- Email: [email protected]
Tape casting of high dielectric ceramic composite substrates for microelectronics application
Journal of Materials Processing Technology 89±90 (1999) 508±512
Tape casting of high dielectric ceramic composite substrates for microelectronics application Alfred I.Y. Tok*, Freddy Y.C. Boey, K.A. Khor Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore
Abstract The current device of miniaturization and higher device counts in IC (integrated circuit) packages has signi®cantly increased the use of both multilayer ceramic packages (MLCP) and multilayer capacitors (MLC). Crucial to the production of these ceramic substrates is the use of a tape cater with precise control of tape thickness, uniformity and surface ®nishes. The control of green tape thickness is crucial in governing the properties of tape cast products. Currently, the production and accurate reproduction of smaller and thinner substrates are limited by several factors, one of which is the precise control of the green tape thickness. This is in turn in¯uenced by the various processing parameters and formulation of the materials used. Tape casting was performed on a batch process type caster, and the tapes were cast using varying blade gaps and casting speeds. It can be concluded that a higher degree of accuracy in tape thickness can be obtained when casting at lower casting speeds. However, this accuracy decreases when thinner tapes are required, and the optimized use of an increasing casting speed would be needed to maintain this accuracy. An optimal setting between the casting speed and the blade gap setting is, therefore, needed in order to achieve accurate and reproducible results when performing tape casting. Determination of this optimum combination would depend on the required thickness of the ®nal product. # 1999 Published by Elsevier Science S.A. All rights reserved. Keywords: Ceramics; Tape cast; Alumina; Tape thickness; Doctor blade
1. Introduction The trend in the microelectronics packaging industry towards greater densi®cation, miniaturization and higher computing speeds has imposed increasing demands on materials for higher thermal properties, better dielectric strength and improved reliability, leading to the increased usage of ceramics in the form of multilayer substrates, for example in pin grid array (PGA) and ball grid array (BGA) applications. Current usage of high performance systems such as high purity alumina substrates has resulted in the use of thinner substrates. Currently, one of the main methods used for the manufacture of ¯at ceramics substrates with precisely controlled thickness and consistency is the tape casting technique. This method basically starts with a specially formulated slurry which can be cast by a blade to a ¯at sheet or layer, then dried into a ¯exible solid tape which can be sintered subsequently into a hard ceramic substrate layer.
*Corresponding author. E-mail: [email protected]/[email protected]
The applications of tape casting are numerous, some of which are Al2O3 and AlN substrates for thin and thick ®lm circuitry [1], piezoelectric ceramics for actuators [2], photovoltaic solar cells (Silicon-On-Ceramics: SOC) ± using mullite substrates for supports due to their thermal expansion coef®cient being similar to that of silicon [3]. A more comprehensive listing of its applications can be found in reference [4]. A customized tape caster with a special pressure control unit which is able to produce thin but consistent and very ¯at substrates have been developed for this project. Current research work is comprehensively separated into three areas, material and process formulation for a composite slurry based on alumina/aluminium nitride, development of a control system and a process model to produce required tape cast sheets with the required dimensional accuracy and consistency, and ®nally development of an optimal debinding and sintering process for the ®nal composite multilayered substrates. In the tape casting process, ceramic particles and other essential additives are dispersed in a solvent to form a slurry. The slurry is layered onto a carrier by passing it under a doctor blade of pre-set gap either onto a moving carrier
0924-0136/99/$ ± see front matter # 1999 Published by Elsevier Science S.A. All rights reserved. PII: S 0 9 2 4 - 0 1 3 6 ( 9 9 ) 0 0 1 3 1 - 4
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
509
Fig. 1. A continuous tape caster.
