- Email: [email protected]
Comminution of silicon carbide powder in a planetary mill
Powder Technology 169 (2006) 84 – 88 www.elsevier.com/locate/powtec
Comminution of silicon carbide powder in a planetary mill Maria Aparecida P. dos Santos a , Célio A. Costa b,⁎ a b
Instituto de Pesquisas da Marinha (IPqM), Grupo de Materiais-Ilha do Governador, CEP 21931-090, Rio de Janeiro/RJ, Brazil Programa de Engenharia Metalúrgica e de Materiais, COPPE/UFRJ, PO Box 68505, CEP 21945-970, Rio de Janeiro/RJ, Brazil Received 2 June 2005; received in revised form 25 July 2006; accepted 25 July 2006 Available online 8 August 2006
Abstract Silicon carbide powder (SiC) was comminuted in a planetary mill during time intervals of 0.5, 2, 4 and 6 h. The wet milling media was ZrO2 spheres in isopropyl alcohol. The powders were then characterized with respect to chemical composition, particle size distribution, surface area and density for each milling time. The average particle size was reduced from 1.8 μm to 0.4 μm in 30 min and the amount of ZrO2 increased linearly with milling time. The result was a homogeneous combination of submicrometer and nanometric SiC + ZrO2 powder, which possessed good sinterability in liquid phase and high fracture toughness. © 2006 Elsevier B.V. All rights reserved. Keywords: Comminution; Silicon carbide; Planetary mill; Ultra-fine powder; Nano-powder
1. Introduction The properties of an advanced ceramic product depend upon the purity of the raw materials, the green body density and the sintering process used. Highly covalent ceramics, such as carbides and nitrides, have inherent low sinterability as a consequence of the atomic bond. A very efficient way to increase density is through the reduction of the particle size, tight distribution and increase of surface area, which strongly influence the overall processing route and the final properties of the product [1,2]. A large number of mechanical methods are available for preparing fine and ultra-fine powders. Typical high-energy processes include planetary, attritor and jet milling [2]. The jet grinding process is the collision among particles propelled by a high-pressure gas jet, which has the main virtues of preventing contamination, precise mean particle size and tight distribution. The drawbacks are elevated loss of product; requires a highquality classification system for fine particles; high equipment costs; and the presence of electrostatic charges can be a problem [3,4]. On the other hand, the planetary and attritor milling processes are via mechanical shock between the grinding bodies ⁎ Corresponding author. Tel: +55 21 2562 8533. E-mail address: [email protected] (C.A. Costa). 0032-5910/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.powtec.2006.07.019
and powder particles in a wet media. The characteristic of both processes are low material loss; the cost of the equipment is significantly lower than jet milling; very good homogenization and particles with high superficial area are easily generated in a very short time; however, contamination may occur [4,5]. The work of Brito [6], which compares the planetary and attritor mill in reducing silica particle size, shows the planetary mill to be more efficient, even though no data on contamination have been reported. Tavares et al. [7], studying the influence of seven types of milling media on the comminution of a silicon carbide powder using a planetary mill, showed that sodium pyrophosphate was the most efficient media and ethyl alcohol was the least efficient, and for short periods of time (about 2 h) the mechanical effect was preponderant over the chemical one, but no data concerning contamination and surface area were reported. Dense silicon carbide is an advanced ceramic in high demand and very small particles are needed to fulfill this goal. The understanding of the comminution process of SiC is very important but data in literature are difficult to find; for instance, the ISI WEB of Science database shows approximately 300 articles published about high-energy milling since 1960, but only one involving silicon carbide. To contribute to the knowledge of milling SiC, this study was done in a planetary mill and the results showed the process to be very fast in reducing the
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88 Table 1 Chemical composition of the received powder Composition
wt.%
SiC Si + SiO2 Fe Al Free carbon (CL) S, Ca, V, Ni, Cu and Zr
98.17 0.60 0.30 0.19 0.20 Trace
particle size and increasing the surface area, with a tight distribution; on the other hand, improvements must be made to reduce contamination. 2. Materials and methods The silicon carbide grade SiC-1000 (Alcoa S.A.- Brazil) produced by the Achesson process was used as the raw material. The chemical composition is specified in Table 1 and the physical characteristics measured were mostly alpha phase, bimodal particle size distribution (d99 and d50 are, respectively, equal to 14.22 and 1.77 μm), density of 3.211 g/cm3 and superficial area of 3.483 m2/g. The milling process was done in a planetary mill (PM-4 model, Retsch). The jars used were specially fabricated to support the elevated wear and to comminute 120 g of material. They were made of stainless steel with an inner diameter of 100 mm, 500 ml capacity and internally coated with tungsten carbide (WC–Co) by HVOF (High-Velocity Oxygen Fuel). The milling conditions were 300 rpm, isopropyl alcohol P.A. (IPA), zirconia milling balls stabilized with ceria (80% ZrO2 + 20% CeO2, Zirconox®, Netsch) with a diameter in between 0.7 and 1.2 mm, density of 6.1 g/cm3 and Vickers hardness of 1200 kgf mm2.
