Manufacturing - Processing MECHANISTIC STUDIES OF CANE MUD FLOCCULATION E. Whayman and 0. L. Crees I Sugar Research Institute, Australia ABSTRACT A series of partially hydrolysed polyacrylamide flocculants was evaluated in batch settling tests on limed cane juice. The dependence of settling rate on polymer molecular weight was determined, together with the variation of optimum flocculant chemical composition with particle zeta potential. It was found that adsorption occurs through the acrylate group in the copolymers and the experimental results are interpreted in this light. INTRODUCTION The use of polymeric materials to aid clarification is now a familiar part of raw sugar manufacture. In recent years, their importance in increasing liquidsolid separation equipment capacity and reducing sucrose losses in filtration has become well established whilst the number of materials available has increased enormously. Of the bewildering array of materials confronting consumers in the raw sugar industry, the partially hydrolysed polyacrylamides (Fig. 1) have proved the most successful. Even so, there is frequently a need for considerable guesswork in selecting flocculants from within this category. The results presented later represent part of a systematic investigation commenced in 1971 aimed at eliminating this guesswork through an understanding of the mode of action of these flocculants and the factors controlling their effectiveness. 1 ooy 1.i Degree of hydrolysis = ---% FIGURE 1. Structure of Partially (X f Y) -, Hydrolysed Polyacrylamides. According to the established theory of colloid stability, small particles are attracted to one another by van der Waals forces and therefore always tend to aggregate unless kept apart by electrostatic repulsive forces arising from the presence of electrical charges on the particles. The aggregation process is aided by the addition of simple electrolytes which reduce the electrostatic repulsion; this process is commonly called "coagulation". Aggregation can also be brought about by various polymeric substances which bind the particles together, with the chain molecules forming interparticle bridges. Following La Mer's suggestion, the term "flocculation" refers specifically to the latter process, and Fig. 2 shows the basic difference between the two phenomena.
MANUFACTURING - PROCESSING: FIGURE 3. Particle Coated with Flocculant. The bridging theory of flocculation was first suggested by Ruehrwein and Ward.2 The theory postulates that the polymer molecules attach themselves to the surface of the suspended particles at one or more adsorption sites, and that part of the chain extends out into the bulk of the solution. When these extended chain segments make contact with vacant adsorption sites on other particles, bridges are formed. The particles are thus bound into small Aocs which can grow to a size limited by the degree of agitation and the amount of polymer initially adsorbed on the particle surfaces. If too many adsorption sites are occupied, bridging will be hindered, and wholly inhibited if all are occupied (Fig. 3). If too few sites are occupied, bridging may be too weak to withstand the shearing forces imposed by even mild agitation. Flocculation is therefore a two-stage reaction comprising an initial primary floc formation step followed by a second growth step. For a polymer to be a successful flocculant it must have some means available for adsorption of particles and it must also be large enough to bridge between them. Materials of molecular weights less than lo6 generally are not large enough for bridging and tend to act as stabilisers. It has already been shown that molecular weights of the order of 10' are necessary for good clarification of cane juices and that lower molecular weights produce changes in the observed pattern of behavio~r.~ With partially hydrolysed polyacrylamides there are two possible sites for adsorption. Adsorption through the amide group by a hydrogen bond has been postulated in some systems and, in fact, Groit and Kitchener4 have shown that this occurs in the adsorption of polyacrylamide onto colloidal silica. It has also been suggested6 that, in such a case, the inclusion of ionic groups along the polymer molecule serves only to provide electrostatic repulsion thus creating molecular extension and facilitating bridging. Adsorption through the acrylate group may seem unlikely since most natural colloids, including cane muds,3 are negatively charged. However, Davies and RidealG cite examples where cation bridges particularly through calcium, are responsible for bonding clay and carbon black particles onto carboxylate groups on fibres. H. G. Bungenberg de Jong7 has shown that similar, though weaker, calcium bridges are formed between the carboxylates on proteins and those on other polymers. Furthermore, Bennett8 provides evidence that cane mud particles behave as though they are coated with denatured pro-
