Rheological Investigation of Suspensions and Ceramic Pastes: Characterization of Extrusion Properties

A large number of complicated catalyst geometries are produced by extrusion of plastic ceramic materials. The demands for high precision in the forming process and adequate formability of the materials are extremely stringent. As a first approximation, plastic ceramic materials can be treated as ordinary concentrated suspensions. Rheometric methods, in particular capillary rheometry, are especially suitable for testing of these materials. However, the flow processes occurring during extrusion are very complex, with many special effects such as wall slip, shear-thinning, shear hardening and high entrance pressure loss. As a result, apparent viscosity functions are not material functions. In spite of these difficulties, capillary rheometry, when critically applied, is an advantageous tool in the development of easily extrudable ceramic materials. The correlation of rheometrical test results with the extrusion process during production of honeycomb geometries is presented using aluminium oxide ceramics as an example.


Introduction
A large number of products based on ceramic materials are shaped by extrusion.This is normally a continuous manufacturing process with its related technical and economic advantages.Traditional building trade products such as bricks and pipes are formed by extrusion of ceramic pastes.
The precision demands on these products are not very high, ranging up to tenths of a centimeter.Because ceramic pastes possessing natural formability have been handled for thousands of years, such low-precision requirements are easy to achieve.In recent years, ceramics have become important high-tech-materials in mechanical, electrical and chemical engineering as well as in the field of medicine.The precision demanded for such products is the same as that demanded for parts made from metals, i.e. in the range of microns.Many such products are formed by extrusion.
The typical objects made in enormous numbers by continuous extrusion include honeycomb-formed ceramic catalyst carriers for automobiles.An example of such a catalyst which has a cell wall thickness of less than two hundred microns is illustrated in Fig. 1.One can imagine that the development of the extrusion technology specially for the production ------• ---------------* Postfach 6980, D-76131 Karlsruhe, Germany t Received June 1st, 1993 KONA No.ll (1993) of this highly complex item was extremely expensive and time consuming, even if the extruded material possessed a natural formability.
All clays based on flat, disc-like mineral particles possess good formability, even when tempered with pure water.These are materials with ''natural formability''.Other ceramic materials such as metal oxides possess no natural formability.The attempt to extrude catalysts from these materials may result in defective parts, such as shown in Fig. 2. Extrudable pastes require plasticizers as flow additives such as, for example, high-molecular-weight polymers.
At first glance, extrusion appears to be an ordinary flow process.Thus it seems reasonable to use rheometric methods for the straightforward development of the formability behaviour of pastes which do not possess natural formability.
Typical instruments for the investigation of the flow properties of pure and complex fluids such as particleliquid systems are shown in Fig. 3.Because of its correspondence to extrusion machines, the capillary rheometer seems to be the preferable instrument to characterize the formability of ceramic pastes.The velocity distribution within the measuring capillary is thought to be similar to the schematic in Fig. 4 which corresponds to adhesion of the fluid to the capillary wall (v (R) = 0).It is important to note that all rheological quantities calculated from capillary  From a rheological point of view, ceramic pastes are particle-fluid suspensions, with a flow behaviour which depends on the viscosity function of the liquid itself, the solid-liquid ratio (volume fraction of the solid), the particle size, or particle size distribution and the particle shape.It is generally well-known that the viscosity of a suspension increases with increasing solids content.Recent investigations show that at high shear stresses (or shear rates) the shear stress function of suspensions is primarily determined by the shear stress function of the suspending liquid (for both Newtonian and non-Newtonian liquids) [1].This means that the flow behaviour of the suspension is controlled by the hydrodynamic forces within the pure liquid (i.e. between the particles).