85
The milling media occupied half of the jar volume (250 ml), where 60% of this volume (150 ml) was filled by the grinding balls and the other 40% (100 ml) composed of the pulp (100 g of SiC powder and IPA). The amount of IPA varied a little, since the milling balls had to be slightly covered by the IPA. The milling was done in batches for time periods of 0.5, 2, 4 and 6 h. For periods longer than 6 h, the milling efficiency does not compensate the energy spent, as verified by Matos [8]. The materials were then poured into a Pyrex container, dried in an oven at 70 °C for 24 h, deagglomerated in a mortar and sieved. The powder of each batch was then characterized by X-ray fluorescence spectrometer (Philips PW 2400), particle size distribution (Malver Mastersizer Micro Plus, MAF 5001), superficial area by BET (Gemini III 2375, Micromeritics) and density (Helium Pycnometer, ACCUPYC 1330 Micromeritics). 3. Results and discussion The main object of this study was to produce a highly sinterable submicrometer silicon carbide powder. The first step to look at was the milling efficiency, since as the particles approach submicrometric (b 1 μm) or nanometric (b0.3 μm) dimensions, there is a reduction in the milling efficiency as a result of two competing phenomena, namely, comminution and agglomeration [1]. The second step was to observe if the conditions of the high-energy milling process used generated contamination. The comminution process was executed in a planetary mill, operated at 300 rpm with milling media composed of SiC powder, spheres of ZrO2 + CeO2 and isopropyl alcohol as the dispersant. These processing characteristics resulted in a fast size reduction, but slow change in distribution, as shown in Fig. 1. The as-received powder (Fig. 1a) had a bimodal distribution, with peaks centered at 0.2 and 5 μm. The peak situated at 5 μm
Fig. 1. Particle size distribution: (a) as-received SiC, (b) after 4 h and (c) after 6 h of milling.
86
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88
of milling d90 ≅ 1 μm. This shows the high efficiency of the milling process in producing submicrometric particles. Another important feature is the pivot point for all milling times over 0.5 h. This loco is represented by coordinates 0.3 μm and 38% (marked by the arrows in Fig. 2), which means that 38% of all particles had diameter smaller than 0.3 μm, indicating that there is an expressive quantity of nanometric particles. The milling curve for 4 and 6 h is practically identical but, below the pivot point, the particles seem to be larger than those corresponding to the shorter milling time. This behavior was due to the strong agglomeration between the very small particles, as shown in Figs. 3 and 4, since reduction in size leads to an increase in Fig. 2. Influence of the milling time on the particle size distribution.
was a little higher than the 0.2-μm peak, indicating that the volume of particles larger than 1 μm was greater than the volume of particles smaller than 1 μm. After milling for 0.5, 2 and 4 h, the distribution continued to be bimodal, with the peak centered at 0.2 μm steady and increasing in height, while the second peak was being reduced both in height as in its center position, as the milling time increased (Fig. 1b). This means that the volume of particles smaller that 1 μm was increasing at the expense of the comminuted larger particles. At 6 h of milling, the distribution became unimodal and centered at 0.2 μm, as shown in Fig. 1c. These data pointed out that the planetary milling operation worked well on breaking large particles but not the submicrometer ones, since the peak centered at 0.2 μm kept steady during the whole period. Also, for the milling conditions used, the value of 0.2 μm can be considered as a threshold value, below which particles cannot be broken. A similar result was reported by Matos et al. [8] working with a planetary mill, but using different media. It is possible that this threshold value might be limited by the equipment, the milling media used or the processing characteristics themselves. When the particle size for all the milling times were closely observed, as shown in Fig. 2, the increase in quantity of particles having smaller diameters as the milling time increased was reconfirmed. For instance, 90% of the as-received particles had a diameter of below 7.8 μm, represented by d90 ≅ 7.8 μm. After 0.5 h of milling, the value d90 falls to 4.2 μm and after 4 h
Fig. 3. Effect of the milling time on the surface area measured by BET.
Fig. 4. SEM photos of the received powder (a) and the milled for 4 (b) and 6 h (c).
