Fractal theory study on morphological dependence of viscosity of semisolid slurries R. S. Qin and Z. Fan A theoretical model is developed which enables the effect of particle morphology on the viscosity of semisolid slurries to be assessed. Hausdroff dimensionality is introduced as a means of characterising the morphology of a solid particle. The effective solid volume fraction, defined as the volume fraction of actual solid plus the entrapped liquid, is derived analytically. The viscosity of semisolid slurries is calculated by the adaptation of Thomas’s empirical approximation. The viscosity of a semisolid slurry is found to be a non-linear function of the Hausdroff dimensionality. Slurries with particle morphologies of lower Hausdroff dimensionality have a higher viscosity for a given solid fraction. The external variables, such as shear rate and cooling rate, affect the flow behaviour of semisolid slurries primarily by changing the Hausdroff dimensionality and the linear size of solid particles. The effect of particle size on the viscosity is also discussed. MST/4694 At the time the work was carried out the authors were in the Department of Materials Engineering, Brunel University, Uxbridge, Middx UB8 3PH, UK. Dr Qin ([email protected]) is now in the Department of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, Cambridge, CB2 3QZ, UK. Manuscript received 9 May 2000; accepted 27 November 2000. # 2001 IoM Communications Ltd. Introduction Semisolid slurry (at a solid fraction less than 0.6) is a special kind of solid – liquid suspension. Its main characteristics are the morphological complexity of the solid phase and a changeable degree of agglomeration among the solid particles. As a result of the environment sensitive morphology of the solid particles and a stirring dependent agglomeration reaction among the particles, semisolid slurry behaves as a history dependent and non-Newtonian fluid.1 There has been quite a lot of theoretical work which has addressed the viscosity of semisolid slurries during the last two decades.2 – 4 However, the physical understanding of experimental evidence is still a daunting challenge. The difficulties in modelling the viscosity of semisolid slurries are manifold. The bottleneck is how to characterise the morphology of the solid particle. As is well known, the shape of a solid particle may be in the form of a dendrite, rosette, or spheroid, depending on environmental conditions.5 Furthermore, the subdivisional characteristics of dendrites are also environment sensitive according to pattern formation theory.6 Neglecting the morphological detail of the solid phase makes it impossible to understand the whole flow behaviour of the semisolid slurry. The literature on this topic is comprised of two different types of contribution. The first focuses only on the effect of agglomeration, such as in the models proposed by Mada et al.2 and Kumar et al.,1 where the effect of particle morphology is neglected. The solid particles are assumed to be dense spheroids in those models. The second class of contribution does include some effects of particle structure, but these works do not concern themselves with the method of characterising the morphology of the solid particle. For example, Quemada7 considered the contribution of the morphology using the concept of the effective solid volume fraction. However, the model introduced an immeasurable structural parameter S and did not specify the method to determine its value. The research in this aspect has stayed at a stage of phenomenological and qualitative description. On the other hand, the highly developed fractal geometry has proved to be a powerful method by which to describe irregular patterns.8 The evolution of the particle morphology in semisolid slurries is from dendrite via rosette to more or less spheroid as the shear rate increases.5 ISSN 0267 – 0836 In most cases the geometry of the solid phase is irregular. By noting that the dendritic form is a typical fractal structure6 and that the solid – liquid interface morphology in directional solidification can be described quantitatively by the fractal method,8 one may think that the irregular geometry of the particle morphology in a semisolid slurry can be quantitatively characterised by fractal theory.9 The Hausdroff dimensionality of the solid particle can be measured by many different methods, such as slit island analysis,8 profile analysis,10 direct surface area analysis,11 and variation-correlation analysis.12 The aim of the present work is to illustrate how the viscosity is affected by the particle morphology by means of the fractal theory. Theoretical considerations For a general solid – liquid suspension, the viscosity is a function of the fraction of solid. A formula which has been obtained from regression analysis of the empirical data of Thomas13 can be expressed in the following form m ~m=m ~1z2:5Wz10:05W2 zA exp(BW) : : : (1) 0 where m* is the