于氨基酸在水中的扩散系数的模拟,目前尚未见报道。本文通过分子动力学模拟的手段模拟计算了甘氨酸、丙氨酸等12种氨基酸在水中的扩散系数。模拟值与实验值的误差小于7%,证明了分子模拟方法应用在扩散系数计算方面的有效性。3)本论文通过分子模拟的手段模拟了氧气等9种气体在水溶液内扩散过程,并利用聚类分析方法计算了气体在水中的扩散系数。对于O2,CO2,CH4,NH3四种气体在水中的扩散系数的模拟,已有文献的模拟值与实验值的误差最大超过50%,而本论文的模拟数据与实验值的误差小于30%。充分证明了本文所采用的分子模拟途径的可靠性。4)本文以8种反渗透复合膜为主要研究对象,模拟了海水中主要成分(H2O,NaCl, MgCl2,CaSO4和K2SO4)在膜内的扩散过程并计算了各粒子在膜内的扩散系数。①模拟了TMC/MPD膜在体系中的扩散系数,模拟结果为0.0115A2/ps,与已有的对于该膜的模拟结果0.015 A2/ps基本吻合。证明了本文模拟过程中对于模型建立,模拟方法以及数据分析方法的有效性。②水分子在TMC/MPD膜内的扩散状态表明水分子与高分子链上的功能基团有明显的分子间相互作用。水分子在靠近苯环的区域内与苯环有排斥运动;在靠近酰胺基的区域内有吸引运动。③在所有膜中,水分子在膜HT/MMPD中的扩散系数最大,远远高于在其他膜内的扩散系数。对比水分子在膜TMC/MMPD和TMC-HT/MMPD的扩散系数发现水分子在加入了HT的膜中扩散系数要高一些了;对比水分子在膜TMC/MPD和膜TMC/MMPD内的扩散系数,水分子在膜TMC/MMPD的扩散系数要小很多。由于MMPD单体在苯环上含有一个甲基,在同样数量和比例生成聚合物膜的情况下,MMPD比MPD的刚性要强一些,从而不利于水分子的通通过。对比水分子在膜CFIC/MMPD和膜CFIC/MPD的扩散系数也可得出同样的结论。④NaCl分子在膜HT/MMPD中的扩散系数最大,而在膜ICIC-HT/MMPD中的扩散系数则最小,其次分别是膜ICIC-IPC/MPD ,膜TMC-HT/MMPD,膜CFIC/MPD以及膜TMC/MMPD。模拟结果中只有ICIC-IPC/MPD与实验结果有较大差距。⑤在所模拟的8种膜内,Na+和Cl-在膜内的扩散系数值相差较大,NaCl分子的运动受限于运动较慢的离子。膜的种类不同,制约NaCl在膜内扩散的离子便不同:对于膜TMC/MPD以及膜CFIC/MMPD,NaCl分子的扩散受Cl-的制约;而其他6种类型的膜则受Na+的制约。⑥对于膜ICIC-HT/MPD,水分子在MgCl2,CaSO4和K2SO4体系中的扩散系数与在NaCl体系中的结果基本一致,该结果表明在同种膜中,水分子的扩散过程不受体系中盐分子类型的影响。⑦在ICIC-HT/MPD反渗透膜中,阴离子(Cl-,SO42-)的扩散系数均远大于阳离子(Mg2+,Ca2+,K+)的扩散系数。这与NaCl水体系中,Cl-的扩散系数远大于Na+的扩散系数的运动状况一致,该结果说明当盐分子单独存在时,其在膜内的扩散过程受哪个离子制约只与膜种类有关,与盐分子本身无关。

【Abstract】 This paper used molecular Dynamic Simulation method to detailedly study the diffusion process and diffusion coefficient of small molecules in reverse osmosis membrane and water solution, and got some inaccessible results with traditional experimental method. There are mainly four aspects discussed in the paper: (1) Research of molecular Dynamic Simulation method for molecular diffusion coefficients. (2) The simulation of diffusion coefficients of Glycine and other 11 kinds of amino acids in water solution. (3) The simulation of diffusion coefficients of oxygen and other 8 kinds of gas molecules in water solution. (4) The diffusion process and diffusion coefficients of H2O, NaCl, MgCl2, CaSO4, K2SO4 in 8 kinds of reverse osmosis membrane, and relationships of the membranous structure monomer and the diffusion coefficients.1) This paper firstly researched the factors which affect diffusion coefficients in the system from the molecular number sample and simulation time step of the model, and advanced a new analytic method with the Dynamic Simulation data.①This paper simulated diffusion process of Glycine solutions of 0.7mol/L and 0.9mol/L. In the Dynamic Simulation, the models of 0.7mol/L Glycine solution were constructed by adding 3, 5, 7, 10, 12 Glycine molecules into the cell, and 4,7,9,13,15 Glycine molecules were added into the cell of 0.9mol/L Glycine solution models. Comparing the simulation values with the experimental results, the sample number of molecules had some effect on the simulation of diffusion coefficients, which was closer to the experimental results with more molecules added into the cell, and approached a steady value when the molecule number exceeds a line.This paper also simulated the diffusion process of oxygen in water of different temperature, such as 303K and 313K. The models were separately constructed by adding 1,2,3 oxygen molecules into the cell . Simulation values were also found approaching experimental results with more molecules in the cell.②Through comparing the use of traditional analysis method in MSD curves with differentiation interval variation method, this paper advanced a new data analysis method-clustering method, which overcome shortcomings of traditional method used in some MSD curves with bad linear direction and modifies differentiation interval variation method. Using clustering method and differentiation interval variation method separately in the analysis of diffusion coefficient of oxygen in the water, the results of them and the experimental value were separately 2.00×10-9m2s-1,1.80×10-9m2s-1,2.10×10-9m2s-1, which obviously proved clustering method effective.③This paper increased simulation time of 0.1fs time step to 2~3 times longer than before in the simulation of diffusion process in water of methane and ammonia, the results showed the increasing time of simulation did not have obvious effect. Then several different time step between 0.1fs to 0.5 fs were used in the simulation of diffusion process in water of methane, ammonia and carbon dioxide, results showed time step was the effective factor to the diffusion coefficient, which rose to a maximum with the increasing of time step. The diffusion coefficients corresponding to the two time steps close to the maximum were most close to the experimental value. This two time steps were 70 and 150 percent of the time step corresponding to the maximum diffusion coefficient value. Considering the effect of time step in the simulation, this paper advised several time steps between 0.1fs to 0.5 fs should be taken in the simulation of a special system, and took 70 percent of the time step corresponding to maximum diffusion coefficient value as the reasonable value.2) The paper dealt with for the first