(Fig. 1), or onto a ®xed casting surface using a traversing casting head. The slurry is then allowed to dry, and a thin ceramic green tape is formed. After drying, a high concentration of organic components remain in the tape, which must be removed by pyrolysis. This burning out of organic compounds generates an open porosity in the green tape, which must be eliminated by sintering, a high diffusion process. The control of green tape thickness is crucial in governing the properties of the tape cast products. Currently, the production and accurate reproduction of smaller and thinner substrates are limited by several factors, one of which is the precise control of the green tape thickness. This is in turn in¯uenced by the various processing parameters and formulation of the materials used. The processing parameters and formulation for an Al2O3 substrate has been optimized [4], and this was later adapted to the tape casting of a composite Al2O3/AlN substrate [5]. In order to predict the tape thickness more accurately, theoretical modeling of the ¯ow characteristics has been used. Chou et al. [6] analyzed the tape casting ¯ow behavior using the principles of ¯uid dynamics. They assumed a Newtonian ¯ow for the slurry in a parallel channel consisting of both a Couette and Poiseuille ¯ow, and obtained the following equation for predicting tape thickness: ho 1 h2o P (1) tape 2 tape 6UL In their analysis, their volumetric ¯owrate was derived from the linear addition of a Pressure (Poiseuille) and Drag (Couette) ¯ow within the tape casting channel. Their derived volumetric ¯owrate was: 1 h2o P (2) Qlinear ho W U 2 6UL The tape thickness equation basically indicates a polynomial relationship between the tape thickness, tape and the blade gap, ho. Decrease or increase in the blade gap can be effected linearly by changing the slurry pressure head, P or the casting speed. As a tape casting slurry seldom behaves as a Newtonian ¯uid, other attempts at modeling tape casting ¯ow behavior have been done. These include work by Ring [7], who used the Bingham constitutive equation for a non-Newtonian
Fig. 2. Tape casting flowchart.
slurry, and Huang et al. [8] who assumes the slurry to behave according to the Herschel±Bulkley model. 2. Experimental work The overall process steps for producing the ceramic tapes have been summarized in the ¯ow chart in Fig. 2. Tape casting was performed using the formulation as listed in Table 1 below. A dual solvent was chosen to improve the organic solubility, and prevent preferential volatilisation and polymeric surface skin formation. The sequence of component addition affects the viscosity of the slurry, due to preferential adsorption and water adsorption effects [9,10]. It was critical that the de¯occulant be introduced independent of the other polymers, to prevent competition for adsorption sites on the ceramic particulate surface, and therefore produce a more uniform slurry viscosity [11]. The plasticizer was added to the slurry before the binder as this aided the dissolution of the binder into the overall slurry, due to the preferred solubility of the binder in the plasticizer relative to the solvent system. The alumina powder was mixed with a compatible sintering aid, magnesium oxide and ball milled at 60 rpm for 24 h Table 1 Tape casting formulation Component
Function
Aluminium oxide Magnesium oxide Trichloroethylene Ethanol Corn Oil Polyethylene glycol Benzyl butyl phthalate Polyvinyl Butyral
Ceramic substrate Sintering additive Solvent Solvent Deflocculant Plasticizer Plasticizer Binder
510
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
Fig. 4. Plot of the green tape thickness obtained at different casting speeds for three different blade gaps. Fig. 3. Tape casting in NTU.
in a corundum milling jar with high-purity alumina ball charges. Ball-to-weight ratio used was 15 : 1. After ball milling, the def¯oculant and solvents were added to the mixed powders to form a slurry. This slurry was then placed in an ultrasonic bath and ultrasonically agitated for 45 min at 47 kHz, 35 W. Sequential addition of plasticizer and binder to the slurry was performed after ultrasonic deagglomeration and the slurry was mechanically mixed for 2 h to dissolve the organic additives in the solvents, and to achieve a homogeneous mix. After mixing was complete, the homogenous slurry was then transferred to a vacuum chamber where it is de-aired at 700 mmHg (90 kPa) for 20 min. Finally, the slurry was sieved into the casting chamber via two nylon sieves of openings 0.032 and 0.020 mm to remove any remaining agglomerates, aggregates, milling debris and contaminants. Tape casting was performed on a batch process type caster, where the casting head and doctor blade traversed over a stationary ¯oating glass slab, discharging slurry onto the surface (Fig. 3). The doctor blade gap was varied by a micrometer screw gauge which could achieve an accuracy of upto 0.01 mm. Accurate casting velocity control was achieved via a stepping motor controlled by a motor controller. Tapes were cast using varying blade gaps and casting speeds. The slurry formulation and other processing parameters such as casting temperature and pressure for all the casts were kept constant. The tape thickness was accurately measured at varying locations on the cast tape and an average value was derived. Accurate measurement was achieved by sandwiching the tape between two pieces of thin glass of known thickness, and using a micrometer to measure the total thickness. This method of measurement eased pressure from the micrometer jaws, which might cause that small portion of tape to compress and give inaccurate results.
