Development of surface activated granular media for effective adsorption and filtration

... being ineffective to the removal of suspended particles and dissolved substances It is therefore desirable to develop more effective granular media as both adsorption and filtration materials for. .. experimental parts of coating PPy on the surfaces of glass beads and nylon 6,6 granules and immobilizing chitosan on the surfaces of PET and nylon 6,6 granules, as well as the adsorption and filtration. .. stages) and deposition of colloid particles 2.3 Adsorption and Granular Filtration Separation processes like adsorption and filtration refer to the operations that transform a mixture of substances

Chapter 1 Chapter 1 Introduction and Research Objectives Water or wastewater is a complex system that often contains various kinds of dissolved substances (organic or inorganic) and suspended/colloidal particles. Organic matters found in water or wastewater may include such diverse species as humic substances, carbohydrates, lignin, fats, soaps, synthetic detergents, proteins and their decomposition products, as well as various synthetic organic chemicals from the process industries. Examples of suspended matters in water and wastewater can include various inorganic and organic particles, such as soil or clay particles, immiscible liquids, metal hydroxides or microorganisms, etc. Many of the dissolved or suspended matters may cause aesthetical problems and/or be toxic or hazardous to our human beings and other aquatic lives. Therefore, appropriate treatment of water or wastewater is necessary in order to make water suitable for dinking or wastewater suitable for discharging or reuse. The unit processes used in water and wastewater treatment to remove dissolved or suspended matters include coagulation/flocculation, sedimentation, granular media filtration, adsorption, and biological treatment, etc. Among these processes, granular media adsorption and filtration have widely been used. While the two processes are both concerned with separating certain species present in a fluid stream, the sizes of the species to be separated by the two processes are often different. The molecular sizes of the dissolved species removed in adsorption are usually in the order of angstroms or 1 Chapter 1 nanometers, whereas the sizes of particles to be separated in granular filtration are in the range of submicrons or microns. Nevertheless, the operation of fixed-bed adsorption is quite similar to that of granular filtration. In both operations, granular media are placed in a bed and the fluid to be treated is allowed to flow through the granular media. It is also believed that the same types of interaction forces (such as the London-van der Waals and electrostatic double-layer forces) are active in both adsorption and filtration (Tien, 1989). Thus, many similarities exist between adsorption and granular filtration processes, in terms of equipment configuration, mode of operation, and the respective underlying phenomena. Because of these similarities, the words adsorption and filtration have sometimes become interchangeable (Tien, 1989; 1994). The removal of colloidal particles from a fluid phase to a solid phase may be described as either adsorption or filtration (Hirtzel and Rajagopalan, 1985). In engineering practice, granular carbon columns used to remove organic compounds in drinking water supplies are often referred to as carbon filter. For economic reasons, sand and anthracite have been widely used as granular filter media in almost all water and wastewater filtration in municipalities and various industries. To remove dissolved hazardous species, granular activated carbon is often applied as the adsorbent in an adsorption process (Clark et al., 1989). Sand and anthracite themselves are, in fact, not particularly efficient to remove suspended particles, especially those fine submicron particles, such as colloids, bacteria (pathogens) and viruses, etc., nor are they effective to remove any dissolved organic and inorganic substances. However, because of 2 Chapter 1 the difficulty and high cost in regeneration, it is impractical to use activated carbon for the purpose of suspended particle removal. Moreover, activated carbon is inefficient in an adsorption process to remove those dissolved organic matters with larger molecular mass and wide molecular mass distribution, e.g. natural organic matters (NOM). Nowadays, water and wastewater treatment systems often have to apply sand filtration followed by activated carbon adsorption in order to achieve the desired level of purification. This approach not only complicates the treatment process but also results in significant cost increase. The surface properties of granular media play an important role in the removal of dissolved substances and suspended particles. The common mechanisms of particle removal in granular filtration include interception, sedimentation and adsorption. Interception and sedimentation are significant for large particles. For most of the fine particles or colloids in water and wastewater, adsorption is the key mechanism for their removal, which relies on the surface interactions between the media grains and the particles to be removed. A classic theory for analysis and prediction of colloid adsorption and deposition is the Derjaguin-Landau-Verwey-Overbeek (DLVO) model, which was the postulate of Landon-ver der Waals and electrostatic double-layer interactions. In general, the Landon-ver der Waals force is always attractive, but the double-layer force can be either attractive or repulsive. Extensive theoretical calculations and experimental investigations have demonstrated that colloid adsorption/deposition efficiency can be enhanced significantly under attractive double-layer interactions. 3 Chapter 1 In most water and wastewater treatment processes, the dissolved substances and suspended colloids and particles to be removed have negative surface charges. The conventional filter media, such as sand, are usually negatively charged under normal water and wastewater treatment conditions (Redman et al., 1997). Due to the force of electrostatic repulsion, the removal of most of the fine particles, such as colloidal particles, bacteria and viruses, has been difficult. Similarly, activated carbon also carries negative surface charges in this pH range (Wu et al., 2001), thus decreasing the adsorption efficiency greatly. Another important factor for effective adsorption or filtration is the surface morphology of the granular media. Rough surfaces with micropores provide large specific surface area for particle deposition. The heterogeneity of the surface morphology also possibly changes surface charge distributions, with certain surface patches favorable for particle deposition even though the overall interactions between the media grains and the particles to be removed may be unfavorable (Tien, et al., 1979; Choo and Tien, 1995). The rough surface will also change the hydrodynamics of colloid adsorption/deposition. The smooth surface of the conventional filter media (e.g. sand) is another reason for their being ineffective to the removal of suspended particles and dissolved substances. It is therefore desirable to develop more effective granular media as both adsorption and filtration materials for high efficient water and wastewater treatment and for simplification of the treatment system. In the present study, the objectives are to develop granular media with positive surface charges, which would provide attractively electrostatic surface interactions between the granular media and the dissolved organic substances or suspended particles to be removed in water or wastewater, thus enhance 4 Chapter 1 their removal in an adsorption or filtration process. Granular media with positive surface charges are obtained through modifying the surface of inorganic and organic granules with polypyrrole (PPy) or chitosan. The modified granules are then used to study the performance and mechanisms in removing organic pollutant (use humic acid as a model organic compound) and inorganic colloid (use clay particle as a model colloid). Attempts are also made to understand the role of surface interactions in the removal of dissolved and suspended substances in a granular media adsorption or filtration system. This thesis is organized as follows. Chapter 2 presents background information on granular adsorption or filtration in water and wastewater treatment, and the typical pollutants in water system, and an overview of PPy and chitosan that were used in the present study. The experimental parts of coating PPy on the surfaces of glass beads and nylon 6,6 granules and immobilizing chitosan on the surfaces of PET and nylon 6,6 granules, as well as the adsorption and filtration experiments, are discussed in Chapter 3 – Chapter 6. Chapter 7 discusses the qualitative and semiquantitative analyses of electrochemical properties of PPy-water interface and presents a site-binding model for humic acid adsorption onto PPy surface. 5 Chapter 2 Chapter 2 Literature Review 2.1 Surface Charges and Electrical Double Layer Most solid surfaces will be charged when they are brought into contact with aqueous solutions. The possible charging mechanisms include (1) ionization of the surface groups, (2) specific adsorption of ions or surfactants from the solutions, and (3) lattice imperfections at the solid surface or isomorphic substitution in the crystal lattice (Stumm, 1992; Koopal, 1993; Myers, 1999). Although the presence or absence of surface charge may often be neglected in the macroscopic systems, in the microscopic systems of colloids and interfaces, the surface charge is a critical factor and plays an important role in many applications, such as coagulation, flocculation, adsorption and filtration. The overall arrangement of the electric charge on the solid surface, together with the balancing charge in the bulk solution, is often referred to as the electrical double layers (EDL), or just double layers (Bockris and Reddy, 1970). The EDL can be regarded as consisting of two regions: an inner region which may include adsorbed ions (i.e. the adsorption layer), and a diffuse region in which ions are distributed according to the influence of electrical forces and random thermal motion (i.e. the diffusion layer). The Gouy-Chapman theory (Figure 2.1a) is the one-dimensional analysis of the diffusive double layer based on Poisson’s equation and Boltzmann equation, and gives (Hunter, 1991) 6 Chapter 2 dψ 2κkT zeψ sinh =− dx ze 2kT (2.1) where ψ is the electrical double layer potential at a distance x from the surface, k is the Boltzman constant, T is the absolute temperature, z is the valency of the counterions, e is the coulombic charge, and κ is the reciprocal of the electrical double layer thickness which is defined as κ= 1000 N A e 2 ∑ ci z i2 εkT (2.2) where NA is the Avogadro’s number, ci is the molar concentration of the counterions of type i, zi is the valency of counterion i, ε is the solution permittivity (equal to ε0εr, ε0 being the permitivity of vacuum and εr the relative permittivity of the liquid). The total charge σd, per unit area of surface, in the diffuse layer can be calculated from Eq. (2.1) as (Hunter, 1991) σd = − zeψ d 2κkTε sinh 2kT ze (2.3) where ψd is the electrical potential at the onset of the diffuse layer. To improve the Gouy-Chapman model, Stern proposed to divide the double layer into two parts separated by a plane (the Stern plane) located at about a hydrated ion radius from the surface (Figure 2.1b). Specific ion adsorption may take place at the Stern plane when electrostatic and/or van der Waals forces are strongly enough to overcome the thermal agitation. Across the Stern plane, the potential drops from the surface potential ψs to the 7 Chapter 2 potential at the Stern plane, ψd, and further decays from ψd to zero in the diffusion layer. The potential drop (ψs- ψd) is related to the capacity of Stern layer, C1, as (Koopal, 1993) ψ s −ψ d = σs (2.4) C1 and C1 is defined as C1 = ε 0ε r δ (2.5) where δ is the distance from the surface to the Stern plane. Stern plane Surface of shear Diffuse layer Stern layer ψ0 Potential ( ψ) Potential (ψ ) ψ0 ψd 1/κ Distance (x) (a) according to Gouy-Chapman model ζ δ 1/κ Distance (x) (b) according to Stern model Figure 2.1 Schematic representation of the structure of the electrical double layers (EDL). 8 Chapter 2 ψd can be estimated from electrokinetic measurements. Electrokinetic behavior depends on the potential at the surface of shear between the charged surface and the electrolyte solution. This potential is called the electrokinetic or ζ (zeta) potential. The concept of the surface of shear implies some idealization and the exact location of the shear plane is an unknown feature of the electrical double layer (Shaw, 1970; Kohler, 1993; Kosmulski, 1995). Usually the shear plane is supposed be located at a small distance further out from the stern plane. In general, ζ potential may not be too different from ψd (usually a little smaller in magnitude than ψd), but ψd can often be considerably smaller than ψs. It is customary to assume identity of ψd and ζ potential for low values of the ionic strength (Koopal, 1993; Kosmulski, 1995). 2.2 Surface Interactions 2.2.1 Short-Range Forces and Long-Range Forces Surface interaction forces can generally be subdivided into two types: short-range forces and long-range forces. The action distance of short-range forces is usually no more than 0.1-0.2 nm (Garbassi et al., 1998). The typical short-range force is covalent force. When two atoms bind to form a non-ionized molecule, the force involved in bond formation is referred to as covalent force, and the resulting bond is covalent bond. Covalent bond has certain characteristic bond length and bond angle which depends on the atoms involved. Hence covalent force is directional. Other short-range forces include hydrogen-bonding interactions and Lewis acid-base interactions. Hydrogen bond can be formed between a proton covalently bonded to a 9 Chapter 2 highly electronegative atom (e.g. O, N, F) and another electronegative atom nearby. The hydrogen bond energy depends, in a rather complex way, on the distance between the participating atoms and the angle between the atoms. Fowkes (1985) defined the interactions between electron acceptors (Lewis acids) and electron donors (Lewis bases) as Lewis acid-base interactions. The interaction strength of a basic or acidic site depends not only on the ability to donate or accept electrons, but also on the polarizability. Van Oss (1994) classified hydrogen-bonding interactions and Lewis acid-base interactions as non-covalent interactions, whilst other researchers (Drago et al., 1977) suggested that these interactions have, at least partly, a covalent character. As far as surface and colloidal phenomena are concerned, one often only considers the long-range forces or physical interactions which act between discrete, non-bonded atoms or molecules over distances significantly greater than molecular bond dimensions and are generally nondirectional. The two fundamental long-range forces include two kinds: coulombic or electrostatic interactions, and van der Waals forces. For two point charges, Q1 and Q2, the free energy of electrostatic interaction, wel, may be given by (Myers, 1999) wel = Q1Q2 z z e2 = 1 2 4πε 0 εr 4πε 0 εr (2.6) where r is the distance between the two charge, and z is the valency of each ion. For two charges of the same sign, wel is positive, which means that the interaction is repulsive. In contrast, wel is attractive for two unlike charges. 10 Chapter 2 Forces between molecules caused by permanent and induced dipoles and other multipoles are collectively known as van der Waals forces. The fluctuating dipole-induced dipole interactions, described by London, is called London-van der Waals (dispersion) interactions, which often makes the most important contribution to the total van der Waals interactions, especially in the aqueous media containing electrolytes (Chaudhury, 1984). Hamaker (1937) calculated the London-van der Waals force between individual atoms of spherical particles. His treatment was rather coarse, but the concept of a Hamaker constant has still been used. A more accurate calculation was developed by Lifshitz and co-workers (Dzyaloshinskii et al., 1961), who described the van der Waals force as originating from spontaneous electromagnetic fluctuations at interfaces. The London-van der Waals forces are characterized as being universal and almost always attractive over relatively long distances. The simplest situation in analyzing London-van der Waals force is of two hard, flat, and infinite surfaces separated by a distance, d, in a vacuum. The free energy of attraction per unit area, wvdw, in such a case is given by (Myers, 1999) wvdw = − A 12πd 2 (2.7) where A is the Hamaker constant, which depends on the dielectric properties of the two interacting plates and the intervening medium and typically amounts to about 10-20 J. Although short-range forces are regarded as inferior factors to long-range physical forces in the systems of colloids and surfaces, short-range forces may have indirect effects on longer-range interactions. For example, solvation layer is created through the short-range 11 Chapter 2 interactions such as hydrogen bonds or chemical bonds between the surface and the solvent molecules, and the solvation layer influences the long-range forces between two approaching surfaces greatly. Moreover, adsorbed macromolecule layer strongly affects the interaction between surfaces (Fleer et al., 1993a; Claesson, 1998). 2.2.2 DLVO Theory The combined action of van der Waals forces and electrostatic or electrical double layer forces in aqueous systems is described by the DLVO theory, which was developed by two groups of researchers, Derjaguin and Landau, and Venwey and Overbeek, independently in the late 1940s (Shaw, 1970). For two symmetrically charged plates in an electrolyte solution, the total free energy of interactions in the DLVO theory is expressed by the sum of the van der Waals and electrostatic contributions (Kohler, 1993; Garbassi et al., 1998): w = wel + wvdw = 64c0 kT κ [tanh( eψ s 2 −κd A )] e − 4kT 12πd 2 (2.8) where c0 is the bulk concentration of the electrolyte solution, and ψs is the surface potential. In Eq. (2.8), the term of van der Waals forces tends to infinity as separation distance, d, approaches to zero, while the electrostatic term remains finite. Therefore, the van der Waals attractive interactions always prevail at small distances. On the other hand, the van der Waals’ term is insensitive to the change of ionic concentration or solution pH, which greatly affects the electrostatic term. For a given colloid suspension system, van der Waals attraction outweighs electrostatic repulsion at small separations, while at intermediate separations electrostatic repulsion predominates and the high energy barrier 12 Chapter 2 prevents the colloid surfaces from contacting each other. As the ionic concentration increases, the thickness of the double layers, κ-1, decreases (see Eq. (2.2)), and thus the repulsion between the colloid surfaces are reduced (decrease of e-κd outweighs the increase of c0/κ). Hence, increasing ionic concentration favors flocculation and coagulation of colloids. DLVO theory represents the pillar of colloid science. Ninham (1999) asserted that DLVO stands at the same level of importance for colloid science as does Darwin’s theory of the origin of species in biology. However, DLVO theory failed at short distance from the surface (Sposito, 1984). Non-DLVO forces, or so-called “extra-DLVO” forces (Pashley, 1981; Israelachvili, 1992), such as solvation, hydration, hydrophobic, oscillator, capillary and water structure forces may become operative at separations below 5 nm. Moreover, the agreement between DLVO theory and experimental measurement is less satisfactory for biological regime or for any system where the ionic concentration is in the order of 0.1 M or higher. Boström et al. (2001a, 2001b) pointed out that this might be, in part, due to the reason that the dispersion forces of specific ion effects are ignored. Despite these limitations, the classical DLVO theory has appeared to work reasonably well at separations of intermediate distance and for low ionic concentrations ( 1 is called "unfavorable". Although the Freundlich equation was developed empirically, it can also be derived theoretically using a model in which it is assumed that the heat of adsorption is not constant but varies exponentially with the extent of surface coverage (Halsey and Taylor, 1947; Halsey, 1948), which is more reasonable than the Langmuir assumption in most cases. The Freundlich equation applies very well for solids with heterogeneous surface properties. However, this equation cannot be applied to all values of C. As C increases, q increases (according to Eq. (2.12)), only until the adsorbents reach saturation. At saturation, q is a constant and is independent of further increase in C, thus the Freundlich equation no longer applies. In addition, there is no assurance that adsorption data will be in a good agreement with the Freundlich equation over all concentrations less than saturation. 2.3.2 Granular Filtration 2.3.2.1 Definition and Applications Granular filtration is a fluid-solid separation process commonly applied to remove minute quantities of small particles (liquid or gas suspensions) from various kinds of fluid by passing through granular media (Aitken, 1969; Tien, 1989). As the fluid or suspension is forced through the voids or pores of the granular media, the solid particles are retained on the medium’s surface or, in some cases, on the walls of the pores, while the fluid (i.e. filtrate) passes through (Cheremisinoff, 1998). 18 Chapter 2 Both Sanskrit medical lore and Egyptian inscriptions give clear evidence that granular filtration was used for water treatment as early as 200 B.C. The versatility of granular filtration is evident from its scope of application as well as from the manner in which it is carried out. Besides water or air, systems that may be treated by granular filtration include such diverse substances as flue gas, molten metal, petrochemical feedstocks, polymers, and alcoholic or nonalcoholic beverages. Although granular filtration is frequently carried out in the fixed-bed mode, it may also be conducted in a moving-bed or fluidized-bed mode so that the operation is continuous. The most commonly known application of granular filtration is for water and wastewater treatment to remove solids, including bacteria present in surface waters, suspended clay particles, precipitated hardness from lime-softened waters, and precipitated iron and manganese (Uchrin, 1983). It has been reported that various forms of sand filtration have been used to purify water for centuries (Montiel et al., 1988). Slow sand filtration was first developed to purify surface water for drinking purposes. Since the mid-19th century, slow sand filtration has been widely employed in treating community water supplies in many countries against water-borne disease. A significant improvement to water treatment in the 1880s and 1890s was the development of rapid sand filters, which could handle considerably larger volumes of water. At present, granular filtration of water and wastewater is inevitably applied in conjunction with sedimentation and/or coagulation. Granular filtration is widely used to remove 19 Chapter 2 residual biological floc in settled effluents from secondary treatment by trickling filters or activated sludge process, to remove residual chemical-biological floc after alum, iron, or lime precipitation of phosphates in secondary settling tanks of biological treatment processes, and to remove solids remaining after the chemical coagulation of wastewaters in tertiary or independent physical-chemical waste treatment (Cheremisinoff, 1995). 