time the diffusion process of amino acid in the water solution by the way of molecular simulation, at the same time, the diffusion coefficient of amino acid in water was calculated by the method of differentiation interval variation. At present, the method of gaining the diffusion coefficient of amino acid in the water mainly based on the experiments, on the contrary, the way of simulation for the diffusion coefficient of amino acid in the water wasn’t published preciously. In this thesis, the diffusion coefficient of amino acid, such as Glycine and other 11 kinds of amino acid, was calculated using the method of molecular dynamic simulation. The error between the simulated results and the experimental results was less than seven percent, and therefore the way of calculating diffusion coefficient by molecular dynamic simulation was proved to be effective.3) In this paper, the diffusion process of oxygen and other 8 kinds of gases in the water solution was simulated by the way of molecular simulation, meanwhile, the diffusion coefficient of gas in the water was calculated by the method of clustering method. The work present in this thesis about the simulation of getting diffusion coefficient of the oxygen, carbon dioxide, methanol and ammonia in the water solution was enough to prove the reliability of the method of molecular simulation, in which the error between the simulated data in this paper and the experimental results was less than 30 percent, however, the biggest error between simulated data in the literature and experimental result was more than 50 percent.4) In this paper, eight kinds of reverse osmosis compound membrane were used as the main base case, in which the diffusion process of the main components (water, sodium chloride, magnesium chloride, calcium sulphate and kalium sulphate ) in the seawater within the membrane were simulated and the diffusion coefficient of every particle in the membrane is calculated.①The diffusion coefficient of TMC/MPD membrane in the system was simulated with the result of 0.0115A2/ps, which was almost inosculated with the existing simulated result 0.015 A2/ps for this membrane. In a word, the foundation of the model, the method of simulation and the analytic method for the data were effective in the simulation process for this paper.②The diffusion state of water molecules in the TMC/MPD indicates that there were obvious interaction between water molecules and Functional groups of the polymer. When the Water molecules were in the field nearing benzene fragments there was a significant repulsion action. On the other way, when water molecules near the amido-group there was an attracting tend movement.③The diffusion coefficient of water molecule in the membranes of HT/MMPD was the biggest in all of the membranes we studied. It will be founded that the diffusion coefficient of water molecule in the membrane of TMC-HT/MMPD was higher somewhat than TMC/MMPD which was added TH; Analysis of the water molecule diffusion coefficient in the membrane of TMC/MPD and TMC/MMPD displayed that the water molecule diffusion coefficient in the membrane of TMC/MPD was much bigger than in TMC/MMPD. For the monomer of the membrane MMPD had a methyl substitution in the benzene ring, the rigidity of the membrane MMPD was stronger than MPD when the membranes consist of the same account and scale of the monomer. As a result, this did not benefit the molecules of water transitting through the membrane MMPD. The same conclusion was obtained by comparing the diffusion coefficient of the molecules of water in the membrane CFIC/MMPD and CFIC/MPD.④NaCl molecule had the biggest diffusion coefficient in the membrane of HT/MMPD. Then it took turns to the membrane of ICIC-IPC/MPD, TMC-HT/MMPD, CFIC/MPD , TMC/MMPD, and the least in the membrane of ICIC-HT/MMPD. Only ICIC-IPC/MPD had the bigger difference between simulation result and the experiment result.⑤In eight kinds of membranes we simulated, the diffusion coefficient value of Na+ and Cl- had much difference. The movement of NaCl molecule was restricted by the ions which moves slowly. The kind of the ions which constrained the molecule NaCl was variety in different membranes: in the membranes of TMC/MPD and CFIC/MMPD, the constrain ion was Cl-, while was Na+ in other six membranes.⑥For membrane ICIC-HT/MPD ,the diffusion coefficients of water in the systems containing MgCl2,CaSO4 and K2SO4 were consistent with that in the NaCl system, this result demonstrated that movement of molecule of water was not affected by the kinds of salt molecules in the same membrane. ⑦The diffusion coefficients of anions (Cl-,SO42-) in the reverse osmosis membrane ICIC-HT/MPD were much bigger than cations (Mg2+,Ca2+,K+). This result was consistent with the diffusion coefficient of the anions Cl- and cations Na+ in the NaCl-H2O system, which indicated that diffusion process of salt molecule in the membrane was only restricted by the membrane species,without relationship with salt molecules itself.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络