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88
Fig. 5. The amount of element introduced during the milling process along the milling time.
surface energy and the powder naturally agglomerates to stabilize this, as will be demonstrated below. To confirm that the agglomeration phenomenon was taking place, the surface area of the milled particles was measured by BET. The results shown in Fig. 3 demonstrated a fast comminution rate up to 4 h of milling, where the surface area of the as-received powder changed from 3.4 to 10 m2/g, an increase of about 190%. After 4 h, the surface area still increases, demonstrating that milling still continues, albeit at a reduced rate. The fast reduction in particle size and the agglomeration effect was confirmed by the SEM photos shown in Fig. 4, where the as-received powder (a) and the milled for 4 (b) and 6 h (c) are shown. This confirms the generation of agglomerates as particles become smaller. Another possibility that had to be considered was contamination that could occur in the processes, namely, the increase in surface area could be the result of the breaking of SiC particles plus the addition of other small particles from the milling media. The chemical characterization of the powder for each milling time was done by X-ray fluorescence, as shown in Fig. 5. As can be observed, the amount of ZrO2 increased linearly with the milling time, reaching 15 wt.% for 6 h of milling (all the Zr was considered to come from the ZrO2 balls, since it was the only Zr phase observed in the X-ray diffraction). The element Fe increased to 5 wt.% after 2 h of milling and, thereafter, remained practically constant. The WC content increased in a discontinuous way up to 6.5 wt.% for 4 h of milling and then, the value reduced to 2 wt.%. The other compounds observed were Ce, Co and Cr and their behavior was as expected, since these elements were already part of the milling spheres (80% ZrO2 + 20% CeO2), HVOF coating (WC + Co) and the stainless steel itself (Cr), respectively. Consequently, the Ce, Co and Cr curves were similar to those observed in the parent element, i.e., Zr and Ce curves increased continuously, WC and Co increased up to 4 h and then decreased, while Fe and Cr reached a maximum in 2 h of milling and remained practically constant. These data demonstrated that the SiC powder was somewhat contaminated, especially by the milling spheres. The milling behavior in Fig. 5 might be explained as follows: until 2 h, the elements/compounds taking part in the milling process continuously increased in quantity due to the strong
87
friction between the milling spheres, the powder, the coating and the very small regions of the vessel wall not coated, which introduced small fragments originating from each of those places. For time periods above 2 h, the WC + Co coating was polished and its contribution decreased since no more sharp wedges were available (it is also possible that a SiC layer was built up on the polished coating wall preventing its further contamination into the milling media), the Fe + Cr stabilized because the region from where they came did not enlarge as the milling time increased, and the amount of ZrO2 + CeO2 keeps increasing as a combination between wearing and impact of spheres with the vessel wall. The above results pointed out that the vessel design needs improvement, at least to reduce the contamination from the stainless steel vessel and so SiC spheres shall be used instead of ZrO2. Regarding the WC + Co coating, as it became polished, its contribution tended to null. The amount of elements added to the slurry during the milling process affected the original density of the SiC powder, as shown in Table 2. A 26% increase in density after 6 h of milling was observed, measured by He pycnometry, but the data became asymptotic around 6 h, similar to what happened in Fig. 3 for the same period of time. The contribution of each compound in Fig. 5 to the density can be confirmed by applying the lever rule, which was shown in the Fluorescence column (Table 2). Here, an excellent agreement between the measured and the calculated values was observed. For instance, the smallest difference was 0.1%, the highest 2.3% and the average about 0.5%. These measurements were important because they showed the real density of the milled powder, which has to be used to verify the density of the sintered material. Also, in the absence of a He pycnometer, fluorescence data could be used for calculation of densities, giving reliable values. The data in Table 2 pointed out that the contamination process reached a limit around 6 h of milling, which was coincident with the end of the milling efficiency, since the particle size distribution (Fig. 1) and the BET (Fig. 3) showed almost no variation in their measurements for that same time. In summary, for the milling conditions used, the milling process ceased to be operative when a unimodal distribution was completed (centered at 0.2 μm) and the increase in the surface area tended to null, even though agglomeration of small particles was taking place. The powder generated here possesses average particle size (d50 = 0.4 μm), surface area (BET = 11.5 m2/g) and distribution (d100 b 1.1 μm) comparable or better than commercial sinterable SiC powders, which have d50 and BET between 0.6 and 1.2 μm Table 2 Density of the α-SiC powder, measured by the He pycnometry and calculated from X-ray fluorescence, as a function of the milling time Milling time (h)
Pycnometry (g/cm3)
Fluorescence (g/cm3)
Difference |Modulus| (%)
0 1/2 2 4 6