2.3.2.2 Filtration Theory Although granular filtration is one of the most widely used processes in a variety of applications, the design and operation of filters is still carried out on an almost entirely empirical basis. The reason for this is that a filter itself may be a relatively simple device, but the process of filtration is quite dynamic and extremely complex. In general, filtration process involves two sequential steps, i.e. transportation and attachment (O’Melia, 1985). Particles in the suspension to be filtered are first transported from the bulk of the fluid to the vicinity of the stationary surface of the filter media by physical forces. Then attachment of the particles to the collectors (filter media grains) occurs through various physical and chemical interactions (Amirtharajah and Westein, 1980). The major surface interaction forces between the collector grains and the suspended or colloidal particles include the London-van der Waals force and the double layer force. The former is attractive and remains essentially constant during particle deposition. The double layer force can be either attractive or repulsive and may change during particle deposition (Bai and Tien, 1997; 2000a). While much progress has been made in the area of filtration modeling, no generally applicable models for the process have been developed yet. 20 Chapter 2 Until about 1970, most of the modeling work in filtration had followed what is known as the phenomenological approach. In its simplest form, the process is described by two equations (Iwasaki, 1937) ∂c = −λc ∂Z (2.14) ∂σ ∂c +u =0 ∂t ∂Z (2.15) where c is the concentration of particles, Z the depth of filter bed, λ the filter coefficient, σ the (absolute) specific deposit and u the approach velocity. Eq. (2.14) basically describes the rate of removal and Eq. (2.15) is a mass balance equation. The solution of Eqs. (2.14) and (2.15) will depend on the filter coefficient which is usually taken to be a function of σ. Various attempts have been made to find a correlation between the initial filter coefficient λ0 and the various system variables such as media grain size, approach velocity, etc., and to determine a relationship between λ and σ. From about 1970 onwards, the trajectory analysis approach has been used extensively for the study of λ0 and received much success (Yao et al., 1971; Tien et al., 1979). Predicting the development of removal (λ against σ) is however a much harder task. O’Melia and Ali (1978) recognized that deposited particles can act as additional collectors, and thus enhance particle removal in a filter (increase in removal efficiency or so-called ripening). Chang and Tien (1985) developed a dendrite model to quantify the increase in removal efficiency due to the presence of deposited particles. As deposition progresses, the process becomes increasingly complex. Some pores in the filter may be blocked off, and therefore become unavailable for deposition (Tien et al., 1979). The increase of interstitial velocity in the filter pores can 21 Chapter 2 also result in detachment of particles that deposited previously (Bai and Tien, 1999). In addition, Choo and Tien (1995) have considered that the deposit may be porous for the flow, whereas most others have assumed that the deposit layer is impermeable. More recently, Bai and Tien (2000b) have proposed the rate expressions that distinguish different types of deposition in filtration. They assume that the media grain surface may be divided into two parts: one covered with deposited particles, and the other without particles. The nature of surface interactions between the particles and either of those two parts of grain surfaces are therefore characterized as particle-particle and collector-particle types. The overall filter coefficient is given as λ = (λ ) c − p (1 − f ) + (λ ) p − p f (2.16) where (λ)p-p is the filter coefficient between grains covered with deposited particles and particles to be collected, (λ)c-p the filter coefficient between clean grains and particles to be collected, and f is the fraction of the grain surface covered with deposited particles. In many cases, (λ)p-p and (λ)c-p can be expected not to vary with time since the condition determining the surface interactions remains the same. Bai and Tien (2000b) also divide the deposition process into two types: deposition on grains and deposition on previously deposited particles. The first type is named as the monolayer deposition while the latter as multilayer deposition. They demonstrated that Eq. (2.16), although simple, together with Eq. (2.14) and (2.15), is quite versatile to describe the various filtration behaviors. 2.4 Granular Media for Adsorption and Filtration 2.4.1 Conventional Granular Adsorbents and Filter Media 22 Chapter 2 Adsorbents and filter media represent the hearts of adsorption and filtration devices. All practical adsorbents have large specific surface areas and are therefore highly porous or composed of fine particles. The old types of industrial adsorbents (e.g. activated carbons and silica gels) are generally non-crystalline and their surface and pore structure therefore tend to be ill-defined and difficult to characterize. Many new adsorbents possessing intracrystalline pore structures have been developed over the past 20 years, including carbon molecular sieves, new zeolites and aluminophosphates, pillared clays and model mesoporous solids (Rouquerol et al., 1999). In addition, various modern spectroscopic and microscopic techniques can now be employed for studying the surface state and the microstructure of the adsorbents (Dobiás et al., 1999). Although several materials may be applied in the adsorption process for water and wastewater treatment, including alumina, silica gel, Fuller’s earth and diatomaceous earth, granular activated carbon (GAC) has been by far the most widely used adsorbent which provides tertiary treatment for water contaminated with organic matters. By the structure of materials, filter media may be classified into two types: flexible and rigid media (Cheremisinoff, 1995). Rigid filter media are commonly used for granular filtration. Ceramic filter media for example are widely used in gas filtration, and in separation of dust and liquid droplets from gases (Loff, 1981). The relatively uniform particle size of diatomaceous media achieves high efficiency of filtration in retaining solid particles of sizes less than 1 µm, as well as certain types of bacteria. Plastic granules (e.g. polyvinyl chloride and nylon) have gained growing attention as filter media in recent years because of their relatively low cost (Driscoll, 1977; Loff, 1981). In water and 23 Chapter 2 wastewater treatment, the most commonly used filter media include silica sand (specific gravity =2.65), garnet sand (specific gravity = 4.0-4.4) and anthracite coal (specific gravity = 1.35-1.75). The effectiveness of granular adsorption and filtration processes largely relies on the surface interactions between the substances to be removed and the granular media. As discussed early, DLVO forces (London-van der Waals force and electrostatic double layer force) play the most important role in colloid adsorption and deposition processes. In general, the London-van der Waals force is always attractive, but the double layer force can be either attractive or repulsive. Hence, the surface charges of impurities or pollutants in water systems and those on the granular media have an impact on the efficiency of water and wastewater treatment. In face, most impurities or pollutants in waters carry negative surface charges in the pH range of natural waters (Ives, 1990; Chen et al., 1998). On the other hand, the activated carbon, sand and anthracite that are used to remove these impurities or pollutants also carry negative surface charges in that pH range. Therefore, the conventional filter media or adsorbents are not particularly efficient to remove suspended particles, nor are they effective to remove dissolved organic and inorganic substances. This problem may be solved by surface modification of the traditional granular media to obtain the desired surface properties, i.e., positive surface charges. 2.4.2 Surface Modification of Granular Media Several studies have examined the possibility of modifying granular media to improve their ability to remove dissolved matters and suspended particles. These modifications 24 Chapter 2 included impregnation of coal with metallic hydroxides (Chaudhuri and Sattar, 1986; Lukasik et al., 1999), addition of positive charges to silica using organosilane derivative (Zerda et al., 1985), incorporation of metallic hydroxides onto the surfaces of sand using in situ precipitation of metallic hydroxides (Farrah and Preston, 1985; Lukasik et al., 1996), adsorption of metallic flocs onto the surfaces of sand (Edwards and Benjamin, 1989), and modification of diatomaceous earth by precipitation of metallic peroxides (Farrah et al, 1988). Among these developments, coating sand with ferric and aluminum oxide or hydroxide has received much attention. Because the surface of granules became electrically positive, it facilitated the attachment of the negatively charged particles or other pollutants found in water and wastewater (Farrah and Preston, 1985). The coated sand was also found to effectively adsorb microorganisms (Lukasik et al., 1996; Mansoor and Chaudhuri, 1996). The experimental and theoretical studies by Truesdail et al. (1998) demonstrated the benefit of granule surfaces coated with metal oxides, or hydroxide rich oxide/hydroxide mixtures in increasing the efficiencies of commercial filtration systems. The electropositive surface coatings from aluminum or mixed (hydr)oxides, had similar average kinetic rate constants and were five times greater than the rate constants for the uncoated sand. In particular, the adsorption of natural organic matters (NOM) onto iron oxide coated sand has been actively studied recently. A new granular adsorbent based on β-FeOOH was developed by Teermann and Jekel (1999), and was shown to have high adsorption capacities for both smaller and higher molecular weight humic substances. It was reported that coating iron oxide particles with cationic polymer significantly 25 Chapter 2 increased adsorption of humic acid at high pH, but had less effect at low pH (Kim and Walker, 2001). Lai et al. (2001; 2002) found that copper ions, lead ions and humic acid could be removed efficiently from water by the iron oxide coated sand, and the adsorption of cadmium ions and humic acid onto goethite-coated sand was highly pH-dependent. Although the coatings of metallic hydroxides, oxides, or peroxides on filter media enhanced the removal of bacterial, viruses and turbidity from water and wastewater, the coated layer was however subject to dissolution when the coated filter media were placed in service (Chen, et al., 1998; Ahammed and Chaudhuri, 1996). The effect of continuous exposure of coated media to water and wastewater treatment therefore has been difficult to assess. 2.5 Impurities and Pollutants in Water and Wastewater Water is one of the most important requirements of life. Without it, neither the individual nor the organized community can survive. Absolutely pure water is never found in nature. When the natural waters contact with their surroundings, they leach and dissolve various minerals and salts from the components of the earth and rocks. Natural waters may also serve to nurture organisms such as bacteria and viruses. The source waters therefore are always aqueous solutions and suspensions of various compositions (Meltzer, 1993). Municipal wastewater consists of various pollutants from homes and commercial establishments, including various suspended and dissolved organic and inorganic substances, and harmful viruses or bacteria. 26 Chapter 2 The impurities and pollutants in water and wastewater may be divided into two groups, dissolved matters and suspended solids, based on the fact that many of the water and wastewater treatment processes are only effective against one of them (Henze et al., 2001). Natural organic matters (NOM) and suspended particles will be discussed in the following sections because traditional water and wastewater treatment processes are not able to effectively remove low concentration of these matters completely. 2.5.1 Humic Substances 2.5.1.1 Characterization of Humic Substances Natural organic matters in the environment (soils, sediments and natural waters) can be divided into two classes of compounds: non-humic materials (e.g. protein, polysaccharides, nucleic acids and small molecules such as sugars and amino acids) and humic substances (Hayes et al., 1989; Perdue and Gjessing, 1990; Rowell, 1994; Ziechmann et al., 2000). Humic substances are structurally complex, heterogeneous, and have a yellow to black appearance. They consist mainly of carbon, oxygen, hydrogen and, sometimes, small amounts of nitrogen, and occasionally phosphorous and sulphur. The amount of humic substances in natural waters can range from less than 2 mg/L up to 40 mg/L, and typically be in the range of 5-15 mg/L (Aiken et al., 1985). Humic substances result from biological decomposition of natural materials, including forest residues, grasses, food crops, animal remains and dead microorganisms, with molecular weights from a few hundreds for the simple ones to the thousands for the large polymeric materials. Humic substances predominantly carry negative charges in liquid solutions because of the abundance of carboxylic and phenolic groups (Jones and Bryan, 1998). 27 Chapter 2 Due to the hydration of the charged groups and electrostatic repulsion between the charges, the dissolved humic substances can be distributed in an extended conformation that adjusts themselves due to changes in the environmental conditions. Humic substances affect water quality adversely in several ways: undesirable color, complexation with metals to yield high metal concentrations, and reaction with chlorine to produce trihalomethanes (Ruohomäki et al., 1998). Hence it is desirable to minimize the concentration of humic substances in drinking water supplies and other process waters. Humic substances may be further divided into three fractions that can be isolated from soil: humic acids, fulvic acids and humans. The main fraction of humic substances is humic acids and their salts - humates. A rich soil with a near neutral pH would contain a high level of humates. Whereas the same soil with a low pH would be replete with humic acids (Schnitzer and Khan, 1972). The main chemical functions in humic acid include carboxylic, phenolic, amino and quinone with aromatic nucleuses of low degree of condensation. The presence of aromatic nucleuses with mobile p-type electrons and the various functional groups cause humic acid to have the ability of ionic exchange, complex formation and oxidization-reduction reactions. At present, the structure of humic acid is still ill defined despite many decades of research. A proposed building block for humic acid suggested by Jansen et al. (1996) is shown in Figure 2.2. The building block contains carboxylic groups connected to various alkyl chains and aromatic rings, phenolic groups, quinone structure, aromatic amine groups, seven chiral centers and thus 128 stereoisomers. 28 Chapter 2 HO . O O H2N OH ... ... O O HO COOH OH HOOC COOH OH Figure 2.2 Proposed building block of humic acids. 2.5.1.2 Removal of Humic Substances Humic substances have several characteristics that influence on how they may be removed from water and wastewater. Humic substances carry negative charges, which makes them often be removed through coagulation. Because of the negative charge property, humic substances are sometimes removed by ion exchange. Humic substances can also be adsorbed on some types of sorbent material, such as activated carbon (Ødegaard et al., 1999). Coagulation/Direct Filtration. Removal of humic substances by coagulation/direct filtration process is normally carried out by addition of coagulant, such as aluminum sulphate, ferric chloride, and subsequent floc separation through direct filtration. The task of the coagulant is to neutralize the negative charges of the humic substances to form flocs. The tiny flocs can then be transported and attached to the surfaces of filter grains and be separated from water. Although coagulation/direct filtration for the removal of humic substances in water and wastewater treatment has gained a lot of attention in resent years, low molecular weight humic substances are difficult to coagulate. In addition, the 29 Chapter 2 process incurs high chemical operation cost and generates high volume of extra sludge (from filter washing) that is difficult to handle. Adsorption by Activated Carbon. Activated carbon is the sorbent material most widely applied in drinking water treatment. Due to the composition of the raw material and the production process, between 5% and 20% by weight of activated carbon consists of elements other than carbon, primarily metals and surface bound oxygen (Sontheimer et al., 1988). The latter can be present in both acidic and basic surface functional groups as well as in metal oxides. As a result, the hydrophobicity of the ‘clean’ carbon surface is reduced and the conditions for adsorption of hydrophilic solutes are improved. There are several physical and chemical parameters that affect the adsorption of humic substances. Summers and Roberts (1986) showed that a positively charged carbon surface is favorable for the binding of the negatively charged humic substances. The ultimate carbon capacity, however, is more strongly influenced by the effects of pore structure and size. Lee et al. (1981) found that the molecular weight distribution of humic substances in relation to the pore size distribution of activated carbon is of great importance. Humic substances with larger molecular weight are excluded from the smaller pores of activated carbon, and thus have a lower volume available for adsorption. On the other hand, although humic substance fractions with smaller molecular weight generally adsorb to a great extent, they have larger polarity than larger molecules, thus decreasing the adsorption on the hydrophobic activated carbon surface. These two factors, as well as the high cost of regeneration of activated carbon, limited the application of activated carbon to the removal of humic substances from water and wastewater. 30 Chapter 2 Anion Exchange Resin. Since a large fraction of humic substances can be characterized as anionic polyeletrolytes, macroporous anion exchange resins in principle can be used for humic substance removal. Specifically, strong base resins have been found to be superior to weak-base resins with respect to capacity (Brattebф et al., 1987). Moreover, a resin skeleton which allows a high degree of swelling is more favorable for humic substances removal than a rigid skeleton (Fu and Symons, 1990). These studies concluded that ion exchange was the dominant mechanism for humic substance removal and surface adsorption only contributed to the removal of humic substances with smaller molecular weight. The break-through curves of humic substances in ion exchange columns are not practically favorable after a short time (Ødegaard et al., 1989), resulting in relatively poor utilization of the capacity of the resin and thus requiring frequent regeneration. The waste regeneration solution has a disposal problem due to its high pH (12.5-14) and salt concentrations (conductivity 10000-20000 mS/m) (Ødegaard et al., 1999). 2.5.2 Suspended Particles in Water and Wastewater 2.5.2.1 Sources of Suspended Particles By convention, suspended particles are defined as the particles which have a diameter larger than 0.4 µm (Eisma, 1993). Modern technologies allow particles as small as 0.02 µm to be observed, but these particles have often been considered as ‘dissolved’. The upper size limit of suspended particles is difficult to define, because small but heavy particles can rapidly undergo sedimentation, while large but low-density particles can stay in suspension for a long time. Suspended particles are present in almost all natural waters 31 Chapter 2 and wastewater, even although sometimes they are in very limited amounts. The universe presence and the complex physical and chemical properties of the suspended particles have made their removal a subject of significant scientific and engineering importance. Suspended particles present in natural waters include inorganic materials like clay, silt, sand, silica and calcium carbonate, or organic remnants of plant and other substances like fats, greases, microorganisms, oils, etc, entrained from the surface by rain water or thawed snow and carried into basins, rivers and lakes (Sastry and Agamuthu, 1996). Microorganisms and organic growths are common in natural surface and river waters, including normal soil bacteria, iron or manganese bacteria, algae and crustacea. Suspended particles in municipal wastewater include organic materials such as dead animal matter, plant tissue, organisms, synthetic (artificial) organic compounds, and inorganic solids which may include a number of potentially toxic elements such as arsenic, cadmium, chromium, copper, lead, mercury, zinc, etc. Pathogenic viruses, bacteria, protozoa and helminths may be suspended in raw municipal wastewater and will survive in the environment for long periods. The presence of them may cause human infections and be an important public health problem (Payment et al., 1991). The fine suspended mineral particles in the aquatic environment are mostly clays. It is difficult to remove clay particles from suspension because their negative charges repel each another, and, as a result, clay particles do not agglomerate to sizes large enough to be responsive to the sedimentation induced by gravitational forces (Meltzer, 1993). Neither is sand filtration particularly efficient to remove fine clay particles because sand also 32 Chapter 2 carries negative surface charges in water systems. Furthermore, it is generally know that clay particles are highly surface-active, and easily adsorb dissolved or colloidal materials onto their surfaces (Eisma, 1993). It has also been suggested that bacterial surfaces are not homogeneous with regard to charge distribution. Bacteria therefore have electrostatic and other interactions with clays in aquatic systems and tend to attach on the clays’ surface (Beveridge et al., 1993). Clay-bacterial composites are common to most aquatic environments (Beveridge, 1989). Even though most suspended particles may not be hazardous, their presence can cause color, odor or taste, making water aesthetically unpleasant for drinking, and unsuitable for many industrial applications. Suspended particles in the water can also interact with the disinfection processes in water and wastewater treatment, making them less effective. 