Adsorption and desorption of cellulose derivatives

Cellulose derivatives, in particular carboxymethyl cellulose (CMC) are used in many (industrial) applications. The aim of this work is to obtain insight into the adsorption mechanism of cellulose derivatives on solid-liquid interfaces.In chapter 1 of this thesis we discuss some applications of cellulose derivatives. Application of CMC in pelleting of iron ore and in papermaking and the role of adsorption are given in more detail. Further we present a short introduction in the adsorption of polyelectrolytes.A set of 20 CMC samples was used in this study. Samples with four different degrees of substitution (ds=0.75, 0.91, 0.99, and 1.25) were prepared by AKZO Nobel by reaction of cellulose with NaOH and sodium monochloro acetate (ClCH 2 COONa). Samples were subjected to a random cleavage reaction with hydrogen peroxide yielding samples with molar masses ranging from M w =30 to 10 3kg mol -1. Characterisation of the CMC samples by size exclusion chromatography in combination with multi-angle laser light scattering (SEC-MALLS) and potentiometric titrations has been described in chapter 2 . Size exclusion chromatography separates molecules according to their size.The radius of gyration (R g ) and the molar mass (M w ) of each eluted fraction are then obtained on-line by multi-angle laser light scattering (via a Zimm-plot), yielding information about the molecular mass distribution of each sample. SEC-MALLS characterisation has been carried out at pH=7 in 0.02 and 0.1 mol l -1NaNO 3 . It turns out that the distribution depends on the salt concentration and the method of extrapolation of the scattered intensity to zero scattering angle in the Zimm-plot (i.e. using a linear or a non-linear extrapolation). Such a non-linearity is often attributed to the presence of aggregates. Because a CMC solution is supposed to contain more aggregates at high electrolyte concentration it is expected that the molar mass distribution will be shifted to higher molar mass when obtained in 0.1 mol l -1NaNO 3 in comparison with 0.02 mol l -1NaNO 3 . However, the experiments show the opposite trend, indicating that the lack of coincidence of distributions obtained at both salt concentrations is probably not caused by the presence of aggregates.Because SEC-MALLS gives both the molar mass and the radius of gyration of each eluted fraction, it is a highly suitable experimental technique to obtain the relation between M and R g . We applied the electrostatic wormlike chain model as well as Odijk's theory concerning the dimension of a polyelectrolyte to analyse this relation (both M and R g were obtained from the non-linear extrapolation method). A meaningful parameter in these models is the persistence length (L p ) of a polymer, L p characterising the length scale on which a polymer may be considered as rigid. The persistence length of a polyelectrolyte has two additive contributions. The first is the intrinsic or bare persistence length (L p0 ) which characterises the stiffness of the polymer backbone, the second accounts for the stretching of the chain due to electrostatic repulsion (electrostatic persistence length L pe ). Using the electrostatic wormlike chain theory, L p0 is assessed at 16 nm, indicating that CMC can be considered as a semiflexible polymer. A somewhat lower value (12 nm) has been obtained from the theory of Odijk. The value of L p0 does not depend on ds. The difference in L p0 between both models arises from the fact that in the Odijk model the contribution of L pe to L p is higher as compared to the electrostatic wormlike chain model. Furthermore the Odijk model assumes the chain as infinitely long. The electrostatic wormlike chain theory gives a more complete description of a polyelectrolyte chain as it takes molecular properties (such as the length and the cross-section of the molecule) and the details of the electrostatics into account.Potentiometric titrations were used to characterise the dissociation behaviour of CMC as a function of the NaCl concentration and pH. From the titration data the cross-section (radius) of CMC was obtained. Considering CMC as a uniformly charged cylinder radii of 0.95 nm (ds=0.75) up to 1.15 nm (ds=1.25) were obtained. Applying Katchalsky's theory for the dissociation of a polyelectrolyte, L p0 could be also determined from the titration data. In comparison to the analysis of the SEC-MALLS data Katchalsky's model gives a lower value (L p0 =5.9 nm). The difference is probably related to an incorrect evaluation of the electrostatic energy in Katchalsky's model.In chapter 3 the kinetics of polyelectrolyte adsorption has been investigated theoretically. Analogous to Kramers' rate theory for chemical reactions a model is presented which is based on the assumption that a polyelectrolyte encounters a barrier in its motion towards an adsorbing surface. The barrier is composed of the resistance due to transport in solution and to the presence