3.211 3.435 3.695 3.940 4.066
3.215 3.517 3.736 3.898 4.057
0.12 2.33 1.10 1.08 0.22
88
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88
and 5 and 17 m2/g, respectively, and distribution below 5 μm [1,9]. In fact, if the center of the distribution (Fig. 1) is taken as the average (0.2 μm), the particles can be classified as nano, but having a larger surface area than the ones measured by Quanli et al. [9], which was 6.71 m2/g. The above discussion must take into account the contamination observed during milling since it altered the original material significantly. The powder was received with a purity of 98.7%, with Si + SiO2 of 0.6 wt.%, but the oxygen on the particle surface was not determined and it is desirable to be lower than 1.5 wt.% [1]. The amount of oxides added during the milling increased the quantity of oxygen and other metals in the powder, and therefore, the final milled powder is not appropriate for solid-state sintering. Usually, this kind of material requires purity levels higher than 99%, low oxygen and addition of C, B or B4C [1,10–12]. On the other hand, when the milled powder was liquid phase sintered (eutectic Al2O3 + Y2O3) the result was a 99% dense material [13] with fracture toughness of 5.5 MPa m1/2 [14], which are excellent results. These data showed that the contamination introduced in the SiC powder had particles equal or smaller than the particles being comminuted, since the distribution finished as unimodal and surface area increased up to 6 h of milling. Also, even though contamination is not desirable, in this case, it was not exactly harmful, since the liquid phase sintering (LPS) gave high density and the fracture toughness was improved. 4. Conclusions The planetary mill easily reduced the average particles size (d50) of the α-SiC from 1.77 to 0.4 μm in only 0.5 h of grinding. The values of d90 for 4 and 6 h of grinding were 1.20 and 0.67 μm, respectively, showing that the process is efficient to obtain fine powders. The planetary mill (300 rpm), with the ZrO2 spheres (0.7 b ϕ b 1.2 mm) and IPA, changed the particle size distribution from bimodal to unimodal, centered at 0.2 μm, in 6 h. For the present conditions, it seems that the process ceases its efficiency in 6 h. In fact, 4 h of milling would have been enough for powders with high sinterability in the liquid phase. When very fine particles were generated (after 4 h), they agglomerated and analysis should be conducted by both particle size analyzer and surface area (BET), since they are complementary and allow good understanding of the comminuting process. The contamination was high, especially from the milling spheres. Consequently, it is suggested that SiC spheres should be used instead of ZrO2 ones, since they would lead to an
autogenous condition. Also, under such conditions, the real density of the milled powder must be checked, either by pycnometry or X-ray fluorescence, which was shown to be a reliable technique to calculate density when a pycnometer is not available. Lastly, even though there was elevated contamination, high density and high fracture toughness was obtained via liquid phase sintering. Acknowledgement We thank the Brazilian Navy (IPqM) for supporting this research and Dr. Francisco C.L. Melo (CTA/Air Force) for the technical collaboration. References [1] G. Schwier, I. Teusel, M.H. Lewis, Characterization of SiC powders and the influence of powder properties on sintering (technical report), Pure Appl. Chem. 69 (6) (1997) 1305–1316. [2] D.W. Richerson, Modern Ceramic Engineering: Properties, Processing and Use in Design, Ed. Marcel Dekker, 2 Ed. Revised and Expanded (1992). [3] H. Berthiaux, J. Dodds, Modelling fine grinding in a fluidized bed opposed jet mill—part I: batch grinding kinetics, Powder Technol. 106 (1999) 78–87. [4] H.W. Hennicke, J. Stein, Process of fine milling for ceramics materials, Mater. Sci. Eng., A 109 (1989) 3–7. [5] C. Suryanarayana, Mechanical alloying and milling, Prog. Mater. Sci. 46 (2001) 1–184. [6] F.V. Brito, Procesamento de Meios Porosos à Base de Sílica, M.Sc Thesis, PEMM-COPPE-UFRJ, Rio de Janeiro-RJ-Brasil-Dezembro (2004). [7] L.M.M. Tavares, B.B. Matos, C.A. Costa, Influência do Meio na Moagem Ultrafina do Carbeto de Silício, Proceedings of the 46 Annual Meeting of the Brazilian Ceramic Society, 26–29 May, 2002, São Paulo, SP, Brazil. [8] B.B. Matos, Influência do meio na moagem ultrafina de carbeto de silício, Projeto de Formatura, Departamento de Engenharia Metalúrgica e de Materiais da Escola de Engenharia da Universidade Federal do Rio de Janeiro, Rio de Janeiro, Abril, (2002). [9] J. Quanli, Z. Haijun, L. Suping, J. Xiaolin, Effect of particle size on the oxidation of silicon carbide powders, Ceram. Int., in press. [10] S. Prochaska, The role of boron and carbon in sintering of silicon carbide, Special Ceramics, vol. 6, The British Ceramic Research Association, Stoke-on-Trent, U.K, 1975, pp. 171–181. [11] K. Negita, Effective sintering aids for silicon carbide ceramics: reactivities for silicon carbide with various additives, J. Am. Ceram. Soc. 69 (1986) 308–310. [12] M. Giuseppe, B. Giancarlo, Pressure-less sintering and properties of (– SiC–B4C) composite, J. Eur. Ceram. Soc. 21 (5) (2001) 633–638. [13] M.A.P. Santos, Processamento e Sinterização de Carbeto de Silício Nacional, DSc. Thesis , PEMM/COPPE/UFRJ, Rio de Janeiro–RJ–Brasil (2003). [14] E.S. Lima, L.P. Brandão, L.H.L. Louro, C.A. Costa, M.A.P. Santos, Mechanical behavior of alpha-SiC based nanocomposites, Proceeding of the 2nd Brazilian MRS Meeting, vol. V.1, 2003, pp. 15–20, Rio de Janeiro.
