Journal of the European Ceramic Society 17(1997)299-5 Printed in Great b itin ea s richt reserved PII:s0955-2219(96)00131-8 0955-221997517 Tape-Cast Alumina-Zirconia Laminates Processing and mechanical Properties T. Chartier &t. rouxel LMCTS, URA CNRS 320, ENSCI, 47 Av. Albert Thomas, 87065 Limoges Cedex, France (Received 15 September 1995; revised version received 19 June 1996; accepted 15 July 1996) abstract aim is to improve both the toughness and the strength. This type of composite allows the associ Alumina-zirconia laminar ceramics made from lay- ation of two reinforcing mechanisms, the first act ers of different compositions with diferent stacking ing at the scale of the microstructure, inside the sequences were fabricated by tape casting and com- layers, due to the stress-induced transformation of pared. A phosphate ester dispersant was optimized zirconia particles. the second acting at a macro in MEK/EtOH and an optimum formulation of scopic scale, due to the interfaces between the lay organic components for tape casting was defined. ers. In these laminar structures, residual stresses The fracture resistance, toughness and elastic prop- develop during cooling from the sintering temper erties were characterized. A significant improvement ature because of the differences in thermal expa of both the fracture resistance and the toughness, sions between layers of different compositions from 380 t0 560 Pa and from 3.7 to8 MPavm The sign and the magnitude of these stresses may respectively, was gained between the pressed alu- be adjusted through the compositions, the stacking mina monolith and the tupe-cast AlO -ZrO, comm- sequences, and also through the layer thicknesses posites. The improvement was tentatively related to Hence, it is possible to develop high compressive The presence of residual stresses at both the micro- stresses in thin layers whereas the tensile stresses copic scale( phase transformation toughening )and remain low in the associated thick layers at the macroscopic scale( interface effects). C1996 Tape casting, -4 which is extensively used for Elsevier Science limited electronic ceramics. is well suited to fabricate homogeneous wide and thin ceramic sheets that can be reinforced by zirconia particles. Multilayer 1 Introduction systems are made by stacking green sheets, lami nating, removing the organic components and sin- Engineering applications require improved mechan- tering. Tape casting involves the dispersion of th cal properties of ceramics, particularly fracture ceramic powder in a solvent (typically organic) toughness. A noticeable increase in toughness can with the aid of a dispersant, followed by the addi obtained through ceramic composites. At the tion of binders and plasticizers to ensure the cohe- microscopic scale, composites can be roughly sion, fexibility and workability of the green tape classified into particle-, platelet-and whisker-rein- when the solvent is evaporated forced materials. At the macroscopic scale, com s paper posite materials can be arranged according to nated AlO3-ZrO, composites by tape casting and various configurations, among which are lami- the optimization of the dispersion which is a cru nated composites, consisting of alternate layers cial step. The influence of organic compounds on with different compositions. These structures pro- properties of green tapes, cracking sensitivity, den ide the opportunity for tailoring the properties sity and thermocompression ability will also be by stacking layers of different compositions in a reported. The room temperature fracture char suitable sequence. It is then possible to produce teristics (strength, toughness, fracture path)and functionally gradient ceramics to meet specific elastic properties were investigated to give an requirements- insight into the complex nature of the mechanical The laminated ceramic composites which are behaviour of ceramic laminates and to illustrate studied here consist of alternate layers of alumina the advantages of the composite materials over reinforced by various amount of zirconia. The monoliths. 299
Printed in Great Britain. All rights reserved PII: SO955-2219(96)00131-8 0955-22191971Sl7.00 Tape-Cast Alumina-Zirconia Laminates: Processing and Mechanical Properties T. Chartier & T. Rouxel LMCTS, URA CNRS 320, ENSCI, 47 Av. Albert Thomas, 87065 Limoges Cedex, France (Received 15 September 1995; revised version received 19 June 1996; accepted 15 July 1996) Abstract Alumina-zirconia laminar ceramics made from layers of dtflerent compositions with dtyerent stacking sequences were fabricated by tape casting and compared. A phosphate ester dispersant was optimized in MEWEtOH and an optimum formulation of organic components for tape casting was defined. The fracture resistance, toughness and elastic properties were characterized. A signt$cant improvement of both the fracture resistance and the toughness, from 380 to 560 MPa and from 3.7 to 8 MPadm respectively, was gained between the pressed alumina monolith and the tape-cast A120J-Zr02 composites. The improvement was tentatively related to the presence of residual stresses at both the microscopic scale (phase transformation toughening) and at the macroscopic scale (interface eflects). 