relative or dimensionless viscosity, m is the viscosity of the solid – liquid suspension, and m0 is the viscosity of the pure liquid, W is the volume fraction of solid in the suspension, and A and B are constants which are found to be 0.00273 and 16.6 respectively. The last two terms in equation (1) represent the effect of particle agglomeration (doublet, triplet, and so on) due to particle collision. Equation (1) is suitable for solid – liquid suspensions where the morphology of the solid particle is spheroid. For semisolid slurries, the solid particle morphology is not always so. Figure 1a is a possible example of a semisolid slurry where the morphology of the solid particle is dendritic. However, when the suspension is sheared, the non-spheroidal particle will rotate and the entrapped liquid will be forced to move together with the solid particle. The result is that the entrapped liquid plays the role of a solid. The solid particles with trapped liquid contained in them are approximately spherical in morphology as illustrated in Fig. 1b. The solid plus the entrapped liquid is defined as the effective solid. Replacing the solid fraction W by the effective solid fraction Weff leads to a condition in Materials Science and Technology September 2001 Vol. 17 1149 1150 Qin and Fan Fractal theory study on viscosity of semisolid slurries (3{DH )=DH N fs3=DH : Q Weff ~k(3)k(DH ){3=DH : : (5) Equation (5) shows that the effective solid fraction is affected by the dimensionless factor Q/N, the real solid fraction fs, and the Hausdroff dimensionality DH. In the extreme case of DH~3, which means no liquid is entrapped in the solid, equation (5) reduces to Weff ~fs a real (dendritic) solid particle; b effective solid particle 1 Schematic diagram illustrating concept of effective solid fraction which equation (1) can be used to describe the viscosity of semisolid slurries. The task now is to derive a expression for Weff. Assume that the considered system contains N atoms. Neglecting volume shrinkage during solidification, the solidified atom number M at a temperature T is given by M~fs N : : : : : : : : : : : : : : : (2) where fs is the solid fraction which is a function of temperature and can be obtained from consulting the phase diagram. The M atoms are assumed to form Q particles of the same size l. The length unit in this work is defined as the atomic diameter. The actual volume of a solid particle equals the number of atoms in the considered particle. The effective solid fraction in such a length unit is obviously pQl 3 Weff ~ : 6N : : : : : : : : : : : : : (3) Here one uses the Hausdroff dimensionality DH to characterise the morphology of the solid particles.8 DH is sensitive to the external variables, such as the shear rate and the cooling rate. Numerical simulation shows that DH~2.5 for a particle formed by a diffusion controlled mechanism which corresponds to a zero shear rate.14 Also, DH~3 for spheroids which formed at a very high shear rate. The analytical relationship between the Hausdroff dimensionality and the external variables is not available in the literature at present, but it is known that the higher the shear rate (or the lower the cooling rate) the larger the Hausdroff dimensionality. For a particle of Hausdroff dimension DH, the volume V(l) of the region bounded by the solid/liquid interface scales with the increasing linear size l of the particle in a non-trivial way15,16 V(l)~k(DH )l DH where k(DH )~pDH =2 V(l)~ fs N Q : : : : : : : DH ! 2DH | 2 : : : : : : : : : : : : (4) : : : : : : (4a) : : : : : : (4b) The factorials in equation (4a) can be calculated approximately by using Stirling’s formula17 DH =2 DH DH !~exp(DH =2) (pDH )1=2 : : : : (4c) 2 2 Bringing equation (4) and (4b) into (3) and noting that k(3)~p/6, the effective solid fraction takes the following expression Materials Science and Technology September 2001 Vol. 17 : : : : : : : : : : : : : (5a) : The dimensionless factor Q/N has the indirect physical meaning of the number of solid particles per unit ‘volume’. This factor can be calculated approximately according to the physical and mathematical assumptions in the treatment of Avrami.18 The essence of the assumptions in Avrami’s treatment arises from the following:19 (a) as time proceeds during particle growth, some grains grown from a site active at various times in the past, may impinge upon grains grown from other sites, and (b) active nucleation sites are randomly distributed throughout the volume. Therefore, one gets the following formula to estimate Q/N Q=N~v=d 3 : : : : : : : : : : : : : (6) where v is the volume of an atom and d is the average grain size when the solidification is completed. Q/N can also be measured experimentally by counting the number of solid particles using SEM or optical microscopy. Replacing the solid fraction W in equation (1) with the effective volume fraction Weff given in equation (5), the viscosity of the semisolid slurry can be