of an electrical field. As soon as one segment touches the surface the chain is assumed to be adsorbed, i.e. the resistance that a chain encounters in the process of spreading out is neglected.We consider the motion of a strong polyelectrolyte with only one segment positioned at the front of the moving chain, all other segments are lagging behind the front segment. At each distance the chain explores all possible configurations, i.e. one needs to calculate the partition function of a chain with one segment at z=z* and all other segments at z > z* (Q(z*)). Such a partition function is readily evaluated from the numerical procedure proposed by Scheutjens and Fleer. Using this self-consistent-field (SCF) lattice model the resistance of an entering polyelectrolyte chain is calculated as a function of the distance from the surface. It turns out that the profile of the potential energy felt by the moving chain shows a strong resemblance with the interaction curve of colloidal particles, i.e. we observed a resemblance between the attachment process and the classical DLVO theory.Summing the contributions over the entire trajectory yields the barrier for adsorption R b . The barrier is calculated as a function of the adsorbed amount, and the results are inserted in the equation for the rate of the adsorption process. Finally, integration at a fixed concentration of polyelectrolyte leads to the time dependent adsorption. Endpoints in the calculated time dependent adsorption refer to equilibrium at that particular polyelectrolyte concentration.Parameters that affect the height of the barrier are the net charge at the interface (i.e. the surface charge plus the charge of the adsorbed polyelectrolyte), the charge density of the chain and the electrolyte concentration. Consider the adsorption of a polyelectrolyte on an oppositely charged surface. As long as the surface charge is not compensated there is no electrostatic barrier for adsorption, i.e. the rate of adsorption is determined by the rate of transport in solution to the surface. The height of the barrier strongly decreases with the electrolyte concentration. Consequently, the time needed to reach adsorption equilibrium also strongly depends on the electrolyte concentration. For low electrolyte concentration (0.01 mol l -1) an extremely long time is needed (15s), at a moderate concentration (0.2 mol l -1) it takes about 10 s. Hence, compared to the time scale of an experiment (around 10 5 s) adsorption equilibrium will not be accomplished for low electrolyte concentrations.The adsorption of carboxymethyl cellulose on rutile (TiO 2 ) and hematite (α-Fe 2 O 3 ) is discussed in chapter 4 . Data were obtained by batch adsorption experiments (depletion method) and by reflectometry, the latter yielding information about the kinetics of the adsorption. Systematically, we examined the influence of pH (pH=3 to 11), electrolyte concentration (c NaCl = 0.01 to 1 mol l -1) , molar mass (M w =35 to 1200 kg mol -1) and degree of substitution (ds=0.75 to 1.25). Adsorption isotherms are of the high affinity type and have well-defined plateau values. Plateau values in the adsorption decrease with increasing pH and increase with salt concentration. The adsorbed amount depends neither on ds nor on M w , the latter indicating a (rather) flat conformation of adsorbed CMC. The non-dependence on ds is possibly related to the fact that counter ions in the proximity of the polyelectrolyte chain lower the effective charge in such a way that CMCs varying in ds can have an identical effective charge density.On both surfaces a strong hysteresis in the adsorption with respect to pH is observed: in the high pH range a substantially higher adsorbed amount can be obtained by initially adsorbing at low pH and subsequently increasing the pH than by measuring the adsorption directly at any specified pH value. Desorption of CMC only takes place after the pH is increased substantially, which indicates a (very) strong interaction between CMC and the surface. Strong binding is likely related to the formation of ion pairs between the carboxylic groups of CMC and positively charged surface groups. Furthermore the desorption becomes even more difficult due to the chain rigidity of the CMC backbone, i.e. several bonds to the surface need to be broken simultaneously.Both the dissociation of the OH groups of the mineral surfaces and of the carboxylic groups of CMC depends on pH and electrolyte concentration. Furthermore the adsorption of a weak polyelectrolyte on such variable charged surfaces induces additional charges on the surface as well as on the polyelectrolyte. These characteristics cause the adsorption of a weak polyelectrolyte on a mineral surface to be very complicated. The model as presented in chapter 3 is used to elucidate this kind of adsorption. We calculated the adsorption and the charge of the surface at 10 5 s (a time which is comparable to the duration of an experiment). Because the