0 1996 Elsevier Science Limited. 1 Introduction Engineering applications require improved mechanical properties of ceramics, particularly fracture toughness. A noticeable increase in toughness can be obtained through ceramic composites. At the microscopic scale, composites can be roughly classified into particle-, platelet- and whisker-reinforced materials. At the macroscopic scale, composite materials can be arranged according to various configurations, among which are laminated composites, consisting of alternate layers with different compositions. These structures provide the opportunity for tailoring the properties by stacking layers of different compositions in a suitable sequence. It is then possible to produce functionally gradient ceramics to meet specific requirements. I-9 The laminated ceramic composites which are studied here consist of alternate layers of alumina reinforced by various amount of zirconia.” The aim is to improve both the toughness and the strength. This type of composite allows the association of two reinforcing mechanisms, the first acting at the scale of the microstructure, inside the layers, due to the stress-induced transformation of zirconia particles, the second acting at a macroscopic scale, due to the interfaces between the layers. In these laminar structures, residual stresses develop during cooling from the sintering temperature because of the differences in thermal expansions between layers of different compositions. The sign and the magnitude of these stresses may be adjusted through the compositions, the stacking sequences, and also through the layer thicknesses. Hence, it is possible to develop high compressive stresses in thin layers whereas the tensile stresses remain low in the associated thick layers. Tape casting,’ ‘-I4 which is extensively used for electronic ceramics, is well suited to fabricate homogeneous wide and thin ceramic sheets that can be reinforced by zirconia particles. Multilayer systems are made by stacking green sheets, laminating, removing the organic components and sintering. Tape casting involves the dispersion of the ceramic powder in a solvent (typically organic) with the aid of a dispersant, followed by the addition of binders and plasticizers to ensure the cohesion, flexibility and workability of the green tape when the solvent is evaporated. This paper describes the processing of laminated A&O,-Zr02 composites by tape casting and the optimization of the dispersion which is a crucial step. The influence of organic compounds on properties of green tapes, cracking sensitivity, density and thermocompression ability will also be reported. The room temperature fracture characteristics (strength, toughness, fracture path) and elastic properties were investigated to give an insight into the complex nature of the mechanical behaviour of ceramic laminates and to illustrate the advantages of the composite materials over monoliths. 299
T. Chartier. T. Rouxel 2 Experimental Procedure green density, (ii)good microstructural homo- geneity, and (iv) good thermocompression ability 2. 1 Starting materials Two parameters were chosen to improve the Tape casting slurries are complex, multicompo urry composition with regards to the crack sensi- nent systems, which contain ceramic powders tivity, density and thermocompression ability of (including sintering aids), solvents, dispersants, the green tapes. The first is the volume solids binders and plasticizers. The starting powders are ratio(X= vol% solids= powder/powder disper 997 wt% purity, 0.5 um grain sizc alumina sant t binder plasticizer) and the sccond is the (P172SB, Pechiney, France)and 97.5 wt% purity, volume binder to plasticizer ratio(r binder/ 0 4 unl grain size zirconia(UPH 12, Criceram, plasticizer). Alunina+10 vol% zirconia slurries were fran prepared and tape cast with volume solids values Tape casting slurries were prepared with an varying from 0 6 to 0.9 with steps of 0.05 and azeotropic mixture of methyl ethyl ketone(MeK) binder/plasticizer values varying from 0.3 to 1. 9 and ethanol (EtOH)(40/60 vol%o), which is a with steps of 0-4. In all cases, the viscosities rather low polarity solvent (dielectric constant of slurries were adjusted to I Pa s by addition of 20). The use of an efficient dispersant is necessary vent to obtain an homogeneous and stable dispersion of ceramic particles in the solvent, and to achieve 2.3 Tape casting a low viscosity with a high ceramic/organic ratio. Tape casting was performed with a laboratory a stable dispersion of deagglomerated particles tape casting bench(Cerlim Equipement Limoges, leads to a dense particle packing and to an homo- france). Slurries were tape cast onto a fixed glass geneous microstructure. According to previous plate with a moving double blade device at a con studies, 5-18 phosphate esters were chosen to dis- stant speed of I m min. The thickness of the perse Al,O3 and ZrO, powders in the MEK/EtoH green tapes was 160 um solvent. The effectiveness of different phosphate sters was evaluated using the viscosity of the 2. 