Analysis of fouling mechanisms in TiO₂ embedded high density polyethylene membranes for collagen separation

Highlights

• HDPE/TiO₂ membranes were successfully fabricated via TIPS method.

• Incorporation of TiO₂ nanoparticles increased surface roughness of the membranes.

• Fouling properties of membrane were analyzed using classic and combined models.

• Membrane containing 0.50–1 wt.% TiO₂ showed reversible fouling properties.

• Neat HDPE membrane showed irreversible fouling property.

Abstract

In this study, the fouling behavior of TiO₂–polyethylene hybrid membranes was analyzed. Initially, high density polyethylene (HDPE) membranes embedded with TiO₂ nanoparticles were fabricated via thermally induced phase separation (TIPS) method. FESEM images showed that the membranes had leafy structure indicating solid–liquid phase separation mechanism. The results of XRD analysis confirmed the presence of TiO₂ nanoparticles in the polymer matrix. AFM images showed that the surface roughness of TiO₂ embedded membranes were higher than that of neat HDPE membrane. Pure water flux of membranes improved as the TiO₂ content increased. The fouling behavior of membranes was investigated by filtration of collagen protein solution. The governing fouling mechanisms of membranes were also investigated using classic models as well as combined fouling models. The results showed that the best model fitted into the experimental data for 0.50, 0.75 and 1.0 wt.% TiO₂ embedded HDPE membranes was the cake filtration model. For neat and 0.25 wt.% TiO₂ embedded membranes, however, cake filtration-complete blockage model was in good agreement with the experimental data. Moreover, addition of TiO₂ nanoparticles increased reversible portion of fouling.

Introduction

Collagen proteins have recently attracted attention because of their unique characteristics (Duan et al., 2009, Shen et al., 2009). As a scaffold, collagen is one of the most important biological components which are widely used in biotechnological applications such as tissue engineering (Xiong et al., 2009). The major sources of collagen used currently for biomedical and tissue engineering applications are porcine, bovine skin and equine Achilles tendons (Alizadeh et al., 2013). Like other proteins such as BSA and whey, membrane separation processes can be applied to purify and concentrate collagen extraction. However, there are few reports about collagen proteins purification using membranes (Cao and Xu, 2008, Shen et al., 2009).

Despite the advantages of membrane processes, the application of these processes is limited due to the fouling phenomenon in membranes. Fouling is a complex physicochemical phenomenon which usually involves the deposition and adsorption of foulants on the surface and inside the pores of the membrane. This phenomenon leads to a dramatic decrease in flow through the membrane (Balta et al., 2012, Hilal et al., 2005) and consequently, operating pressure should be increased to maintain the permeate flux constant, which, in turn, results in higher energy demand. In addition, fouling may cause a decrease in the rejection of the target components of the feed (Hoek et al., 2011, Van der Bruggen et al., 2002, Van der Bruggen et al., 2008). Therefore, from an economic point of view, elucidation of membrane fouling is essential. To do so, an insight into fouling mechanisms is required.

The governing mechanisms of fouling are usually explained by fouling models. Classically, there are four different models used to describe membrane fouling, including standard blockage, intermediate blockage, complete blockage and cake filtration (Charfi et al., 2012). In the standard blockage or pore constriction model, foulants accumulate on the walls of inner pores and constrict their radius. In the intermediate blockage model, a portion of the particles seal off some pores and the rest of them settle on each other. Foulants seal off pore entrances completely in the complete blockage model. Finally, in the cake formation model, a layer of particles and foulants is formed on the surface of membrane (Bolton et al., 2006, Rezaei et al., 2011). These models have been used individually or in combinations to investigate the fouling behavior of membranes in different filtration processes. The classical models are the simplest models that identify the fouling mechanisms, but it should be noted that these models predict just a simple mechanism for the whole membrane filtration process. This is due to the complexity of the fouling phenomenon. In addition to the classical models, some combined models have been developed to predict flux behavior in membrane filtration processes. Ho and Zydney (2000) developed a combined model consisted of a two-stage mechanism. This mathematically complex model accounts for initial fouling due to the pore blockage and subsequent fouling due to the cake formation. Bolton et al. (2006) developed some combined models describing fouling mechanisms with two fitted parameters. These models are mathematically simpler than the model developed by Ho and Zydney, therefore have attracted most attentions, recently.