Comminution of silicon carbide powder in a planetary mill Maria Aparecida P. dos Santos a , Célio A. Costa b,⁎ a b
Instituto de Pesquisas da Marinha (IPqM), Grupo de Materiais-Ilha do Governador, CEP 21931-090, Rio de Janeiro/RJ, Brazil Programa de Engenharia Metalúrgica e de Materiais, COPPE/UFRJ, PO Box 68505, CEP 21945-970, Rio de Janeiro/RJ, Brazil Received 2 June 2005; received in revised form 25 July 2006; accepted 25 July 2006 Available online 8 August 2006
Abstract Silicon carbide powder (SiC) was comminuted in a planetary mill during time intervals of 0.5, 2, 4 and 6 h. The wet milling media was ZrO2 spheres in isopropyl alcohol. The powders were then characterized with respect to chemical composition, particle size distribution, surface area and density for each milling time. The average particle size was reduced from 1.8 μm to 0.4 μm in 30 min and the amount of ZrO2 increased linearly with milling time. The result was a homogeneous combination of submicrometer and nanometric SiC + ZrO2 powder, which possessed good sinterability in liquid phase and high fracture toughness. © 2006 Elsevier B.V. All rights reserved. Keywords: Comminution; Silicon carbide; Planetary mill; Ultra-fine powder; Nano-powder
1. Introduction The properties of an advanced ceramic product depend upon the purity of the raw materials, the green body density and the sintering process used. Highly covalent ceramics, such as carbides and nitrides, have inherent low sinterability as a consequence of the atomic bond. A very efficient way to increase density is through the reduction of the particle size, tight distribution and increase of surface area, which strongly influence the overall processing route and the final properties of the product [1,2]. A large number of mechanical methods are available for preparing fine and ultra-fine powders. Typical high-energy processes include planetary, attritor and jet milling [2]. The jet grinding process is the collision among particles propelled by a high-pressure gas jet, which has the main virtues of preventing contamination, precise mean particle size and tight distribution. The drawbacks are elevated loss of product; requires a highquality classification system for fine particles; high equipment costs; and the presence of electrostatic charges can be a problem [3,4]. On the other hand, the planetary and attritor milling processes are via mechanical shock between the grinding bodies ⁎ Corresponding author. Tel: +55 21 2562 8533. E-mail address: [email protected] (C.A. Costa). 0032-5910/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.powtec.2006.07.019
and powder particles in a wet media. The characteristic of both processes are low material loss; the cost of the equipment is significantly lower than jet milling; very good homogenization and particles with high superficial area are easily generated in a very short time; however, contamination may occur [4,5]. The work of Brito [6], which compares the planetary and attritor mill in reducing silica particle size, shows the planetary mill to be more efficient, even though no data on contamination have been reported. Tavares et al. [7], studying the influence of seven types of milling media on the comminution of a silicon carbide powder using a planetary mill, showed that sodium pyrophosphate was the most efficient media and ethyl alcohol was the least efficient, and for short periods of time (about 2 h) the mechanical effect was preponderant over the chemical one, but no data concerning contamination and surface area were reported. Dense silicon carbide is an advanced ceramic in high demand and very small particles are needed to fulfill this goal. The understanding of the comminution process of SiC is very important but data in literature are difficult to find; for instance, the ISI WEB of Science database shows approximately 300 articles published about high-energy milling since 1960, but only one involving silicon carbide. To contribute to the knowledge of milling SiC, this study was done in a planetary mill and the results showed the process to be very fast in reducing the
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88 Table 1 Chemical composition of the received powder Composition
wt.%
SiC Si + SiO2 Fe Al Free carbon (CL) S, Ca, V, Ni, Cu and Zr
98.17 0.60 0.30 0.19 0.20 Trace
particle size and increasing the surface area, with a tight distribution; on the other hand, improvements must be made to reduce contamination. 2. Materials and methods The silicon carbide grade SiC-1000 (Alcoa S.A.- Brazil) produced by the Achesson process was used as the raw material. The chemical composition is specified in Table 1 and the physical characteristics measured were mostly alpha phase, bimodal particle size distribution (d99 and d50 are, respectively, equal to 14.22 and 1.77 μm), density of 3.211 g/cm3 and superficial area of 3.483 m2/g. The milling process was done in a planetary mill (PM-4 model, Retsch). The jars used were specially fabricated to support the elevated wear and to comminute 120 g of material. They were made of stainless steel with an inner diameter of 100 mm, 500 ml capacity and internally coated with tungsten carbide (WC–Co) by HVOF (High-Velocity Oxygen Fuel). The milling conditions were 300 rpm, isopropyl alcohol P.A. (IPA), zirconia milling balls stabilized with ceria (80% ZrO2 + 20% CeO2, Zirconox®, Netsch) with a diameter in between 0.7 and 1.2 mm, density of 6.1 g/cm3 and Vickers hardness of 1200 kgf mm2.