At low shear rates, low shear stresses are generated within the suspension.In this case the relatively weak particle-particle interaction forces, which are generally independent of the relative velocity between the single particles, begin to dominate the total stress state.Therefore, in the case of low shear stresses, particle properties such as size, shape and surface activity determine the flow behaviour.The flow behaviour of suspensions is quite different from that of the pure liquid.Suspension behaviour is often significantly non-Newtonian, (especially at high solids concentrations) and a yield stress approaching 7 0 may exist.At stresses lower than 7 0 the fluid is not deformable.
As shown in Fig. 5, one generally finds that particle size affects the flow behaviour of concentrated suspensions in a manner comparable to solids concentration [2,3].
From the shear stress function 7()') of the pure silicone oil, one observes a slight deviation from Newtonian behaviour (pseudoplasticity) at high shear rates.At high shear rates, the shear stress functions of the suspensions filled with limestone par-KONA No.ll (1993) tides of various size form a unique curve.In the direction of the shear rate axis, a constant distance exists between this curve and the shear stress function of the suspending fluid despite an increase in the average particle size in the different suspensions from x = 1.7 to x = 20.5 ,um.The volume fraction solids is held constant at Cv = 0.30.This means that the flow behaviour is dominated by the shear stress function of the suspending fluid and remains independent of particle size, which is a surprising result.At low shear rates (and therefore low shear stresses), a significant influence of particle size on the shear stress function of the suspensions becomes apparent.With decreasing particle size, the shear stress measured at the same shear rate increases.This increase correlates with the increase in the number of single particles with decreasing particle size at a constant solids volume fraction.
-The influence of the particle collective (i.e.degree of polydispersity) on the flow behaviour of suspensions is (relatively) stronger at low than at high shear stresses.-The shape of the particles, however, exerts a significant influence on the flow behaviour in both the high, as well as the low, shear rate region.When investigating the formability of ceramic pastes by rheometric methods, different ''shear stress functions'' are usually measured when using different types of rheometers, as shown in Fig. 6.The ''shear stress functions" 7( )'*) of the same kaoline paste calculated from measurements with cone and plate-, couette and capillary-rheometers differ, in some instances, by a factor of ten.Even when 7()'*) is determined using a single capillary rheometer with different capillaries, one obtains functions which differ significantly.A prediction of the flow behaviour in shear rate .Y • /s• 1 Fig. 6 "Shear stress functions" of a china clay determined with different types of rheometers a real process is therefore not possible.These "shear stress functions'' are called apparent and are not real material functions.This uncertainty is a general problem in the rheology of highly filled suspensions which is often not taken into consideration in published measurements.
One reason for this uncertainty is that a velocity profile v(r), as demonstrated in Fig. 4, only exists in the case of fluid adhesion to the wall.If the paste slips at the wall of the shear gap, a complex velocity distribution emerges and the shear rate )'* calculated without considering slip is not the true wall shear rate.The quantity )'* is normally termed the apparent shear rate.The velocity distribution across the capillary when considering slip and internal deformation is shown in Fig. 7.
Complete velocity distribution v(r) in a circular capillary for a fluid with wall slip effects.v G is the slip velocity and v* (r) the velocity as a function of r due to the internal deformation of the fluid alone (the fluid has no yield stress) The velocity distribution is a superposition of the velocity profile v*(r) due to the internal deformation of the fluid and a constant velocity v G which represents the slip velocity of the fluid in the proximity of the wall.The slip velocity v G has no influence on the internal deformation.The total volumetric flow rate V is composed of tw? parts: a cylindrical ~ontribution V G from vc and a parabolic contribution Vs from v* (r).The real internal shear rate 'Y must be calculated from v* (r) alone.-Forthe calculation of the total volumetric flow rate V, at least two separate material functions should be known: the slip velocity as a functi?n of the shear stress vc(7), in order to calculate Vc; and the well known shear stress function 7()'), which describes th~ internal deformation for the calculation of v*(r) or V 5 .Experimental methods have been developed to separate vc from v*(r) [4 -6].