established. The following results were obtained from recent work done. [12]. Fig. 4 plots the tape thickness obtained as the casting speed was changed, for three different blade gaps. The thickness was seen to decrease exponentially with increases in the casting speed, for all three blade gaps. It is noted that this change in the blade gaps did not result in a proportional change in the tape thickness. This is more clearly shown when the tape thickness is plotted against the blade gap, as shown in Fig. 5 for the three casting speeds of 5, 10 and 20 mm/s. Fig. 5 clearly shows that there was a maximum achievable tape thickness for a given casting speed. The tape thickness increased signi®cantly for very small blade gaps, but rapidly leveled off to a constant thickness past a certain blade gap value. It is also noted that this limiting blade gap was smaller for faster casting speeds. This has been plotted in Fig. 6. The above results highlight the dif®culty involved in achieving consistent tape thickness, especially for thin tapes. As shown in Fig. 7, ideally, the desired tape thickness required should be obtained when the thickness becomes independent of the blade gap, and depends only on the casting speed. Fig. 6 also shows that a faster casting speed will lower the tape thickness obtainable. However, it would appear that there is a limiting thickness that can be achieved, even at very fast casting speed, for a given pressure and slurry viscosity. Further work was done by analyzing the physics of the ¯uid ¯ow process. It was suggested in the Newtonian ¯uid ¯ow model [6] that the total volumetric ¯owrate can be obtained by the linear addition of both the pressure
3. Results and discussion In order to build on an accurate model to describe the ¯ow behavior and predict tape thickness, various trends had to be
Fig. 5. Plot of the green tape thickness obtained at different blade gaps for three different casting speeds.
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
511
Fig. 8. Schematic diagram of fluid velocity profile. Fig. 6. Plot of the maximum green tape thickness obtained at different casting speeds.
(Poiseuille) and drag (Couette) ¯ow.A simple pressure ¯ow model between parallel plates was analyzed. It has been well established that the pressure ¯ow (without drag) ¯uid velocity pro®le between two parallel plates would be parabolic (Fig. 8), and this ¯owrate would be described by the equation: Q
h3o WP 12L
(3)
It is noted that the above schematic could also represent the ¯uid ¯ow velocity pro®le in a tape casting system from y ho/2 to y ho, where the belt is moving at a velocity of exactly Umax at y ho/2. The volumetric ¯owrate for this tape casting system would be exactly half the pressure ¯owrate as depicted in Eq. (3). Using this analysis, the volumetric ¯owrate for a blade gap of Xo was derived and compared to the linear derived volumetric ¯owrate, both at a belt velocity of Umax. From results, it is seen that there is a difference of 12.5% between the two predictions as shown in Eqs. (4) and (5). Qlinear
7h3o WP 12L
(4)
Qlinear
8h3o WP 12L
(5)
The above equations thus, propose that the linear addition of a pressure (Poiseuille) and drag (Couette) ¯ow [6] does not hold for a tape casting system. Eq. (5) which was derived based on fundamental ¯uid ¯ow offers a more logical description of the physics of the tape casting system.
Fig. 7. Plot of the blade gap setting at which the maximum green tape thickness occurs for different casting speeds.