4 Processing of alumina-zirconia laminar ceramic/dispersant/solvent systems composites The binder used is a polyvinyl butyral(PVB) Laminar composites consisting of stacked layers and the plasticizer a mixture of polyethylene gly- of alumina with various zirconia contents(Al,O3 col(PEG)and dibutyl phthalate(DBP) A, AL,O, 5 vol% ZrO,= AZ5. Al 10 vol% ZrO,= AZlO)were fabricated by 2.2 Slurry preparation thermocompression, pyrolysis of the organic com The preparation of slurries is carried out in two ponents and sintering. Two series of composites stages, namely (i) the deagglomeration and disper- were prepared by thermocompression of 19 to 22 sion of powders in the solvent with the aid of the single layers of different compositions with differ dispersant, and (ii) the homogenization of the ent stacking sequences the number of layers slurry with binders and plasticizers. The sequence depends on the stacking sequence. The first series f component addition is critical. The dispersant is devoted to the influence of the processing has to be added before the binders to prevent routes, transformation toughening and interfacial competitive adsorption. The initial adsorption effects. The second series consists of composites of of the binder on the particle surfaces would a and Azio layers with different stacking prevent the dispersant from being adsorbed subse- sequences. Both series are designed according to quently, thereby decreasing its effectiveness. Fur- the schematic drawings of Fig. 1. A/A. aZS/AZ5 thermore, the deagglomeration is morc cfficicnt in and AZ1O/AZ10 arc not true laminar composites a low-viscosity system (i.e. without binders and however these materials were fabricated for com- plasticizers) and the mechanical damage of parison with the properties of laminar composites binder molecules is minimized by this seq and with monolithic materials prepared by dry of addition. The deagglomeration is carried out pressing (i.e. A and AZIO by ultrasonic treatment. The second stage of The thermocompression was performed at homogenization is performed by milling for 24 h. 110 C under a pressure of 60 MPa The slurry is rotated continuously at a slow speed for de-aeration and to prevent settling 2.5 Pyrolysis and sintering All the organic components affect the rheolog Thermal debinding remains one of the most criti cal behaviour of the slurry and therefore affect the cal steps of ceramic processing and requires an properties of the green tapes. An optimized slurry efficient heating cycle to prevent stresses and the should lead to tapes which satisfy the following formation of defects in ceramic parts. According criteria: (i)no cracking during drying, (i1) high to the thermogravimetric analysis of tape-cast
300 T. Chartirr. T. Rouxrl 2 Experimental Procedure 2.1 Starting materials Tape casting slurries are complex, multicomponent systems, which contain ceramic powders (including sintering aids), solvents, dispersants, binders and plasticizers. The starting powders are 99.7 wt% purity, 0.5 pm grain size alumina (P172SB, Pechiney, France) and 97.5 wt% purity, 0.4 ,um grain size zirconia (UPH 12, Criceram, France). Tape casting slurries were prepared with an azeotropic mixture of methyl ethyl ketone (MEK) and ethanol (EtOH) (40/60 vol%), which is a rather low polarity solvent (dielectric constant = 20). The use of an efficient dispersant is necessary to obtain an homogeneous and stable dispersion of ceramic particles in the solvent, and to achieve a low viscosity with a high ceramic/organic ratio. A stable dispersion of deagglomerated particles leads to a dense particle packing and to an homogeneous microstructure. According to previous studies,‘5m’8 phosphate esters were chosen to disperse A&O, and ZrO, powders in the MEWEtOH solvent. The effectiveness of different phosphate esters was evaluated using the viscosity of the ceramic/dispersant/solvent systems. The binder used is a polyvinyl butyral (PVB) and the plasticizer a mixture of polyethylene glyco1 (PEG) and dibutyl phthalate (DBP). 2.2 Slurry preparation The preparation of slurries is carried out in two stages, namely (i) the deagglomeration and dispersion of powders in the solvent with the aid of the dispersant, and (ii) the homogenization of the slurry with binders and plasticizers. The sequence of component addition is critical. The dispersant has to be added before the binders to prevent competitive adsorption.17 The initial adsorption of the binder on the particle surfaces would prevent the dispersant from being adsorbed subsequently, thereby decreasing its effectiveness. Furthermore, the deagglomeration is more efficient in a low-viscosity system (i.e. without binders and plasticizers) and the mechanical damage of the binder molecules is minimized by this sequence of addition. The deagglomeration is carried out by ultrasonic treatment.” The second stage of homogenization is performed by milling for 24 h. The slurry is rotated continuously at a slow speed for de-aeration and to prevent settling. 