Typical flow properties of ceramic pastes
Ceramic pastes are, in principle, suspensions of very small particles dispersed at a high concentration in a liquid which may even be pure water.All of the previously discussed problems concerning high concentrations, influence of particle size and wallslip effects must be considered if the real flow functions of such materials are to be determined for the calculation of flow through complicated dies.This was not possible until now.
Bearing in mind the complicating effects which influence results obtained when testing concentrated suspensions, rheometric measurements with capillary rheometers were conducted to optimize the formability of ceramic pastes.An example test procedure is outlined in Fig. 8 [7, 8].
The piston of a capillary rheometer can be moved at different constant speeds Vst• This !llovement produces constant volumetric flow rates vi of paste through the capillary which ?re recorded with time t.The volumetric flow rate Vi generates.a pressure drop Pi• measured simulta~eously to Vi and also recorded with time.From Vi and Pi• the apparent shear stress function 7 ( )'*) is calculated normally without considering the non-Newtonian flow profile and slip effects.For ideal ceramic pastes, a unique shear stress function is expected when using a single capillary.In the case of bad formability, the real rheometric measurements for ceramic pastes appear quite different, and no well-defined shear stress functions can be recorded.Typical shear stress functions of real pastes are presented in Fig. 9.

I:
If the ceramic paste is inadequately prepared and the suspension thus possesses an inhomogeneous water-solid distribution, no strict correlation will be obtainable between the pressure and the volumetric flow rate.The pressure oscillates with varying solid concentration and the shear stress function degenerates into a shear stress spectrum.

II:
The consistency of the paste can be influenced by the extrusion procedure itself.Because of the pressure drop, a disproportionately high volume flow rate of water through the capillary is possible (dewatering) and the solids concentration in the remaining paste is elevated.A steady increase of the pressure will be recorded even at constant volume flow rates.The shear stress function is only the lower boundary of an infinite shear stress range of infinite height.

III:
A pressure trend similar to that which results from dewatering can also be caused by micropore effects.

KONA No.ll (1993)
Because of the elevated pressure, a certain amount of the suspending liquid can be pressed into micropores.As the free water content is reduced, the paste becomes stiffer, and a pressure increase is recorded.Contrary to the dewatering process, a steady-state pressure is reached and the suitable shear stress function occupies a range between lower and upper limits (which depend on the duration of the experiment).
IV: A very disadvantageous flow behaviour which causes severe damage to the extrudate is indicated in the pressure plot of example IV.At elevated volumetric flow rates, extremely high pressure values can be measured and consequently, the shear stress function becomes very steep at high shear rates, with a broad range of scattering.This effect is caused by a layer of hardened dry mate,rial which forms at high flow rates and high extrusion pressures along the capillary wall.

V:
A pressure plot which indicates excellent extrusion properties is demonstrated in example V.The steadystate pressure is nearly independent of the volumetric flow rate, whereby maximal and minimal values of the pressure only exist at the transition point to a higher or lower flow rate.This is the behaviour of a nearly ideal plastic material such as ceramic pastes with natural formability, e.g.china clays.The shear stress function has a plateau range in which the shear stress is essentially independent of the shear rate.