85
The milling media occupied half of the jar volume (250 ml), where 60% of this volume (150 ml) was filled by the grinding balls and the other 40% (100 ml) composed of the pulp (100 g of SiC powder and IPA). The amount of IPA varied a little, since the milling balls had to be slightly covered by the IPA. The milling was done in batches for time periods of 0.5, 2, 4 and 6 h. For periods longer than 6 h, the milling efficiency does not compensate the energy spent, as verified by Matos [8]. The materials were then poured into a Pyrex container, dried in an oven at 70 °C for 24 h, deagglomerated in a mortar and sieved. The powder of each batch was then characterized by X-ray fluorescence spectrometer (Philips PW 2400), particle size distribution (Malver Mastersizer Micro Plus, MAF 5001), superficial area by BET (Gemini III 2375, Micromeritics) and density (Helium Pycnometer, ACCUPYC 1330 Micromeritics). 3. Results and discussion The main object of this study was to produce a highly sinterable submicrometer silicon carbide powder. The first step to look at was the milling efficiency, since as the particles approach submicrometric (b 1 μm) or nanometric (b0.3 μm) dimensions, there is a reduction in the milling efficiency as a result of two competing phenomena, namely, comminution and agglomeration [1]. The second step was to observe if the conditions of the high-energy milling process used generated contamination. The comminution process was executed in a planetary mill, operated at 300 rpm with milling media composed of SiC powder, spheres of ZrO2 + CeO2 and isopropyl alcohol as the dispersant. These processing characteristics resulted in a fast size reduction, but slow change in distribution, as shown in Fig. 1. The as-received powder (Fig. 1a) had a bimodal distribution, with peaks centered at 0.2 and 5 μm. The peak situated at 5 μm
Fig. 1. Particle size distribution: (a) as-received SiC, (b) after 4 h and (c) after 6 h of milling.
86
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88
of milling d90 ≅ 1 μm. This shows the high efficiency of the milling process in producing submicrometric particles. Another important feature is the pivot point for all milling times over 0.5 h. This loco is represented by coordinates 0.3 μm and 38% (marked by the arrows in Fig. 2), which means that 38% of all particles had diameter smaller than 0.3 μm, indicating that there is an expressive quantity of nanometric particles. The milling curve for 4 and 6 h is practically identical but, below the pivot point, the particles seem to be larger than those corresponding to the shorter milling time. This behavior was due to the strong agglomeration between the very small particles, as shown in Figs. 3 and 4, since reduction in size leads to an increase in Fig. 2. Influence of the milling time on the particle size distribution.
was a little higher than the 0.2-μm peak, indicating that the volume of particles larger than 1 μm was greater than the volume of particles smaller than 1 μm. After milling for 0.5, 2 and 4 h, the distribution continued to be bimodal, with the peak centered at 0.2 μm steady and increasing in height, while the second peak was being reduced both in height as in its center position, as the milling time increased (Fig. 1b). This means that the volume of particles smaller that 1 μm was increasing at the expense of the comminuted larger particles. At 6 h of milling, the distribution became unimodal and centered at 0.2 μm, as shown in Fig. 1c. These data pointed out that the planetary milling operation worked well on breaking large particles but not the submicrometer ones, since the peak centered at 0.2 μm kept steady during the whole period. Also, for the milling conditions used, the value of 0.2 μm can be considered as a threshold value, below which particles cannot be broken. A similar result was reported by Matos et al. [8] working with a planetary mill, but using different media. It is possible that this threshold value might be limited by the equipment, the milling media used or the processing characteristics themselves. When the particle size for all the milling times were closely observed, as shown in Fig. 2, the increase in quantity of particles having smaller diameters as the milling time increased was reconfirmed. For instance, 90% of the as-received particles had a diameter of below 7.8 μm, represented by d90 ≅ 7.8 μm. After 0.5 h of milling, the value d90 falls to 4.2 μm and after 4 h
Fig. 3. Effect of the milling time on the surface area measured by BET.
Fig. 4. SEM photos of the received powder (a) and the milled for 4 (b) and 6 h (c).
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88
Fig. 5. The amount of element introduced during the milling process along the milling time.