Work is currently being done to derive the ¯uid ¯ow equations using the approach as proposed above using a nonNewtonian (shear thinning) constitutive model, as rheology experiments performed show that the tape casting slurry behaves as a non-Newtonian shear thinning ¯uid [12]. This will then be veri®ed against experimental results. 4. Conclusions From the results, it can be concluded that a higher degree of accuracy in tape thickness can be obtained when casting at lower casting speeds. However, this accuracy decreases when thinner tapes are required, and the optimized use of an increased casting speed would be needed to maintain this accuracy. An optimal setting between the casting speed and the blade height setting is, therefore, needed in order to achieve accurate and reproducible results when performing tape casting. Determination of this optimum combination would depend on the required thickness of the ®nal product. Theoretical modeling can be used to verify the reliability of tape thickness predictions in the tape casting system. Therefore, work is currently focused upon this area to derive accurate models to describe the ¯uid behavior, and to accurately predict the cast tape thickness. References [1] G. Fischer, Electronic products weathering market swings, Am. Ceram. Soc. Bull. 66 (1987) 636. [2] S. Takahashi, Longitudinal mode multilayer piezoelectric actuators, Am. Ceram. Soc. Bull. 65 (1986) 1156±1159. [3] C. Fiori, G. De Portu, Tape casting: A technique for preparing and studying new materials, British Ceram. Proc. 38 (1986) 213± 225. [4] A.I.Y Tok, F.Y.C. Boey, K.A. Khor, Tape casting of high dielectric ceramics composite/hybrid substrates for microelectronics application, in: Proc. 6th Int. Conf. on Processing and Fabrication of Advanced Materials, Singapore, 1997. [5] A.I.Y. Tok, F.Y.C. Boey, K.A. Khor, Development of High Dielectric Al2O3/AlN Composite Substrates for Microelectronics Application, Materials & Manufacturing Processes, in press. [6] Y. Chou, Y. Ko, M. Yan, Fluid flow model for ceramic tape casting, J. Am. Ceram. Soc. 70(10) (1987) C280±282. [7] T.A. Ring, A model of tape casting Bingham plastic and Newtonian fluids, Adv. Ceram. 26 (1989) 569±576.
512
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
[8] X.Y. Huang, C.Y. Liu, H.Q. Gong, A viscoplastic flow modeling of ceramic tape casting, Materials and Manufacturing Processes, 12(5) (1997). [9] W.R. Cannon, R. Becker, K.R. Mikeska, Interactions Among Organic Additives Used for Tapecasting, Advances in: M.F. Yan (Ed.), Ceramics, vol. 26, Am. Ceram. Soc. Ohio, 1989, pp. 525±541. [10] D.J. Shanefield, Competing adsorptions in tape casting advances in ceramics, Am. Ceram. Soc 19 (1988) 155±160.
[11] L. Braun, J.R. Morris Jr., W.R. Cannon, Viscosity of tape-casting slips, Am. Ceram. Soc. Bull. 64(5) (1985) 727±729. [12] A.I.Y. Tok, F.Y.C. Boey, K.A. Khor, Tape casting of high dielectric ceramics substrates for microelectronics packaging, J. Mater. Eng. Performance, ASM Int'l, in press.
Tape casting of high dielectric ceramic composite substrates for microelectronics application Alfred I.Y. Tok*, Freddy Y.C. Boey, K.A. Khor Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore
Abstract The current device of miniaturization and higher device counts in IC (integrated circuit) packages has signi®cantly increased the use of both multilayer ceramic packages (MLCP) and multilayer capacitors (MLC). Crucial to the production of these ceramic substrates is the use of a tape cater with precise control of tape thickness, uniformity and surface ®nishes. The control of green tape thickness is crucial in governing the properties of tape cast products. Currently, the production and accurate reproduction of smaller and thinner substrates are limited by several factors, one of which is the precise control of the green tape thickness. This is in turn in¯uenced by the various processing parameters and formulation of the materials used. Tape casting was performed on a batch process type caster, and the tapes were cast using varying blade gaps and casting speeds. It can be concluded that a higher degree of accuracy in tape thickness can be obtained when casting at lower casting speeds. However, this accuracy decreases when thinner tapes are required, and the optimized use of an increasing casting speed would be needed to maintain this accuracy. An optimal setting between the casting speed and the blade gap setting is, therefore, needed in order to achieve accurate and reproducible results when performing tape casting. Determination of this optimum combination would depend on the required thickness of the ®nal product. # 1999 Published by Elsevier Science S.A. All rights reserved. Keywords: Ceramics; Tape cast; Alumina; Tape thickness; Doctor blade
1. Introduction The trend in the microelectronics packaging industry towards greater densi®cation, miniaturization and higher computing speeds has imposed increasing demands on materials for higher thermal properties, better dielectric strength and improved reliability, leading to the increased usage of ceramics in the form of multilayer substrates, for example in pin grid array (PGA) and ball grid array (BGA) applications. Current usage of high performance systems such as high purity alumina substrates has resulted in the use of thinner substrates. Currently, one of the main methods used for the manufacture of ¯at ceramics substrates with precisely controlled thickness and consistency is the tape casting technique. This method basically starts with a specially formulated slurry which can be cast by a blade to a ¯at sheet or layer, then dried into a ¯exible solid tape which can be sintered subsequently into a hard ceramic substrate layer.