Experimental results from exemplary ceramic pastes
In order to correlate rheometric measurements with the real extrusion behaviour, the same material has to be tested in a complex forming tool.In Fig. 10, a tool for the extrusion of honeycombshaped catalyst carriers is shown [8].The alternately  The reduction of the volumetric flow rate in the last test range induces a pressure minimum followed by attainment of the original steady-state pressure.
The shear stress function 7 ( )') of the clay in Fig. 12 has a plateau range which extends over nearly two decades of the shear rate.This is the typical trend of a shear stress function for a material possessing good formability.This material exhibits virtually no shear deformation in a capillary flow and the shape of the coloured discs remains uncleformed as they pass through the cylinder of the test extruder [7, 8] (left-hand illustration of Fig. 13).The honeycomb catalyst carriers which were extruded from this material are exactly formed and possess a good surface quality (right-hand picture in Fig. 13).
This material shows a nearly pure slip behaviour without any internal deformation when pressed through a capillary.A true shear rate cannot be calculated for such ceramic pastes.
An apparent shear stress function with a steep increase of T at high shear rates was found in rheametric measurements of a paste based on a solid material called Pural SB (catalyst carrier material) with Luviskol as plasticizer in the liquid phase, Fig. 14.The pressure plot at high volumetric flow rates reveals very high maximas and extremely scattered values, as displayed in example IV of Fig. 9.The extrudate which emerges from the rheometer capillary is irregularly shaped and has a smaller diameter than the capillary itself (see inset diagram in Fig. 14).
A cross-section of the coloured paste sample reveals very irregular deformations, as seen in Fig. 15.The flow patterns reveal three distinctly separate regions: 1) a narrow range near the wall without  yields ''negative'' slip velocities as predicted by Schlegel and Weller [8].The extrusion of a regularlyshaped honeycomb catalyst is not possible as the photo on the right-hand side of Fig. 15 demonstrates.This Christmas-tree shaped body was intended to become a honeycomb catalyst carrier.The same ceramic material Pural SB can be the basis of a paste with good extrusion properties provided that suitable plasticizers are used.The advantageous flow properties are evident from the shear stress functions shown in Fig. 16 which are flat in comparison with the one shown in Fig. 14 for the same basic ceramic material.Flat shear stress functions, with a smoothly increasing shear Fig. 15 Flow patterns of a ceramic paste possessing unfavourable extrusion properties [7,8] axial deformation; 2) a rather small range with extended axial deformation where shear flow occurred; and 3) a broad core region with only small deformations.The undeformed paste near the wall (where the highest shear stress exists) indicates a shearhardened material layer.Most of the flow deformation occurs in a rather thin region whereas the core remains as undeformed as the narrow wall region.The attempt to separate the shear and slip flows stress for an increasing shear rate, are similar to shear rate functions of natural clay, and generally indicate suitable extrusion properties.The reason for these good flow properties is that such materials form a plug-like velocity profile in a tube, with most of the deformation occurring in a thin flow layer near the wall, so that the main volume of the ''flowing'' material remains almost undeformed.This can be seen in the flow patterns on the left-hand picture of Fig. 17.Contrary to the flow of a pure fluid, most of the paste transport occurs by slip effects in an extremly thin film at the wall.Only slight deformations can be identified across the bulk of the sample.The catalyst carrier depicted in the right-hand illustration demonstrates the formability of this paste.
Rheometric measurements on ceramic pastes with good natural or artificial formabilities lead to rather flat shear stress functions which sometimes include a plastic plateau range.In a capillary flow, only small internal deformations occur within the bulk of sample resulting in "plug flow" (like a rigid rod) through the capillary.