surface energy and the powder naturally agglomerates to stabilize this, as will be demonstrated below. To confirm that the agglomeration phenomenon was taking place, the surface area of the milled particles was measured by BET. The results shown in Fig. 3 demonstrated a fast comminution rate up to 4 h of milling, where the surface area of the as-received powder changed from 3.4 to 10 m2/g, an increase of about 190%. After 4 h, the surface area still increases, demonstrating that milling still continues, albeit at a reduced rate. The fast reduction in particle size and the agglomeration effect was confirmed by the SEM photos shown in Fig. 4, where the as-received powder (a) and the milled for 4 (b) and 6 h (c) are shown. This confirms the generation of agglomerates as particles become smaller. Another possibility that had to be considered was contamination that could occur in the processes, namely, the increase in surface area could be the result of the breaking of SiC particles plus the addition of other small particles from the milling media. The chemical characterization of the powder for each milling time was done by X-ray fluorescence, as shown in Fig. 5. As can be observed, the amount of ZrO2 increased linearly with the milling time, reaching 15 wt.% for 6 h of milling (all the Zr was considered to come from the ZrO2 balls, since it was the only Zr phase observed in the X-ray diffraction). The element Fe increased to 5 wt.% after 2 h of milling and, thereafter, remained practically constant. The WC content increased in a discontinuous way up to 6.5 wt.% for 4 h of milling and then, the value reduced to 2 wt.%. The other compounds observed were Ce, Co and Cr and their behavior was as expected, since these elements were already part of the milling spheres (80% ZrO2 + 20% CeO2), HVOF coating (WC + Co) and the stainless steel itself (Cr), respectively. Consequently, the Ce, Co and Cr curves were similar to those observed in the parent element, i.e., Zr and Ce curves increased continuously, WC and Co increased up to 4 h and then decreased, while Fe and Cr reached a maximum in 2 h of milling and remained practically constant. These data demonstrated that the SiC powder was somewhat contaminated, especially by the milling spheres. The milling behavior in Fig. 5 might be explained as follows: until 2 h, the elements/compounds taking part in the milling process continuously increased in quantity due to the strong
87
friction between the milling spheres, the powder, the coating and the very small regions of the vessel wall not coated, which introduced small fragments originating from each of those places. For time periods above 2 h, the WC + Co coating was polished and its contribution decreased since no more sharp wedges were available (it is also possible that a SiC layer was built up on the polished coating wall preventing its further contamination into the milling media), the Fe + Cr stabilized because the region from where they came did not enlarge as the milling time increased, and the amount of ZrO2 + CeO2 keeps increasing as a combination between wearing and impact of spheres with the vessel wall. The above results pointed out that the vessel design needs improvement, at least to reduce the contamination from the stainless steel vessel and so SiC spheres shall be used instead of ZrO2. Regarding the WC + Co coating, as it became polished, its contribution tended to null. The amount of elements added to the slurry during the milling process affected the original density of the SiC powder, as shown in Table 2. A 26% increase in density after 6 h of milling was observed, measured by He pycnometry, but the data became asymptotic around 6 h, similar to what happened in Fig. 3 for the same period of time. The contribution of each compound in Fig. 5 to the density can be confirmed by applying the lever rule, which was shown in the Fluorescence column (Table 2). Here, an excellent agreement between the measured and the calculated values was observed. For instance, the smallest difference was 0.1%, the highest 2.3% and the average about 0.5%. These measurements were important because they showed the real density of the milled powder, which has to be used to verify the density of the sintered material. Also, in the absence of a He pycnometer, fluorescence data could be used for calculation of densities, giving reliable values. The data in Table 2 pointed out that the contamination process reached a limit around 6 h of milling, which was coincident with the end of the milling efficiency, since the particle size distribution (Fig. 1) and the BET (Fig. 3) showed almost no variation in their measurements for that same time. In summary, for the milling conditions used, the milling process ceased to be operative when a unimodal distribution was completed (centered at 0.2 μm) and the increase in the surface area tended to null, even though agglomeration of small particles was taking place. The powder generated here possesses average particle size (d50 = 0.4 μm), surface area (BET = 11.5 m2/g) and distribution (d100 b 1.1 μm) comparable or better than commercial sinterable SiC powders, which have d50 and BET between 0.6 and 1.2 μm Table 2 Density of the α-SiC powder, measured by the He pycnometry and calculated from X-ray fluorescence, as a function of the milling time Milling time (h)
Pycnometry (g/cm3)
Fluorescence (g/cm3)
Difference |Modulus| (%)
0 1/2 2 4 6
3.211 3.435 3.695 3.940 4.066
3.215 3.517 3.736 3.898 4.057
0.12 2.33 1.10 1.08 0.22
88
