運用氫鍵特性預測耦合系統之相行為

Abstract

活性係數對於了解混合物的熱力學性質與相平衡來說相當重要。COSMO-SAC活性係數模型藉由量子力學計算取得分子資訊 (如分子表面屏蔽電荷分佈)，因此可不需仰賴實驗值迴歸分子相關參數，對一般系統均可提供可靠的流體相行為預測。然而在過往研究中也發現COSMO-SAC模型對於耦合系統的預測準確性較差，有鑑於此本研究提出以下方法來改善COSMO-SAC模型在耦合系統上的預測能力。 首先由於過往COSMO-SAC方法中對於氫鍵作用的描述僅利用分子表面屏蔽電荷密度的強度，而未考慮到在形成氫鍵時施體和受體所具有的特殊方向性，換言之在形成氫鍵時施體(連接於氟、氧、氮等上的氫原子)傾向與受體(氟、氧、氮等原子)的孤對電子對齊。由此特性，本研究在第一部分提出新的COSMO-SAC(DHB)模型，將氫鍵在空間上特殊的方向性納入考慮。此模型利用價層電子對互斥理論 (VSEPR)理論，將受體的孤對電子投影至分子表面上作為在空間上氫鍵作用範圍的限制。研究結果顯示在引入氫鍵的方向性後，能夠減少模型參數並且提升耦合系統在汽液相平衡、無限稀釋下活性係數和水與正辛烷分布係數等係質預測上的精準度與可靠性。 然而COSMO-SAC(DHB)模型應用於有機酸的相平衡時，表現與以往的COSMO-SAC模型一樣仍有極大的誤差存在。對此從過往文獻的研究中發現對於氫鍵作用力較強的系統，分子間容易發生聚合現象，產生二聚體或分子簇(molecular cluster)等局部流體結構(local fluid structure)。因此本研究的第二部分以乙酸為例，利用PR+COSMOSAC狀態方程式，探討局部流體結構對於乙酸在相平衡上的影響。本研究中提出利用反應平衡的方式同時考慮乙酸的單體(monomer)、環狀二聚(cyclic dimer)、鏈狀碎片(chain fragment)以及與水(或醇)形成的共聚體(cross-associating structure)。研究結果顯示出只有在同時考慮乙酸各種局部結構，才能一致地預測出乙酸在各種環境下的相行為，因此也證實了局部結構對於乙酸相平衡的重要性。 除了以上針對耦合系統的研究外，本研究的第三部分著重在不同量子化學方法，在虛擬溶劑模型(COSMO solvation)計算上造成分子表面屏蔽電荷分佈的差異，對於COSMO-SAC模型的影響。此研究對常見的量子化學方法重新優化模型的參數，並利用大量的實驗資料測試三種模型的表現，包括汽液相平衡、液液相平衡、無限稀釋下活性係數和水與正辛烷分布係數等總共超過22,000個數據點。結果顯示COSMO-SAC 2010和COSMO-SAC(DHB)在理論上的修正，皆適用於各種量子力學方法，而其中以COSMO-SAC(DHB)模型表現最好。此外COSMO-SAC(DHB)模型會敏感地受到量子化學方法的影響，一般而言使用描述分子屏蔽電荷越極化程度越好的基底函數組(如b3lyp/6-31+G(d,p))，COSMO-SAC(DHB)模型具有較佳的預測精準度度。此結果也顯出除了在COSMO-SAC模型本身的開發上，量子力學計算上的進展對於提升COSMO-SAC模型的精準度也相當重要。

The activity coefficient of a chemical in a mixture is important to understand the thermodynamic properties and the phase behaviors of the mixture. The COSMO-SAC model based on the result of quantum mechanical implicit solvation calculations has been shown to provide reliable predictions of activity coefficients for mixture fluids. However, it is found that the prediction accuracy is in general inferior for associating fluids. This work aims to improve the prediction of phase behaviors of associating fluids through consideration of hydrogen bond direction and the formation of local fluid structures. First existing COSMO-SAC methods for describing the hydrogen-bonding interaction consider the strength of interaction based only on the polarity of the screening charges, neglecting the fact that the formation of hydrogen bonds require specific orientation between the donor and acceptor pairs. We develop a new approach that takes into account the spatial orientational constraints in hydrogen bonds. Based on the Valence Shell Electron Pair Repulsion (VSEPR) theory, the molecular surfaces associated with the formation of hydrogen bonds are limited to those in the projection of the lone pair electrons of hydrogen bond acceptors, in additional to the polarity of the surface screening charges. Our results show that the directional hydrogen bond approach, denoted as the COSMO-SAC(DHB) model, requires fewer number of universal parameters and is significantly more accurate and reliable compared to previous models for a variety of properties, including vapor-liquid equilibria (VLE), infinite dilution activity coefficient (IDAC) and water-octanol partition coefficient (Kow). With the increasing strength of interactions between hydrogen bonded molecules, the formation of local fluid structure, such as the dimers and/or hydrogen bonded clusters, can strongly affect the phase behavior of fluids. For acetic acid, we identify four local fluids structures, including the acetic acid monomer, cyclic dimer and chain fragment and the cross-associating structure with water (or alcohol), and develop a novel approach to explicitly consider these local fluid structures in the predictive thermodynamic model, PR+COSMOSAC equation of state. The transition of acetic acid in these local fluid structures are considered via chemical reaction. The results show that the phase behaviors of pure acetic acid and its mixtures with various chemicals can be described only when all the significant local fluid structures are included. In addition to the efforts on the development of a better description for associating fluids, the utilization of different quantum calculation methods for COSMO solvation calculation is carefully examined for different COSMO-SAC models. The model parameters are reoptimized for each quantum calculation method, and the performance is evaluated using a large set of experimental databases covering VLE, liquid-liquid equilibrium (LLE), IDAC and Kow (containing more than 22,000 data points). The results show that the modification introduced in either COSMO-SAC 2010 or COSMO-SAC(DHB) is applicable for all quantum calculation methods and the performance of COSMO-SAC(DHB) model is generally better than others. Besides, the COSMO-SAC(DHB) model is sensitive to the quantum chemical method used. The use of a basis set that allows for higher molecular polarity, such as b3lyp/6-31+G(d,p), often results in a better prediction accuracy. The finding implies apart from the refinement in COSMO-based method itself, the improvement in the COSMO solvation calculation is also important for the development of COSMO-based model.