*Corresponding author. E-mail: [email protected]/[email protected]
The applications of tape casting are numerous, some of which are Al2O3 and AlN substrates for thin and thick ®lm circuitry [1], piezoelectric ceramics for actuators [2], photovoltaic solar cells (Silicon-On-Ceramics: SOC) ± using mullite substrates for supports due to their thermal expansion coef®cient being similar to that of silicon [3]. A more comprehensive listing of its applications can be found in reference [4]. A customized tape caster with a special pressure control unit which is able to produce thin but consistent and very ¯at substrates have been developed for this project. Current research work is comprehensively separated into three areas, material and process formulation for a composite slurry based on alumina/aluminium nitride, development of a control system and a process model to produce required tape cast sheets with the required dimensional accuracy and consistency, and ®nally development of an optimal debinding and sintering process for the ®nal composite multilayered substrates. In the tape casting process, ceramic particles and other essential additives are dispersed in a solvent to form a slurry. The slurry is layered onto a carrier by passing it under a doctor blade of pre-set gap either onto a moving carrier
0924-0136/99/$ ± see front matter # 1999 Published by Elsevier Science S.A. All rights reserved. PII: S 0 9 2 4 - 0 1 3 6 ( 9 9 ) 0 0 1 3 1 - 4
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
509
Fig. 1. A continuous tape caster.
(Fig. 1), or onto a ®xed casting surface using a traversing casting head. The slurry is then allowed to dry, and a thin ceramic green tape is formed. After drying, a high concentration of organic components remain in the tape, which must be removed by pyrolysis. This burning out of organic compounds generates an open porosity in the green tape, which must be eliminated by sintering, a high diffusion process. The control of green tape thickness is crucial in governing the properties of the tape cast products. Currently, the production and accurate reproduction of smaller and thinner substrates are limited by several factors, one of which is the precise control of the green tape thickness. This is in turn in¯uenced by the various processing parameters and formulation of the materials used. The processing parameters and formulation for an Al2O3 substrate has been optimized [4], and this was later adapted to the tape casting of a composite Al2O3/AlN substrate [5]. In order to predict the tape thickness more accurately, theoretical modeling of the ¯ow characteristics has been used. Chou et al. [6] analyzed the tape casting ¯ow behavior using the principles of ¯uid dynamics. They assumed a Newtonian ¯ow for the slurry in a parallel channel consisting of both a Couette and Poiseuille ¯ow, and obtained the following equation for predicting tape thickness: ho 1 h2o P (1) tape 2 tape 6UL In their analysis, their volumetric ¯owrate was derived from the linear addition of a Pressure (Poiseuille) and Drag (Couette) ¯ow within the tape casting channel. Their derived volumetric ¯owrate was: 1 h2o P (2) Qlinear ho W U 2 6UL The tape thickness equation basically indicates a polynomial relationship between the tape thickness, tape and the blade gap, ho. Decrease or increase in the blade gap can be effected linearly by changing the slurry pressure head, P or the casting speed. As a tape casting slurry seldom behaves as a Newtonian ¯uid, other attempts at modeling tape casting ¯ow behavior have been done. These include work by Ring [7], who used the Bingham constitutive equation for a non-Newtonian
Fig. 2. Tape casting flowchart.
slurry, and Huang et al. [8] who assumes the slurry to behave according to the Herschel±Bulkley model. 2. Experimental work The overall process steps for producing the ceramic tapes have been summarized in the ¯ow chart in Fig. 2. Tape casting was performed using the formulation as listed in Table 1 below. A dual solvent was chosen to improve the organic solubility, and prevent preferential volatilisation and polymeric surface skin formation. The sequence of component addition affects the viscosity of the slurry, due to preferential adsorption and water adsorption effects [9,10]. It was critical that the de¯occulant be introduced independent of the other polymers, to prevent competition for adsorption sites on the ceramic particulate surface, and therefore produce a more uniform slurry viscosity [11]. The plasticizer was added to the slurry before the binder as this aided the dissolution of the binder into the overall slurry, due to the preferred solubility of the binder in the plasticizer relative to the solvent system. The alumina powder was mixed with a compatible sintering aid, magnesium oxide and ball milled at 60 rpm for 24 h Table 1 Tape casting formulation Component
Function
Aluminium oxide Magnesium oxide Trichloroethylene Ethanol Corn Oil Polyethylene glycol Benzyl butyl phthalate Polyvinyl Butyral
Ceramic substrate Sintering additive Solvent Solvent Deflocculant Plasticizer Plasticizer Binder
510
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
Fig. 4. Plot of the green tape thickness obtained at different casting speeds for three different blade gaps. Fig. 3. Tape casting in NTU.