In protein separation, fouling is likely because of the progressive deposition of proteins on the membrane surface or within the internal pores of membrane which leads to pore constriction, pore blockage and formation of a layer of cake or gel (Tung et al., 2008). There is evidence showing that proteins can foul membranes with pore sizes much greater than the size of protein monomers or dimers. This may be due to the production of nucleation sites by some of the fouling species (Velasco et al., 2003) or the presence of protein agglomerates (Mourouzidis-Mourouzis and Karabelas, 2006).

Many efforts have been devoted to reduce fouling in polymer membranes via surface modification using different methods such as grafting, plasma technique for surface treatment and dip-coating (Azari and Zou, 2012, Azari and Zou, 2013, Rahimpour et al., 2008, Ulbricht and Belfort, 1996). Incorporation of inorganic nanoparticles into polymer matrix of membranes is another method to improve membrane antifouling properties and has recently attracted attention due to physicochemical properties of these nanoparticles (Balta et al., 2012, Liang et al., 2012, Nguyen et al., 2014). In this method, nanoparticles with appropriate properties are dispersed through the membrane bulk and therefore, internal pores of membrane could be also concerned. Among different nanoparticles which have been widely used in polymer–inorganic membranes, TiO₂ is perhaps the most convenient one because of its desirable properties such as photocatalytic activities, antifouling abilities, stability and availability. (Bae and Tak, 2005, Kim et al., 2003, Razmjou et al., 2011, Shi et al., 2012). Several studies have investigated incorporation of TiO₂ nanoparticles into different polymers including PSf (Hamid et al., 2011, Yang et al., 2006, Yang et al., 2007), PVDF (Damodar et al., 2009, Razmjou et al., 2011, Shi et al., 2012) and PES (Luo et al., 2005, Rahimpour et al., 2008, Razmjou et al., 2011, Vatanpour et al., 2012, Wu et al., 2008, Yang et al., 2007). However, to the best of our knowledge, TiO₂/polyethylene membranes have not been yet studied. Polyethylene (PE) membranes are widely used in microfiltration processes due to its proper mechanical property, good chemical resistance and low cost in comparison with other commercial polymers. PE membranes are usually fabricated via thermally induced phase separation (TIPS) method (Lloyd et al., 1990).

The scope of the present study is to fabricate TiO₂/PE membranes via TIPS method for the separation of collagen protein solution and to analyze the fouling mechanisms during filtration of collagen protein solution. Different dosages of TiO₂ nanoparticles were added to the PE/mineral oil homogeneous solution. A set of analyses including field emission scanning electron microscopy (FESEM), X-ray diffraction (XRD), contact angle measurement, atomic force microscopy (AFM), pure water flux and mean pore radius was carried out for the characterization of membranes. The fouling mechanism of membranes was analyzed using classic and combined models to investigate the effect of nanoparticles on the fouling of polyethylene membrane.

For a constant pressure filtration system, the flux decline can be expressed in the following mathematical form (Charfi et al., 2012):

$$\frac{d^{2}t}{d{V}^2}=k{\left(\frac{dt}{dV}\right)}^m$$

where t is filtration time, V is the filtrate volume, k is resistance coefficient and m is a constant which characterizes the fouling model. The values of m for cake filtration, intermediate blockage, standard blockage and complete blockage models are 0, 1, 1.5 and 2, respectively (Bolton et al., 2006, Charfi et al., 2012). Having the flux equation on the hand:

$$J=\frac{1}{A}\frac{dV}{dt}$$

Eq. (1) can be written as:

$$\frac{dJ}{dt}=-kJ{(AJ)}^{2-m}$$

The analytical solutions to Eq. (3) for each m value (0, 1, 1.5 and 2) as well as the linear forms of flux expressions are listed in Table 1. These are classic models for membrane fouling. Hermia showed that a plot of the filtrate flux data in an appropriate linearized form could be a useful approach to examine the fouling mechanism. In this traditional approach, the expressions for fouling models (cake filtration, pore blockage, and pore constriction) are rearranged in a way that predicts simply the fouling model. A summary of the approach is given in Table 2.

As mentioned earlier, there are some combined models which account simultaneously for two fouling mechanisms. These models were developed first by Bolton et al. (2006) based on Darcy's law and some of them are listed in Table 3.