2.5.2.2 Removal of Suspended Particles by Filtration Filtration with granular media is most commonly used for suspended particle removal in water and wastewater treatment. Media commonly used in water and wastewater filtration include sand and anthracite. Slow sand filters have been effective in providing a safe, potable water supply. However, there are limitations in the use of this type of treatment. Generally, the average turbidity is limited to 10 ppm, with a maximum concentration of 30 ppm (Cheremisinoff, 1995). Within these limits, effective bacterial and turbidity removals are realized. Rapid sand filters may be used for treatment of waters with highsuspended solids content when preceded by a preliminary solids separation device, e.g. coagulation and sedimentation. Diatomaceous earth filtration units were found to be 33 Chapter 2 effective during World War II in the prevention of amoebic dysentery by the removal of Endomebeia histolytica cysts (Fulton, 2000). The performance of filtration depends on several factors including particle size and physicochemical properties of the particles and the filer media. Larger particles may be filtered by entrapment mechanism. As the sizes of particles decrease, particle removal is usually more difficult and the interactions between the particles and the collectors (filter media) will play the most important role. Onorato and Tien (1980) have found that favorable interactions between particles and the collector can significantly increase particle deposition. They reported a 10-fold increase in the deposition of negatively charged particles onto positively charged collectors, as compared to negatively charged collectors. A notable increase in colloid deposition rates was observed by Elimelech (1991; 1994) when positively charged latex particles and negatively charged glass beads was used, and the enhancement of deposition was attributed to the attractive double layer interactions. In membrane filtration, using a cationic surfactant to create a positively charged surface was also found to increase the removal of negatively charged particles by 10% to 95% (Kang and Shah, 1997). 2.6 Polypyrrole and Chitosan A conducting polymer, polypyrrole (PPy), and a natural biopolymer, chitosan, were chosen to modify the surface of conventional granular media in this study because they possess activated sites (positively charged sites) under the neutral or weak acidic solution pH conditions. 34 Chapter 2 2.6.1 Polypyrrole (PPy) 2.6.1.1 Properties and Applications During the past two decades, intrinsic conducting polymers (ICPs) with conjugated double bonds have attracted a great deal of attention as advanced materials capable of simultaneously presenting the properties of organic polymers and semiconductors. In spite of the active and extensive research spent on ICPs, the ‘applications’ usually still reside in the laboratory rooms. One of the ICPs that actually has had a number of commercial applications is PPy, due to its good environmental stability, facile synthesis, possibility of forming homopolymers or composites, and higher conductivity than many other conducting polymers (Rodríguez et al., 1997; van Hutten and Hadziioannou, 1997; Wang et al., 2001). PPy displays many interesting properties: redox activity (Deronzier and Moutet, 1996), high surface free energy (Chehimi et al., 1999) and polarity (Liu et al., 1994), good ion and proton exchange capacity (Ge et al., 1992), Lewis acid-base interactions (Chehimi et al., 1993), electro-chromatographic stationary phase properties (Deinhammer, 1991), exceptional gas transport properties (Liang and Martin, 1991), good optical properties (De Paoli et al., 1990), strong adsorptive capabilities towards gases and macromolecules (Chehimi, 1999; Azioune et al., 1999). Most of these properties are highly depending on the nature of the counteranion, acid/base treatment, synthesis condition and procedure, handling and history of the polymer. 35 Chapter 2 Due to its unique attractive properties, PPy has been used to prepare high performance materials such as biosensors (Parthasarathy and Martin, 1994), gas sensors (Kincal et al., 1998), microactuators (Smela, 1999), and pH sensing systems (Talaie, 1997). The capability of PPy to be reversibly doped and dedoped by electrochemical methods makes it a suitable candidate for the construction of rechargeable batteries (Alper, 1989; Gemeay et al., 1995). PPy can also be used as functional membranes for liquid separation or gas separation. PPy membranes containing high concentrations of different counterions show exceptional selectivity of transport of ionic species and gas (Liang and Martin, 1991; Zhao et al., 1998). Furthermore, PPy membranes can be reversibly transformed between polyionic and neutral forms, which may switch the transport of ionic species across the membrane (Kyöstikontturi et al., 1998). PPy shows good adhesion to both organic and inorganic substrates. PPy can be stably deposited on a metal surface to reduce the rate of degradation process in an aggressive environment, hence to protect metal from corrosion (Idla et al., 1997; Su and Iroh, 2000). PPy may be coated on porous membranes (Nikpour et al., 1999) or be combined with membranes as composites (Morita, 1998) to improve their performance, and to prepare novel functional membranes. Polymer composites and blends of PPy materials have been widely studied to overcome the relatively poor mechanical properties (Rodríguez et al., 1997). One of the recent interests in research on PPy materials is the preparation of nanocomposite of PPy. PPy was coated on the surfaces of gold (Selvan et al., 1998) or ceramics (Cho et al., 2001) using surfactants as templates for the potential application in the field of rechargeable batteries. PPy/PMMA coaxial nanocable with an electrical 36 Chapter 2 conductivity of 1.7 S/cm was prepared using mesoporous silica as a nanoreactor (Jang et al., 2001). A series of PPy nanocomposites with both electrical and ferromagnetic behaviors were synthesized for applications in electrical-magnetic shields and microwaveabsorbing materials (Jarjayes et al., 1995; Gangopadhyay and De, 1999; Liu and Wan, 2000). 2.6.1.2 Charge Carriers in PPy PPy can be easily prepared via chemical or electrochemical polymerization. The former method usually gives powdered PPy and the latter gives filmy PPy. To prepare a large quantity of PPy, the chemical polymerization is the better method because it is free from the restriction of an electrode shape. N N + _ e- (a) . + (b) (b) N N + (e) . + (b) N + N + _ 2H + (c) ... N N _ e- (d) N n PPy Figure 2.3 Reaction mechanism of chemical polymerization of pyrrole. Figure 2.3 schematically shows the process of chemical polymerization of pyrrole. A neutral molecule of pyrrole monomer (a) changes to its cation radical species (b) due to oxidation, and (b) combines to form a dication of bipyrrole (c), and subsequently, a neutral bipyrrole molecule (d), which can undergo further oxidation into (e), then deprotonation and recombination steps continue and finally lead to PPy as the end product 37 Chapter 2 of the chemical polymerization (Malinauskas, 2001). Coupling of pyrrole monomers during polymerization reaction occurs primarily through the α-carbon atoms of the heterocyclic ring since these are the positions of highest unpaired electron π-spin density and hence highest reactivity. Ideal PPy therefore has α,α’-linked ring (poly(2,5-pyrrole)), with 180o dihedral angles so as to form a planar and linear chain. However, two-dimensional microstructures of PPy through β-linkages have also been detected (Bartl, et al., 1993; Joo et al., 2001). PPy can have two structures of bond alternation (Kikuchi, et al., 1992; Fermín et al., 1996), the aromatic configuration with long bonds between the rings and aromatic structures within the rings, and the quinoid configuration with shortened bonds between the rings and quinoid rings, as shown in Figure 2.4. The nondegenerate fundamental state of PPy is characterized by a unique aromatic structure of lower energy than that of the quinoid structure. H H H N N N N N N H H H aromatic structure H H H N N N N N N H H H quinoid structure Figure 2.4 Two structures of bond alternation for PPy: aromatic and quinoid structures. 38 Chapter 2 H H H N N N N N N H H H Electron acceptor H N H H . N + N N N N H H H polaron Electron acceptor H H H N N N + N N H H + N bipolaron H Figure 2.5 p-Type doping of PPy introducing polarons and bipolarons. The molecular arrangement of quinoid structure of PPy favors the formation of electronic defects, mainly polarons and bipolarons (Heeger, et al., 1988). When the nondegenerate ground state of PPy is oxidized, the structural motif of the chain segment between the cation radical (polaron) and the unpaired electron is that of a quinoid configuration. This polaron structure is higher in energy and confines the charge and spin density to a single self-localized structural deformation that is mobile along the chain. When further oxidation of a nondegenerate PPy chain takes place, a second electron can be removed from a different segment to create a new polaron, or the unpaired electron of previously formed polaron is removed and thus a spinless dication (bipolaron) is produced to confine a single lattice deformation on the chain (Figure 2.5). The oxidation of PPy is a doping process during which the electrons are transferred away from the polymer chain and the counteranions from the solution diffuse into the polymer to ensure the electroneutrality 39 Chapter 2 condition. Two charge transport mechanisms accounting for macroscopic conductivity in PPy polymer have been proposed (Rodríguez et al., 2000): charge propagation along onedimensional conjugated chains, with tunneling across conjugation defects, and inter-chain charge propagation between localized defects, a three dimensional electron-hopping mechanism. Because the oxidation potential of PPy is lower than that of the monomer (pyrrole) (Brédas and Street, 1985), the polymer is simultaneously oxidized during polymerization and thus charge carriers are generated. Hence the PPy prepared by both chemical and electrochemical polymerization is in the oxidation state and the counteranions present in the polymerization solution are incorporated into the growing polymer. The doping densities of PPy synthesized by chemical polymerization are usually in the range of 25 to 30% (Kim et al., 1995). Early studies suggested delocalized positively charged nitrogen atoms or partially charged nitrogen atoms in doped PPy. Kang et al. (1997) revealed the simultaneous presence of neutral and positively charged nitrogen atoms in the N1s corelevel XPS spectrum, an agreement between the proportion of positively charged nitrogen and the counter anions. They thus proposed the presence of unit positive charges on the positively charged nitrogen atoms in doped PPy. Malitesta et al. (1995) suggested the .+ polaron and bipolaron structures in doped PPy as N H and + N H , respectively, on the basis of XPS findings. The electrical and structural properties of chemically and electrochemically synthesized PPy in aqueous solutions are greatly dependent on the solution pH. When PPy is 40 Chapter 2 immersed in an alkaline solution, a deprotonation reaction takes place, resulting in elimination of both protons and dopant anions from the oxidized polymer as well as nonconducting polymer of quinoid structure. The deprotonated PPy can undergo a reprotonation process by treatment with an acid and the polymer chain redopes back to its previous form (Pei and Qian, 1991). The deprotonated PPy may also be readily reduced by phenylhydrazine to form the unique aromatic structure. In addition, cyclic voltammetry studies show that PPy can be cycled between the conducting (oxidized, doped) and the neutral (reduced, undoped) states (Deronzier and Moutet, 1996). The chemical process involved in various redox states of PPy can be represented in Figure 2.6 The parent scheme was originally suggested by Kang et al. (1990) and was later modified to be more consistent with the chemical composition. H N H N N H N H + N H + N H N H N H Reduction (Phenylhydrazine) Base H N H N N N H H N H N Acid H N N H Oxidation H N + N H N N H N H Reduction H N H N H N N H N H N Figure 2.6 Schematic representation of interconversion among the various intrinsic redox states of PPy. 2.6.1.3 PPy as Adsorbents 41 Chapter 2 Inverse gas chromatography (IGC) results revealed that PPy exhibited strong intermolecular interactions either with nonpolar organic vapors such as n-alkanes or polar organic vapors such as Lewis acid chloroform (CHCl3) or Lewis base tetrahydrofuran (THF). IGC also showed that the Lewis acid-base interactions govern adsorption of organic vapors on PPy to an extent of at least 75% (Chehimi et al., 1993). The adsorption of poly(methyl methacrylate) (PMMA) on chloride-doped PPy (PPyCl) and hydrogensulfate-doped PPy (PPyHSO4) was studied by Abel et al. (1994; 1995) with XPS and time-of-flight secondary ion mass spectroscopy (ToF-SIMS). They found that the adsorption isotherms were of the Langmuir type, and the amount of PMMA adsorbed depended greatly on the acid-base properties of the solvents. When PMMA and PVC blends were used, PMMA and PVC were simultaneously adsorbed by PPy at an extent that depended on the initial polymer concentrations, and, more importantly, the properties of solvent (Chehimi et al., 1998; Abel et al., 2002). For the development of novel biomaterials, biological assays, biosensors, and organic semiconductor devices, the adsorption of biomacromolecules onto PPy has been studied in recent years. The adsorption capacity of human serum albumin (HSA) onto PPy core/polyacrolein shell latex was found up to 11 mg/g (Miksa and Slomkowski, 1995). Azioune et al. (1999) increased the adsorption capacity of HSA at pH = 7.4 to 63 mg/g and 147 mg/g when using PPy powder and PPy-silica nano-composite, respectively. Azioune et al. (2002) also applied the van Oss-Good-Chaudhury (VOGC) theory to understand the nature of interfacial interactions between HSA and PPy power. Moreover, PPy was found to be a strong bioadsorbent of DNA fragments and the adsorption was 42 Chapter 2 dominated by the electrostatic interactions between negatively charged DNA and positively charged PPy (Saoudi et al., 1997a). Minehan et al. (1994; 2001) fitted the adsorption data by a simple diffusion-limited binding model, and suggested that the binding of DNA to PPy was directly related to the presence of positive charge carriers in PPy. Saoudi et al. (1997b) found that the solution pH 5-7 and the addition of salt favored the adsorption. However, the application of PPy in water and wastewater treatment as an adsorption or a filtration material has not been explored before the current research. 2.6.2 Chitosan 2.6.2.1 Origin and Properties Chitosan ((1-4)-linked 2-amino-2-deoxy-β-D-glucopyranose, Figure 2.7b) is a polysaccharide derived by the deacetylation of chitin ((1-4)-linked 2-acetamido-2-deoxyβ-D-glucopyranose, Figure 2.7a), which is an abundantly available natural polymer making up the skeletal support and body armor of some sea animals like crab, lobster, shrimp and squid (Tokura and Nishi, 1995; Ravi Kumar, 1999; Ravi Kumar, 2000). Fungal fermentation and exoskeletons of most insects can also be potentially important sources for chitin. Chitin is believed to be the second most abundant biomaterial after cellulose. The annual biosynthesis of chitin has been estimated to be 109 to 1011 tons. As the byproducts of seafood processing, shells of crab, lobsters, shrimp and squid are collected and extracted for chitin. The content of chitin can be up to 20% of these shells while the rest is calcium carbonate and proteins (Ravi Kumar, 1999). 43 Chapter 2 CH2OH H H O OH H H O H NHCOCH 3 CH2OH H H O OH H H (a) Chitin O H NHCOCH 3 n CH2OH H H O OH H H H NH2 O CH2OH H H O OH H H H NH2 O n (b) Chitosan Figure 2.7 Structural formula of chitin and chitosan. Chitosan is a hydrophilic, biocompatible, biodegradable, and non-toxic polymer. It is commercially available in a range of molecular weights, degrees of deacetylation and types of salts such as glutamate, hydrochloride and lactate. Because of the stable, semicrystalline structure, chitosan is normally insoluble in aqueous solutions above pH 6.5. However, in dilute acids, the free amino groups are protonated and the molecules become partially soluble below pH 3 (Suh and Matthew, 2000). The pH-dependent solubility of chitosan provides a convenient mechanism for processing under mild conditions. The OH and NH2 groups in chitosan molecules can give rise to hydrogen bonding, and the linear molecular structure expresses sufficient chain flexibility (He et al., 1998). Viscous chitosan solutions can be extruded and gelled in high pH solutions or baths of nonsolvents e.g. methanol. Such gel fibers can be subsequently drawn and dried to form highstrength fibers. 2.6.2.2 Applications Since chitosan has unique physiological and biological properties, it is regarded as a versatile starting material for the preparation of various products. Chitosan is an ideal biomaterial because of its excellent features such as hydrophilicity, biocompatibility, 44 Chapter 2 biodegradability and non-toxicity. For example, chitosan and chitosan derivatives may represent potentially safe and efficient cationic carriers for gene delivery (Borchard, 2001) and gene transfection (Liu and Yao, 2002). A number of researches have been directed towards the development of safe and efficient chitosan-based drug delivery systems (Dodane and Vilivalam, 1998). Chitosan is also a widely used adsorbent for adsorption of metal ions because the amino and hydroxy groups on chitosan chain can act as chelation and reaction sites for metal ions (Guibal et al., 1998; Inoue et al., 1999). Another main interest in chitosan arises from its cationic nature in acidic solutions, which provides unique properties relative to other polysaccharides which are usually neutral or negatively charged. The cationic nature allows chitosan to be used in other applications such as water and wastewater treatment. Positively charged chitosan reacts with organic solids and cell surfaces that are usually negatively charged. Chitosan can be sued as a coagulant or filtration aid in water and wastewater treatment (Kawamura, 1995). It was found that chitosan was not only effective in treatment, but also competitive in price. 45 Chapter 3 Chapter 3 Preparation of PPy-Coated Glass Beads 3.1 Introduction PPy has received much attention as a novel functional material because it has various unique electronic and electrochemical properties. However, the π-electron system along the polymer backbone makes PPy insoluble and infusible, and thus not processible, which largely limits its applications (Frommer and Chance, 1985). To overcome this limitation, many researchers focused on the preparation of conducting PPy-based composite materials. PPy can be prepared either by electrochemical or chemical oxidation of pyrole. In consequence, two approaches have been used to coat various materials with PPy. Electropolymerization of pyrole at electrodes is a widely used method to prepare PPy film. However, electropolymerization of pyrrole is greatly restricted by the use of conducting substrates, such as metals or carbon materials, and no electrochemically prepared PPy layers can be obtained on insulating surfaces. Adhesion between the polymer and the conducting substrate (electrode) is also quite poor because of the weak physical interactions between them (Wu and Chen, 1997). PPy hence can be easily peeled off from the substrate. Moreover, electropolymerizatin is not a well-defined process. It is difficult to find the relationship between the PPy’s properties and the conditions of electrochemical polymerization (Rodríguez et al., 1997). 46 Chapter 3 Chemical polymerization is another method for coating various substrates with PPy. Usually, Fe3+-salt or ammonium peroxydisulfate is used as the oxidant, but polymerization with other oxidant like Cu2+ (Khedkar and Radhakrishnan, 1997) and hydrogen peroxide has also been reported. Chemical polymerization occurs in the bulk of the solution, and the resulting polymer, PPy, precipitates as insoluble solids. Part of PPy, formed by chemical polymerization, can deposit spontaneously on the surface of various materials, including filter paper, quartz and glass fiber fabrics, polymer films and carbon fibre, etc. Coating inorganic substrates with PPy is of current interest in research on intrinsic conducting polymers (Beleze and Zarbin, 2001). A major advantage of this type of material is the possibility of combining the properties of organic and inorganic compounds into a unique one. The inorganic substrates may also provide an environment that will favor polymer ordering from chain to chain. Both the characteristics of the conducting polymers and the nature of the inorganic substrates will dictate the properties of the prepared hybrid composite materials. Some studies have been performed to coat glass materials with thin layers of conducting polymers, chiefly PPy and polyaniline (PANI). The chemical coating of PANI on glass support such as textile gave conductive materials with excellent stability, mechanical and electrical properties (Genies, et al., 1991). Adhesion of PANI layers in situ synthesized on the substrate of glass fiber fabric was reported to be sufficiently strong (Forveille and Olmedo, 1994). A linear relation of the thickness and optical density of PANI coatings on 47 Chapter 3 glass substrates was found through optical interferometry (Stejskal et al., 1999). A series of PPy/glass composites was obtained by in situ oxidative polymerization of pyrrole inside the pores of porous glass (Maia et al., 1995; Zarbin et al., 1999; Gimenez and Alves, 1999). Solid state 13C and 29Si CP-MAS-NMR spectrometry showed that PPy was in their oxidized conductive state with Si-O- groups of the glass surface acting as counter anions. Marchant et al. (1998) demonstrated that glass fibre fabric can be coated with PPy through chemical polymerization in an aqueous solution, and the electrical resistivity as well as microwave reflectivity of the glass fibre fabric can be controlled by varying the concentrations of the chemicals used in the chemical polymerization reaction. Some attempts have been made to increase the adhesion of PPy layers and the surfaces of glass. For example, glass fibers were treated with pyrrole-substituted organotrialkoxysilanes, and PPy was deposited onto the modified glass surface, resulting in unusual high electric conductivity (Faverolle, 1998). Multilayer organic xerogel-PPy composites were prepared by alternate successive treatments of a glass substrate with a siloxane functionalized pyrrole, pyrrole and a sol precursor of the xerogel to protect the conducting layer (Cattey et al., 1998). In this study, PPy-coated glass beads were prepared by chemical polymerization of pyrrole on the surface of glass beads with FeCl3 as an oxidant. Optical microscope, scanning electron microscope (SEM) and atomic force microscope (AFM) were used to examine the surface morphologies of the uncoated and coated glass beads. Physical and chemical surface properties of uncoated and coated glass beads were characterized by X- 48 Chapter 3 ray photoelectron spectroscopy (XPS), Fourier transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), and ζ-potential measurements. 