in a corundum milling jar with high-purity alumina ball charges. Ball-to-weight ratio used was 15 : 1. After ball milling, the def¯oculant and solvents were added to the mixed powders to form a slurry. This slurry was then placed in an ultrasonic bath and ultrasonically agitated for 45 min at 47 kHz, 35 W. Sequential addition of plasticizer and binder to the slurry was performed after ultrasonic deagglomeration and the slurry was mechanically mixed for 2 h to dissolve the organic additives in the solvents, and to achieve a homogeneous mix. After mixing was complete, the homogenous slurry was then transferred to a vacuum chamber where it is de-aired at 700 mmHg (90 kPa) for 20 min. Finally, the slurry was sieved into the casting chamber via two nylon sieves of openings 0.032 and 0.020 mm to remove any remaining agglomerates, aggregates, milling debris and contaminants. Tape casting was performed on a batch process type caster, where the casting head and doctor blade traversed over a stationary ¯oating glass slab, discharging slurry onto the surface (Fig. 3). The doctor blade gap was varied by a micrometer screw gauge which could achieve an accuracy of upto 0.01 mm. Accurate casting velocity control was achieved via a stepping motor controlled by a motor controller. Tapes were cast using varying blade gaps and casting speeds. The slurry formulation and other processing parameters such as casting temperature and pressure for all the casts were kept constant. The tape thickness was accurately measured at varying locations on the cast tape and an average value was derived. Accurate measurement was achieved by sandwiching the tape between two pieces of thin glass of known thickness, and using a micrometer to measure the total thickness. This method of measurement eased pressure from the micrometer jaws, which might cause that small portion of tape to compress and give inaccurate results.
established. The following results were obtained from recent work done. [12]. Fig. 4 plots the tape thickness obtained as the casting speed was changed, for three different blade gaps. The thickness was seen to decrease exponentially with increases in the casting speed, for all three blade gaps. It is noted that this change in the blade gaps did not result in a proportional change in the tape thickness. This is more clearly shown when the tape thickness is plotted against the blade gap, as shown in Fig. 5 for the three casting speeds of 5, 10 and 20 mm/s. Fig. 5 clearly shows that there was a maximum achievable tape thickness for a given casting speed. The tape thickness increased signi®cantly for very small blade gaps, but rapidly leveled off to a constant thickness past a certain blade gap value. It is also noted that this limiting blade gap was smaller for faster casting speeds. This has been plotted in Fig. 6. The above results highlight the dif®culty involved in achieving consistent tape thickness, especially for thin tapes. As shown in Fig. 7, ideally, the desired tape thickness required should be obtained when the thickness becomes independent of the blade gap, and depends only on the casting speed. Fig. 6 also shows that a faster casting speed will lower the tape thickness obtainable. However, it would appear that there is a limiting thickness that can be achieved, even at very fast casting speed, for a given pressure and slurry viscosity. Further work was done by analyzing the physics of the ¯uid ¯ow process. It was suggested in the Newtonian ¯uid ¯ow model [6] that the total volumetric ¯owrate can be obtained by the linear addition of both the pressure
3. Results and discussion In order to build on an accurate model to describe the ¯ow behavior and predict tape thickness, various trends had to be
Fig. 5. Plot of the green tape thickness obtained at different blade gaps for three different casting speeds.
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
511
Fig. 8. Schematic diagram of fluid velocity profile. Fig. 6. Plot of the maximum green tape thickness obtained at different casting speeds.
