and low organic levels in the dried green ceramics. Therefore, a final mechanical properties of Si3 N4 parts: i)the surface quality wide variety of ceramic materials have been prepared using gel of the mold that is transferred into the final part. Therefore, the casting process including Si3 N4, SiC, AlO3 and zro,, among others. higher this surface roughness, leads to lower the final mechanical In gel casting, slurry made from ceramic powder and a water-based strength. The defects in the mold surface can cause notches which monomer solution is poured into a mold, polymerized in situ to lead to stress concentration in the final part and ii)with perfectly mobilize the particles in a gelled part, removed from the mold smooth mold surface, the difference between bulk microstructure while still wet, then dried and fired. If the solvent for the monomers and surface microstructure has to be considered when describing is organic, it is no aqueous gel casting: if is water, it is aqueous gel- the mechanical properties of silicon nitride parts. The maximum casting [94, 96, 98 The development of an aqueous process using strength is achieved with polished samples. Similarly, Stampfl et acrylamide as monomer was completed in 1988[95, 99]. However, al. [93 obtained strength values of 414, 950, and 983 MPa in ncerns regarding health, safety and disposal of acrylamide caused Si3N4 unpolished, Si3N4 polished and Si3 N4 GPS and polished industrial rejection of the process because the acrylamide is a neu- respectively, where all samples were sintered at 1750 C in nitro- toxin. Therefore, the development of a low toxicity process was gen atmosphere. On the other hand, materials such as aluminum, initiated to deal with the lack of acceptance and it was fully demon anodized aluminum, brass, glass, graphite indium alloys, neoprene trated in 1990 90 rubber, plaster and polyethylene are commonly used in construct- eramic parts from different ceramics such as aluminum oxide ing gelcasting molds. Al203, and high-performance silicon nitride Si3 N4, have been pro- injection molding, before a ceramic body can be sintered, luced by gelcasting ranging in size from <l g to >6 kg with thin the binder added must be removed. One advantage of gelcasting sections as small as 0. 2 mm and solids loading as high as 55-60 vol% is the small amount of polymer that remains in the green bod in alumina slurries and 45-57 vol% in silicon nitride suspensions after drying [101]. The dried gelcast ceramic contains only about [99 Although gelcast bodies typically shrink -23% during densifi- 2-6wt% polymer, which depends on the solids loading of the slurry, tion and resulting sometimes in distortion in the densified parts, the concentration of monomers in the premix, and the density this technique starts to be used not only for manufacturing of com- of powder. For comparison, a 45 vol% solids silicon nitride with plicated shaped dense products such as Si3N4 parts of turbines but the composition Si3N4+5 wt% Al203 +5 wt% Y203 made using a also for manufacturing of porous ceramic[ 90, 100). The most 15 wt% 4: 1 Methacrylamide-N, N'-Methylene bisacrylamide(MAM- dvanced works in this field are already in the phase or commer- MBAM) premix contains about 5.5 wto polymer in the dried part, alization and, Allied Signal Ceramics Components (torrance CA. meanwhile an injection molded silicon nitride with the same com- USA)working together with ORNL (Oak Ridge, TN, USA)has devel- position and solid loading would contain about 27 wt% polymer, ped and automated gelcasting fabrication process of production nearly five times as much. During burn out binder process, a lower Si3 N4 ceramic turbine rotors [100 temperature is required to remove the polymer carefully or else To optimize gelcasting of silicon nitride, Omatete[98 reported the final product may have defects and cracks. Heating rates on the he optimal gelcasting condition for the AlliedSignal Ceramic com- order to 0.5-1oCmin- to temperatures as high as 650C have been onents GN-10 silicon nitride formulation in a near-production used successfully for both silicon nitride and alumina gelcast parts. nvironment. The principal criterion used to determine optimum On the other side the AlliedSignal engineers [98, 99, 103, 104] noted design was the green strength. The investigation predicted 80% that gelcasting did not have two problems that plague injection increase in green strength(4.3 MPa versus 2.4 MPa )but the results molding such as the binder that can be as high as 20% of the weight howed only 60%increase(3.8 MPa). It is preferred that gelcasting of the ceramic versus to 4% in gelcasting and the other problem is slurries be at least 50 vol% solids: higher solids loadings are desir- related to the burning out the binder in injection molding that may ables. In most of the cases, solids loadings above 501 ompared to less than a day for a gelcast ceramic Addi- achieved and some cases, solids loading above 60 vol% are attain- tionally, defects and cracks can also develop in other stages of the ble. Notwithstanding, sometimes these requirements cannot be injection molding process such as drying. Such problems are rarely and Si3N4(Ube E10)is an example of a ceramic powder in which it seen in gelcast ceramics if they are properly dried [101] is nearly impossible to achieve a solids loading above 45 vol% in a In the manufacture of Si3 Na gelcast gelcast slurry due to that this powder is very difficult to disperse for turbine engines, the reproducibility of the parts is a