3.2 Experimental Section 3.2.1 PPy Coating Pyrrole (99%) and FeCl3·6H2O (97%) were purchased from Aldrich chemical company. Ballotini glass beads, with specific weight of 2.55, were obtained from Jecons Scientific Ltd (UK) and were used as the model granular media for surface modification. 13.5 g of FeCl3·6H2O (0.05 mol) were dissolved in 50 mL deionized (DI) water with stirring. 100 g of glass beads were then added into the solution. After a few minutes, 7 mL of pyrrole (0.10 mol) diluted with DI water was added in droplets into the mixture with vigorous stirring. The colour of the system rapidly became black. The system was continuously mixed for 3 hours to allow the growth of PPy coating on the surface of glass beads. The product was then thoroughly washed with DI water and alcohol, dried in air for 24 hours, and stored in a desiccator for analysis or experiments 3.2.2 Morphology Observation The surface morphologies of the coated and uncoated glass beads were examined by SEM (JEOL JSM-6400) in involving bombardment of the filter surface with an electron beam emitted from a tungsten wire at 10~20 kV. The granules were mounted on a metal sample substrate by double-sided adhesive tapes; surface was coated with platinum and then scanned in the microscope. 49 Chapter 3 AFM measurements were made with a multimode Nanoscope III at ambient pressure, room temperature and humidity. The topographic imaging was done with the taping mode (scanning size 15 µm, set point 3.34V, scan rate 1.0 Hz). To determine the size of the uncoated and coated glass beads, the granules were observed under an optical microscope (BX60, Olympus, Japan) and the sizes were measured with the software AnalySIS®3.0 (Soft Imaging System GmbH). 3.2.3 ζ-Potential Measurements A Zeta Plus4 Instrument (Brookhaven Instruments Corp.) was used to determine the surface ζ-potentials of the uncoated and coated glass beads. Measurements of the uncoated glass beads were performed by the method described by Bai and Tien (1999). The coated glass beads were placed in a 100-mL vial with 50 mL DI water and the vial was vibrated in a sonic bath for 30 minutes. The liquid in the vial with small fragments from the PPy coating in it was then decanted and the solution pH was adjusted for measurements. The ζ-potentials so determined were found to be similar to those determined using PPy particles from the chemical polymerization reaction, indicating that the fragments in the decanted liquid were essentially from the PPy coating and were completely made of PPy. The pH of the samples was adjusted with 0.1 M HCl and 0.1 M NaOH. All ζ-potential measurements were carried out with a background ionic concentration of 0.001 M NaCl. 50 Chapter 3 3.2.4 XPS Measurements XPS measurements were made on a VG ESCALAB MKII spectrometer with an Al Kα Xray source (1486.6 eV photos). To compensate for surface charging effects, all binding energies were referenced to the C1s neutral carbon peak at 284.6 eV. Surface elemental stoichiometries were determined from sensitivity-factor corrected peak area ratios, and the software XPSpeak 4.1 was used to fit the XPS spectra peaks. A pure Gaussian lineshape was applied to the spectra in peak fit. The line-width (full width at half maximum, or FWHM) was maintained constant for all the components in a particular spectrum. The intensities were calculated from the peak areas given by the fitting program. 3.2.5 FTIR Measurements FTIR measurements were performed using a Bio-Rad FTS-135 FTIR spectrometer. The surplus PPy powder in the coating solution was collected, thoroughly washed with D.I. water and alcohol, and dried in vacuum desiccator for 24 hours. Pressed pellets of so prepared PPy powder samples were ground with KBr powder (Aldrich, IR grade) and used for the FTIR data acquisition. A simple transmission mode was employed and the spectral resolution was set at 4 cm-1. 3.2.6 XRD Measurements XRD measurements were performed on a Shimadzu XRD-6000 diffractometer using a Ni-filtered Cu (Kα) radiation operating at 40 kV and 30 mA. The 2θ range started from 2θ of 1.5°. PPy powder collected from the polymerization solution was dried, and carefully ground through hand grinding with an agate mortar and pestle before the measurement. 51 Chapter 3 3.3 Results and Discussion 3.3.1 Coating Thickness The average diameters of the coated and uncoated glass beads were determined through measuring 50 individual grains using an optical microscope. The uncoated glass beads had an average diameter of 549.96 µm, and the glass beads after one-time coating and twice coatings had average diameters of 562.33 µm and 590.36 µm, respectively. The thickness of the PPy coating was therefore estimated to be 6.2 µm for one-time coating, and 20.2 µm for twice coatings. The results reported below used glass beads from onetime coating of PPy. 3.3.2 Surface Morphologies The morphologies of the uncoated and coated glass beads were examined using SEM. The images are shown in Figure 3.1. The surface of glass beads was relatively smooth (Figure 3.1a). Polymerization of pyrrole generated a large number of fine PPy particles (ranging from a few tens to a few hundred nanometers) which deposited onto the surfaces (Figure 3.1b). Both SEM (Figure 3.1c) and AFM results (Figure 3.2) show that the coated surface was rough with a considerable number of ring-like structures. These structures may be attributed to the adsorption and release of gas bubbles (i.e., O2) during the oxidation polymerization reaction of the pyrrole (Unsworth et al., 1992). The oxidation of the pyrrole monomer was found to compete with the oxidation of water to evolve O2 gas, which adsorbed at the polymerization surface to act as a polymerization local point, and 52 Chapter 3 later released, which resulted in the ring-like “defects”. The rough surface due to the formation of the ring-like structures has been considered as surface defects in other applications, but it is actually beneficial for particle removal through adsorption or filtration in water and wastewater treatment. (a) (b) (c) Figure 3.1 SEM images showing the surface morphologies of (a) uncoated glass beads (magnification 1500×), (b) PPy-coated glass beads (magnification 7000×), (c) PPy-coated glass beads (magnification 1500×). (a) (b) Figure 3.2 AFM images of PPy-coated glass beads (a) 2-D image, (b) 3-D image. 53 Chapter 3 3.3.3 ζ-Potentials In general, two methods are widely used to determine the ζ-potentials of solid-liquid interfaces: (1) to crush the solid material into fine particles and disperse them in an aqueous solution, and (2) to use the streaming potential technique (Gu and Li, 2000). The first method was applied in this study. ζ-Potential were calculated by the Smoluchowski equation (Hunter, 1981) ζ = ηµ e / ε (3.1) where η and ε are the absolute viscosity and dielectric permittivity of the liquid medium, respectively, and µe is the electrophoretic mobility of the particles. Smoluchowski equation is only valid for the Debye length κ-1 much smaller than the mean radius of the particles, R0, i.e. κR0 >> 1. The known κ-1 value for 0.001M NaCl is 9.6 nm (Hunter, 1991). The sizes of the glass and PPy fragments from the granules in the solutions, as estimated from a Coulter Series 230 Particle Sizer (Coulter Counter Corp.) and a Zeta Plus4 Instrument, were larger than 3 µm. So it is valid to use the Smoluchowski equation. Figure 3.3 shows the ζ-potentials of the uncoated glass beads and PPy-coated glass beads as a function of the solution pH. The uncoated glass beads are negatively charged at the pH range investigated (pH = 2~12), with the zeta potential becoming more negative as the pH increases. For the coated glass beads, the surface ζ-potentials are positive when pH below 10, and, over the ranges of 3 to 10, it remains relatively constant (approximately +37 mV). Because of the positive surface charges, the coated glass beads can be expected to improve filtration or adsorption performance for all negatively charged pollutants in water and wastewater. 54 Chapter 3 120 PPy-coated glass beads ζ-Potential (mV) 100 Uncoated glass beads 80 60 40 20 0 -20 -40 -60 -80 0 2 4 6 8 10 12 14 pH Figure 3.3 ζ-Potentials of the uncoated and PPy-coated glass beads. 3.3.4 XPS Results XPS analysis was used to characterize the PPy coatings on the glass beads. XPS is the fundamental technology used to identify the elemental composition and oxidation state of the surfaces and to characterize the interactions of the interfaces in this study. The principle of XPS analysis is briefly described here. An atom is made of a nucleus and a number of electrons that orbit the nucleus with defined orbits. The analysis of XPS involves catching the electrons from the matter being studied in order to find out from which atom they are coming. To be able to free the electrons from the nucleus attractive force, the sample to be analyzed is exposed to an Xray source. This particular type of source brings the electrons enough energy to free them from the nucleus. Once they have been freed up, some of them carrying enough energy 55 Chapter 3 may leave the solid matter and are collected by an electron analyzer. The most important information in the analysis is the so-called binding energy (BE) which an electron had before leaving the atom. By counting the electrons for various BEs, a corresponding spectrum is obtained. Since each value of the BE is characteristic for a given element, and each peak area is proportional to the number of atoms being present in the solid matter studied, the chemical elemental composition of the sample surface is therefore determined by calculating the respective contribution of each peak area. Appropriate data processing then leads to the determination of the chemical-bound nature that exists between these elements. In addition, the BE of an element increases with increasing its oxidation state and the energy shifts are typically in the range of 0–3 eV. C1s INTENSITY (ARB. UNITS) INTENSITY Fe2p 700 N1s 710 720 730 740 BINDING ENERGY (eV) O1s O KLL Cl2p 0 220 440 660 880 1100 BINDING ENERGY (eV) Figure 3.4 XPS survey scan and narrow spectra of PPy coating on the coated glass beads. Figure 3.4 shows a typical wide survey spectrum of the PPy coating. The Cl 2p signal at 56 Chapter 3 around 197 eV was due to the chloride dopant anions. A fairly strong O1s signal at 532 eV was found in Figure 3.4. The excess of oxygen in PPy prepared by chemical or electrochemical polymerization has been widely observed in other studies (Lei and Martin, 1992). It has been proposed that this excess oxygen in PPy is caused by surface oxidation since PPy has a relatively low oxidation potential. For FeCl3 and FeCl2, the Fe2p3/2 XPS signals centered at around 710 eV (Muilenberg, 1977). Wide and narrow BE scans of PPy coating do not reveal the presence of iron, suggesting that Fe was completed INTENSITY (ARB. UNITS) removed by DI water and alcohol washing in the preparation process of PPy. N =N- + -NH28% 393 396 399 402 405 BINDING ENERGY (eV) Figure 3.5 N1s XPS spectrum of PPy coating on the coated glass beads. The N 1s core level spectrum from XPS analysis for the PPy coating is presented in Figure 3.5. The BE of the main amine-like nitrogen component (–NH–) was at about 57 Chapter 3 399.6 eV, and that of imine-like nitrogen components (=N-) is at about 397.7 eV. The higher BE tails in the N 1s spectrum at 401.1 and 402.7 eV were attributed to the two high oxidation states of the nitrogen atoms with positive charges, i.e. NI+ and NII+ (Malitesta et al. (1995) assigned them to polaron and bipolaron structures, respectively.). The proportion of the positively charged nitrogen atoms in the PPy coating was found to be 28% (in terms of [N+]/[N]). It is these protonated nitrogen atoms that contribute to the positive surface charges of the PPy-coated glass beads. The reaction stoichiometries of pyrrole with FeCl3 have been proposed, respectively, as in Eq. (3.2) (Pron at al., 1985) and Eq. (3.3) (Machida et al., 1989) 4n(C4H3N) + 9nFeCl3 → [(C4H3N)4+Cl-]n + 8nHCl + 9nFeCl2 (3.2) 3n(C4H5N) + 7nFeCl3 → [(C4H3N)3+Cl-]n + 6nHCl + 7nFeCl2 (3.3) where Cl- is the dopant anion. Eq. (3.2) indicates a pyrrole/Cl- ratio of 4:1 (25% [N+]/[N]) while Eq. (3.3) a ratio of 3:1 (33% [N+]/[N]) in the PPy complex. The PPy synthesized by oxidative polymerization in other studies was reported to have a proportion of positively charged nitrogen in the range of 25 to 30% (Kim et al., 1995; Deronzier and Moutet, 1996), which was comparable to the 28% determined in this study. The 25–30% of [N+]/[N] ratio suggests that in the polymerization of pyrrole into PPy complex, both Eq. (3.2) and Eq. (3.3) reactions are likely to be coexisting, and, on an average, at least one pyrrole monomer among every four pyrrole monomers in the PPy complex is protonated. Cl 2p core-level spectrum (Figure 3.6) was fitted with three spin-orbit split doublets (Cl2p3/2 and Cl2p1/2) of a fixed doublet separation of 1.7 eV (Hasik et al., 2002). The BE 58 Chapter 3 values of Cl2p3/2 at about 197.1 and 201.1 eV were suggested to be due to the presence of the ionic chloride (Cl-) and covalent chloride (-Cl), respectively. The third component, intermediate chloride (Cl*), located at 198.6 eV, was not definitively explained in the literature. Kang et al. (1991) reported that the proportion of Cl* species was in the order of 20% of the total chlorine incorporated for PPy prepared with oxidative Fe(III) salts, and explained that the high Cl* content was probably associated with a small amount of ironbased chloride species in PPy. The proportion of Cl* shown in Figure 3.6 is lower than that reported by Kang et al. (1991). This may indicate that the residual Fe(III) in PPy was removed during the washing process. - INTENSITY (ARB. UNITS) Cl component * Cl component -Cl component 192 195 198 201 204 207 210 BINDING ENERGY (eV) Figure 3.6 Cl 2p XPS spectrum of PPy coating on the coated glass beads. 59 Chapter 3 Table 3.1 Results of qualitative and quantitative XPS analysis of the PPy coating Binding Energy (eV) Assignment % 397.7 =N- 6 399.6 -NH- 66 401.1 NI+ 21 402.7 NII+ 7 197.1 Cl- 66 ([Cl-]/[N] = 25) 198.6 Cl* 13 ([Cl*]/[N] = 5) 200.1 -Cl 21 ([-Cl]/[N] = 8) 198.8 Cl- 66 200.3 Cl* 13 201.8 -Cl 21 N1s Cl2p3/2 Cl2p1/2 The distributions of the nitrogen and chloride components in the PPy coating from the XPS results are summarized in Table 3.1. [Cl-]/[N], [Cl*]/[N] and [-Cl]/[N] refer to the ratios of ionic chloride to total nitrogen, intermediate chloride to total nitrogen, and covalent chloride to total nitrogen, respectively. The close balance between the amount of chloride anion (Cl- and Cl* species) and positively charged nitrogen, i.e. ([Cl]/[N]+[Cl*]/[N]) ≈ [N+]/[N], implies that the positive charges probably present in unit on positively charged nitrogen atoms in the oxidized PPy. 3.3.5 FTIR Results 60 Chapter 3 A typical FTIR spectrum of the PPy fragment (Figure 3.7) shows a featureless absorption tail from 4000 to 2000 cm-1 because the absorbance in this region arises from the free charge carriers present in the doped material (Cai et al., 1991; Lei et al., 1992). Hence, the N-H and C-H stretching bands are not visible as they are obscured by the tail of the 4000- 972 1095 1710 1555 1475 1305 1185 1045 915 795 668 TRANSMISSION 2000 cm-1 peaks. 3900 3400 2900 2400 1900 1400 900 400 -1 WAVENUMBER (cm ) Figure 3.7 FTIR absorption spectrum of PPy fragment. The band positions and their assignments are given in table 3.2. The weak band at 1710 cm-1 was attributed to the carbonyl defects. Carbonyls may be formed by overoxidation of PPy or as the product of chain termination by nucleophilic attack by H2O on the pyrrole rings. This band has been observed in both chemically prepared PPy and electrochemically prepared PPy (Lei et al., 1992; Liang et al., 1992). It has been suggested that the bands at 1555 and 1475 cm-1, corresponding to C=C and C-C stretching 61 Chapter 3 of pyrrole rings, respectively, are especially affected by changes in the extent of delocalization. The stronger the peak intensity at 1555 cm-1, as compared with the peak intensity at 1475 cm-1, the longer the conjugation length of the PPy backbone (Liang et al., 1992; Yan et al., 2000). Pyrrole ring vibrations, characteristic of PPy, are dominant in the region of 1600-600 cm-1. Table 3.2 Characteristic FTIR absorption bands (cm-1) of PPy fragment Wavenumber (cm-1) Assignments (Allen et al., 1997; Du Pasquier et al., 1999; Nicho and Hu, 2000) 1710 C=O stretching 1555 C=C stretching of pyrrole rings 1475 C-C stretching of pyrrole rings 1305 C-N stretching of pyrrole rings 1185 C-H in-plane deformation 1095 C-OH; C-O stretching 1045 N-H in-plane deformation 972, 915, 795 668 C-H out-of plane deformation C-H bending 3.3.6 XRD Analysis The structure of doped PPy has been reported to be of low order levels (Cheah et al., 1998). Figure 3.8 shows the typical X-ray diffraction spectrum of the PPy powder 62 Chapter 3 collected from the coating solution. There are three broad peaks at around 2θ = 9o, 21.5o and 26o. Applying Bragg’s law, nλ = 2d sin θ (3.4) where λ is the X-ray wavelength, d is the spacing of the lattice atom layers, θ is diffraction angel, and n is an integer (often taken to be 1), these three peaks correspond to spacing of approximately 9.8 Å, 4.1 Å and 3.4 Å, respectively. Figure 3.8 XRD pattern of PPy powder. Geiss et al. (1983) proposed a layered structure of reduced PPy according X-ray studies (see Figure 3.9), in which the polymer chains lie in the (a, b) plane and the interlayer spacing observed is 3.41 Å. In this model structure, the c value of the unit cell is twice of the spacing value since the stacking is of the type ABABAB… The crystallographic parameter given as a = 8.2 Å, b = 7.35 Å, c = 6.82 Å. In our prepared PPy, 3.4 Å therefore 63 Chapter 3 is identified as the polymer chain stacking distance, i.e. interlayer spacing between the (a, b) plane of the polymer. 9.8 Å, corresponding to the spacing between neighboring polymer chains in the plane of the polymer, is a little larger than that of reduced PPy in order to accommodate the dopant ions (Cl-). The repeat distance along their axis (b), 4.1 Å, however, is quite smaller than the idealized value, 7.35 Å. Some researchers explained it by amorphous ring or mechanical stretching of linear organic polymer (Van Hutten and Hadziioannou, 1997). N N H N N N H H H N H H H H H a = 8.2 A H N H H N H H H N N H N H H N b = 7.35 A Figure 3.9 Model arrangement of the chains in a layer of undoped PPy. As discussed in Chapter 2, it is generally assumed that PPy consists of one-dimensional chains through α,α’-linkages. The real PPy synthesized by oxidized polymerization however has a highly defective molecular structure. Two-dimensional structure of PPy was proposed on the basis of calculations and spectra caused by β-linked pyrrole rings (Paasch et al., 1994). Strong disorder will result from the structural defects and will force the chain to depart from planarity, which can also be influenced by the preparation condition greatly. This may be the reason of considerable discrepancies in the XRD 64 Chapter 3 studies of PPy. 3.3.7 Interactions between Glass and PPy The solid/solid interaction is concerned with the nature and properties of the adhesion between the glass surfaces and the surfaces of PPy coatings. Adhesion is defined by American Society for Testing and Materials (ASTM) as “a condition in which two surfaces are held together by either valence forces or by mechanical anchoring or by both together” (Pulker, 1984). The bonding forces can be van der Waals forces, electrostatic forces, and/or chemical bonding forces. XPS technology was used to examine the interactions between the surfaces of glass beads and PPy coatings. Figure 3.10a shows the XPS survey scan of the glass beads, which is dominated by silica (SiO2) shown by the three peaks of Si2p, Si2s and O1s. The C1s peak is caused by contaminated carbon on the sample surface. The composition of Ballotini glass beads purchased from Jecons Scientific Ltd is defined as: SiO2, 72.0%; Na2O, 13.0%; CaO, 9.0%; MgO, 4.0%; Al2O3, 1.0%; Fe2O3+K2O, 1.0%. The Ca2p, Mg2p and Al2p peaks in Figure 3.10a however are quite weak, due to their low proportion. For the coated glass beads after stripping PPy coating by ultrasonic, Ca2p, Mg2p and Al2p peaks are nearly disappeared but a new peak exists at around 399 eV, which indicates the PPy residual on the surface (Figure 3.10b). 65 0 O KLL Na KLL Ca 2p C 1s 220 Na 1s O 1s (a) O 2s Mg 2p Al 2p Si 2p Si 2s INTENSITY (ARB. UNITS) Chapter 3 440 660 880 1100 O 2s O KLL Na KLL Ca 2p N 1s Si 2p Si 2s C 1s INTENSITY (ARB. UNITS) (b) 0 Na 1s O 1s BINDING ENERGY (eV) 220 440 660 880 1100 BINDING ENERGY (eV) Figure 3.10 XPS survey scan of (a) bare glass beads, and (b) coated glass beads after stripping the PPy coating through ultrasonic. 66 Chapter 3 When glass surfaces contact with aqueous solutions, water molecules may react with silicon-oxygen bonds, yielding hydroxide groups (Pulker, 1984; Doremus, 1994), as expressed below: Si O Si + H 2O Si O H + HO Si (3.5) The alkali component on the glass may also react with the water molecules to form hydroxide groups: _ Si O Na + + H 2O Si O H + Na + + OH _ (3.6) Hence, the surface of glass acquires negative charges via Si O H + H 2O Si O _ + H3O + (3.7) PPy prepared by oxidation polymerization is known to possess 25-30% of positively charged nitrogen atoms in its molecular backbone. The negative Si-O- sites on the glass surface may interact with the positive N+ sites of PPy through electrostatic forces. Zarbin et al. (1997; 1999) proposed that the Si-O- groups of the glass surface acted as counteranions when PPy was formed on the surface of Porous Vycor Glass. XPS results in this study show that the proportion of positively charged nitrogen atoms on the PPy from the coated glass surface ([N+]/[N] = 20 %) is lower than that of pure PPy ([N+]/[N] = 29%). The Si2p spectrum of the glass beads with PPy residual can be decomposed by curve fitting into two peaks at 102.8 and 103.7 eV, in comparison with the Si2p signal for the clean glass beads with a single peak at a binding energy of about 102.8 eV (see Figure 67 Chapter 3 3.11). These results imply that there is charge transport between the N+ in PPy and the Si on the glass beads. N+ atoms are proton donors and the Si-O- groups accept protons partially. Hence, the interactions between the PPy coatings and the glass surfaces are more likely to be the acid-base interactions. (b) INTENSITY (ARB. UNITS) (a) 99 102 105 108 99 102 105 108 BINDING ENERGY (eV) Figure 3.11 Si2p XPS spectra of (a) bare glass beads, (b) glass beads with PPy residual. 