short-range interaction between CMC and the surface is strong, the charge of adsorbed CMC can exceed the surface charge. The amount of overcompensation (or excess adsorbed chargeσ exc ) depends on the possibility that molecules reach the surface, i.e. on the height of the barrier for adsorption. As this barrier is a function of the net charge at the interface,σat a fixed electrolyte concentration does not depend on the pH. Increasing the electrolyte concentration lowers the barrier which allows higherσ exc . The calculations in chapter 4 show that at pH values where a weak polyelectrolyte is fully dissociated (i.e. acts as a strong polyelectrolyte) the adsorbed amount decreases linear with pH. Our experiments are in qualitative agreement with these calculations. The shape of the calculated time dependent adsorption curves also shows qualitative agreement with reflectometry experiments.In chapter 5 we discuss the adsorption of hydroxyethyl cellulose and quaternary ammonium substituted HEC (QNHEC) on silica and titanium dioxide. The adsorption has been investigated as a function of pH (pH=2 to 12) and electrolyte concentration (c NaCl =0.01 and 0.5 mol l -1) by means of reflectometry.The adsorption of HEC on SiO 2 shows a strong resemblance with the adsorption of polyethylene oxide. The adsorption is constant up to pH=5 in both 0.01 and 0.5 mol l -1NaCl, albeit in the latter case the adsorption is higher. At pH > 5 the adsorption decreases, which is most pronounced at the high salt concentration, reaching the level of zero adsorption at pH≈9. On TiO 2 the adsorption decreases monotonously with pH in 0.01 mol l -1NaCl. At high salt concentration it is constant up to pH=10, beyond which it decreases rapidly. The adsorption of HEC on SiO 2 is facilitated by hydrogen bonding between HEC ether groups and Si-OH surface groups, whilst the mechanism on TiO 2 is probably an interaction between non-substituted glucose hydroxyl groups and Ti-OH surface groups. The latter involves a chemical reaction, which may account for the fact that the time dependent adsorption of HEC on TiO 2 lacks a region where the adsorption increases linearly in time (i.e. mass transport in the solution is not the rate determining step even when the adsorption is low).Just as for CMC, in the adsorption of QNHEC there is an electrostatic barrier for adsorption. We compared the time dependent adsorption of QNHEC with calculations obtained from the model presented in chapter 3. It appears that in the case of QNHEC equilibrium is very likely not reached in 0.01 mol l -1NaCl, whereas in 0.5 mol l -1NaCl equilibrium is reached. As the charge density of QNHEC is lower (0.4 charged groups per glucose unit) than for CMC, the electrostatic barrier is also lower. In 0.01 mol l -1NaCl both on SiO 2 and TiO 2 the adsorption increases linearly with pH up to pH=10. This linearity is interpreted in analogous to the CMC adsorption. The adsorption reaches a maximum at pH≈12, then it decreased rapidly. According to the classification of van de Steeg the adsorption of QNHEC on SiO 2 in 0.5 mol l -1NaCl is of the screening-enhanced type up to pH ≈10, whereas at higher pH it is of the screening-reduced type. On TiO 2 the adsorbed amount is low and does not depend on pH.In chapter 6 the diffusion of spherical silica particles (with radii ranging from 12 to 510 nm) in dilute CMC solutions (M w =180 to 1200 kg mol -1, c CMC =5 to 1000 mg l -1) was investigated by means of dynamic light scattering. From the diffusion coefficient the viscosity as experienced by these inert probes (the "microscopic" or effective viscosity) is obtained. The smallest particles experience a viscosity which is slightly higher than the solvent viscosity, which may be interpreted in terms of the motion of these particles hardly being affected by the presence of polymer. The effect of polymer on the motion of the probes increases with the size of the probes. However, the value of the viscosity as obtained from capillary viscosimetry (bulk viscosity) is still not reached for the largest sphere, albeit for CMC M w =180 kg mol -1the effective viscosity comes rather close to the bulk viscosity.The thickness of the CMC/HEC layer adsorbed on Fe 2 O 3 /SiO 2 is also investigated in chapter 6. The layer thickness as obtained using the bulk viscosity shows a maximum as a function of the polymer concentration. The origin of the maximum is a consequence of an incorrect choice of the viscosity. Using the viscosity as obtained from the inert probe diffusion the layer thickness increases monotonously with polymer concentration.The diffusion behaviour of the inert probes is discussed in terms of a model in which the particles are surrounded by a layer of polymer free solution. This layer is assumed to be equal to the thickness of the depletion layer. According to this model the thickness of the depletion layer decreases with the CMC concentration, at low concentration approaching the radius of gyration of CMC.