critical any application inasmuch as possess a high surface area. In this con- tor. Janney et al. [101]reported the fabrication of a series of silicon text and with the purpose to obtain Si3 N4 ceramics with excellent nitride batches prepared under identical conditions and gelcast properties, a series of commercial dispersants was evaluated for Eleven batches were prepared for the repeatability study and each heir efficacy in dispersing silicon nitride in water[ 90]. The follow- batch was prepared on a different day over a period of 18 days. The ng conditions should be obtained for a good dispersant in silicon results were favorable and the uniformity of the castings is shown nitride: i)The 24-h sedimentation height should be high: i. e the clearly by the standard deviation values which indicate a varia- powder should not have settled out of a suspension in a short time tion of only 0. 1-0.3% about the average value for the dimensions should pack very well when is does finally settle out of suspension, bodies show that gelcasting process is reproducible. A standard and iii) the 3-week cloudy /clear interface height should be high; deviation of only 0.02 mm or about 0. 1% was obtained i.e. the finest of the particles should stay in suspension for a very As outlined above, gelcasting provides an excellent alternative long time and should not reagglomerate and settle. Therefore, the to manufacturing large, complex-shaped components such as tur- evelopment of excellent dispersant system is critical. As example, bine rotors, valves and cam followers for gasoline and dieseleng the Table 3 of [101 summarizes the dispersants cor In reason of this, the high degree of homogeneity required to for gelcasting ceramics including Si3N produce excellent parts can be retained. Janney et al. [101 taking The mold selection. mold fabrication and mold use are the crit- data from Pollinger [105 ]illustrated this point wherein after drying. ical aspects of successful gelcasting 93, 101. Proper selection of the green density of several sections of the rotor was determined mold material, fabrication method, filling method and mold release by the Archimedes immersion method. They observed that the In make the difference between producing excellent parts and variation in green density was extremely low and all the sections producing"also rans"Stampfl et al. [93, 102] have concluded that measured except one were within 0. 2% of the average green density two parameters in addition to the bulk properties determine the of 57.77% of the theoretical density. This is an especially significant134 M.H. Bocanegra-Bernal, B. Matovic / Materials Science and Engineering A 500 (2009) 130–149 and low organic levels in the dried green ceramics. Therefore, a wide variety of ceramic materials have been prepared using gel casting process including Si3N4, SiC, Al2O3 and ZrO2, among others. In gel casting, slurry made from ceramic powder and a water-based monomer solution is poured into a mold, polymerized in situ to immobilize the particles in a gelled part, removed from the mold while still wet, then dried and fired. If the solvent for the monomers is organic, it is no aqueous gel casting; if is water, it is aqueous gel￾casting [94,96,98]. The development of an aqueous process using acrylamide as monomer was completed in 1988 [95,99]. However, concerns regarding health, safety and disposal of acrylamide caused industrial rejection of the process because the acrylamide is a neu￾rotoxin. Therefore, the development of a low toxicity process was initiated to deal with the lack of acceptance, and it was fully demon￾strated in 1990 [90]. Ceramic parts from different ceramics such as aluminum oxide Al2O3, and high-performance silicon nitride Si3N4, have been pro￾duced by gelcasting ranging in size from <1 g to >6 kg with thin sections as small as 0.2 mm and solids loading as high as 55–60 vol% in alumina slurries and 45–57 vol% in silicon nitride suspensions [99]. Although gelcast bodies typically shrink ∼23% during densifi- cation and resulting sometimes in distortion in the densified parts, this technique starts to be used not only for manufacturing of com￾plicated shaped dense products such as Si3N4 parts of turbines but also for manufacturing of porous ceramic objects [90,100]. The most advanced works in this field are already in the phase or commer￾cialization and, Allied Signal Ceramics Components (Torrance, CA, USA) working together with ORNL (Oak Ridge, TN, USA) has devel￾oped and automated gelcasting fabrication process of production Si3N4 ceramic turbine rotors [100]. To optimize gelcasting of silicon nitride, Omatete [98] reported the optimal gelcasting condition for the AlliedSignal Ceramic com￾ponents’ GN-10 silicon nitride formulation in a near-production environment. The principal criterion used to determine optimum design was the green strength. The investigation predicted 80% increase in green strength (4.3 MPa versus 2.4 MPa) but the results showed only 60% increase (∼3.8 MPa). It is preferred that gelcasting slurries be at least 50 vol% solids; higher solids loadings are desir￾ables. In most of the cases, solids loadings above 50 vol% can be achieved and some cases, solids loading above 60 vol% are attain￾able. Notwithstanding, sometimes these requirements cannot be and Si3N4 (Ube E10) is an example of a ceramic powder in which it is nearly impossible to achieve a solids loading above 45 vol% in a gelcast slurry due to that this powder is very difficult to disperse for any application