3.4 Conclusions A new granular media with positive surface charges and porous surface morphologies were prepared by chemical polymerization of pyrrole on the surface of glass beads. The surface-modified glass beads were characterized by optical microscopy, SEM, AFM, FTIR, XPS, XRD, and ζ-potential measurements. The PPy-coated glass beads were found to have positive surface ζ-potentials for pH below 10. The surface morphology of the 68 Chapter 3 coated glass beads was considerably rough and many ring-like structures appeared on the coated surface. XPS results revealed that about 28% of the nitrogen atoms in the PPy molecules were positively charged. XPS results also suggested that there were strong chemical interactions between the Si-O- groups of the glass beads and the N+ of the PPy coatings. FTIR and XRD spectra indicated that the surface composition and molecular order of the PPy coating were rather complicated. 69 Chapter 4 Chapter 4 Removal of Humic Acid and Clay Particles by PPy-Coated Glass Beads 4.1 Introduction One of the challenges in water supply is the removal of humic substances as the unit treatment processes in a conventional water treatment plant are usually not effective for this purpose. Moreover, humic substances present in the water resources facilitate the transport of many organic and inorganic pollutants through the water treatment facilities, and, also by reacting with chlorine during water treatment, produce trihalomethanes, which are known or suspected as human carcinogens. Humic acid is the major component of humic substances, and behaves like flexible, linear, and synthetic polyelectrolytes. The structure of humic acid varies with salt contents, pH values and concentrations. At low concentrations with only little salt present, its structure is rather linear at pH values from 6.5 to 9.5 (Ghosh and Schnitzer, 1980). At high concentration and together with salt it forms a sphero-colloidal structure. This is mainly because of the electrostatic repulsion between negatively charged groups of humic acid. Due to its size differences at different ionic strengths and pH, its diffusion properties vary. At high ionic strengths it diffuses faster due to its coiled structure. Also, at low pH it diffuses faster, and this difference is more remarkable at low ionic strength (Cornel, et al., 1986). Avena et al. (1999) used viscometry to evaluate the effects of pH and ionic strength on the volume and structure of humic acid, and obtained the similar results. The direct information on the macromolecular structure changes of humic substances under 70 Chapter 4 different pH, ionic strength and complexing cations conditions with the in situ spectromicroscopic technology was reported by Myneni et al. (1999). There have been a number of studies in the adsorption of humic acid on oxide or clay surfaces (Davis and Bhatnagar, 1995; Vermeer and Koopal, 1998; Vermeer et al., 1998; Bolto et al., 2001). Solution pH, ionic strength, and the electrostatic properties were found to play an important role in humic acid adsorption. For the adsorption of humic acid on positively charged minerals, an increase in adsorption is observed with decreasing pH and increasing salt concentrations (Summers and Roberts, 1988; Zhou et al., 1994; Vermeer et al., 1998). A 13 C-NMR spectra study by Marshall et al. (1998) suggested the chemical interactions between the functional groups of humic acid and the mineral surfaces. Gu et al. (1995) investigated the binding of natural organic matter (NOM) by hematite using FTIR spectroscopy, and found that both carboxylic and hydroxyl groups are involved in the binding of NOM. Vermeer et al. (1998) studied the adsorption isotherms of humic acid onto hematite particles based on the Self-Consistent Field (SCF) theory and found that the adsorbed amounts were determined by two opposite effects: charge compensation and specific attraction enhance the adsorption, whereas lateral electrostatic repulsion and loss of entropy inhibit the adsorption. Suspended or colloidal particles such as kaolin clay and other mineral particles are other important constituents of natural waters. Clay particles are weathering products of feldspars and other minerals present in rocks and soils, having very small sizes and with an equivalent diameter smaller than 2 µm (Pédro, 1997; Yariv and Michaelian, 2002). 71 Chapter 4 Clay particles and other mineral materials in the environment are responsible for most surface and colloid reactions on the earth, including adsorption of organic matters. They are the most important components and have great influence on the chemical and physical properties of soils and natural waters. In soils, clay dispersion can result in surface crusting, reduced water infiltration, and increased surface runoff and erosion. Clay colloids that remain suspended in surface runoff or in water infiltration in the soil can also enhance the translocation of adsorbed contaminants (Kretzschmar et al., 1997). The unit processes to remove clay particles in water and wastewater treatment include sedimentation and sand filtration. However, the negative charges of clay particles make neither sedimentation nor sand filtration be particularly efficient to remove fine clay particles (see discussions in Chapter 2). Membrane separation technology may be an alternative process for this purpose. But it is difficult to control the fouling of membranes by clay particles (Jones et al., 1999). In this study, adsorption and filtration experiments were conducted with humic acid and kaolin suspension, respectively, to evaluate the performance of the surface-modified glass beads as an adsorbent or filter medium. SEM was used to examine the surface morphology of the coated granules before and after humic acid adsorption and kaolin particle deposition. ζ-potential and XPS were applied to assess the specific interactions between the PPy coating and the adsorbed humic acid/deposited kaolin particles. A preliminary study of cleaning/regeneration of the coated granules by DI water and acid/base were also carried out. 72 Chapter 4 4.2 Experimental Section 4.2.1 Preparation of Solutions Humic acid (defined as sodium salt with C 39.03%, H 4.43% and N 0.68%) was purchased from Aldrich chemical company. XPS and FTIR analyses were made for the dry Aldrich humic acids according to the method described in Chapter 3. Kaolin particles, obtained from Kaolin Sdn Bhd (Malaysia), were used as model colloidal particles. Humic acid solution was prepared by dissolving a certain amount of humic acid in a known volume of DI water. The solution was then mixed on a magnetic stirrer for an hour and filtered through a Whatman membrane filter (0.45 µm) before conducting the adsorption tests. Humic acid concentrations in the experiments were determined by an ultraviolet-visible (UV-VIS) spectrometer (Hitachi UV-2000) at 408nm. Kaolin particles were first suspended in DI water and ultrasonically bathed for 10 minutes and then fractionated by sedimentation for half an hour. The large particles were settled out and discarded, and the supernatant was used to make the test suspension by diluting it with DI water. The size distributions of the particles in the test suspensions were analyzed with a Coulter Series 230 Particle Sizer (Coulter Counter Corp.). The morphology of the kaolin particles as observed under an optical microscope (BX60, Olympus, Japan) was quite irregular. 4.2.2 Adsorption of Humic Acid 73 Chapter 4 Batch adsorption measurements were made by placing 15 g of PPy-coated glass beads in 250 mL humic acid solution (15 mg/L concentration in a 500 mL glass beaker) at room temperature with slow shaking for a period of 24 hours. The solution concentration history was determined by taking and analyzing solution samples periodically. Similar measurements using uncoated glass beads were also made to provide a basis of comparison. The amount of humic acid per unit weight of the beads adsorbed at time t, q(t) (mg/g), was calculated from the mass balance equation q (t ) = (C 0 − C t )V m (4.1) where C0 and Ct (mg/L) are the initial humic acid concentration and the humic acid concentration at any time t, respectively, V is the volume of the solution, and m is the weight of the granules. To study the adsorption equilibrium of humic acid onto the PPy-coated glass beads, batch adsorption experiments were made by placing 9 g of the PPy-coated glass beads in 75 mL humic acid solution with different initial humic acid concentrations (10 mg/L, 20 mg/L, 30 mg/L, 40 mg/L, 50 mg/L, 60 mg/L concentration in 250 mL glass beakers respectively) at room temperature for 3 days. To study the effects of pH and ionic concentrations on the adsorption, batch adsorption experiments were also made at different pH values and ionic concentrations. HCl (0.1 M) and NaOH (0.1 M) were applied to adjust the solution pH values. NaCl was added to the humic acid solution to vary its ionic concentrations. 3 g of the PPy-coated glass beads was 74 Chapter 4 placed into 25 mL of humic acid solution with a known initial concentration until the adsorption equilibrium was established. 4.2.3 Removal of Humic Acid and Kaolin Particles in a Fixed Bed The experimental set-up consisted of a suspension tank with a stirrer, a pump, a filter column, and sample collectors. The filter column, a 25mm i.d.× 60cm height acrylic column, was packed with glass beads or PPy-coated glass beads. Glass beads and PPycoated glass beads were rinsed with DI water before experiments. Humic acid solutions or kaolin suspensions were fed to the column by a pump at filtration velocity of 2.93 m/hr or 1.47 m/hr. The turbidity of kaolin suspension samples was measured using a turbidity meter (Model 800, Spectral Tecknik Pte Ltd, Singapore). NaCl was added to the kaolin suspension to vary their ionic concentrations. The histories of the effluent concentrations in the experiments were monitored through taking and analyzing samples periodically. 4.2.4 Regeneration Studies Two ways to regenerate the media grains after humic acid adsorption were conducted. Regeneration using D.I. water. After an equilibrium adsorption, the solution was carefully decanted and the granules were placed in 500 mL of D.I. water for desorption. The test was carried out for three cycles of 12 hours each. Batch adsorption experiments were conducted at room temperature with the regenerated granules to assess the regeneration 75 Chapter 4 efficiency (10 g of regenerated granules, 250 mL of humic acid with initial concentration of 15 mg/L). Regeneration using base and acid. 10 g of humic acid saturated granules were immersed in 250 mL of 0.5 M NaOH solution for 12 hours, followed by in 0.5 M HCl solution for another 12 hours. After 3 runs, the treated granular media were washed with DI water to about neutral pH values (pH = 6.5). Batch adsorption experiments were again conducted to assess the regeneration efficiency. 4.3 Results and Discussion 4.3.1 Batch Adsorption of Humic Acid INTENSITY (ARB. UNITS) N1s 392 397 402 407 O KLL Na KLL Na 2s Na 1s BINDING ENERGY (eV) O 1s C 1s INTENSITY (ARB. UNITS) 4.3.1.1 Characterization of Humic Acid 0 220 440 660 880 1100 BINDGING ENERGY (eV) Figure 4.1 XPS survey and narrow spectra of Aldrich humic acid. 76 Chapter 4 The XPS survey spectrum for humic acid (Figure 4.1) reveals the presence of three major elements: carbon, oxygen and sodium. Although humic acid obtained from Aldrich was specified to contain a trace amount of nitrogen (the supplier specified at 0.68% by weight nitrogen component), XPS narrow BE scan of N1s does not reveal obvious signal of nitrogen. This may be due to the fact that the nitrogen component in the Aldrich humic acid is very small and no nitrogen atoms could be detected. 2.5 Absorbance 0.15 ABSORBANCE 2 0.1 0.05 1.5 0 0 5 10 15 Concentration (mg/L) 1 254 nm 0.5 0 190 390 590 790 990 WAVELENGTH (nm) Figure 4.2 UV-Vis spectrum of Aldrich humic acid (50 mg/L) at pH = 6.5. The inset graph shows the calibration of the UV-Vis spectrometry at 408 nm and pH = 6.5. The UV-Vis absorption spectrum of humic acid was recorded in the range of 190 ~ 1100 nm, as shown in Figure 4.2. The spectrum is rather featureless and increases in absorbance nearly monotonously from 1100 nm to 190 nm. Absorbance may be mainly caused by the aromatic ring structure (Bloom and Leenheer, 1989), internal vibration and 77 Chapter 4 rotation of the molecules, and intermolecular interactions (Korshin et al., 1997). An indistinct absorption peak at around 254 nm is found in Figure 4.2, which can be attributed to the charge transfer transition of aromatic rings (Reckhow et al., 1990; Kumke et al., 2001). Korshin et al. (1997) suggested the use of the absorption at λabs = 253 nm for an estimation of the degree of functionality of the aromatic ring in humic substances. Measuring the UV-Vis absorbance of water is a largely accepted and very popular way to estimate the content of humic substances in natural waters. Various wavelengths have been proposed for measuring the UV-Vis absorbance of humic substances with spectrphotometer: 240 nm (Davis and Bhatnagar, 1995), 250 nm (Pettersson et al., 1993), 254 nm (Yoon et al., 1998; Lai and Chen, 2001), 280 nm (Perminova et al., 1999; Seida and Nakano, 2000), 400 nm (Liu and Gonzalez, 2000; Bolto et al, 2001), 456 nm (Bennett and Drikas, 1993) and 465 nm (Shawwa et al., 2001). A relatively short wavelength such as 410 nm has been recommended because the absorbance of humic substances increases with decreasing wavelength (Hautala et al., 2000). At long wavelength the absorbance is too small, whilst the absorbance may be very sensitive to other compositions at short wavelength. The inset graph of Figure 4.2 shows the calibration curve of humic acid at pH 6.5 without salt present, obtained by plotting the absorbance readings versus the concentrations of humic acid solution (mg/L). Since the absorbance of humic acid is pH and salt dependent, the calibration lines were made for each experimental pH and salt concentration. 78 Chapter 4 4.3.1.2 Adsorption Performance Typical batch adsorption results of humic acid are shown in Figure 4.3. For the uncoated glass beads, the adsorption of humic acid was very limited, probably due to the repulsive interaction between the negatively charged glass beads and the humic acid molecules. The PPy-coated glass beads, however, had considerably higher adsorptive capacity for humic acid. After 10 h, adsorption equilibrium appeared to be established. The specific amount of humic acid adsorption was calculated from mass balance to be 0.17 mg (humic acid)/g (coated glass beads) in this case. Amount adsorbed (mg/g) 0.20 0.16 0.12 PPy-coated glass beads Uncoated glass beads 0.08 0.04 0.00 0 4 8 12 16 20 24 Time (hr) Figure 4.3 Adsorbed amount of humic acid as a function of time (initial humic acid: 15 mg/L, solution volume: 250 mL, quantity of filter media: 15 g, pH = 6.50, T = 25.0 oC). 4.3.1.3 Interaction between Humic Acid and PPy 79 Chapter 4 A typical XPS result for the N1s spectrum of the PPy-coated glass beads with humic acid adsorption is shown in Figure 4.4. In contrast with the results of Figure 3.5 in Chapter 3, the [N+]/[N] ratio of the PPy reduced to about 21%, in comparison with 28% of [N+]/[N] ratio when PPy was absent of humic acid. This result indicates that a portion of the positively charged nitrogen atoms in the PPy coating layer was neutralized by humic acid. The reaction pathway may be proposed as shown in Figure 4.5, where R-COO- stands for INTENSITY (ARB. UNITS) humic acid macromolecule. N + =N-NH21% 393 396 399 402 405 408 BINDING ENERGY (eV) Figure 4.4 N1s XPS spectra of PPy-coated glass beads adsorbed with humic acid. The percentage of the N+ atoms reacted with humic acid may be calculated, as in this case, to be only 25% of the total positively charged N+ atoms presented before humic acid adsorption. Compared with the size of a pyrrole monomer in the PPy polymer molecule, humic acid is believed to have a much larger molecular size. Therefore, only some of the 80 Chapter 4 protonated pyrrole monomers in the PPy molecules may contact the reactive groups (i.e., COOH, etc.) of a humic acid molecule and thus result in reaction. H N + H N + N H H N H N N H + N H N H N H + RCOO _ R C H H N + N H O _ O N H H N H N N + N H N H + N H Figure 4.5 A schematic representation of humic acid adsorption on PPy. Special attention was paid in obtaining the results of Figure 4.4. XPS analyses of the Aldrich humic acid dry particles didn’t show the N1s signal, as shown in Figure 4.1, which makes sure the N1s spectrum in Figure 4.4 being representative of the nitrogen in the PPy layer but not that of humic acid. ζ-potential measurements were performed to study the effect of humic acid adsorption on the electrical properties of the PPy coating. These results under various solution pH values are given in Figure 4.6. The most noticeable difference is that the ζ-potentials of the PPy coating (PPy fragments from the coated glass beads) adsorbed with humic acid were mostly negative. Adsorption of humic acid therefore not only neutralized the positive charges but also reversed the surface charges of the PPy coating. The mechanisms of 81 Chapter 4 humic acid removal by the PPy-coated glass beads therefore include at least chemical reaction between the reactive groups, which is responsible for charge neutralization, and macromolecular adsorption, which is responsible for charge reversal. The first mechanism may act quickly but the second mechanism can be a slow process. 100 PPy-coated glass beads PPy adsorbed humic acid Pure humic acid Kaolin particle 80 ζ-Potential (mV) 60 40 20 0 -20 -40 -60 0 2 4 6 8 10 12 14 pH Figure 4.6 ζ-Potentials of the PPy-coated glass beads, humic acid (15mg/L), kaolin particle (45 mg/L) and PPy-coated glass beads with adsorbed humic acid. 4.3.1.4 Adsorption Isotherms During the batch experiments, isotherms were used to evaluate adsorption properties. The Langmuir and Freundlich models were used to fit the data of humic acid adsorption by the PPy-coated glass beads. 82 Chapter 4 The experimental values of the isotherm were used in the linearized form of the Langmuir equation, which was previously given in Eq. (2.11), i.e. 1 1 1 = + q qm K a qm C (2.11) with q and qm expressed in mg/g, Ka in L/mg and C in mg/L 40 Experiment 1/q (g/mg) Langmuir model 30 20 10 0 0 0.04 0.08 0.12 0.16 0.2 1/C (L/mg) Figure 4.7 Adsorption isotherms of humic acid on PPy-coated glass beads, linearised according to the Langmuir Eq. (2.11). Figure 4.7 shows the experimental adsorption isotherm data fitted to the linearized form of Langmuir equation. A regression analysis was conducted and from the slope and interception values, the Langmuir isotherm constants were calculated to be qm = 0.396 mg/g, and Ka = 0.0117 L/mg. Hence the Langmuir equation can be given as 83 Chapter 4 1 215 = 2.525 + q C (4.2) The experimental data of the isotherm were also used in the linearized form of the Freundlich equation expressed in the following format: log q = 1 log C + log K f n (2.13) The plot of log q versus log C based on Eq. (2.13) to the same experimental results in Figure 4.7 is shown in Figure 4.8, which gives log q = 0.7993 log C − 2.179 (4.3) The Freundlich isotherm constants were determined to be Kf = 6.622×10-3, and n = 1.25. -0.6 Experiment Freundlich model Log q -0.9 -1.2 -1.5 -1.8 0.6 0.8 1 1.2 1.4 1.6 1.8 Log C Figure 4.8 Adsorption isotherms of humic acid on PPy-coated glass beads, based on the linearized form of the Freundlich Eq. (2.13). 84 Chapter 4 Correlation coefficients were 0.9797 and 0.9945 for the linearized Langmuir equation and Freundlich equation, respectively. This indicates that the Freundlich relationship gives a slightly better fit to the experimental data than the Langmuir equation. 4.3.1.5 Effect of pH and Ionic Concentrations Solution pH and ionic concentrations are often shown to be the important parameters determining adsorption performance because of their influence on the nature of surface interactions between the media grains and the substances to be removed. The results on the effect of the solution pH values are given in Figure 4.9, in which the amounts of adsorbed humic acid versus pH values are presented. It is clear that the extent of adsorption improved with a decrease of pH, and there was hardly any adsorption at pH > 12. It is also interesting to note that the amount of humic acid adsorbed by the PPycoated glass beads decreased from pH = 3.1 to pH = 1.1 although the ζ-potential of PPy is more positive at pH = 1.1 than that at pH = 3. The results can be explained by the interactions between the PPy coating and the humic acid to be adsorbed. For pH > 10.5, both the PPy coating and the humic acid had negative ζ-potentials. Due to the strong electrostatic repulsion, no humic acid adsorption therefore took place at pH > 12. The pKa of the carboxylic groups in humic acid molecules was at around pH 3 (Kinniburgh et al., 1996; 1999), and the Aldrich humic acid possessed an isoelectric point (IEP) at around pH 2.0 (Fairhurst et al., 1995; Fairhurst and Warwick, 1998). Hence, for pH = 1.1, the PPy coating had strong positive ζ-potential and the humic acid also became positively charged. Again, the electrostatic repulsion developed between the PPy molecules and the 85 Chapter 4 humic acid limited the adsorption of humic acid on the coated glass beads, leading to the observed decrease of humic acid adsorption at pH 1.1, as compared to that at pH 3.1. From pH 10 to 3.1, the adsorption of humic acid showed a gradual increase with decreasing pH. Since the negative ζ-potentials of the humic acid reduced with decreasing pH and the positive ζ-potentials of the PPy coating were relatively constant, humic acid adsorption on the PPy surface was therefore in a more compact pattern in the lateral direction due to reduced repulsion between the adsorbed humic acid molecules, which would result in more humic acid molecules being adsorbed on the surface at a lower pH. Amount Adsorbed (mg/g) 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0 2 4 6 8 10 12 14 pH Figure 4.9 Effect of initial solution pH values on the adsorption of humic acid by the PPycoated glass beads (initial humic acid concentration = 15mg/L, solution volume = 25 mL, quantity of granular media = 3 g). 86 Chapter 4 Amount Adsorbed (mg/g) 0.09 0.085 0.08 0.075 0.07 0.065 0.06 0 0.1 0.2 0.3 0.4 0.5 Ionic Concentration (M) Figure 4.10 Effect of ionic concentrations on the adsorption of humic acid by PPy-coated glass beads at pH 6.5 (3 g grains in 25 mL solution). The effect of ionic concentrations on humic acid removal at a near neutral pH is shown in Figure 4.10. Improved removal with increasing ionic concentrations (NaCl) was observed, but the effect was not significant. It is important again to consider the electrostatic interactions between the humic acid molecules themselves as well as between the humic acid molecules and the PPy coating, as adsorption performance depends on both of these interactions. Generally speaking, an increase in the ionic concentration would reduce the electrostatic repulsion between the adsorbed humic acid molecules and the molecules in the solution, thus leading to increased adsorption. Increasing the ionic concentration is also believed to reduce the size of humic acid molecules and hence to increase the amount of humic acid adsorption. On the other hand, when the ionic concentration is increased, some of the charges are screened and the attractive force between the humic acid 87 Chapter 4 molecules and the PPy molecules is reduced. The positive ions in the solution will also compete with the PPy coating for the negative sites in the humic acid molecules. These last two effects will contribute to reduced adsorption of humic acid. An experimental comparison was conducted for the adsorption performance of PPy powder collected from the polymerization reaction (0.5-1 mm) and activated carbon (0.43-0.85 mm purchased from Aldrich Chemical Company). A 0.1 g amount of PPy powder or activated carbon was put into 25 mL of humic acid solution (15 mg/L humic acid, pH = 6.5) at room temperature for 24 h with slow shaking. The amounts of humic acid adsorbed onto the PPy powder and activated carbon were determined to be 3.69 mg/g and 1.82 mg/g, respectively. The specific surface areas of the PPy powder and the activated carbon were determined to be 7.8 m2/g and 1008 m2/g, respectively. The results therefore indicate that most of the surface area of the activated carbon is not available for humic acid adsorption, due to the large molecular size of the humic acid. 4.3.2 Removal of Humic Acids in a Fixed Bed As humic acid molecules used in this study had much smaller sizes ( 6.6, the factor of flopped-down water dipoles as well as the specific adsorption of OH- ions may overwhelm the protonation of amino groups, resulting in the negative ζ-potentials. 