Membrane fouling by Fe-Humic cake layers in nano-scale: Effect of in-situ formed Fe(III) coagulant

Abstract

The application of in-situ Fe(III) coagulant, formed by reacting Fe(II) and KMnO₄ (Fe/Mn), before membrane ultrafiltration (Fe/Mn-CUF) has been compared with conventional dosing of an Fe(III) coagulant (Fe-CUF). The performance of the alternative coagulants has been evaluated in terms of membrane fouling using model waters containing humic acid. A lower development of trans-membrane pressure (TMP) suggested that in-situ formed Fe(III) caused a lower rate of fouling compared to the dosed Fe(III). The lower TMP development with in-situ formed Fe(III), compared to dosed Fe(III), was associated with smaller, fewer and less compact coagulant flocs and differences in the membrane cake layer; in both cases influent flocs were aggregates of nanometer-scale primary particles.

Interestingly, TEM and SEM images showed that the size of nanometer-scale particles formed by in-situ Fe(III) was slightly larger than dosed Fe (III), and this contributed to a more porous cake layer. In addition, the Fe/Mn flocs formed a thinner cake layer than that formed by Fe flocs. Overall, the effect of decreasing the thickness and density of the cake layer and increasing the size of the nanometer-sized primary particles was significantly beneficial in reducing membrane fouling. Also NOM analysis by size exclusion chromatography (SEC) indicated that the in-situ Fe(III) coagulant achieved greater NOM removal than the dosed Fe(III), both in the UF influent and filtrate waters. In summary, the results have shown that in-situ Fe(III) is a superior pretreatment to conventional dosed Fe(III), in terms of significantly improving membrane performance (reduced fouling) with a little higher improvement in water quality.

Introduction

Associated with the dramatic increase in the application of low-pressure membrane filtration in drinking water treatment in the past decade [1], [2] is the continuing problem of membrane fouling, causing higher energy. A number of studies have reported that humic substances, as the main component of natural organic matter (NOM), play an important role in membrane fouling [3], [4], [5]. An effective method of removing humic substances is by the application of hydrolyzing coagulants at pH values near to neutral, where dissolved organic substances can be adsorbed on to the amorphous hydroxide precipitate (‘sweep flocculation’) and then physically separated by sedimentation and/or filtration [6]. Therefore, a combined process train comprising either microfiltration (MF) or ultrafiltration (UF) and pre-coagulation, especially in-line coagulation, is particularly advantageous since it is expected to combine satisfactory operational efficiencies by controlling membrane fouling and the necessary treatment performance [7], [8], [9], [10], [11], [12]. In this arrangement both the coagulation conditions and type of coagulant are expected to influence the membrane fouling and treatment performance.

Cho et al. [13] observed that coagulation with a rapid mix followed by a slow mix gave lower specific cake resistance than that with only a rapid mix for a submerged MF membrane. The formation of loose and porous flocs and the reduction of small colloidal particles at longer flocculation time have been shown to lead to a higher membrane flux [13], [14], [15], which indicates that the inclusion of a flocculation process after rapid mixing is a preferred arrangement.

The influence of the type of coagulant used has been widely reported, where the nature of the hydrolysis products and their interaction with contaminants determines the structural characteristics of the resulting aggregates and their porosity, which in turn affects the efficiency of filtration [16], [17], [18], [19], [20]. Thus, the permeate flux has been shown to depend on the coagulant used, as illustrated by one study which found some coagulants had no influence on permeate flux, another gave a 20% increase in flux, while another coagulant gave a 50% decrease in flux [8]. Wang et al. [21] showed that MF membrane flux deteriorated much more severely when pre-coagulated by aluminum chlorohydrate (ACH) and polyaluminum chloride (PACl), than by alum. For iron coagulants, Shon et al. [22] reported that adding FeCl₃ significantly improved membrane fouling in wastewater treatment, and another study considered the feasibility of applying polymeric ferric sulfate (PFS) to mitigate membrane fouling in a long-term running MBR [23]. A comparison of three conventional coagulants showed that the relative treatment performance of organic matter by coagulation-UF was in the order Al₂(SO₄)₃>Fe₂(SO₄)₃>FeCl₃ [12], and the comparative flux decrease was Fe₂(SO₄)₃<Al₂(SO₄)₃<FeCl₃.

In recent years, the application of combined oxidants and coagulants has been the subject of growing research interest in water treatment involving either separate chemicals, such as ozone and alum [24] or permanganate and alum/PAC [25], or dual role single chemicals, such as ferrate [26]. In this arrangement the impact of the oxidation on the contaminants and coagulants species significantly influences the nature of the resulting coagulant flocs produced. For the case of a combination of separate chemicals, the combining of an oxidant with a Fe(II) salt allows the Fe(III) coagulant species to be formed in-situ through competitive oxidation of the Fe(II) to Fe(III). This arrangement is advantageous since the cost of Fe(II) salts may be considerably lower than Fe(III) salts (as in the UK), and the action of the oxidant can achieve other water quality objectives (e.g. pre-disinfection and micro-pollutant degradation). Previous studies have indicated that in-situ formed Fe(III) was more effective than dosed Fe(III) [e.g., Fe₂(SO₄)₃] in removing contaminants such as arsenic [27] and phosphate [28]. Most recently, in-situ formed Fe(III), produced from the oxidation of Fe(II) by KMnO₄ and dissolved oxygen, was found to be a major reason for the high degree of treatment of the KMnO₄–Fe(II) process with respect to the removal of Microcystis aeruginosa [29].

To-date, there has been little reported in the literature concerning the effectiveness of in-situ formed Fe(III) as a pretreatment coagulant prior to ultrafiltration in drinking water treatment. Therefore, in this paper we present some new results concerning the effectiveness of coagulation before ultrafiltration with in-situ formed Fe(III), produced by combining FeSO₄ and KMnO₄ (Fe/Mn-CUF), in comparison with a typical ferric salt, Fe₂(SO₄)₃ (Fe-CUF). The impact and mechanisms involved of both pretreatment methods on membrane fouling are described with reference to the floc properties and the nature of the cake layer formed on the membrane surface. The results obtained in this study are expected to provide a better understanding of the effectiveness of in-situ formed Fe(III) for controlling membrane fouling and improving organic matter removal.