2.4 Processing of alumina-zirconia laminar composites Laminar composites consisting of stacked layers of alumina with various zirconia contents (Al,O, = A, A&O, + 5 ~01% ZrOz = AZ5, A&O3 + 10 ~01% ZrO, = AZlO) were fabricated by thermocompression, pyrolysis of the organic components and sintering. Two series of composites were prepared by thermocompression of 19 to 22 single layers of different compositions with different stacking sequences. The number of layers depends on the stacking sequence. The first series is devoted to the influence of the processing routes, transformation toughening and interfacial effects. The second series consists of composites of A and AZ10 layers with different stacking sequences. Both series are designed according to the schematic drawings of Fig. 1. AIA, AZ5iAZ5 and AZlO/AZlO are not true laminar composites: however these materials were fabricated for comparison with the properties of laminar composites and with monolithic materials prepared by dry pressing (i.e. A and AZlO). The thermocompression was performed at 110°C under a pressure of 60 MPa. 2.5 Pyrolysis and sintering All the organic components affect the rheologi- Thermal debinding remains one of the most critical behaviour of the slurry and therefore affect the cal steps of ceramic processing and requires an properties of the green tapes. An optimized slurry efficient heating cycle to prevent stresses and the should lead to tapes which satisfy the following formation of defects in ceramic parts. According criteria: (i) no cracking during drying, (ii) high to the thermogravimetric analysis of tape-cast green density, (iii) good microstructural homogeneity, and (iv) good thermocompression ability. Two parameters were chosen to improve the slurry composition with regards to the crack sensitivity, density and thermocompression ability of the green tapes. The first is the volume solids ratio (X = ~01% solids = powder/powder + dispersant + binder + plasticizer) and the second is the volume binder to plasticizer ratio ( Y = binder/ plasticizer). Alumina+ 10 vol”% zirconia slurries were prepared and tape cast with volume solids values varying from 0.6 to 0.9 with steps of 0.05 and binder/plasticizer values varying from 0.3 to 1.9 with steps of 0.4. In all cases, the viscosities of slurries were adjusted to 1 Pa s by addition of solvent. 2.3 Tape casting Tape casting was performed with a laboratory tape casting bench (Cerlim Equipement, Limoges, France). Slurries were tape cast onto a fixed glass plate with a moving double blade device at a constant speed of 1 m min ‘. The thickness of the green tapes was 160 pm
Tape-cast alumina-zirconia laminates processing and mechanical properties FIRST SERIES PRESSED LAMINATED (21 LAY Load roller A210 A/A AZ5/AZ5 AZ10/AZ10AZ10/A25 Support rolle A25 +10 vol Zr02 SECOND SERIES 3A/A210A Fig. 2. Chevron-notched(CN)specimens used for the fracture 5A10 10A210 toughness and fracture energy determination 2A10 3AAz1o】A 3n5/3 2/2/2n02/2/2 span length and a cross-head speed of 0.1 mm Fig. I. Schematics of the two series of laminar composites min-l. The toughness was measured by four-point bending of4×3×50mm( height x width length)chevron-notched(CN) specimens( with a 60 V-notch angle, and a 0.15 mm samples, the pyrolysis was carried out with heat- width, inner and outer span of 20 and ing at rate of 1C min up to 120C, then with respectively, and cross-head speeds between 0-001 heating at rate of 0.lC min up to 550C with and 0.I mm min. The V-notch is cut perpendicu dwell time of 4 h. The samples were sintered in an lar to the laminae. However, two different orienta electric furnace with MoSi, heating elements, with tions of the V, relative to the orientation of the heating at rate of 5C min up to 1600C with laminae, were investigated to make the crack dwell time of 3 h propagate normal to the layers(N specimens)or transverse to them (T specimens). The mode I 2.6 Characterization fracture toughness Kic is then calculated from the Rheological measurements were performed using maximum load measured on the experimental a rotating cylinder viscometer(Rotovisco RV12, curves, using the polynomial equation of munz Haake)at a shear rate of 28s". The shear rate et al. 2 Chevron-notched specimens were used as evaluated according to the gap between because of their suitability for fracture toughnes he casting support and the moving blade, and to determination in brittle materials. Sharpness of the casting speed. the notch is not critical and in the absence of the The apparent densities of green samples, exclud- R-curve effect, Kc values are relatively indepen ing organic phases, were determined by measuring dent of the initial crack length and agree with the ht (mcal ) of