Part of the total pressure drop occurs immediately at the capillary inlet to shape the slim rod for the capillary.The degree of this inlet pressure loss can Fig. 17 Left: Flow patterns of a ceramic based on plasticized Pural SB paste with rather good extrusion properties Right: Honeycomb catalyst carrier extruded from this material be quantified by a method developed by Bagley [9] for the investigation of the true shear stress functions of polymer melts, as shown in Fig. 18.The total pressure loss of a fluid flowing at a constant volume rate must be measured with capillaries of different length.If the flow is isothermal and the viscosity independent of the hydrostatic pressure, then the total pressure loss itself increases linearly with the capillary length.The extrapolation of this straight line to zero capillary lengths yields the entrance pressure loss (as demonstrated in Fig. 18).This entrance pressure loss is sometimes predominant in the capillary flow of pastes.The true shear stress within the capillary can only be determined from the pressure gradient within the capillary and is equal to the slope of the straight lines in Fig. 18.If the shear stress from capillary experiments is calculated from the total pressure loss disregarding KONA No.ll (1993) the entrance pressure loss, then an "apparent" shear stress is obtained, resulting in a separate shear stress function for each capillary length (even for a constant capillary diameter), as shown in Fig. 19.Since the entry pressure loss is a relatively high share of the total pressure drop (as can be seen in Fig. 19 for the example of the ceramic paste "A"), higher apparent shear stresses will be calculated from the total pressure drop when measured with a short capillary.In order to derive the real "viscous" pressure loss over the capillary, the entrance pressure must be subtracted from the total pressure.This procedure is known as the Bagley correction, and leads to "true" shear stresses.The true shear stress (lowest shear stress function in Fig. 19) of the ceramic paste "A" is independent of the shear rate or volumetric flow rate.Thus, the ceramic paste "A" (Pural SB with hydroxyethylcellulose as plasticizer) is a material with a good formability.The increase of the apparent shear stress with increasing extrudate velocity is only caused by the increase of the entrance pressure loss (Bagley pressure).
If the Bagley pressure measured for material "A" is plotted as a function of the total extrusion pressure p, one obtains a straight line with a 45 degree slope as can be seen in Fig. 20.The increase of the extrusion pressure is equal to the increase in the Bagley pressure.For this ceramic paste, the internal capillary pressure loss is constant, and is independent of the total extrusion pressure.The flow patterns of the ceramic paste are shown in Fig. 21.They exhibit nearly ideal plastic behaviour similar to that obtained with natural clay, shown in Fig. 13.This flow behaviour is caused by a special type of plasticizer.The ceramic paste "B" is also based on Pural SB but with NAL (ammonium alginate) as the plasticizer.The apparent shear stress function measured with two capillaries of different length also yields higher apparent shear stresses for the shorter capillary.The increase of 7 (Fig. 22) is stronger with increasing extrudate velocity in comparison with ceramic paste "A" .
The essential difference in the results for paste "B" compared with paste "A" is that not only the apparent shear stress, but also the true shear stress increases significantly with extrusion velocity or shear rate.Fig. 20 shows that the Bagley pressure increases nearly proportionally to the total extrusion pressure.At all extrusion velocities, the total pressure drop consists of proportional contributions from 134 10' ,---,-----,--,-,-,----,-- the deformation processes at the capillary inlet (entrance pressure losses) and from losses within the capillary.The flow patterns in Fig. 23 indicate that within the capillary, the ceramic paste is subjected to weak shear deformations different from those exhibited by ceramic paste ''A'', which passes through the capillary in a rod-like shape.Good honeycomb catalysts may be extruded from both materials.The flow behaviour of the ceramic pastes "A" and "B", which are based on the same solid material, was altered by changing the type of liquid phase.Separation of the entrance pressure losses from the ''flow'' pressure drop within the capillary can help in distinguishing the internal deformation behaviour of "externally" similar materials.