1 1 E. whayman AND 0. L. CREES 1177 teins and acidic polysaccharides. Favourable conditions therefore exist for the formation of calcium bridges between cane mud particles and partially hydropolyacrylamides. This paper presents evidence that the acrylate group is the active site for adsorption of partially hydrolysed polyacrylamides onto cane mud particles, and experimental results are interpreted in this light. 1i I EXPERIMENTAL Partially hydrolysed polyacrylamides were especially produced for this study by the Research Laboratory of BTI Chemicals Ltd, Bradford (now a wholly owned subsidiary of the American Cyanamid Company) and analysed at Sugar Research Institute as described elsewhere.3 The flocculants used in the work had molecular weights of 4 x lo6 (referred to as low molecular weight) and 12 x lo6 (high molecular weight). At each molecular weight, the degree of hydrolyses varied in 5% steps from 10 to 50%. Flocculants were evaluated in batch tests in which settling rates and also optical densities at 800 nm were determined. Calibrated tubes with 5,5 cm diameter and 40 cm settling depth were used, with the equipment and techniques described previously.3 Unless specifically stated otherwise, all tests were conducted at a dosage rate of 2 ppm flocculant on juice. In any test series, the highest initial mud settling rate was designated as 100% and other rates were compared to this on a percentage "relative efficiency" basis. Flocculant Adsorption Due to difficulties in determining very low concentrations of flocculant in cane juices, an inferential method was used to determine the extent of flocculant adsorption by the mud suspensions. Part of a sample of limed cane juice was centrifuged at 900G for 15 min in a 4L Mistral to remove suspended matter. Two sub-samples were taken from this clean juice, 2 ppm of a high molecular weight non-ionic polyacrylamide added to one, 2 ppm of a 50% hydrolysed flocculant added to the other, and the samples centrifuged again. Sub-samples of the uncentrifuged limed juice were treated with 2 ppm of the same flocculants, settled for half an hour, then the supernatant solutions were centrifuged to remove any residual suspended matter. Filterabilities were determined on the two flocculated settled samples, the clear juices with added flocculant, and the blank. The method used was the standard CSR sugar filterability test,g modified for juices by first bringing the brix up to- 60 with refined sugar. RESULTS en reported in a previous paper3 that there is a characteristic pattern of flocculant efficiencies through the range of hydrolyses. As Fig. 4 illustrates, there is an optimum degree of hydrolysis corresponding to a maximum in both settling rate and juice clarity. It is also characteristic that the clarity remains high over a considerable range of hydrolyses above the optimum.
1178 MANUFACTURING - PROCESSING In the same paper it was also reported that the optimum degree of hydrolysis is different in different mill areas. Fig. 5 indicates that the optimum hydrolysis increases with increasing zeta potential. The effect of molecular weight was previously illustrated as being approximately linear. This is not strictly true, as subsequent work has shown that there is also a strong dependence on degree of hydrolysis. It is apparent from Fig. 6 that molecular weight is much less critical at high than at low hydrolyses. Degree of hydrolysis (%) FIGURE 4. Effect of Hydrolysis on Settling Rate and Clarity. 20 25 30 35 40 FIGURE 5. Effect ofhydrolysis and Degree of hydrolysis (%) Zeta Potential on Settling Rate. 0 1... 0 I FIGURE 6. Effect of Molecular 15 20 25 30. 35 40 Weight and Hydrolysis on Settling Degree of hydrolysis Rate.