obtained. Numerical study and discussion The Hausdroff dimensionality of a solid particle in a semisolid slurry is related to the distribution of constituents, the solid – liquid interfacial energy, undercooling, cooling rate, shear rate, and many other factors. However, the Hausdroff dimensionality is bounded by two limits which are named the upper limit and lower limit. This is due to the limitation that fs ¡Weff ¡1 : : : : : : : : : : : : : : (7) The upper limit is equal to 3 according to equation (5a). The lower limit can be calculated by letting Weff~1 in equation (5). The upper limit implies that no liquid is trapped in the solid particle while the lower limit implies a solid skeleton is formed at that solid fraction. A numerical analysis of the relationship between the lower limit of Hausdroff dimensionality and the actual solid fraction fs is illustrated in Fig. 2, where it is assumed that N/Q~1016 corresponding to an average grain size in a completely solidified cast of about 100 mm. Figure 2 shows that a network of solid phase can be formed even at a very small solid fraction. For example, the pattern in Meakin’s14 simulation has only a solid fraction of about 0.004. Of course the solid network formed is very weak when the solid fraction is small, due to the very thin solid arm. The strength of the solid skeleton will develop as the solid fraction increases. That is the reason for the observation that a solid – liquid suspension exhibits an apparent yield point of 106 dyne cm22 (1 dyne cm22~ 1021 Pa) at a small solid fraction of 0.15 which increases with increasing fraction of solid.20 Figure 2 also shows that the critical solid fraction for the formation of a solid network depends on the particle morphology. For a morphology of DH~2.808 a network is formed from a critical solid fraction of 0.1. While DH~2.92 corresponds to a critical solid fraction of 0.4. It is well known that shearing the slurry from a temperature above the melting point can Qin and Fan Fractal theory study on viscosity of semisolid slurries 1151 Figure 3 shows the results of numerical calculation. There are two limiting curves in the figure. One limiting curve of DH~2.95733 corresponds to the extreme external condition that a solid skeleton will be formed at a solid fraction of 0.6. In terms of shear rate it is the lowest shear rate which enables the slurry to behave as a fluid at a solid fraction of less than 0.6; In terms of the cooling rate it is the highest cooling rate. The other limiting curve of DH~3 corresponds to the opposite extreme case, e.g. a high enough shear rate to form a spheroid. The curve of DH~2.95733 shows that the relative viscosity begins to increase sharply from a solid fraction of about 0.3, while the curve of DH~3 shows that the relative viscosity is very low even for a solid fraction of near to 0.6. At fs~0.4 the apparent viscosity of a slurry with DH~2.95733 is 60 times larger than that of a slurry with DH~3. At fs~0.5 the ratio is 180. Comparing the results in Fig. 3 with the experimental results reviewed by Flemmings,6 the main characteristics coincide fairly well. The size of the solid particles may also affect the viscosity of the solid – liquid suspension for a given solid fraction. Thus the effect of the particle size on the viscosity requires to be studied. For a certain solid fraction, the particle size is inversely proportional to the cube root number of particles. Figures 4 and 5 show the numerical results for the size related relative viscosity. In these figures the particle size is normalised. The normalised particle size is taken to mean the ratio between the true particle size and a standard particle size. The standard particle size is defined as the particle size which enables a number density of 1 cm23 at various solid fractions. Figure 4 shows the results of relative viscosity v. normalised particle size at a solid fraction of 0.43 and Hausdroff dimensionality of 2.95841, 2.975, 2.99, and 3. It can be seen that the size effect is more important for a lower Hausdroff dimensionality. For DH~3 the size of the solid particle has no effect, just as suggested by Einstein’s work. For a given fractal morphology of the solid phase, a smaller particle size will result in a smaller relative viscosity. Another calculation demonstrates the effect of particle size on the viscosity of slurry at different solid fractions as shown in Fig. 5. The Hausdroff dimensionality is chosen to be 2.975 and the solid fractions are 0.2, 0.4, and 0.6. It is found that the smaller the solid fraction the smaller the contribution of the size effect. The figure shows that the size effect cannot be neglected where the solid fraction is relatively high. 3 Relationship between relative viscosity and solid fraction for different morphologies: illustrated for Hausdroff dimensionalities of 2.95733, 2.975, 2.99, and 3 4 Effect of particle size on relative viscosity for different values of Hausdroff dimensionality