samples values obtained from straight-through crack speci before and after calcination, respectively. The mens with notch thickness as low as 66 um(single apparent density is expressed by the Mcal/V ratio. edge notched beam specimens ). The elastic mod- Evaporation of solvent can causc visible crack- uli were measured by means of an ultrasonic tech ing of tapes. The tape shrinkage, and then the nique, using 10 MHz piezo-electric transducers in shrinkage rate, during drying were measured by a contact with specimens. Youngs modulus(E)and laboratory-made detector using a laser system.20 the shear modulus (G)are calculated from the ers in the green composite was detected using an shear(V) wave velocities according to 2O)and The presence of any delaminations between lay- measured values for the compressional (V)and ultrasonic method. 2 (3v2-4V3) The mechanical testing was conducted in bend E P and G=pV (1) ing on an INSTRon 8562 testing machine equipped with a differential measurement device, oy means of a Linear Variable Displacement where p is the specific mass, as measured by Transducer(LvdT)and mechanical contact with Archimedes mcthod using distilled watcr Poissons he specimen, to accurately measure the specimen ratio (v)is then given by deflection. Fracture tests were performed in three- point bending,on3×4×25mm( height width x length) rectangular bars with a 20 mm
Tape-cast alumina-zirconia laminates: processing und mechanical properties 301 FIRST SERIES PRESSED LAMINATED (21 LAYERS) A AZ10 AIA AiWAZ5 A21 OlAzlO AZ1 OIAZS A= Al203 A25= Al203+ 5VOl% 2fO2 AZ1 0 = A1203 + 10 VOl% LfD2 SECOND SERIES 3lAIAZl OIIA 2A 3(A/AZlOl/A 2A 7/5/7 3n 513 2/2/211012/212 Fig. 1. Schematics of the two series of laminar composites tested. samples, the pyrolysis was carried out with heating at rate of 1°C min’ up to 120°C then with heating at rate of O.l”C min’ up to 550°C with a dwell time of 4 h. The samples were sintered in an electric furnace with MoSi, heating elements, with heating at rate of 5°C min’ up to 1600°C with a dwell time of 3 h. 2.6 Characterization Rheological measurements were performed using a rotating cylinder viscometer (Rotovisco RV12, Haake) at a shear rate of 28 s&. The shear rate was evaluated according to the gap between the casting support and the moving blade, and to the casting speed. The apparent densities of green samples, excluding organic phases, were determined by measuring the volume (V) and weight (Mea,,) of samples before and after calcination, respectively. The apparent density is expressed by the M,,,,/V ratio. Evaporation of solvent can cause visible cracking of tapes. The tape shrinkage, and then the shrinkage rate, during drying were measured by a laboratory-made detector using a laser system.20 The presence of any delaminations between layers in the green composite was detected using an ultrasonic method.2 The mechanical testing was conducted in bending on an INSTRON 8562 testing machine equipped with a differential measurement device, by means of a Linear Variable Displacement Transducer (LVDT) and mechanical contact with the specimen, to accurately measure the specimen deflection. Fracture tests were performed in threepoint bending, on 3 X 4 X 25 mm (height X width X length) rectangular bars with a 20 mm / Load roller Fig. 2. Chevron-notched (CN) specimens used for the fracture toughness and fracture energy determinations. span length and a cross-head speed of 0.1 mm min-‘. The toughness was measured by four-point bending of 4 X 3 X 50 mm (height X width X length) chevron-notched (CN) specimens (Fig. 2), with a 60” V-notch angle, and a 0.15 mm notch width, inner and outer span of 20 and 40 mm respectively, and cross-head speeds between 0.00 1 and 0.1 mm min’. The V-notch is cut perpendicular to the laminae. However, two different orientations of the V, relative to the orientation of the laminae, were investigated to make the crack propagate normal to the layers (N specimens) or transverse to them (T specimens). The mode I fracture toughness K,, is then calculated from the maximum load measured on the experimental curves, using the polynomial equation of Munz et ~1.~’ Chevron-notched specimens were used because of their suitability for fracture toughness determination in brittle materials. Sharpness of the notch is not critical and in the absence of the R-curve effect, K,, values are relatively independent of the initial crack length and agree with the values obtained from straight-through crack specimens with notch thickness as low as 66 pm (single edge notched beam specimens).2’ The elastic moduli were measured by means of an ultrasonic technique, using 10 MHz piezo-electric transducers in contact with specimens. Young’s modulus (E) and the shear modulus (G) are calculated from the measured values for the compressional (V,) and shear (V,) wave velocities according to:22 and G = pV,’ (1) where p is the specific mass, as measured by Archimedes method using distilled water. Poisson’s ratio (v) is then given by:23 E v=__l 2G