100 Chitosan PET Nylon 6,6 80 ζ-Potential (mV) 60 40 20 0 -20 -40 -60 -80 0 2 4 6 8 10 12 14 pH Figure 6.4 Influence of pH on the ζ-potentials of nylon 6,6 granules, PET granules and the chitosan layer. 128 Chapter 6 Τhe ζ-potentials of the nylon 6,6 granules and PET granules are also shown in Figure 6.4. The results indicate that the zero points of ζ-potentials of the nylon 6,6 granules and PET granules were at pH = 2.3 and 3.4, respectively. 6.3.3 XPS Analysis The XPS spectra of the nylon 6,6 and PET granules agree well with those published in the literature (Burrel and Chera, 1999; Uchida et al., 2000; Vasquez-Borucki et al., 2001). For nylon 6,6 granules, the C1s line in Figure 6.5a can be fitted with three peaks at 284.6, 285.6 and 287.7 eV, respectively. The lowest binding energy peak at 284.6 eV is due to the C−C or C-H carbon atoms in the nylon 6,6 polymer chain while the higher binding energy peaks at 285.6 and 287.7 eV are attributed to the C-N and –CO-NH- carbon atoms, respectively. Only one peak (at 399.6 eV) was observed in the N1s core-level spectrum of the nylon 6,6 granules (see Figure 6.6a). This peak is due to the acylamino nitrogen atoms (¯CO¯NH¯) present in the nylon 6,6 polymer. For PET granules, as shown in Figure 6.5d, the C1s line can be fitted with three peaks at 284.6 eV for the phenyl carbon atoms (C6H4-, C6H4 stands for benzene ring), 286.4 eV for the methylene carbon atoms bound to one oxygen (-CH2-O-), and 288.9 eV for the ester carbon atoms (-CO-O-). The area ratio of these peaks was 67:23:20, which was close to the theoretical value from its molecular structure (60:20:20). In the hydrolysed nylon 6,6 granules (Figure 6.5b), a new peak appeared at around 289 eV in the C1s spectrum, which can be attributed to the carboxyl carbon (-CO-O-), and thus suggests the generation of the carboxylic acid groups on the surface of the nylon 6,6 129 Chapter 6 granules by the hydrolysis. It was also found that the proportion of the peak at 287.7 eV, according to –CO-NH- carbon, decreased, in comparison with that of pure nylon 6,6 granules, and the relative intensity of the peaks at 287.7 eV and 289 eV was about 2:1. Hence 33% of the acylamino groups were hydrolysed and turned into the carboxyl groups. The C1s spectrum of the hydrolysed PET granules (result not shown in Figure 6.5), however, was almost the same as that of untreated granules. The reason can be that the binding energies of the C1s, based on the carboxyl groups on the hydrolysed PET granules, overlapped with that of the ester groups in the untreated granules (Huh et al., 2001). After the nylon 6,6 or PET granules immobilized with chitosan, the C1s spectrum had three components at 284.6, 286.3 and 288.0 eV with relative intensities of 1:1.3:0.3, which are attributed to C-C or C-H carbon, C-N or C-O carbon and O-C-O carbon, respectively (Figure 6.5f). These results are in agreement with the reported XPS spectra of chitosan (Arof et al., 1998, Dambies et al., 2001). XPS with Al Kα X-ray source usually can only penetrate less than 20 nm in depth (Zhang and Gonsalves, 2001), while the average thicknesses of the chitosan layer immobilized on nylon 6,6 granules and PET granules were observed as 42.36 µm and 48.97 µm, respectively. Therefore, only the chitosan layer can be observed in the XPS analysis for the nylon 6,6 or PET granules immobilized with chitosan. The N1s peak of chitosan layer was located at 399.5 eV (Figure 6.6c), which is attributed to the amino nitrogen (-NH2) in the chitosan. 130 Chapter 6 (a) (b) (c) (d) (e) (f) 278 280 282 284 286 288 290 292 294 BINDING ENERGY (eV) Figure 6.5 C1s XPS spectra of (a) nylon 6,6, (b) hydrolysed nylon 6,6, (c) nylon 6,6 after peeling off chitosan, (d) PET, (e) PET after peeling off chitosan, and (f) chitosan layer. 131 Chapter 6 (a) (b) (c) (d) 392 394 396 398 400 402 404 406 408 BINDING ENERGY (eV) Figure 6.6 N1s XPS spectra of (a) nylon 6,6, (b) nylon 6,6 after peeling off chitosan, (c) chitosan layer, and (d) PET after peeling off chitosan. 132 Chapter 6 When the surface of the nylon 6,6 granules was carefully peeled off the immobilized chitosan layer (Figure 6.5c) and was analyzed with XPS, the proportion of C1s peak at around 286 eV, according to C-N carbon, increased greatly, as compared with that of the hydrolysed nylon 6,6 granules, due to the existence of immobilized chitosan residue on the nylon 6,6 surface. The carboxylic carbon peak (-CO-O-) of the hydrolysed nylon 6,6 at 289 eV disappeared completely after immobilization of chitosan. This is an indication that the carboxylic acid groups of the hydrolysed nylon 6,6 surface bound with the amino groups of chitosan and salts were formed. The negative charges transferred from the nitrogen atoms of chitosan to the oxygen atoms of the carboxylic groups as a result of acid-base ionic binding. Hence, the negative charging densities on the carbon atoms of the bonded carboxylic groups increased, which caused a decrease of the binding energies. This can also be supported by the N1s spectrum. A new peak appeared in the N1s spectrum at 401.6 eV (Figure 6.6b), indicating a salt structure (-NH3+…-OOC-) generated between the hydrolysed nylon 6,6 surface and the chitosan during the immobilization process. The hydrolysis and immobilization process of the nylon 6,6 can be suggested in Figure 6.7a. For the PET granules after peeling off the immobilized chitosan layer, the C1s peak at around 289 eV was largely reduced while a new peak appeared at around 288 eV (Figure 6.5e). This peak may be caused by the acid-base ionic binding between the carboxyl groups in the hydrolysed PET granules and the amino groups in the chitosan. A new peak, similar to that of nylon 6,6 immobilized with chitosan, appeared at 401.6 eV in the N1s spectrum (Figure 6.6d), again indicating the salt structure (-NH3+…-OOC-) between the 133 Chapter 6 hydrolysed PET and the chitosan. The hydrolysis and immobilization process of the PET granules may be suggested in Figure 6.7b. ~ R1_ _ _ _ _ _ _ CO O C2H4 O CO _ R2 ~ ( PET granules) OH - Hydrolysis H ~ R1_ _ ~ R1_ _ CO O- + _ _ _ CO OH + ~R2_ CO OH + HO_ C2H _ OH 4 Chitosan Immobilization _ NH3+ Chitosan ~R _ 2 + _ CO O_ NH3+ Chitosan (a) O H+ _ ~ R1 C _ NH_R ~ ~R1_COOH + NH _ R2 ~ 2 2 Hydrolysis ( nylon granules) Chitosan ~ R1_CO O- NH 3+ _ Chitosan (b) Figure 6.7 Schematic representations of the immobilization of chitosan on (a) PET granules, and (b) nylon granules. 6.3.4 Adsorption of Humic Acid 6.3.4.1 XPS Analysis 134 Chapter 6 INTENSITY (ARB. UNITS) (a) _NH 2 _NH+ 3 15.3% 392 394 396 398 400 402 404 406 BINDING ENERGY (eV) INTENSITY (ARB. UNITS) (b) 275 280 285 290 295 BINDING ENERGY (eV) Figure 6.8 XPS spectra of the chitosan layer immobilized on PET granules with adsorbed humic acid (a) N1s, (b) C1s. 135 Chapter 6 Chitosan-immobilized PET granules after adsorption of humic acid in a 15mg/L humic acid solution under a near neutral pH condition were measured using XPS. After adsorption of humic acid, a new peak appeared at 401.6 eV in the N1s spectrum (Figure 6.8a)), indicating the reaction of the amino groups of chitosan with humic acid. The amino groups of chitosan are the only adsorptive groups because other functional groups on the glucosidic rings of chitosan are stable in neutral and weak acidic/basic solutions. Since sodium salt of humic acid was used in our adsorption experiments and the carboxyl and phenolic groups are the reactive groups in humic acid, the amino groups of chitosan may be protonated in the aqueous solution at first, and then react with humic acid to form the organic complexes –NH3+…-OOC-R or –NH3+…-O-C6H4-R (C6H4 stands for benzene ring), as shown in the following equations (A stands for humic acid): CH2OH H H O O O OH H H NH2 H CH2OH O O H H O OH H H N+ H HH H + H+ CH2OH H H O O OH H H _ + A O H (6.3) N+ HHH CH2OH H H O O O OH H H _ N+A H H HH (6.4) Protonation of the amino groups leads to the decrease of the electron densities on the nitrogen atoms and thus increases the binding energy of N1s. The proportion of the protonated nitrogen atoms was found to be around 15.3%, which is indicative that 15.3% of the amino groups in chitosan reacted with humic acid. 136 Chapter 6 After the adsorption of humic acid, the C1s spectrum of the chitosan layer can be decomposed into three peaks at 284.6, 286.4 and 288.1 eV (Figure 6.8b). The relative intensities became 1:0.65:0.2, in contrast to 1:1.3:0.3 before the adsorption of humic acid. C-C or contaminated carbon became the main component after adsorption. This phenomenon was caused by humic acid adsorbed on the surface of chitosan. 6.3.4.2 Effect of pH and Ionic Concentrations Figure 6.9a shows the removal of humic acid (initial concentration of 15 mg/L) by chitosan-immobilized PET granules at solution pH values of 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0 and 12.0. It was found that the adsorption capacity of chitosan was largely dependent on the pH values of the aqueous solutions. A decreased removal of humic acid was observed as the pH increased from 2 to 12. Figure 6.9a also showed a significant change in the adsorption capacities at pH between 6-7. From the results of XPS analysis, the adsorption process can actually be a formation process of organic complexes between humic acid and the chitosan layer. The adsorption performance and adsorption capacity are therefore dominated by the equilibrium of protonation of the amino groups and the formation of the organic complexes together. The adsorption capacity can be expected to improve if the amount of protonated amino sites increases. According to Eq. (6.3), high concentrations of H+ at low pH will drive the equilibrium to the right and therefore increase the amount of protonated amino sites. With the increases of pH, the increases of the OH- concentration can force the reaction in Eq. 137 Chapter 6 (6.3) shift to the left, and thus decrease the amount of protonated amino sites. As a result, an increase in adsorption was observed with decreasing pH. Based on the DLVO theory, the adsorption performance can be influenced by the electrostatic interactions between the humic acid and the chitosan layer. As mentioned earlier, zero ζ-potential of chitosan measured in this study was at around pH = 6.6. Chitosan had a positive ζ-potential at pH = 6. 0, caused by the protonated amino groups (NH3+), which can enhance the adsorption capacity greatly due to the favorable electrostatic interactions between the humic acid and the chitosan surfaces because humic acid was negatively charged at pH = 6.0. At pH = 7.0, the chitosan-immobilized granules had negative ζ-potentials, and the electrostatic interactions between the humic acid to be adsorbed and the chitosan surface were unfavorable. With the increase of pH, the electrostatic repulsion can become significant, which reduced the adsorption amount of humic acid onto the chitosan layer. The variations in the amounts of humic acid adsorbed onto chitosan-immobilized PET granules as a function of the ionic concentrations (NaCl) in the humic acid solution are shown in Figure 6.9b. It can be observed that the removal of humic acid increased with an increase in the ionic concentrations, but the effect is not significant in comparison with the effect of pH. 138 Chapter 6 Percentage of humic acid removed (%) 80 (a) 60 40 20 0 2 4 6 8 10 12 14 0.2 0.25 0.3 Percentage of humic acid removed (%) pH 80 (b) 60 40 20 0 0 0.05 0.1 0.15 NaCl concentration (M) Figure 6.9 Effect of (a) pH, and (b) ionic concentrations on humic acid adsorption (initial concentration of humic acid: 15mg/L, solution volume: 75 ml, quantity of granules: 10 g, temperature: 25 oC). 139 Chapter 6 It is important again to consider the electrostatic interactions between the humic acid molecules themselves as the well as electrostatic interactions between the humic acid molecules and the chitosan surfaces. Since adsorption depends on both of these interactions, an increase in the ionic concentration would reduce the lateral electrostatic repulsion between the adsorbed humic acid molecules and the humic acid molecules in solution, thus leading to increased adsorption. At the same time, when the ionic concentration is increased, some of the charges are screened and the electrostatic interaction between the humic acid molecules and the chitosan layer is reduced. At pH > 6.6, the force between the humic acid molecules and the chitosan layer is repulsive, the reduction of the force between them can result in an increase of adsorption. However, at pH < 6.6, the force between the humic acid molecules and the chitosan layer is attractive. The effect of screened charges will decrease the adsorption. At high ionic concentrations, humic acid can form a sphero-colloidal structure in stead of a linear structure at low ionic concentrations (Ghosh and Schnitzer, 1980). With the increase of ionic concentrations, the reduction of humic acid size can also be expected to increase the amount of humic acid adsorption. 6.3.4.3 Adsorption Isotherms Batch adsorption isotherm experiments were conducted for initial humic acid concentration from 10 to 60 mg/L at pH around 6.5 and temperature of 25oC. The adsorption isotherm results are analyzed by the linearized Langmuir equation (Figure 6.10) which was previously given in Chapter 2, i.e.: 140 Chapter 6 1 1 1 = + q qm K a qm C (2.11) where q is the amount of humic acid adsorbed per unit weight of granules at equilibrium concentration (mg/g), C is the final concentration (mg/L), qm is the maximum adsorption at momolayer coverage (mg/g), and Ka is the adsorption equilibrium constant (mL/mg). The numerical values of the Langmuir isotherm constants were calculated as qm = 0.407 mg humic acid/g granules, and Ka = 0.01757 L/mg humic acid. 30 Experiment 25 1/q (g/mg) Langmuir model 20 15 10 5 0 0 0.04 0.08 0.12 0.16 1/C (L/mg) Figure 6.10 Adsorption isotherms of humic acid on chitosan-immobilized PET granules according to the linearized Langmuir equation. The isotherm experimental values were also used in the linearized Freundlich equation (Figure 6.11), expressed in the following format: log q = 1 log C + log K f n (2.13) 141 Chapter 6 The Freundlich isotherm constants were determined to be Kf = 0.01055 and n = 1.32. -0.7 Experiment Freundlich Log q -0.9 -1.1 -1.3 -1.5 0.7 0.9 1.1 1.3 1.5 1.7 Log C Figure 6.11 Adsorption isotherms of humic acid on chitosan-immobilized PET granules according to the linearized Freundlich equation. 6.3.5 Adsorption Kinetics Since chitosan is vulnerable to dissolution in acid solution, we focused the study in the neutral and basic pH environments (initial pH = 6.50, 8.50, 10.01 and 12.01), as shown in Figure 6.12. At pH = 12.01, the adsorption was a very slow process and reached saturation in a few days. In contrast, the adsorption at lower pH values reached adsorption equilibrium in much shorter time, say only 1~2 hours. At the initial stage of adsorption, the solid surface of chitosan was bare, and all of the humic acid molecules that arrived at the surfaces can be assumed to immediately react with the protonated amino groups, -NH3+. At pH = 6.50, the ζ-potential of chitosan142 Chapter 6 immobilized granules was positive and the number of protonated amino sites was great enough. Hence, the adsorption rate can be dominated by the diffusion of molecules from the bulk solution to the surface. The mass transport can be interpreted as a Fickian diffusion. The absorbed amount of humic acid at the surface as a function of time can be expressed as Γ(t) = 2 π C 0 Dt (5.1) where Γ(t) is the amount of humic acid adsorbed per unit area of chitosan-immobilized granules at time t, C0 is the initial concentration of humic acid, and D is the diffusion coefficient. The specific surface area of the chitosan-immobilized granules in this experiment measured by a Quantachrome NOVA 1000 analyzer was 4.82 m2/g. Therefore, Eq. (5.1) can be converted to q(t) = 2 π C 0 S Dt (6.5) where S is the specific surface area. Figure 6.13 shows the plot of q(t) versus t1/2 for the experimental results. A linear relationship of q(t) against t1/2 is indeed observed at each pH condition from the beginning till almost the adsorption equilibrium. The results suggest the existence and the importance of diffusion-controlled transport mechanism in humic acid adsorption. Figure 6.13 also reveals the decreases of D as pH increases in the initial stage, according to Eq. (6.5). For humic acid molecules, the solution pH values and electrolyte concentrations will influence D values. A decrease in pH will increase D because the hydrodynamic volume of the humic acid molecules decreases (Avena et al., 1999). At pH = 12.01, a 143 Chapter 6 linear relationship of q(t) and t1/2 was observed during all the adsorption process. However, this phenomenon cannot be explained by assuming that the adsorption rate was dominated by mass transport and the small value of D was caused by swelling of the humic acid molecules at a high solution pH. Because the Stokes-Einstein equation may be applied to obtain the hydrodynamic radius, RH, of the humic acid molecules (Jones and Bryan, 1998) RH = kT 6πηD (6.6) where k is the Boltzmann constant, T is the temperature, and η is the viscosity of the aqueous solution. 0.08 Amount adsorbed (mg/g) 0.07 pH = 6.50 pH = 8.55 pH = 10.01 pH = 12.05 0.06 0.05 0.04 0.03 0.02 0.01 0 0 500 1000 1500 2000 2500 3000 Time (min) Figure 6.12 Adsorbed amounts of humic acid at different pH values, as a function of time (humic acid: 15mg/L, solution volume: 75 mL, quantity of granules: 10 g, initial solution pH: 6.50, 8.55, 10.01, and 12.05, temperature: 25 oC). 144 Chapter 6 0.08 Amount adsorbed q(t) (mg/g) 0.07 0.06 0.05 pH = 6.50 pH = 8.50 pH = 10.01 pH = 12.01 0.04 0.03 0.02 0.01 0 0 10 20 30 40 50 60 t1/2 Figure 6.13 Variation of the adsorbed amounts of humic acid, q(t), as a function of the square root of time t. If we assume that the adsorption rate was controlled by mass transport at all the pH values, from Figure 6.13, the diffusion coefficient D at pH = 6.50 will be around 560 times higher than the D at pH = 12.01, calculated according to Eq. (6.5). Therefore, the initial hydrodynamic radius of humic acid at pH = 12.01 should be 560 times higher than the hydrodynamic radius at pH = 6.50 on the basis of Eq. (6.6). Such an increase in molecule volume is impossible because the maximum variation in the volume with a variation in pH at 0.001 M electrolyte concentration was only a factor of 3 (Avena et al., 1999). Hence, other mechanism, in addition to the transport-limited mechanism, may be expected to play a role in the low adsorption rate of humic acid at pH = 12.01. 145 Chapter 6 As discussed in Chapter 5, two processes, i.e. mass transport and attachment, are important in the adsorption of humic acid on a solid surface. At pH = 12.01, the negative charges of the humic acids and the chitosan surfaces generate a strong electrostatic barrier for adsorption and the adsorption process may become attachment limited. Two reactions have been suggested as in Eqs. (6.3) and (6.4). Once a humic acid molecule is transported to the chitosan surface, the step in Eq. (6.4) can be a fast process if protonated –NH3+ is available. The protonation of –NH2 to –NH3+ however can be a slow process in humic acid attachment at higher pH values as the concentration of H+ is very low. Therefore it is reasonable to assume that the attachment process is controlled by reaction (6.3), and the kinetic expression may be given as d [ChH + ] d [Ch ] =− = k[Ch][H + ] dt dt (6.7) where [ChH+] represents the concentration of the protonated amino groups on the chitosan layer and [Ch] the concentration of amino groups on the chitosan layer available for protonation (i.e., the first term on the left hand side of Eq. (6.3)), at time t, and k is the reaction constant. The amount of humic acid adsorbed on the chitosan-immobilized granules, q (mg/g), may be given as q =α V ([A − ]0 − [A − ]) V ([Ch ]0 − [Ch ]) = w w (6.8) where [A-] (mg/L) stands for the concentration of humic acid in the solution at time t, and [A-]0 the initial humic acid concentration. V (L) and w (g) are the volume of the solution and the weight of the chitosan-immobilized granules in the solution, respectively, and α is a conversion constant. Assuming that [H+] is constant at pH = 12.01 because of the buffer effect of high alkaline concentration, one can obtain the following 146 Chapter 6 ln[(q e − q) / q e ] = − k ' t (6.9) by applying the initial condition of q(t) = 0 at t = 0 and the finial condition of q(t) = qe at t = ∞, where qe is the amount of humic acid adsorbed at adsorption equilibrium and k’ = k[H+]. 0.2 0.18 0.16 -ln((qe-q)/qe) 0.14 0.12 pH = 6.50 pH = 8.50 pH = 10.01 pH = 12.01 Eq.(6.9) 0.1 0.08 0.06 0.04 0.02 0 0 500 1000 1500 2000 2500 3000 Time (min) Figure 6.14 Variation of -ln((qe-q)/qe) as a function of time t. Figure 6.14 shows the variation of -ln((qe-q)/qe) as a function of time t. A straight line can be adequately fitted to the experimental data at pH = 12.01, except in the initial short period of time during which transport mechanism may be the sole controlling mechanism for the adsorption process. Hence, the adsorption dynamics of humic acid adsorption to the chitosan layer can be considered to be a transport-limited process under low pH 147 Chapter 6 conditions, but to be a process controlled by both transport-limited and attachment-limited mechanisms under high pH conditions. Although the SEM images showed that there were numerous open pores on the surface of the chitosan-immobilized granules, internal pore diffusions of humic acid did not appear to be a control mechanism for the adsorption, possibly due to the sizes of these open pores being significantly larger than the sizes of humic acid macromolecules. 