where solid fraction is assumed to be 0.43 2 Relationship between lower limit of Hausdroff dimensionality and solid fraction increase the Hausdroff dimensionality. So a solid network will begin to form at a higher solid fraction. Therefore, the reported observation that the shear stress at a solid fraction of 0.4 is quite low for non-dendritic materials is, according to the present model, due to the absence of a solid skeleton at a high Hausdroff dimensionality with that solid fraction.21 The viscosity of semisolid slurry can be calculated under the present theory. From the literature, the semisolid slurry behaves as a fluid at a solid fraction of less than 0.6. Otherwise the yield stress of the slurry would be very strong and the slurry would behave as a viscoplastic solid. This experimental phenomenon enables the suggestion that the critical solid fraction for the formation of a solid network is 0.6. According to the numerical result found in Fig. 2, the Hausdroff dimensionality for this slurry satisfies 2:95733¡DH ¡3 Materials Science and Technology September 2001 Vol. 17 1152 Qin and Fan Fractal theory study on viscosity of semisolid slurries the flow behaviour of semisolid slurries primarily by changing the Hausdroff dimensionality of solid particles. 2. A solid skeleton can be formed at a very small solid fraction if the Hausdroff dimensionality of the solid particle is low. The critical solid fraction to form a solid network depends on the Hausdroff dimensionality. 3. The particle size may affect the viscosity of semisolid slurry at a high solid fraction and a small Hausdroff dimensionality. Acknowledgements The authors wish to thank Brunel University and Ford Motor Co. for financial support to this and other related research projects. Thanks also go to Professor M. J. Bevis at Brunel University for his encouragement. References 5 Effect of particle size on relative viscosity at different solid fractions (0.2, 0.4, and 0.6) where Hausdroff dimensionality is chosen to be 2.975 Concluding remarks A theoretical model based on fractal theory has been developed to assess the effect of particle morphology on the viscosity of semisolid slurries. A solid particle with a Hausdroff dimensionality of less than 3 will entrap liquid in it. This will increase the effective volume fraction of the solid. By applying a fractal analysis the effective solid volume fraction, defined as the volume fraction of actual solid plus the entrapped liquid, has been derived theoretically. Analysis of the viscosity is enabled by adaptation of an empirical expression of viscosity as a function of the volume fraction of solid. It is found that the viscosity is affected by the number of solid particles per unit volume, the actual solid fraction, and the Hausdroff dimensionality of the solid particle. Numerical calculation using the developed model allows the following conclusions to be made. 1. The viscosity of a semisolid slurry is a non-linear function of the Hausdroff dimensionality. A slurry with a particle morphology of lower Hausdroff dimensionality exhibits higher viscosity for a given solid fraction. The external variables, such as shear rate and cooling rate, affect Materials Science and Technology September 2001 Vol. 17 1. p. kumar, c. l. martin, and s. brown: Acta Metall. Mater., 1994, 42, 3595 – 3602. 2. m. mada and f. ajersch: Mater. Sci. Eng. A, 1996, A212, 157 – 170; 1996, A212, 171 – 177. 3. h. k. moon: PhD thesis, Massachusetts Institute of Technology, Cambridge, MA, 1990. 4. g. k. sigworth: Can. Metall. Q., 1996, 35, 101 – 122. 5. m. c. flemings: Metall. Trans. A, 1991, 22A, 957 – 981. 6. j. s. langer: Rev. Mod. Phys., 1980, 52, 1 – 28. 7. d. quemada: ‘Rheology of heterogeneous fluids’, 21st Ann. Colloq. of Groupe Francais de Rheologies, Strassbourg: 1986; 1.1 – 1.11. Paris, Scrivener. 8. b. b. mandelbrot, d. e. passoja, and a. j. paullay: Nature, 1984, 308, 721. 9. l. l. sun, l. k. dong, j. h. zhang, and z. q. hu: J. Mater. Sci. Lett., 1997, 16, 505 – 507. 10. k. j. MÅLǾY, a. hansen, e. l. hinrichsen, and s. roux: Phys. Rev. Lett., 1992, 68, 213 – 216. 11. j. j. friel and c. s. pande: J. Mater. Res., 1993, 8, 100. 12. j. m. li, l. lu, y. su, and m. o. lai: J. Appl. Phys., 1999, 86, 2526 – 2532. 13. d. g. thomas: J. Colloid Sci., 1975, 11, 33. 14. p. meakin: Phys. Rev., 1983, 27A, 1495 – 1507. 15. t. vicsek: ‘Fractal growth phenomena’, 3; 1989, London, World Scientific. 16. k. j. falconer: ‘The geometry of fractal sets’, 13; 1986, Cambridge, Cambridge University Press. 17. r. c. weast: ‘Handbook of tables for mathematics’, 129; 1975, Cleveland, OH, CRC Press. 18. a. j. avrami: Chem. Phys., 1939, 7, 1103; 1940, 8, 212. 19. v. erukhimovitch and j. baram: Phys. Rev. B, 1994, 50B, 5854. 20. d. b. spencer, r. mehrabian, and m. c. flemings: Metall Trans., 1972, 3, 1925 – 1932. 21. d. b. spencer: PhD thesis, Massachusetts Institute of Technology, Cambridge, MA, USA, 1971.
© Copyright 2025 Paperzz