T. Chartier T. Rou./ 3 Results and discussion VISCOSITY (mPa. s) 3.1 Tape casting of layers 1500 30vo.%Ak23 3.1. I Selection of the parameters of phosphate esters The effectiveness of the different phosphate esters HLB 1< HLB 2< HLB 3 HLB 4 was evaluated using the viscosity of the alumina HLB 2 dispersant/solvent systems. The phosphate esters used were prepared by reaction between phospho- B3HB4 cid and an ethoxylate. The ethoxy 500 obtained by condensation of ethylene oxide in alcohol. The esters contain a combination of mono- and diesters and the remains of the ethoxylate not combined with the phosphoric acid. The chemical ESTER/AL2O3(wt %) structure of a typical phosphate ester is shown in Fig. 4. Apparent viscosity of a 30 volo alumina susper Fig. 3. The infuence of four parameters was stud versus dispersant concentration for phosphate esters different hlb value ied, namely (i)the molecular structure(aliphatic or aromatic), 5( i) the monoester/diester ratio, (ii) the degree of phosphatization and (iv) the the unreacted ethoxylate disturbs ceramic/disper Hydrophile/Lipophile Balance(HLB). The addition sant interactions and increases the viscosity, and of phosphate ester to alumina results in lowering (iv) a high hLB because adsorption on charged the pH, indicating that the dispersant partially dis- ceramic particles is favoured by more hydrophilic phosphate ester, the alumina dispersants(Fig 4) powder exhibited a slightly negative surface The most efficient phosphate ester, fulfilling all positive after addition of phosphate ester suggest- cialized( Beycostat C213, CECA, France s mmer- MEK/EtOH. The surfacc reversed to these characteristics, was synthesized and ce g that the H*ions liberated on dissociation were adsorbed onto alumina particles. Phosphate esters 3. 1.2 Selection of the slurry formulation act by a combination of electrostatic and steric For Al2O3 +10 vol% ZrO2 green sheets, cracking repulsion. The steric hindrance prevents contact develops for a volume solids ratio(X) higher than between particles. The double layer, which may be 0. 75 and for a binder/plasticizer ratio (r) higher due to net charge on the particle surface and/ than 0-7(Table 1). Cracking depends on shrinkage or charges coming from the dissociation of the rate, which depends itself on the composition of adsorbed polymer, provides repulsion by a poten- the slurry. The maximum shrinkage rate decreases tial energy barrier at larger distances when the volume solids ratio decreases. The high The best dispersion (i.e. the lowest viscosity) is content of organic phase slows down the motion achieved with a phosphate ester having (i)an of particles, then reduces the shrinkage rate aliphatic molecule, (ii)a high diester concentra- Binders are polymeric molecules which adsorb tion, as diesters contain two lipophilic tails, each on the particle surfaces and form organic bridges able to extend into the solvent for steric sta biliza I'able 1. Cracking during drying (C: cracking, N C:non tion. (ii)a high degree of phosphatization because cracking) and green density, excluding the organic phase of alumina+10 vol zirconia tapes (theoretical density = 4. 19 g MONOESTER Composition Cracking Green densit OH HH HH Y HO-P-0-C-C-0)(C)-C-H HH HYDROPHILE LIPOPHILE < CRACKING LIMIT DIESTER OH 2.64 RO-P-OR RACKING LIMIT 9 4 Fig. 3. Typical chemical structure of a phosphate ester
302 T. Churtier. T. Rouse1 3 Results and Discussion 3.1 Tape casting of layers 3. I. 1 Selection of’ the parameters oj’phosphate esters The effectiveness of the different phosphate esters was evaluated using the viscosity of the alumina/ dispersant/solvent systems. The phosphate esters used were prepared by reaction between phosphoric acid and an ethoxylate. The ethoxylate is obtained by condensation of ethylene oxide in alcohol, The esters contain a combination of monoand diesters and the remains of the ethoxylate not combined with the phosphoric acid. The chemical structure of a typical phosphate ester is shown in Fig. 3. The influence of four parameters was studied, namely (i) the molecular structure (aliphatic or aromatic),15 (ii) the monoester/diester ratio, (iii) the degree of phosphatization, and (iv) the Hydrophile/Lipophile Balance (HLB). The addition of phosphate ester to alumina results in lowering the pH, indicating that the dispersant partially dissociates. Without phosphate ester, the alumina powder exhibited a slightly negative surface charge in MEWEtOH. The surface reversed to positive after addition of phosphate ester suggesting that the H’ ions liberated on dissociation were adsorbed onto alumina particles. Phosphate esters act by a combination of electrostatic and steric repulsion. The steric hindrance prevents contact between particles. The double layer, which may be due to net charge on the particle surface and/ or charges coming from the dissociation of the adsorbed polymer, provides repulsion by a potential energy barrier at larger distances. The best dispersion (i.e. the lowest viscosity) is achieved with a phosphate ester having (i) an aliphatic molecule, (ii) a high diester concentration, as diesters contain two lipophilic tails, each able to extend into the solvent for steric stabilization, (iii) a high degree of phosphatization because MONOESTER R I 0,H 7 lyi 7 c;’ ’ HO-P-0-(C-C-0)-U-C-H c; AA “A’A i._ _I I J HYDROPHILE LIPOPHILE DIESTER OIH RO-P-OR t; Fig. 3. Typical chemical structure of a phosphate ester. VISCOSITY 1mPa.s) 2.000 ,,500_ -2 3Ovol.