NMR-Imaging of ceramic pastes
Good knowledge of the flow patterns is extremely helpful in understanding and predicting the extrusion properties of pastes.In order to visualize the defor-mation process, the paste can be marked by alternately coloured layers.This technique has been used successfully in Figs. 10, 15, 17, 21 and 23.A new and very promising alternative method of visualizing paste flow patterns is the NMR-Imaging technique.This has been employed in 1 1 10: to investigate the extrusion behaviour in a ram extruder.Results using limestone pastes with water contents between 23o/o and 29o/o and oxide ceramic pastes (Pural) with water contents of 23o/o to 76o/o have shown that NMR-imaging is a non-invasive technique which allows successive exposures to be taken of a single sample during an experiment, thus providing an animation of the flow process in the interior of the specimen.
Fig. 24 shows a schematic of the ram extruder used.The paste is prepared by marking individual layers with a contrast medium (Magnevist).In addition, 2 mm glass spheres are imbedded, providing information concerning the axial and radial displacement of the paste.Fig. 2 7 shows a series of subsequent images (from left to right) of Pural paste, with hydroxyethylcellulose as an additive, in the immediate vicinity of the right wall.Obviously a thin wall shear layer exists, in which the paste is subjected to extremely large shear deformations.

Final remarks
Ceramic pastes are, in principle, suspensions possessing extremely high solid concentrations.Their flow behaviour at low deformation rates is dominated by particle-particle interactions and is therefore KONA No.ll (1993) The flow laws derived for moderately concentrated suspensions cannot be applied, since they only quantify flow behaviour changes as a function of solid content and/or particle characteristics.Investigations of the extrusion properties of ceramic pastes by rheometric methods must be conducted with care because the paste transport through a capillary is not simply an ordinary flow process.The transport is a mixture of different phenomena such as wall slip, internal shear deformations, shear hardening near the wall, and extraordinarily high capillary entry pressure losses.The combination of these phenomena defines the formability of ceramic pastes.Flow functions calculated from capillary measurements without considering these different effects are not real material functions, and therefore cannot form the basis for quantitative predictions of extrusion processes.
All these phenomena can be separated by applying rheometric methods and may also be separately influenced by the addition of specific plasticizers.Because the basic solids of ceramic material are normally invariant, the formability of such pastes can only be optimized by changing the properties of the suspending fluid (which also changes the particle-particle interactions).The influence of specific particle properties can only now be directly quantified for high solid contents.
Optimization of the extrusion properties of ceramic catalysts will be a future research programme once the fundamental problems have been solved.The formability of ceramic pastes can only be optimized in a reasonable manner by rheometric measurements.
Materials with good extrudability can be identified by their specific apparent shear stress function.Capillary rheometry, when critically applied, is therefore an advantageous tool in the development of easily extrudable ceramic pastes.

Fig. 1 2 ~ 2 -
Fig.1View of the cross-section of a ceramic catalyst for automobiles with a total diameter of 110 mm

pFig. 8
Fig.8Schematic of a capillary rheometer with typical diagrams V(t).p(t) and T(,Y) as can be used for testing ceramic pastes

Fig. 9
Fig. 9 Characteristic pressure-time plots p (t) of real ceramic pastes with the suitable apparent shear stress functions 7(-y*) 130

Fig. 10
Fig. 10 Alternately coloured layers of ceramic pastes for the detection of flow patterns in the honeycomb extrusion coloured ceramic paste in the cylinder can be pressed through the tool by the movement of the piston in the upper right of the picture.The differently coloured paste discs allow the flow patterns to become visible.

Fig. 12
Fig. 12 Shear stress functions of two different ceramic pastes

Fig. 13
Fig. 13 Left: Cross-section through an alternately coloured natural clay sample which has passed through a cylindrical tube Right: Honeycomb catalyst carriers made from the specified materials

Fig. 14
Fig. 14 Shear strees function of a ceramic paste with poor extrusion formability Fig. 16 Shear stress function of a ceramic paste based on Pural-SB with advantageous extrusion properties attained by the additicn of suitable plasticizers Fig. 18 Bagley plot for the determination of entrance pressure losses

10 - 1 Fig. 19
Fig. 19 Apparent shear stress functions for the ceramic paste "A" measured by use of capillaries of different lengths

Fig. 20
Fig. 20 Relation between the extrusion pressure and the Bagley pressure

Fig. 21
Fig. 21 Flow patterns for ceramic paste A

1 Fig. 22
Fig. 22 Apparent shear stress functions for ceramic paste "B" measured with capillaries of different length

Fig. 23
Fig. 23 Flow profiles for the ceramic paste "B" Fig. 24 Schematic of the ram extruder Fig. 25 shows a sequence of 8 successive NMRimages of a limestone paste with 25o/o water during extrusion.Fig. 26 presents particle trajectories constructed from the displacement of the glass spheres.Fig.27shows a series of subsequent images (from left to right) of Pural paste, with hydroxyethylcellulose as an additive, in the immediate vicinity of the right wall.Obviously a thin wall shear layer exists, in which the paste is subjected to extremely large shear deformations.

Fig. 25 Fig
Fig. 25 Limestone with 25% moisture content at six successive instants