inasmuch as possess a high surface area. In this con￾text and with the purpose to obtain Si3N4 ceramics with excellent properties, a series of commercial dispersants was evaluated for their efficacy in dispersing silicon nitride in water [90]. The follow￾ing conditions should be obtained for a good dispersant in silicon nitride: i) The 24-h sedimentation height should be high: i.e. the powder should not have settled out of a suspension in a short time, ii) the 3-week sedimentation height should be low; i.e. the powder should pack very well when is does finally settle out of suspension, and iii) the 3-week cloudy/clear interface height should be high; i.e. the finest of the particles should stay in suspension for a very long time and should not reagglomerate and settle. Therefore, the development of excellent dispersant system is critical. As example, the Table 3 of [101] summarizes the dispersants commonly used for gelcasting ceramics including Si3N4. The mold selection, mold fabrication and mold use are the crit￾ical aspects of successful gelcasting [93,101]. Proper selection of mold material, fabrication method, filling method and mold release can make the difference between producing excellent parts and producing “also rans”. Stampfl et al. [93,102] have concluded that two parameters in addition to the bulk properties determine the final mechanical properties of Si3N4 parts: i) the surface quality of the mold that is transferred into the final part. Therefore, the higher this surface roughness, leads to lower the final mechanical strength. The defects in the mold surface can cause notches which lead to stress concentration in the final part and ii) with perfectly smooth mold surface, the difference between bulk microstructure and surface microstructure has to be considered when describing the mechanical properties of silicon nitride parts. The maximum strength is achieved with polished samples. Similarly, Stampfl et al. [93] obtained strength values of 414, 950, and 983 MPa in Si3N4 unpolished, Si3N4 polished and Si3N4 GPS and polished, respectively, where all samples were sintered at 1750 ◦C in nitro￾gen atmosphere. On the other hand, materials such as aluminum, anodized aluminum, brass, glass, graphite, indium alloys, neoprene rubber, plaster and polyethylene are commonly used in construct￾ing gelcasting molds. As injection molding, before a ceramic body can be sintered, the binder added must be removed. One advantage of gelcasting is the small amount of polymer that remains in the green body after drying [101]. The dried gelcast ceramic contains only about 2–6 wt% polymer, which depends on the solids loading of the slurry, the concentration of monomers in the premix, and the density of powder. For comparison, a 45 vol% solids silicon nitride with the composition Si3N4 + 5 wt% Al2O3 + 5 wt% Y2O3 made using a 15 wt% 4:1 Methacrylamide-N,N’-Methylene bisacrylamide (MAM￾MBAM) premix contains about 5.5 wt% polymer in the dried part, meanwhile an injection molded silicon nitride with the same com￾position and solid loading would contain about 27 wt% polymer, nearly five times as much. During burn out binder process, a lower temperature is required to remove the polymer carefully or else the final product may have defects and cracks. Heating rates on the order to 0.5–1 ◦C min−1 to temperatures as high as 650 ◦C have been used successfully for both silicon nitride and alumina gelcast parts. On the other side, the AlliedSignal engineers [98,99,103,104] noted that gelcasting did not have two problems that plague injection molding such as the binder that can be as high as 20% of the weight of the ceramic versus to 4% in gelcasting and the other problem is related to the burning out the binder in injection molding that may take a week compared to less than a day for a gelcast ceramic. Addi￾tionally, defects and cracks can also develop in other stages of the injection molding process such as drying. Such problems are rarely seen in gelcast ceramics if they are properly dried [101]. In the manufacture of Si3N4 gelcast ceramic components for turbine engines, the reproducibility of the parts is a critical fac￾tor. Janney et al. [101] reported the fabrication of a series of silicon nitride batches prepared under identical conditions and gelcast. Eleven batches were prepared for the repeatability study and each batch was prepared on a different day over a period of 18 days. The results were favorable and the uniformity of the castings is shown clearly by the standard deviation values which indicate a varia￾tion of only 0.1–0.3% about the average value for the dimensions measured. The measurement of diameter both green and sintered bodies show that gelcasting process is reproducible. A standard deviation of only 0.02 mm or about 0.1% was obtained. As outlined above, gelcasting provides an excellent alternative to manufacturing large, complex-shaped components such as tur￾bine rotors, valves and cam followers for gasoline and diesel engines [104]. In reason of this, the high degree of homogeneity required to produce excellent parts can be retained. Janney et al. [101] taking data from Pollinger [105]illustrated this point wherein after drying, the green density of several sections of the rotor was determined by the Archimedes immersion method. They observed that the variation in green density was extremely low and all the sections measured except one were within 0.2% of the average green density of 57.77% of the theoretical density. This is an especially significant