% A’2O3 1 HLE 1 < HLB 2 < “LB 3 < HLB 4 0 0.2 O-4 0.6 0.8 ESTER/Al,O,(wt.%) Fig. 4. Apparent viscosity of a 30 vol’%, alumina suspension versus dispersant concentration for phosphate esters with different HLB values. the unreacted ethoxylate disturbs ceramicidispersant interactions and increases the viscosity, and (iv) a high HLB because adsorption on charged ceramic particles is favoured by more hydrophilic dispersants (Fig. 4). The most efficient phosphate ester, fulfilling all these characteristics, was synthesized and commercialized (Beycostat C213, CECA, France).” 3.1.2 Selection of the slurry ,formulation For Al,O,+ 10 vol’/o ZrO, green sheets, cracking develops for a volume solids ratio (X) higher than 0.75 and for a binder/plasticizer ratio (Y) higher than 0.7 (Table 1). Cracking depends on shrinkage rate, which depends itself on the composition of the slurry. The maximum shrinkage rate decreases when the volume solids ratio decreases. The high content of organic phase slows down the motion of particles, then reduces the shrinkage rate. Binders are polymeric molecules which adsorb on the particle surfaces and form organic bridges Table 1. Cracking during drying (C.: cracking, N.C.: noncracking) and green density, excluding the organic phase, of alumina+ 10 vol”% zirconia tapes (theoretical density = 4.19 g cm ‘) for various values of X and Y X1 O-6 0.7 N.C. 3.78 _- x3 0.7 0.7 N.C. 2.48 x4 0.75 0.7 N.C. 2.52 X5< CRACKING LIMIT > 0.8 0.7 c. 2.56 x7 0.9 0.7 c. 2.60 Yl o-7 0.3 N.C. 2.64 Y2 o-7 o-7 N.C. 2.48 Y3< CRACKING LIMIT > o-7 1.1 C. 2.45 Y4 0.7 I.5 C. 2.43 Y5 o-7 I .9 C. 2.40
Tape-cast alumina-zirconia laminates: processing and mechanical properties Table 2. Composition of the retained tape casting slurry Table 3. Physical properties of the various grades tested. E and G are measured normal to the direction of the layering Function Volv Poissons ratio is equal to 0- 24 for all the grades Alumina t Zirconia Ceramic powder 306 Material E(GPa) G(GPa Solvent Dispersant PVB A 143 PEG 300 Plasticizer Plasticizer AZo 880 AZS/AZ4 9.06 AZ 342 AZIO/AZ5 between them. The shrinkage rate increases when 7/5/7 362 the binder/plasticizer ratio increases because the 3/15/3 high binder content increases the links between 2/2/2102/2/2 particles and then the shrinkage rate. In order to avoid cracking during drying, the formulation of the slurry has to be defined to achieve a low way that compressive residual stresses develop cizer-rich organIc phase 6 igh amount of plasti- the outer layers. These residual stresses were shrinkage rate i.e. with a n the of the azio induced upon cooling from high temperature system, the shrinkage rate does not exceed a value (1400C), as a result of different thermal expan of I um s sion coefficients between the layers. The classical The apparent densities of dried tapes, excluding plate theory(plane stress hypothesis) was used to the organic phase, were determined for various calculate the normal stresses in each layer of the volume solids and binder/plasticizer ratios. When two-phase composite. 4 In its main lines, the cal- the volume solids ratio decreases, the organic phase culation is based on the assumption that a cross- prevent particles from packing together, therefore section before deformation remains a cross-section he density decreases. A low binder/plasticizer after deformation and that layers remain plane ratio leads to green tapes with higher densiti and parallel to each other throughout deformation When the binder/plasticizer ratio decreases, the This results in the following form for the elastic low-viscosity plasticizer-rich organic phase allows displacement vector at any point (x, x2, x3) a good packing of ceramic particles l{(x1,x2) 3. 1.3 Thermocompression ability u=2=2(x1,x2 The thermocompression behaviour of green sheets is sensitive to x and Y. delaminations were numerous in samples with low quantities of organic where (1, 2) subscripts refer to the in-plane axis phases (X4 to X7)(Table 1). a lot of delamina- and (3)is normal to the layering The studied lam tions were observed in samples with a high quan- inates are constructed such that they have com tity of plasticizer (Y1). Plasticizers are low plete symmetry of individual lamina thickness and molecular weight species which can act as a lubri- properties about the middle plane of the laminate cant between the individual layers, limiting the Furthermore, no texture was introduced by the bond strength during thermocompression tape casting process, so that the studied laminates The final formulation of tapc casting slurries can be considered as perfectly orthotropic. Hence, was defined in order to tape cast non-cracked in the case of pure linear elasticity, the thermal tapes with a high green density, which do not lead induced normal(N, and N2) and tangential (T12 to delamination during thermocompression of forces, per unit length, are given by: 4 laminated composites. This formulation (Table 2) was applied to the tape casting of pure alumina N1「A1A1201e and of AlO3+5 vol%o ZrO, with similar green tape A21A220 characteristics and a good thermocompression 00 AbLE ability As= 2Gd 3. 