(Poiseuille) and drag (Couette) ¯ow.A simple pressure ¯ow model between parallel plates was analyzed. It has been well established that the pressure ¯ow (without drag) ¯uid velocity pro®le between two parallel plates would be parabolic (Fig. 8), and this ¯owrate would be described by the equation: Q
h3o WP 12L
(3)
It is noted that the above schematic could also represent the ¯uid ¯ow velocity pro®le in a tape casting system from y ho/2 to y ho, where the belt is moving at a velocity of exactly Umax at y ho/2. The volumetric ¯owrate for this tape casting system would be exactly half the pressure ¯owrate as depicted in Eq. (3). Using this analysis, the volumetric ¯owrate for a blade gap of Xo was derived and compared to the linear derived volumetric ¯owrate, both at a belt velocity of Umax. From results, it is seen that there is a difference of 12.5% between the two predictions as shown in Eqs. (4) and (5). Qlinear
7h3o WP 12L
(4)
Qlinear
8h3o WP 12L
(5)
The above equations thus, propose that the linear addition of a pressure (Poiseuille) and drag (Couette) ¯ow [6] does not hold for a tape casting system. Eq. (5) which was derived based on fundamental ¯uid ¯ow offers a more logical description of the physics of the tape casting system.
Fig. 7. Plot of the blade gap setting at which the maximum green tape thickness occurs for different casting speeds.
Work is currently being done to derive the ¯uid ¯ow equations using the approach as proposed above using a nonNewtonian (shear thinning) constitutive model, as rheology experiments performed show that the tape casting slurry behaves as a non-Newtonian shear thinning ¯uid [12]. This will then be veri®ed against experimental results. 4. Conclusions From the results, it can be concluded that a higher degree of accuracy in tape thickness can be obtained when casting at lower casting speeds. However, this accuracy decreases when thinner tapes are required, and the optimized use of an increased casting speed would be needed to maintain this accuracy. An optimal setting between the casting speed and the blade height setting is, therefore, needed in order to achieve accurate and reproducible results when performing tape casting. Determination of this optimum combination would depend on the required thickness of the ®nal product. Theoretical modeling can be used to verify the reliability of tape thickness predictions in the tape casting system. Therefore, work is currently focused upon this area to derive accurate models to describe the ¯uid behavior, and to accurately predict the cast tape thickness. References [1] G. Fischer, Electronic products weathering market swings, Am. Ceram. Soc. Bull. 66 (1987) 636. [2] S. Takahashi, Longitudinal mode multilayer piezoelectric actuators, Am. Ceram. Soc. Bull. 65 (1986) 1156±1159. [3] C. Fiori, G. De Portu, Tape casting: A technique for preparing and studying new materials, British Ceram. Proc. 38 (1986) 213± 225. [4] A.I.Y Tok, F.Y.C. Boey, K.A. Khor, Tape casting of high dielectric ceramics composite/hybrid substrates for microelectronics application, in: Proc. 6th Int. Conf. on Processing and Fabrication of Advanced Materials, Singapore, 1997. [5] A.I.Y. Tok, F.Y.C. Boey, K.A. Khor, Development of High Dielectric Al2O3/AlN Composite Substrates for Microelectronics Application, Materials & Manufacturing Processes, in press. [6] Y. Chou, Y. Ko, M. Yan, Fluid flow model for ceramic tape casting, J. Am. Ceram. Soc. 70(10) (1987) C280±282. [7] T.A. Ring, A model of tape casting Bingham plastic and Newtonian fluids, Adv. Ceram. 26 (1989) 569±576.
512
A.I.Y. Tok et al. / Journal of Materials Processing Technology 89±90 (1999) 508±512
[8] X.Y. Huang, C.Y. Liu, H.Q. Gong, A viscoplastic flow modeling of ceramic tape casting, Materials and Manufacturing Processes, 12(5) (1997). [9] W.R. Cannon, R. Becker, K.R. Mikeska, Interactions Among Organic Additives Used for Tapecasting, Advances in: M.F. Yan (Ed.), Ceramics, vol. 26, Am. Ceram. Soc. Ohio, 1989, pp. 525±541. [10] D.J. Shanefield, Competing adsorptions in tape casting advances in ceramics, Am. Ceram. Soc 19 (1988) 155±160.
[11] L. Braun, J.R. Morris Jr., W.R. Cannon, Viscosity of tape-casting slips, Am. Ceram. Soc. Bull. 64(5) (1985) 727±729. [12] A.I.Y. Tok, F.Y.C. Boey, K.A. Khor, Tape casting of high dielectric ceramics substrates for microelectronics packaging, J. Mater. Eng. Performance, ASM Int'l, in press.