E. WHAYMAN AND O. L. CREES 1179 drolysis at lower molecular Fig. 7 shows that at hydrolyses well above the optimum, i.e. 40%, it is possible to compensate for low molecular weight by slightly increasing the dose. The dose needed increases as the hydrolysis decreases until, below the optimum, i.e. 20%, no dose will compensate for low molecular weight. ----- Ppm of flocculant 8 FIGURE 7. Dosage Effect on Settling Rate. Although the optimum flocculant for settling is characteristically also the optimum for clarity, there were occasions where this was not so. On these igher hydrolysis, as Fig. 9 The effect of higher doses on high molecular weight materials is shown in Fig. 8. Although an increase in dose at both 25 and 50% hydrolysis will increase the settling rate considerably, even 6 ppm of 50% hydrolysis material will not match 2 ppm of a 25% hydrolysis polymer. Adsorption measurements Typical results of the "filterabilities" of the treated.juices are shown in Table 1. It can be seen that the marked depression caused by presence of flocculant in clear juice is still apparent when the non-ionic material is tried as a flocculant. Flocculation in fact does not occur in this case, and if this were due to excessively strong adsorption, then little or no residual material would be expected in the clear centrate. In the case of the 50% hydrolysed material however, good flocculation was apparent and little polymer was left in the clear supernatant. 50% HMW Flocculant dose (ppm) FIGURE 8. Dosage Effect on Settling Rate.
1180 MANUFACTURING - PROCESSING 0 1 0, 1 0 10 20 30 40 50 FIGURE 9. Clarity Offset from Sett- Degree of hydrolysis (%) ling Peak. TABLE 1. Treatment Filterability Centrifuged 65 Centrifuged - 2 ppm non-ionic - Centrifuged 16 Centrifuged - 2 ppm 50% hydrolysis - centrifuged 28 Flocculated - 2 ppm non-ionic - centrifuged 16 Flocculated - 2 ppm 50% hydrolysis - centrifuged 51 DISCUSSION From the results in Table 1 it appears that the acrylate group is the active site on the polymer for particle adsorption. Such a mechanism provides an adequate explanation of the results presented here and satisfies the scheme for bridging flocculation outlined earlier. In the characteristic flocullant efficiency curve (Fig. 4), the optimum degree of hydrolysis represents a balance between the processes of adsorption and growth. Enough adsorption sites on both the particles and the polymer chains are occupied to provide sufficiently strong bonding to resist agitation and to scavenge the maximum number of particles in the primary "micro floc" stage. The polymer however has not collapsed onto the particle surface, and sufficient loops or tendrils with unoccupied sites project into the liquid phase for maximum "macro floc" growth. Below this hydrolysis, bonding is weaker and insufficient to scavenge all particles, thus giving poorer settling rate and clarity. On the other hand, at hydrolyses above the optimum all the particles are scavenged and the clarity remains high. However, a large number of bonding sites are occupied in the initial step so the ability for secondary growth is reduced. Flocs are therefore smaller and there is a corresponding decrease in settling rate. This optimum bond strength concept is supported by the results shown in Fig. 8 where well above the optimum hydrolysis, higher doses will hot improve the settling rate to match the performance of the optimum flocculant. It has also been found that, below the optimum, higher doses will not improve the clarity sufficiently to approach the maximum obtained at the peak. Although the acrylate group is the active site for adsorption, there is the possibility that it also functions as a means of extending the molecule in solution. However, preliminary studiesl0 of polymer viscosity in solutions with ionic
strengths similar to that of cane juice suggest that this does not occur with flocculants of molecular weights approaching 10. The requirement of higher hydrolysis polymers for optimum flocculation as the zeta potential increases (Fig. 5) is surprising at first sight. It is easier to envisage the required hydrolysis decreasing, so that total repulsion of polymer from the particle surface does not occur. Deeper consideration of the situation however leads to two possible explanations. Using the extension concept, it appears feasible that as the interparticle repulsion increases with increasing zeta potential, the mean distance apart of the particles also becomes greater. Therefore for interparticle bridging to occur, a more extended polymer is required, i.e. one of higher hydrolysis. This does not however explain the marked fall off in performance above the peak, as over-extension is unlikely to inhibit flocculation. With the alternative optimum bond strength concept it appears that as the overall repulsion between particle and polymer increases (or attraction decreases), a greater number of the individual weak bonds is required to maintain the required adhesion between chain and surface. From this viewpoint the increase in the optimum hydrolysis with reduced molecular weight (Fig. 6) would be due to the well-known reduction in extent of adsorption with reducing molecular weight. The relative insensitivity of flocculation to molecular weight variations above the optimum hydrolysis also shown in Fig. 6, means that chain length is no longer as important in the overbonded condition. Extensive secondary macrofloc building cannot occur even with very long chain polymers, presumably due to collapse of potential bridging loops onto the particle surface. Under these conditions increased dosing of low molecular weight flocculants can make up for chain length Fig. 7, though this never occurs with materials near the hydrolysis peak. The offset of clarity and settling peaks illustrated in Fig. 9, is thought to be due to variations in charge distribution in the particle suspension. Bennett's evidence8 that cane mud particles behave as if coated with a proteinacidic polysaccharide complex suggests not only that there can be variations in the nature of the surface coating but also that the charge distribution can vary widely. If sufficient inorganic (e.g. calcium phosphate) surface area is available, then the particle charge distribution may be expected to be comparatively narrow. With restricted adsorbent area, unadsorbed material may have 0-1 - 2-3 - 4 Particle charge (mv) FIGURE 10. Surface Charge Distributions.
