奈米薄膜積垢機制研究-以金門太湖為例

Abstract

近幾年來，薄膜處理技術被廣泛運用在淨水場以去除特定汙染物，如溶解性固體、天然有機物、無機離子以及一些毒性物質。但積垢為目前遭遇到最大且最難克服的問題。薄膜積垢產生後，不僅出水通量會衰減，連帶出水品質也會受到影響。本研究主要以金門太湖水場為例，以太湖原水和該場快濾池出水作為對象，研究奈米薄膜實際發生之積垢現象，並藉此找出相關之特定積垢物種，利用Hermia模式來判斷積垢機制。最後使用反應曲面法(responds surface method) 根據實驗結果求出本實驗之最佳操作條件。本研究使用之薄膜機組為平板掃流式，並在不同操作條件下 (如pH值、壓力及掃流速度) 觀察薄膜通量衰減情形和出水品質。另外利用各種不同鑑定方法分析原水水質基本物化特性，例如以三種不同離子交換樹脂 (陽離子、陰離子及非離子型)、分析水中天然有機物之特性及分子量分布情況、利用傅立葉紅外線轉換光譜儀鑑定其水中天然有機物之特定官能基等。 結果顯示，金門太湖原水及快濾池出水之DOC分別約為8.49 ± 0.22 和6.19 ± 0.22 mg L-1，且兩股水之親疏水性有機物質比例相當。其中又以疏水兼酸性之有機物 (30.2%) 為原水中主要有機物；親水兼酸性 (35%) 和疏水兼中性 (35.7%) 有機物為快濾池出水之主要有機物成分。原水之有機物分子量分部主要集中在1-5k (30.8%) 和低於1k (31.2%)，快濾池出水部分亦集中於1-5k (約49.4%)。 以原水而言，在pH 5 (有機物積垢為主) 及9.5時通量衰減較嚴重。以SEM-EDX及原水特性分析輔助鑑定可知在高鹼性環境下，膜上之積垢物主要以無機鹽類為主 (如硫酸鈣或硫酸鎂)。以快濾池出水而言，除了pH 9.5之外，在其他pH下其通量衰減情形相較於原水而言均輕微許多 (pH5為7.3%; pH6.5為14.6%; pH8為12.9%)。較值得注意的是，比對原水通量衰減隨時間之趨勢圖及NOM去除效率隨時間之趨勢圖，可明顯發現積垢現象有兩階段：0-7小時和7-48小時。在酸性條件下 (pH 5)，前7小時主要是不可逆積垢造成通量衰減，後7至48小時則為可逆積垢形式之通量衰減；反之，在鹼性條件下 (pH 8)，前7小時主要是可逆積垢形式之通量衰減，後7至48小時則為不可逆積垢形式之通量衰減。反觀快濾池出水，其通量衰減隨時間之趨勢圖及NOM去除效率隨時間之趨勢圖則無此特性。 藉由模式的預測可發現，不管是原水或是快濾池出水，標準阻塞 (Standard blocking) 機制無法解釋通量衰減情況，換言之，粒子(汙染物)並未進入奈米薄膜孔洞內且被吸附於孔內膜壁上。然而以原水而言，中間阻塞 (intermediate blocking) 機制為其主要積垢機制；濾餅阻塞 (gel layer formation) 機制為快濾池出水之主要積垢機制。 用實驗結果及反應曲面應用程式推算結果顯示，在本奈米薄膜處理程序其最佳操作條件為操作壓力在556.50 kPa，掃流速度在0.44 m s-1及溶液pH值在7.76時。在此最佳處理條件下，其預估之通量衰減僅為7.95%，而DOC/UV254去除效率可高達約98.29%。

Membrane or pressure-driven processes are used to remove contaminants such as dissolved solids, nature organic matters, inorganic ions, and some other hazardous compounds from water. One problem with this practice is membrane fouling, which causes not only permeate flux decline but also product quality deterioration. This research studied NF membrane fouling, identified associated foulants, assessed fouling mechanisms by the modified Hermia model, and finally developed optimal operation conditions using the respond surface method (RSM). Filtration was conducted with a cross-flow module using membrane (NF270) in plate form. Kim-Men Tai Lake water (natural) and effluent from the rapid sand filter (SF) (treated) of Kin-Men Water Treatment Plant were used at various pH levels, transmembrane pressures and cross-flow velocities. The physico-chemical properties of the raw water and the SF effluent were determined using instruments such as dissolved organic carbon (DOC) analyzer, gel filtration chromatography (GFC), Fourier transform infrared spectroscopy (FTIR) and so on. The results showed that the DOC concentration were 8.49 ± 0.22 and 6.19 ± 0.22 mg L-1 for the raw water and the SF effluent, respectively. The hydrophobic fraction (49.5 and 54.2% for the raw water and the SF effluent, respectively) was approximately the same as the hydrophilic fraction for both water samples. The HPOA fraction (30.2%) of the raw water was the highest, whereas the HPIA and the HPON fraction of the SF effluent (35 and 35.7%, respectively) were the predominant components. The NOMs of the raw water showed that 30.8% of its molecular weight was in the range of 5 to 10k and 31.2% was in the range of less than 1k Da. The major NOMs of the SF effluent had molecular weight 49.4% in the range of 1 to 5k Da. Results showed that sudden flux decline occurred in 7 hours generally and pH had significant influence over flux change. As expected, the flux decline increased with time. At pH 5, the fouling was mainly caused by organic materials; while at pH 9.5, for both water samples, inorganic scaling (e.g., calcium sulfate or magnesium sulfate formation) may be the main cause of flux decline as seen from results of SEM-EDX analysis of the fouled membrane. In the case of SF effluent, there was much less flux decline (7.3, 14.6 and 12.9% for pH 5, 6.5, and 8, respectively) than that for raw water (over 20% for all levels of pH), except at pH 9.5 (e.g., 28.43%). The NOM rejection and flux decline of raw water samples at all pH values could be divided into two phases in time, i.e., 0-7 and 7-48 hours. At pH 5, in 7 hours, irreversible fouling was the main cause of permeate flux decline; while reversible fouling controlled the permeate flux decline at 7 – 48 hours. At pH 8, in 7 hours, reversible fouling was the main cause of permeate flux decline; while irreversible fouling controlled the permeate flux decline at 7-48 hours. However, this tendency was not observed with the SF effluent (i.e., whose flux decline curves did not intersect at the same point). Standard blocking was not the fouling mechanism for both water samples at all pH, transmembrane pressure, and cross-flow velocity. This indicated that there was no adsorption of solute onto the inner walls of the membrane pores. However, for raw water, intermediate blocking may be the dominant fouling mechanism at all pH levels; whereas for SF effluent, gel layer formation may be the major fouling mechanism regardless of pH. Finally, identification of the best operation condition was attempted using RSM program. Transmembrane pressure of 556.50 kPa, cross-flow velocity of 0.44 m s-1, and pH at 7.76 was the optimal condition. Under this optimal condition, it could be predicted that flux decline would be 7.95% and DOC/UV254 removal would be 98.29% (Desirability 0.64).