M.A.P. dos Santos, C.A. Costa / Powder Technology 169 (2006) 84–88
and 5 and 17 m2/g, respectively, and distribution below 5 μm [1,9]. In fact, if the center of the distribution (Fig. 1) is taken as the average (0.2 μm), the particles can be classified as nano, but having a larger surface area than the ones measured by Quanli et al. [9], which was 6.71 m2/g. The above discussion must take into account the contamination observed during milling since it altered the original material significantly. The powder was received with a purity of 98.7%, with Si + SiO2 of 0.6 wt.%, but the oxygen on the particle surface was not determined and it is desirable to be lower than 1.5 wt.% [1]. The amount of oxides added during the milling increased the quantity of oxygen and other metals in the powder, and therefore, the final milled powder is not appropriate for solid-state sintering. Usually, this kind of material requires purity levels higher than 99%, low oxygen and addition of C, B or B4C [1,10–12]. On the other hand, when the milled powder was liquid phase sintered (eutectic Al2O3 + Y2O3) the result was a 99% dense material [13] with fracture toughness of 5.5 MPa m1/2 [14], which are excellent results. These data showed that the contamination introduced in the SiC powder had particles equal or smaller than the particles being comminuted, since the distribution finished as unimodal and surface area increased up to 6 h of milling. Also, even though contamination is not desirable, in this case, it was not exactly harmful, since the liquid phase sintering (LPS) gave high density and the fracture toughness was improved. 4. Conclusions The planetary mill easily reduced the average particles size (d50) of the α-SiC from 1.77 to 0.4 μm in only 0.5 h of grinding. The values of d90 for 4 and 6 h of grinding were 1.20 and 0.67 μm, respectively, showing that the process is efficient to obtain fine powders. The planetary mill (300 rpm), with the ZrO2 spheres (0.7 b ϕ b 1.2 mm) and IPA, changed the particle size distribution from bimodal to unimodal, centered at 0.2 μm, in 6 h. For the present conditions, it seems that the process ceases its efficiency in 6 h. In fact, 4 h of milling would have been enough for powders with high sinterability in the liquid phase. When very fine particles were generated (after 4 h), they agglomerated and analysis should be conducted by both particle size analyzer and surface area (BET), since they are complementary and allow good understanding of the comminuting process. The contamination was high, especially from the milling spheres. Consequently, it is suggested that SiC spheres should be used instead of ZrO2 ones, since they would lead to an
autogenous condition. Also, under such conditions, the real density of the milled powder must be checked, either by pycnometry or X-ray fluorescence, which was shown to be a reliable technique to calculate density when a pycnometer is not available. Lastly, even though there was elevated contamination, high density and high fracture toughness was obtained via liquid phase sintering. Acknowledgement We thank the Brazilian Navy (IPqM) for supporting this research and Dr. Francisco C.L. Melo (CTA/Air Force) for the technical collaboration. References [1] G. Schwier, I. Teusel, M.H. Lewis, Characterization of SiC powders and the influence of powder properties on sintering (technical report), Pure Appl. Chem. 69 (6) (1997) 1305–1316. [2] D.W. Richerson, Modern Ceramic Engineering: Properties, Processing and Use in Design, Ed. Marcel Dekker, 2 Ed. Revised and Expanded (1992). [3] H. Berthiaux, J. Dodds, Modelling fine grinding in a fluidized bed opposed jet mill—part I: batch grinding kinetics, Powder Technol. 106 (1999) 78–87. [4] H.W. Hennicke, J. Stein, Process of fine milling for ceramics materials, Mater. Sci. Eng., A 109 (1989) 3–7. [5] C. Suryanarayana, Mechanical alloying and milling, Prog. Mater. Sci. 46 (2001) 1–184. [6] F.V. Brito, Procesamento de Meios Porosos à Base de Sílica, M.Sc Thesis, PEMM-COPPE-UFRJ, Rio de Janeiro-RJ-Brasil-Dezembro (2004). [7] L.M.M. Tavares, B.B. Matos, C.A. Costa, Influência do Meio na Moagem Ultrafina do Carbeto de Silício, Proceedings of the 46 Annual Meeting of the Brazilian Ceramic Society, 26–29 May, 2002, São Paulo, SP, Brazil. [8] B.B. Matos, Influência do meio na moagem ultrafina de carbeto de silício, Projeto de Formatura, Departamento de Engenharia Metalúrgica e de Materiais da Escola de Engenharia da Universidade Federal do Rio de Janeiro, Rio de Janeiro, Abril, (2002). [9] J. Quanli, Z. Haijun, L. Suping, J. Xiaolin, Effect of particle size on the oxidation of silicon carbide powders, Ceram. Int., in press. [10] S. Prochaska, The role of boron and carbon in sintering of silicon carbide, Special Ceramics, vol. 6, The British Ceramic Research Association, Stoke-on-Trent, U.K, 1975, pp. 171–181. [11] K. Negita, Effective sintering aids for silicon carbide ceramics: reactivities for silicon carbide with various additives, J. Am. Ceram. Soc. 69 (1986) 308–310. [12] M. Giuseppe, B. Giancarlo, Pressure-less sintering and properties of (– SiC–B4C) composite, J. Eur. Ceram. Soc. 21 (5) (2001) 633–638. [13] M.A.P. Santos, Processamento e Sinterização de Carbeto de Silício Nacional, DSc. Thesis , PEMM/COPPE/UFRJ, Rio de Janeiro–RJ–Brasil (2003). [14] E.S. Lima, L.P. Brandão, L.H.L. Louro, C.A. Costa, M.A.P. Santos, Mechanical behavior of alpha-SiC based nanocomposites, Proceeding of the 2nd Brazilian MRS Meeting, vol. V.1, 2003, pp. 15–20, Rio de Janeiro.