2 Laminar composites I du: du 3. 2. I Design and e sical f monoliths and com posites are given in Table 3. These physical prop- One recalls that eqns(4)only stand for the elastic erties were used to design the composites in such a part of the strain components(elastic gtotal-gthermal)
Tape-cast alumina-zirconia laminates: processing and mechanical properties 303 Table 2. Composition of the retained tape casting slurry Component Function Vol’% _ Alumina + Zirconia MEKlethanol Phosphate ester PVB PEG 300 DBP Ceramic powder 30.6 Solvent 57.6 Dispersant 0.8 Binder 4.6 Plasticizer 2.9 Plasticizer 3.5 between them. The shrinkage rate increases when the binder/plasticizer ratio increases because the high binder content increases the links between particles and then the shrinkage rate. In order to avoid cracking during drying, the formulation of the slurry has to be defined to achieve a low shrinkage rate, i.e. with a high amount of plasticizer-rich organic phase. In the case of the AZ10 system, the shrinkage rate does not exceed a value of 1 pm ss’. The apparent densities of dried tapes, excluding the organic phase, were determined for various volume solids and binder/plasticizer ratios. When the volume solids ratio decreases, the organic phase prevent particles from packing together, therefore the density decreases. A low binder/plasticizer ratio leads to green tapes with higher densities. When the binder/plasticizer ratio decreases, the low-viscosity plasticizer-rich organic phase allows a good packing of ceramic particles. 3.1.3 Thermocompression ability The thermocompression behaviour of green sheets is sensitive to X and Y. Delaminations were numerous in samples with low quantities of organic phases (X4 to X7) (Table 1). A lot of delaminations were observed in samples with a high quantity of plasticizer (Yl). Plasticizers are low molecular weight species which can act as a lubricant between the individual layers, limiting the bond strength during thermocompression. The final formulation of tape casting slurries was defined in order to tape cast non-cracked tapes with a high green density, which do not lead to delamination during thermocompression of laminated composites. This formulation (Table 2) was applied to the tape casting of pure alumina and of A&0,+5 vol% ZrO, with similar green tape characteristics and a good thermocompression ability. 3.2 Laminar composites 3.2. I Design Some physical properties of monoliths and composites are given in Table 3. These physical properties were used to design the composites in such a Table 3. Physical properties of the various grades tested. E and G are measured normal to the direction of the layering. Poisson’s ratio is equal to 0.24 for all the grades Material E (GPa) G (GPa) (YXJ IJllV~ (IO” “C 1) A 355 143 8.69 Al A 363 146 AZ10 341 137 8.80 AZSIAZS 349 141 9.06 AZlOlAZlO 342 138 AZIOIAZS 71517 362 146 311513 349 141 212121 Io/21212 377 152 way that compressive residual stresses develop in the outer layers. These residual stresses were induced upon cooling from high temperature (14OO”C), as a result of different thermal expansion coefficients between the layers. The classical plate theory (plane stress hypothesis) was used to calculate the normal stresses in each layer of the two-phase composite.24 In its main lines, the calculation is based on the assumption that a crosssection before deformation remains a cross-section after deformation and that layers remain plane and parallel to each other throughout deformation. This results in the following form for the elastic displacement vector at any point (x,, x2, x3): [ Ul = r&x,, x2) ii = u2 = l&x,, x2) (3) u3 = &X3> where (1,2) subscripts refer to the in-plane axis and (3) is normal to the layering. The studied laminates are constructed such that they have complete symmetry of individual lamina thickness and properties about the middle plane of the laminate. Furthermore, no texture was introduced by the tape casting process, so that the studied laminates can be considered as perfectly orthotropic. Hence, in the case of pure linear elasticity, the thermally induced normal (N, and N,) and tangential (T,?) forces, per unit length, are given by:24 Ed with A,, = - vEd 1 - v2’ A,, = A,, = ~ 1 --I and E.. = ! %! + ?!! ” 2 [ aXj aXi 1 ’ (4) A,, = 2Gd One recalls that eqns (4) only stand for the elastic part of the strain components (.?lastic = E”‘~’ -E~~~““~‘)