6.4 Comparison of the three Prepared Granular Adsorbents Table 6.1 Some adsorption parameters of the three prepared granular media Adsorbents qm (mg/g) Ka (L/mg) PPy-coated glass beads 0.396 0.0117 PPy-coated nylon 6,6 granules 0.275 0.0152 Chitosan-immobilized PET granules 0.407 0.0176 The maximum adsorption quantity qm and the adsorption equilibrium constant Ka of the three surface-modified granular media for adsorption of humic acid, obtained from Langmuir model, are summarized in Table 6.1. It is seen in Table 6.1 that the maximum adsorption capacities of these adsorbents are not very different from each other. In fact, XPS results showed that the adsorption of humic acid by these adsorbents are chemosorption caused by the special site binding between the active sites between the humic acid molecules and the adsorbents. Therefore, the adsorption capacity is determined by the total amount of the active sites on adsorbent surface. For chitosanimmobilized PET granule, its adsorption capacities is nearly the same as that of PPy- 148 Chapter 6 coated glass beads. This is because that monomer size of chitosan molecule is smaller than PPy, which decreases its number of active site per surface area even though its specific surface area (4.82 m2/g) is larger than PPy-coated glass beads (1.46 m2/g). 6.5 Conclusions This Chapter describes the immobilization of chitosan on the nylon 6,6 or PET granules for use as a granular adsorbent. Nylon 6,6 granules were immersed in an acid solution and PET granules immersed in a base solution for hydrolysis to generate functional groups on the granular surface so that chitosan can be immobilized on it. The hydrolysis and immobilization processes were characterized using SEM, XPS and ζ-potential measurements. XPS results indicated that a salt structure was formed between the hydrolysed nylon 6,6 or PET surface and the chitosan layer. The chitosan layer thickness and thickness distribution were analyzed with an optical microscope, and the Lognormal distribution was found to better fit the experimental data than the Gaussian distribution. Batch adsorption of humic acid was carried out with the chitosan-immobilized PET granules. The process appeared to follow that the amino groups in the chitosan were first protonated and then reacted with humic acids. Decreasing the solution pH and increasing the ionic concentration were observed to increase the amounts of adsorption. The adsorption kinetics studies indicated that the initial rate of adsorption was transportlimited, especially under neutral or acidic conditions, and the adsorption rate was attachment-limited due to the high electrostatic repulse force between the chitosan surface and humic acid at high solution pH values. 149 Chapter 7 Chapter 7 Effect of pH on Surface and Adsorption Properties of PPy 7.1 Introduction When adsorption or filtration phenomenon in aqueous solutions is concerned, the surface properties of colloids or particles in contact with liquid phase are of significant importance. A solid surface may be charged and surrounded by the ions of opposite charge (counterions) to form the so-called electrical double layer (EDL) at the solid-liquid interface. The ions in the EDL may be described by the Stern-Gouy-Chapman model, as discussed in Chapter 2. Electrokinetic measurements have been theoretically established and widely utilized for the characterization of surface properties of particles or colloids in aqueous systems. One of the important surface properties is the ζ-potential which is the electrokinetic potential measured at a shear plane within the EDL. ζ-Potential can provide qualitative information related to the complex surface charges as well as the strength and polarity of EDL (Matsumoto, et al., 2001). Different mechanisms of surface charge formation may also be recognized by ζ-potential vs. pH plots at a constant ionic concentration (Schwarz et al., 1998). The analysis of the charging properties on the surface of amphoteric membrane has been extensively studied through ζ-potentials (Jimbo et al., 1999; Uematsu et al., 2002). Usually, semiquantitative methodology is applied to examine the dissociation behavior and the charge states of the membrane surface from the ζ-potential vs. pH profile. The theoretical model is based on the protonation and deprotonation of the surface acidic and 150 Chapter 7 basic groups (Matsumoto, et al., 2001). These surface properties are very important because membrane-fouling phenomena induce and alter the membrane surface charge due to adsorption of organic contaminants. In addition, the ζ-potentials of colloids and mineral grain surfaces are commonly used in quantitative and semiquantitative analyses of colloid deposition in granular porous media (Elimelech et al., 2000). Stephan and Chase (2001) demonstrated the possibility of using the ζ-potential as a predictive tool to analyze filtration performance in a deep bed filter without requiring complicated models and excessive experimentations. Although the adsorption of biomacromolecules onto PPy has been studied in recent years, and some of the surface interactions (e.g. van der Waals forces, Lewis acid-base forces and hydrophobic interactions) influencing the adsorption have been proposed (Abel and Chehimi, 1994; Minehan et al., 2001; Azioune et al., 2002), detailed study about the surface properties of PPy in contact with the aqueous phase has not been reported in literature. In addition, it has been known that the properties of PPy can be altered by treatment with aqueous bases and acids (Pei and Qian, 1991). By treatment with alkalies, PPy can undergo a deprotonation process, causing a significant decrease of the conductivity. The deprotonated PPy can also be reprotonated by strong protonic acids. As a result, the adsorption behavior of colloids onto PPy will be pH-dependent. Hence, it is important to understand the influence of pH on the surface properties of PPy in order to improve the understanding of the mechanisms affecting colloid or particle adsorption/deposition on the surface of PPy. 151 Chapter 7 The aim of this Chapter is to characterize the surface electric properties of chemically synthesized PPy using the ζ-potential technique and to investigate their influence on the adsorption of humic acid. The ζ-potentials of chloride-doped PPy (PPyCl) and deprotonated PPy powders against the solution pH values or against the acid or base treatment times were studied and used in the analysis of the chemical composition change of the surfaces. The apparent surface charge density was estimated from the ζ-potentials and the molecular structure of PPy. As an application, batch adsorption experiments of humic acid onto PPyCl-coated and deprotonated PPy-coated glass beads were investigated at different pH values to analyze the interactions of PPy with humic acid. Finally a specific site binding model is developed to simulate the adsorption of humic acid onto the PPyCl-coated glass beads at different solution pH values. In this model, adsorption was treated as a chemical reaction between surface sites and reactive groups of humic acid. Electrostatic effects were considered in characterizing the structure of the EDL at the PPy-water interface. 7.2 Experimental Section 7.2.1 Synthesis of PPy Powder and PPy-Coated Glass Beads 1.75 mL of pyrrole (0.025 mol) was added in droplets into 150 mL of FeCl3 aqueous solution dissolved with 13.5 g of FeCl3·6H2O (0.05 mol) at room temperature and with stirring. The mixture was then stirred for another 3 hours to allow polymerization of pyrrole to polypyrrole to proceed. The resulting black precipitates (PPyCl powder) were filtered, thoroughly washed with DI water, dried in vacuum desiccator for 24 hours. The 152 Chapter 7 elemental analysis (Perkin-Elmer 2400 elemental analyzer) of the prepared PPyCl showed a composition of (wt): C, 58.67%; H, 3.85%; N, 17.01%. The PPyCl powders were shaken in a NaOH solution (1.0 M) for 12 hours for deprotonation. The deprotonated PPy powder were washed with DI water until the washing liquid reached near a neutral pH value, dried in a vacuum desiccator for 24 hours. XPS analyses were performed to confirm no positively charged nitrogen atoms present in the deprotonated PPy powders. The PPyCl-coated glass beads, prepared according to the procedure given in Chapter 3, were shaken in a NaOH solution (1.0 M) for 12 hours for deprotonation. The deprotonated PPyCl-coated glass beads were washed with DI water until the washing liquid reached near a neutral pH value, and used for the adsorption experiments. 7.2.2 ζ-Potential Measurements To study the deprotonation behavior of PPy using ζ-potential measurements, the PPyCl powder was ground into fine particles and added in a reaction vessel with a 1.0 M NaOH solution. Similar experiments were conducted by adding deprotonated PPy fine particles in 1.0 M HCl solution for reprotonation. The mixture in the reaction vessel was shaken on a stirrer. At desired time intervals, some of the PPy fine particles were taken from the vessel, washed with large quantities of DI water until the washing liquid reached near a neutral pH value, and then dispersed in a NaCl solution (0.001 M, pH = 6.5) for ζpotential measurements. 153 Chapter 7 7.2.3 Adsorption Experiments Batch adsorption experiments using PPyCl and deprotonated PPy coated glass beads were performed according to the method given in Chapter 4. Each experiment used 10 g of coated glass beads in 75 mL humic acid solution (15 mg/L concentration in a 150 mL flask) at room temperature. Small volumes of HCl (0.1 M) and NaOH (0.1 M) solutions were added to adjust the initial solution pH values to a desired level. 7.3 Results and Discussion 7.3.1 ζ-Potential of PPy 100 PPyCl Deprotonated PPy 80 ζ-Potential (mV) 60 40 20 0 -20 -40 -60 -80 0 2 4 6 8 10 12 14 pH Figure 7.1 ζ-Potentials of PPyCl and deprotonated PPy as a function of pH in 0.001 M NaCl solution. The ζ-potentials of PPyCl are positive below pH 10, as shown in Figure 7.1. In general, 154 Chapter 7 the doping density of PPy which was chemically prepared using Fe(III) salts as an oxidant is about 0.25 to 0.3 per monomer unit, although considerable discrepancies in the chemical compositions of PPyCl still exist (Kang et al., 1991). In this study, XPS analysis was made to determine the doping density of PPyCl powders and the PPyCl coatings on the surface of the coated glass beads. The proportion of the positively charged nitrogen atoms was found to be about 26% (in terms of [N+]/[N], see Figure 7.2a). The positively charged nitrogen atoms in PPyCl can be expected to contribute to the positive ζ-potentials at neutral and acidic solution pH because the ionizable counteranions (Cl-) may transfer into the bulk solution, as shown below H N δ + n nδ X _ H N δ+ + n nδX _ (7.1) where δ (= 0.25 ~ 0.30) is the doping density, n is the average number of pyrrole monomers in PPyCl that contain one unit of positive charge (with n⋅δ = 1), and Xrepresents the counter anion, i.e. Cl-. The increase of pH will exponentially increase the OH- concentrations, and can lead to the selective adsorption of the OH- ions at the solid-water interface (Gu and Li, 2000). Consequently, the positive ζ-potentials of the PPyCl decreased at pH > 9 (see Figure 7.1) and the ζ-potentials eventually became negative with the increase of solution pH values. On the other hand, the increase of the H+ concentration in the solution with the decrease of solution pH value may cause the selective adsorption of the H+ ions at the interface. Thus, the ζ-potentials of PPyCl increased with the decrease of pH value, especially at pH < 3. 155 Chapter 7 Pei and Qian (1991) also proposed that PPyCl may be further protonated in a strong HCl as follows: Py+ H Cl _ Py + HH + _ 2 Cl + HCl (7.2) where ─Py+H─ represents a PPy chain segment of 3-4 pyrrole rings with a positive charged nitrogen atom. They considered that the reaction in Eq.(7.2) does not involve the conjugation system and the added proton should locate on the nitrogen atoms in the aromatic structure. Eq. (7.2) can also be used to explain the rapid increase of the ζpotentials obverted at pH < 3 in Figure 7.1. (b) INTENSITY (ARB. UNITS) (a) + [N ]/[N]=26% 396 398 400 402 404 396 398 400 402 404 BINDING ENERGY (eV) Figure 7.2 N1s core-level spectra of (a) PPyCl and (b) deprotonated PPy. 156 Chapter 7 As shown in Figure 7.1, the ζ-potential of deprotonated PPy is negative for pH above 4. This may be due to the presence of oxygen in PPy. Elemental analyses of PPy typically show empirical formulae like C4H3NXzOy, where X- is the dopant counterion and y ranges from 0.3 to 0.7 (Lei and Martin, 1992). As pointed in Chapter 6, the electronegativity of oxides on polymer surfaces may cause the flopped-down water dipoles and create a negative surface potential. Similar variations of the ζ-potentials with the pH values were reported for poly(L-lactide) which had an zero point of ζ-potential at pH 3.0 and had only carbonyl and hydroxyl groups that may be partially charged (but not ionized) (Makino et al., 1986). The protonation/deprotonation behavior of PPy has been schematically explained in Figure 2.6. XPS results showed the complete disappearance of the positively charged nitrogen atoms in deprotonated PPy (Figure 7.2b). In Figure 7.3 the ζ-potentials of PPyCl during the treatment with 1.0 M NaOH solution or depronated PPy during the treatment with 1.0 M HCl solution were plotted versus the treatment time. Since the transfer of H+ or OH- ions from the bulk solution to the solid-liquid interfaces and their adsorption onto the solid surfaces influence the ζ-potentials greatly, these measurements were all performed at a near neutral pH value (pH = 6.5) for comparison purpose. It can be seen from this figure that the protonation or deprotonation process was very fast at the initial stage, and the sign of the ζ-potentials reversed at around 12 second. After about 15 min treatment, the ζ-potentials of PPyCl treated by a base or the ζ-potentials of deprotonated PPy treated by an acid appeared to stabilize at a constant value. 157 Chapter 7 60 Reprotonation 40 20 ζ-Potential(mV) ζ-Potential (mV) 60 0 40 20 0 -20 -20 -40 0 0.25 0.5 Time (min) 0.75 1 -40 Deprotonation -60 0 5 10 15 20 25 Time (min) Figure 7.3 Variations of the ζ-potentials of PPyCl immersed in 1.0 M NaOH and deprotonated PPy immersed in 1.0 M HCl with time up to 24 min (ζ-potentials were measured in 0.001 M NaCl at pH = 6.5, the inset graph showing the initial stage.). 7.3.2 Evaluation of Surface Charge Densities The EDL model was applied to make the correlation between ζ-potential and surface charge density. For overall electrical neutrality within the whole of the EDL, there is σs + σd + σi = 0 (7.3) where σs is the surface charge per unit area on the solid surface, σi is the charge density of the adsorbed counterion i in the adsorption layer, and σd is the charge density of the counter ions in the diffusion layer. The diffusion charge, σd, can be calculated using the Gouy-Chapman theory (Hunter, 1991) 158 Chapter 7 σd = − zeψ d 2κkTε sinh 2kT ze (2.3) where ε is the dielectric permittivity of the aqueous solution, k is the Boltzman constant, T is the absolute temperature, z is the valency of the counterions, ψd is the potential at the interface between the adsorption layer and the diffusion layer (i.e at the Stern plane), and κ is the reciprocal of the EDL thickness which is defined in Eq. (2.2). The condition for Smoluchowski equation (Eq. (3.1)), i.e. κR0 >> 1, also ensures the applicability of the Gouy-Chapman theory (Kohler, 1993). At 25oC, κ (nm-1) is given by (Hunter, 1991) κ = 3.288 I (7.4) where I is the ionic strength ( = 1/2Σcizi2, ci (mol/L) is the concentration of ion i). For a 1:1 electrolyte solution, e.g. NaCl, with the concentration of c, σd (µC/cm2) is σ d = −11.74 c sinh(19.46ψ d ) (7.5) where ψd is in V and c in mol/L. Often, the quantity of ψd is approximated by the ζpotential at a low ionic concentration. Hence, at 25oC and with the ionic concentration of 0.001 M, the relationship between σs, σi and ζ can be derived from Eq. (7.3) and Eq. (7.5) as σ s + σ i = 0.3713sinh(19.46ζ ) (7.6) where σs, σi are in µC/cm2, and ζ in V. Applying Eq. (7.6), one can calculate the charge densities of σs + σi, for example, as 0.353 and –0.279 µC/cm2 for PPyCl and deprotonated PPy at pH = 6.5 with the ζ-potentials 159 Chapter 7 given in Figure 7.1. These charge densities indicate the extent of the surface charges originated from the oxidized or protenated nitrogen atoms in PPy counter-balanced by the dopant anions or the anions adsorbed in the adsorption layer from the solution. The positive value of 0.353 µC/cm2 for PPyCl at pH = 6.5 suggests that, at least, some of the dopant anions of Cl- in PPyCl dissolved into the solution, which led the PPyCl particles in the solution to have more positive charges from the oxidized or protonated nitrogen atoms than the negative charges from the attached Cl-. For deprotonated PPy at pH = 6.5, the negative value of –0.279 µC/cm2 may indicate that either the amount of oxidized or protonated nitrogen atoms in PPy was reduced or some of the anions from the bulk solution was adsorbed into the adsorption layer. The surface charge density, σs, may be further estimated from the molecular structure of PPyCl and the number of positively charged nitrogen atoms per unit surface area of PPyCl particles. The crystal structure of PPy has been discussed in Chapter 3. To simply the problem, an idealized layered structure of PPyCl with a = 9.8 Å and b = 7.35 Å is assumed in this study. Since the percentage of the positively charged nitrogen atoms was 26% (see Figure 7.2a), the idealized positive surface charge density originated from the oxidized or protonated nitrogen atoms is therefore determined to be 12.5 µC/cm2. The value of σi for PPyCl at pH = 6.5 can thus be calculated as 0.353-12.5 = -11.15 µC/cm2. This calculation indicates that most of the positive charges from the oxidized or protonated nitrogen atoms in PPyCl did not contribute to the positive ζ-potentials as they were counter-balanced by the anions (mainly Cl- at pH = 6.5) in the adsorption layer. For deprotonated PPy, its molecular structure indicates no free charges existing on the 160 Chapter 7 polymer surface, i.e. σs = 0, and thus σi can be given as σi = σs + σi = -0.279 µC/cm2. This suggests that the negative ζ-potential was most likely result from the adsorption of Cl- or OH- from the bulk solution. As discussed earlier, the electronegativity of oxides on polymer surfaces may make the hydrogen atoms of the water molecules near the polymer surface face the polymer surface, and this is called the flopped-down water dipoles. The flopped-down water dipoles may cause Cl- or OH- ions specifically adsorbed into the adsorption layer to counter-balance the electropositivity of hydrogen atoms adjacent the PPy surface. 7.3.3 Effect of Electrostatic Properties of PPy on Humic Acid Adsorption The adsorbed amounts of humic acid on PPyCl-coated glass beads and deprotonated PPycoated glass beads at different solution pH values are shown in Figure 7.4. Figure 7.4a indicates that the amounts of humic acid adsorbed onto the PPyCl surface increase with the decrease of solution pH values, and there was hardly any adsorption of humic acid at pH = 12.06. These results are consistent with the ζ-potential results in Figure 7.1. The amounts of humic acid adsorbed by the deprotonated PPy-coated glass beads, shown in Figure 7.4b, were significantly lower than those of PPyCl-coated glass beads at a similar solution pH value. This can be attributed to the lower ζ-potentials of deprotonated PPy than PPyCl in the solution. The electrostatic interactions between the deprotonated PPy-coated glass beads and the humic acid molecules were less attractive or even repulsive, and thus less humic acid was adsorbed. 161 Chapter 7 At pH = 4.05, 6.11 and 8.93, both humic acid and deprotonated PPy had negative ζpotentials, resulting in a possibily strong electrostatic repulsion for adsorption. However, it is found in Figure 7.4b that certain amounts of humic acid were still adsorbed on the deprotonated PPy-coated glass beads. This may be explained by the protonation/deprotonation behavior of PPy. The protonation-deprotonation equilibrium of PPy (see Figure 2.6) actually exists in the water system at any solution pH values. The surface charges of deprotonated PPy hence are heterogeneously distributed, with a small number of protonated nitrogen sites carrying positive charges even though the overall surface charges are negative in neutral and weak acidic/basic pH conditions. The positively charged sites on the PPy layer therefore provide favorable surface interactions for humic acid adsorption. After these positive sites are consumed due to adsorption reaction (Figure 2.6), new positive sites will be generated according to the protonationdeprotonation equilibrium. Since the concentration of H+ is much lower compared to the concentration of humic acid at pH 4.05, 6.11 and 8.93, the whole adsorption rate should be limited by the protonation process. As a result, the adsorption of humic using deprotonated PPy-coated glass beads is a much slower process in comparison with that of using PPyCl-coated glass beads. This is confirmed from the experimental results in Figure 7.4. It can be seen in Figure 7.4a that the adsorption reached saturation after around 600 min when the PPyCl-coated glass beads were used for the adsorption of humic acid. In contrast, a slow adsorption process continued even after 7200 min by using the deprotonated PPy-coated glass beads. 162 Chapter 7 pH = 1.89 pH = 6.55 pH = 12.06 0.16 0.14 pH = 4.03 pH = 8.22 (a) q (mg/g) 0.12 0.10 0.08 0.06 0.04 0.02 0.00 0 300 600 900 1200 1500 Time (min) 0.14 pH = 2.02 pH = 6.11 pH = 12.05 0.12 pH = 4.05 pH = 8.93 (b) q (mg/g) 0.1 0.08 0.06 0.04 0.02 0 0 1800 3600 Time (min) 5400 7200 Figure 7.4 Adsorption of humic acid onto (a) PPyCl-coated glass beads, and (b) deprotonated PPy-coated glass beads (humic acid: 15mg/L, solution volume: 75 mL, weight of coated glass beads: 10 g). 163 Chapter 7 7.3.4 Specific Site Binding Model 7.3.4.1 Theory The model described in this section is based on the following considerations: (1) Humic acid adsorption is through specific binding of the reactive sites of humic acid (carboxylic groups and phenolic groups) to the surface sites (N+) of PPyCl. The complexing molecules reside at the Stern plane. The total amounts of humic acid adsorbed are composed of those with sites bound to the PPyCl surface. (2) The electrostatic Gouy-Chapman equation is applied to calculate the charge densities and potentials of the EDL, and the Boltzmann distribution law is solved for the concentration of humic acid molecules in the EDL. (3) Charge balance and mass balance are considered to calculate the amounts of humic adsorbed. The binding reaction of a reactive site of humic acid (R-COO- or R-C6H4-O-, -C6H4stands for a benzene ring) with a positively charged site of PPyCl (N+) to form a neutral complex is A− + N + ↔ N + A− (7.7) where A- is a reactive site of humic acid and N+ is a positively charged site of PPyCl. The binding coefficient K for this process is given by K= [ N + A− ] [ N + ][ A − ] d (7.8) and the fraction of surface sites bound to specifically adsorbed humic acid, θ, is [ N + A− ] θ= [ N + ]0 (7.9) 164 Chapter 7 where [N+]0, [N+] and [N+A-] are the total concentration of positively charged nitrogen on PPy surface (mol/m2), the concentration of positively charged nitrogen at adsorption equilibrium (mol/m2), and the concentration of bound sites on the surface (mol/m2), respectively, and [A-]d (mol/L) is the reactive site concentration of humic acid at the Stern plane since it is assumed that the complexing molecules reside at the Stern plane. Some properties of the PPyCl-coated glass beads and humic acid are summarized in Table 7.1. Table 7.1 Properties of the adsorbents and adsorbates ———————————————————————————————————— PPyCl-coated glass beads Specific surface area (m2/g)a 1.46 Proportion of N+ ([N+]/N], %)b 26 Number of N+ (sites/m2)c 8.387x1017 Aldrich humic acid Number of carboxylic groups (mmol/g)d 4.2 Number of phenolic groups (mmol/g)d 2.3 ———————————————————————————————————— a measured by Quantachrome NOVA 1000 analyzer. b measured by XPS. c estimated according to XRD data, referring to Chapter 3. d Results are from Monteil-Rivera et al. (2000). 