1182 MANUFACTURING - PROCESSING higher surface charge than the coated particles, extending the charge distribution as shown in Fig. 10. The optimum flocculant for settling is that scavenging and bridging the bulk of the particles, but this will necessarily leave a haze of the high charge tail in suspension due to weak bonding. Practically the solution to this is to sacrifice settling velocity for clarity by using a higher hydrolysis flocculant, or to provide increased adsorbent area during processing. CONCLUSIONS The range of acrylate-acrylamide copolymers used in this work gave flocculation results that fitted reproducible patterns. This is most encouraging in that flocculant selection for the sugar industry need no longer be entirely trial and error or guesswork. Though the adsorption site for polymer-particle bonding seems certain to be the acrylate ion, the role of chain extension and the effect of percentage hydrolysis on molecular dimensions is still unproved. Further work is required on the configuration of these polymers in environments like cane juice, and this is planned for the near future. ACKNOWLEDGMENTS Thanks are due to Mr A. L. Willersdorf of the Sugar Research Institute for polymer analyses and general assistance with the work programme. The invaluable aid of BTI Chemicals Ltd in providing the test polymers is also gratefully acknowledged. REFERENCES 1. La Mer, V. K. (1964). J. Colloid Sci, 19 :291. 2. Ruehrwein, R. A. and Ward, D. W. (1952). Soil Sci, 73:485. 3. Crees, 0. L., Hale, D. J., Whayman, E. and Willersdorf, A. L. (1973). Proc QSSCT, 40:239. 4. Groit, 0. and Kitchener, J. A. (1965). Trans Faraday Soc, 61 :1026. 5. Michaels, A. S. (1954). Ind Eng Chem, 46:1485. 6. Davies, J. T. and Rideal, E. K. (1961). Interfacial Phenomena, Academic Press, London, p 430. 7. Bungenberg de Jong, H. G. (1949). "Colloid Science", Vol I1 ed Kruyt, H. R., Elsevier, Amsterdam. 8. Bennett, M. C. (1957). Int Sug J, 59:176. 9. Nicholson, R. I. and Horsley, M. (1956). Proc ISSCT, 9:271. 10. Crees, 0. L. and Whayman, E., unpublished data. ' 41 ESTUDIOS MECANISTICOS D&, FLOCULACION DE LA CACHAZA E. WHAYMAN Y 0. L. CREES ;' RESUMEN Una serie de floculantes de poliacrilamida parcialmente hidrolizada fue evaluada en pruebas de decantacidn por carga de jugo de caaa encalado. Se determind la dependencia de la velocidad de decantacidn con el peso molecular del polimero, junto con la variacidn de la dptima composicidn quimica del floculante con el potencial zeta de las particulas. Se encontrd que la adsorcidn ocurre a trav6s del grupo acrilato en 10s copolimeros, y 10s resultados de 10s experimentos se interpretan de acuerdo con este fendmeno. d<