165 Chapter 7 Assuming that the maximum monolayer adsorption of humic acid on PPyCl surface equals to all reactive sites of humic (carboxylic and phenolic groups) bound to PPyCl surface and all the N+ sites of PPyCl occupied by the adsorbed humic acid, one can calculate that the maximum adsorption quantity (qmax) to be 0.313 mg/g. However, at solution pH = 6.5, the ionization degree of phenolic groups of humic acid is very low. If we only consider the carboxylic groups at pH = 6.5, the maximum adsorption quantity will be 0.484 mg/g. In comparison with the value qm = 0.396 mg/g obtained from Langmuir equation in Chapter 4, the calculation suggests that carboxylic groups and some of phenolic groups are bound to N+ sites at pH = 6.5. On the other hand, the results also demonstrate that the site binding model is reasonable. According to the Boltzmann distribution law, the concentration of ion i (mol/L), at a given position in an electric filed, ci(x), is given by ci ( x) = c0,i exp( − z i Fψ ( x) ) RT (7.10) where c0,i (mol/L) is the bulk concentration of ion i, zi is the charge of ion i (including sign), ψ(x) (V) is the electric potential at position x, and F (C/mol) is the Faraday’s constant. Following Eq.(7.10), one may similarly define the concentration of ionized groups of humic acid in solution at the Stern plane, [A-]d (mol/L), as [ A − ] d = [ A − ]∞ exp( Fψ d ) RT (7.11) where [A-]∞ (mol/L) is the ionized concentration of the ionized groups of humic acid in the bulk solution, ψd (V) is the potential at the Stern plane, and zi = 1. 166 Chapter 7 If the macromolecular feature of humic acid is considered, the space next to an adsorbing plane is subdivided into discrete lattice layers numbered z = 1, 2, 3, …(see Figure 7.5). Each lattice site has a number of neighboring sites, of which a fraction, λ0, is found in the same layer and a fraction, λ1, is found in each of the adjacent layers (Fleer and Scheutjens, 1993), so that λ0 + 2λ1 = 1 (7.12) z = -1 0 1 2 . . . . z-1 z z+1 Site of PPy Site of humic acid in solution Site with humic acid and PPy bound togerther Figure 7.5 Schematic illustration of lattice. Layers z ≤ 0 are in the adsorbent, layers z ≥ 1 are in the solution, and z = 0 is the surface of adsorbent. 167 Chapter 7 The adsorbed amount of humic acid is estimated to be the sum of the sites bound to the solid surface on one side and in contact with the bulk solution on the other side q= θ [ N + ]0 Ms S (1 + λ1 / λ 0 ) (7.13) where S (m2/g) is the specific surface area of the coated glass beads, and Ms is the number of reactive sites of humic acid per unit weight (mol/g). For adsorption from a dilute solution, λ0 has been shown to be 0.5 (Koopal, 1993). The mass balance of humic acid molecules is described as wtotal = qwg + V [ A − ]∞ α (7.14) where wtotal (g) is the total humic acid added to the solution, wg (g) is the quantity of coated glass beads in system, and V (L) is the volume of the system, and α is the ionization degree of the reactive groups of humic acid. Considering the surface charge of PPyCl neutralized by complexation with humic acid, the surface charge density at adsorption equilibrium is σ s = σ N + − F[ N + A− ] (7.15) where σs (C/m2) is the surface charge density of PPyCl, σN+ (C/m2) is the surface charge density originated from the positively charged nitrogen atoms in PPyCl, and the F (C/mol) is the Faraday’s constant. Humic acid is a much large molecule with a considerable number of negatively charged reactive groups. Therefore, specific adsorption of Cl- ions at the Stern plane at the 168 Chapter 7 adsorption equilibrium of humic acid may be neglected because of the significantly steric effect and electrostatic repulsion of humic acid. Hence, σ i = − Fα [ N + A − ] λ1 λ0 (7.16) where α is the ionization degree of the reactive groups of humic acid. If the thickness of a lattice layer at the Stern plane is d (m), the concentration of ionized humic acid groups at the Stern plane will be [ A − ]d = α d [ N + A− ] λ1 λ0 (7.17) Combining Eqs. (7.3), (7.16) and (7.17), we obtain − σ d = σ N + − F [ N + A − ] − Fα [ N + A − ] λ1 λ0 (7.18) 7.3.5.2 Computational Procedure and Model Results The following procedure is employed: (1) Assume an initial value of θ, for example, θ = 0.1. (2) Calculate [N+A-] using Eq. (7.9), and q using Eq. (7.13). (3) Use Eq. (7.14) to evaluate [A-]∞. (4) Use Eq. (7.18) to calculate σd, and then use Eq .(7.5) to calculate ψd. (5) Use Eq. (7.11) to calculate [A-]d. (6) Recalculate [N+A-] using Eq. (7.17), which gives a new value of θ. (7) Go back to step 2. 169 Chapter 7 For pH > 6.0, phenolic groups will ionize in solution (α = 0.002 for phenolic group of humic acid at pH = 6.0) and participate in the formation of surface complexes with PPy. Hence the values of θ, [N+A-], [A-]∞ and [A-]d of carboxylic groups and phenolic groups are assumed or calculated separately but that of σd and ψd are calculated as a whole for the carboxylic and phenolic groups. Table 7.2 Some parameters used in adsorption modeling ———————————————————————————————————— Initial quantity of humic acid (g) 1.125x10-3 Quantity of coated glass beads (g) 10 Volume of solution (L) 0.075 Background ionic concentration (M) 2.5x10-4 Lattice size (nm)a 0.4 pKa of humic acidb carboxylic group 2.9 phenolic group 8.8 ———————————————————————————————————— a Au et al. (1999) use this value to model the adsorption of Suwannee River humic acid to hematite surfaces. b Results from Kinniburgh et al. (1999) who estimated these values using NICCA- Donnan model. 170 Chapter 7 Simulation of the adsorption experiments of pH effect was carried out. Summaries of the experimental conditions and modeling parameters are given in Table 7.2. Table 7.3 shows the ionization degree of carboxylic and phenolic groups of humic acid in the solution, calculated according to their pKa values. It has been assumed that the medium effects can be negligible, and hence the ionization degree of humic acid adjacent to the adsorbed layer is the same as that in the bulk solution. Table 7.3 Ionization degree of humic acid Ionization degree of humic acid Solution pH Carboxylic group Phenolic group 1.89 0.089 0.000 3.12 0.624 0.000 4.03 0.931 0.000 5.1 0.994 0.000 6.55 1.000 0.006 8.22 1.000 0.208 12.06 1.000 1.000 The calculated amounts of adsorption as a function of solution pH values are given in Figure 7.6, together with corresponding experimental data. The agreement between experimental and predicted amounts is reasonably good under weak acidic, neutral, and basic conditions. For pH ≤ 4, the agreement however becomes much poorer with the 171 Chapter 7 decrease of solution pH although the model may still predict the qualitative trends, i.e. the adsorbed amount increases with the decrease of pH. Supplementary calculations showed that varying the parameters, such as λ and the total acidity of the humic acid, didn’t improve the model prediction noticeably. 0.14 Simulation Amount adsorbed (mg/g) 0.12 Experiment 0.1 0.08 0.06 0.04 0.02 0 0 2 4 6 8 10 12 14 pH Figure 7.6 Adsorbed amounts of humic acid as a function of pH at ionic concentration of 2.5x10-4 M: model calculations (solid curves) versus experiments (symbols). In Figure 7.1, the ζ-potentials of PPyCl are shown to increase with the decreases of solution pH values, especially for pH < 3.0, which is caused by the specific adsorption of H+ ions into the adsorption layer. Humic acid molecules may also be adsorbed into the adsorption layer due to the attractive electrostatic interactions, and these adsorbed humic acid molecules will contribute to the total amounts of humic acid adsorption. However, 172 Chapter 7 only the molecules with sites bound to the adsorbent surface are considered in the specific site binding model in this study. This may be the reason for the poor agreement between the model prediction and experimental results for pH < 4 in Figure 7.6. 7.4 Conclusions ζ-potential measurements were used to study the protonation/deprotonation behavior of PPy and its surface electrical properties. The zero points of ζ-potential of PPyCl and deprotonated PPy obtained in this study were in consistence with the reported pKa values of PPy for deprotonation and protonation reactions, respectively. Analysis of the ζpotential data revealed that the protonation or deprotonation process of PPy can take place fairly fast in the initial stage in a strong acid or strong base. The apparent surface charge densities on PPyCl and deprotonated PPy surfaces were estimated by the Gouy-Chapman theory. Evaluation of the apparent positive charge density shows that only a fraction of the positively charged nitrogen atoms on PPyCl surface contributed to the positive ζpotential of the chemically synthesized PPy. This may be caused by the specific adsorption of Cl-1, OH-, and H+ ions to the solid-liquid interface. The protonation or deprotonation reaction of PPy shows the pseudo first-order kinetics based on the analysis of ζ-potential data. A specific site binding model for the adsorption of humic acid onto PPyCl surfaces was presented. Calculation showed a good agreement of model prediction with the experimental data under weak acidic, neutral, and basic pH conditions, but the agreement was less satisfactory at solution pH < 4. 173 Chapter 8 Chapter 8 Conclusions Removal of suspended particles such as mineral grains and dissolved substances such as humic substances by granular adsorption and filtration plays an important role in water and wastewater treatment. Electrostatic interaction is one of the most important factors influencing the adsorption and filtration efficiencies, especially, when the substances to be removed carry surface charges. Removal of fine suspended particles and dissolved organic substances using conventional granular media is often ineffective, since the pollutants usually carry negative surface charges in the pH range of natural waters, and the conventional granular media used to remove these pollutants also carry negative charges in the same pH range. Although several studies have examined the possibility of modifying converntianal granular media to improve the removal efficiency, the results were often found to be unsatisfactory because the modified surface was not stable. In this study, surface-activated granular media were developed for effective adsorption or filtration for water and wastewater treatment. The research focused on modifying the surface of conventional granular media to obtain positive surface charges which are desired in removing many colloids or organic pollutants that carry negative surface charges in water and wastewater. Two types of granular media with positive surface charges were prepared: (1) glass beads and nylon 6,6 granules coated with PPy, (2) PET and nylon 6,6 granules coated with chitosan. 174 Chapter 8 PPy belongs to the novel class of intrinsically conducting polymers and displays many interesting properties. The application of PPy as adsorption and filtration materials in the field of water and wastewater treatment was firstly explored in this study. PPy-coated glass beads were first prepared by chemical polymerization of pyrrole on the surface of glass beads with FeCl3 as an oxidant. SEM and AFM studies revealed that rough and sponge-like surface morphology of the coated glass beads was obtained. ζ-Potential measurements showed that PPy-coated glass beads possessed high positive ζ-potential in the pH range of 2 to 10, which would promote strong ionic binding with negatively charged molecules or particle in water and wastewater. XPS analysis was applied to characterize the chemical compositions of the PPy coating and the chemical interactions between the PPy coating and the glass surface. It was found that about 28% of the nitrogen atoms in the PPy coating were protonated, resulting in a highly positively charged surface. In addition, strong chemical bonds were formed between the positively charged nitrogen atoms in the PPy coating and the Si-O- groups on the glass bead surface, which ensured the stability of the PPy-coated glass beads. Adsorption experiments of humic acid and filtration experiments of kaolin particles were conducted to examine the adsorption/filtration efficiency using PPy-coated glass beads as an adsorbent or filter medium. The coated glass beads were found to have significantly better removal of both humic acid and kaolin particles than uncoated glass beads. A layer of adsorbed humic acid or clusters of deposited kaolin particles on the surface of the PPycoated glass beads were clearly observed in the SEM imagines after the adsorption or filtration experiments. XPS results revealed that the positively charged nitrogen atoms on 175 Chapter 8 the PPy coating bound to the negatively charged groups of humic acid or kaolin particles, implying the existence of charge neutralization during the adsorption or deposition process. Solution pH and ion concentrations were also found to affect the determine adsorption/filtration performance. The results confirm the importance of electrostatic forces in adsorption/filtration processes. PPy was also coated on the surface of nylon 6,6 granules with (NH4)2S2O8 as an oxidant. The adsorption and filtration results were found to be similar to those using PPy-coated glass beads. This confirms that PPy can be used as a practical and efficient adsorption/filtration material, suitable for being use to modify the surfaces of both inorganic and organic (polymeric) granules. The adsorption kinetics of humic acid onto PPy-coated nylon 6,6 granules was investigated. It was shown again that electrostatic forces play an important role in adsorption process from the analysis of the adsorption kinetics. Chitosan is the other material used to modify the surface of conventional granular media in this study. Chitosan has been studied as an adsorbent for the removal of transition metal ions and organic compounds. Its usage in water and wastewater treatment, however, is limited due to its poor mechanical strength. In this study, chitosan was immobilized on the surfaces of PET granules or nylon 6,6 granules through partial hydrolysis treatment to overcome the poor mechanical property of chitosan. Measurements of ζ-potential indicated that the ζ-potentials of the chitosan-immobilized granules were positive at pH below 6.6. SEM imagines showed that the surface of the immobilized chitosan layer was 176 Chapter 8 porous. The reactions in the hydrolysis and immobilization processes were proposed with the information provided from XPS analyses. Batch adsorption of humic acid was carried out with the chitosan-immobilized PET granules. From the XPS studies, it was found that, under neutral conditions and after adsorption equilibrium, 15.3% of amino groups in the chitosan coating were protonated and reacted with humic acids. The decrease of the solution pH values and the increase of the ionic concentrations in the solution were observed to increase the amounts of adsorption. The adsorption kinetics studies indicated that, in the initial stage, adsorption was mainly transport-limited under neutral pH conditions, but adsorption became attachment-limited at later stages and especially at high solution pH. These results were well explained by the electrostatic properties of the granules and humic acid. To provide a better understanding of the adsorption/deposition processes of colloids onto the charged PPy surfaces, the surface properties of PPy in contact with a liquid phase were further examined. ζ-Potential measurements were used as a tool to study the electrochemical properties of PPy at the PPy-water interface and the protonation/deprotonation behavior of PPy. The zero points of ζ-potential for the PPyCl and deprotonated PPy obtained in this study were consistent with the reported pKa values of PPy for deprotonation and protonation processes, respectively. An evaluation of the apparent positive charge density showed that only a fraction of the positively charged nitrogen atoms on PPyCl surface contributed to the positive ζ-potentials of chemically synthesized PPy. This may be caused by the specific adsorption of Cl-1 and OH- ions at the solid-liquid interface. Batch adsorption experiments of humic acid onto deprotonated 177 Chapter 8 PPy-coated glass beads were conducted at various solution pH values. Although a strong electrostatic repulsion was expected to exist between the deprotonated PPy coatings and the humic acid molecules to be adsorbed under neutral and weak acidic/basic conditions, certain amounts of humic acid were found to adsorb on the deprotonated PPy-coated glass beads. The process of adsorption of humic acid onto the PPy was proposed to follow a sequence of reprotonation of the deprotonated PPy and then adsorption of humic acid to the protonated sites. A specific site binding model for adsorption of humic acid onto PPy surfaces was presented finally, based on the electrostatic theory and the specific complexation of humic acid functional groups with the surface active sites of PPy. Calculated results agreed satisfactorily with experimental data under weak acidic, neutral, and basic pH conditions. The research shows that surface charges of granular media for adsorption or filtration are crucial to the removal of charged organic substances and suspended particles or colloids in water and wastewater treatment. Adsorption/filtration efficiency can be significantly enhanced with surface-modified granular media. It is concluded that surface modification of conventional granular media, i.e. development of surface-activated (charged) materials has a great potential in water and wastewater treatment. Some recommends for the future work: This study focused on examining the possibility of modifying conventional granular media surfaces to enhance adsorption and filtration efficiency, and investigating the adsorption/filtration mechanism using prepared granular media with activated-surfaces. However, some potential factors which may influence the 178 Chapter 8 practical efficiency in drinking water treatment are not emphasized in this study. For example, there is little concern about the safety of the use of these prepared granular. In the future work, we need expand on the previous work to test the efficiency of these prepared granular media to remove dissolved organic matters and suspended particles from water over extended periods of time and under different pH conditions, and to examine the stability of these media. In addition, glass beads and polymer granules with uniform size and shape are used in this study because they are suitable media for adsorption and filtration model research. For economic and practical reasons, sand and anthracite are recommended to be used as the candidate substrate media in the future work. 179 [...]... 2 Adsorbents and filter media represent the hearts of adsorption and filtration devices All practical adsorbents have large specific surface areas and are therefore highly porous or composed of fine particles The old types of industrial adsorbents (e.g activated carbons and silica gels) are generally non-crystalline and their surface and pore structure therefore tend to be ill-defined and difficult... 1977; Loff, 1981) In water and 23 Chapter 2 wastewater treatment, the most commonly used filter media include silica sand (specific gravity =2.65), garnet sand (specific gravity = 4.0-4.4) and anthracite coal (specific gravity = 1.35-1.75) The effectiveness of granular adsorption and filtration processes largely relies on the surface interactions between the substances to be removed and the granular media. .. they effective to remove dissolved organic and inorganic substances This problem may be solved by surface modification of the traditional granular media to obtain the desired surface properties, i.e., positive surface charges 2.4.2 Surface Modification of Granular Media Several studies have examined the possibility of modifying granular media to improve their ability to remove dissolved matters and. .. earth and diatomaceous earth, granular activated carbon (GAC) has been by far the most widely used adsorbent which provides tertiary treatment for water contaminated with organic matters By the structure of materials, filter media may be classified into two types: flexible and rigid media (Cheremisinoff, 1995) Rigid filter media are commonly used for granular filtration Ceramic filter media for example... gas filtration, and in separation of dust and liquid droplets from gases (Loff, 1981) The relatively uniform particle size of diatomaceous media achieves high efficiency of filtration in retaining solid particles of sizes less than 1 µm, as well as certain types of bacteria Plastic granules (e.g polyvinyl chloride and nylon) have gained growing attention as filter media in recent years because of their... colloid adsorption and deposition at solid/liquid interfaces was given by Adamczyk and Weroński (1999) It was demonstrated in this review that the electrostatic interactions played the most important 13 Chapter 2 role in both adsorption (especially at the initial adsorption stages) and deposition of colloid particles 2.3 Adsorption and Granular Filtration Separation processes like adsorption and filtration. .. impregnation of coal with metallic hydroxides (Chaudhuri and Sattar, 1986; Lukasik et al., 1999), addition of positive charges to silica using organosilane derivative (Zerda et al., 1985), incorporation of metallic hydroxides onto the surfaces of sand using in situ precipitation of metallic hydroxides (Farrah and Preston, 1985; Lukasik et al., 1996), adsorption of metallic flocs onto the surfaces of sand (Edwards... cost and generates high volume of extra sludge (from filter washing) that is difficult to handle Adsorption by Activated Carbon Activated carbon is the sorbent material most widely applied in drinking water treatment Due to the composition of the raw material and the production process, between 5% and 20% by weight of activated carbon consists of elements other than carbon, primarily metals and surface. .. acidic and basic surface functional groups as well as in metal oxides As a result, the hydrophobicity of the ‘clean’ carbon surface is reduced and the conditions for adsorption of hydrophilic solutes are improved There are several physical and chemical parameters that affect the adsorption of humic substances Summers and Roberts (1986) showed that a positively charged carbon surface is favorable for the... process of filtration is quite dynamic and extremely complex In general, filtration process involves two sequential steps, i.e transportation and attachment (O’Melia, 1985) Particles in the suspension to be filtered are first transported from the bulk of the fluid to the vicinity of the stationary surface of the filter media by physical forces Then attachment of the particles to the collectors (filter media