UDK 621.74:622.362 ISSN 1580-2949 Original scientific article/Izvirni znanstveni članek MTAEC9, 45(5)451(2011) INFLUENCE OF THE GRANULATION AND GRAIN SHAPE OF QUARTZ SANDS ON THE QUALITY OF FOUNDRY CORES VPLIV GRANULACIJE IN OBLIKE ZRN KREMENOVEGA PESKA NA KAKOVOST LIVARSKIH JEDER Marjan Marinšek, Klementina Zupan University of Ljubljana, Faculty of Chemistry and Chemical Technology, Aškerčeva 5, 1000 Ljubljana, Slovenia marjan.marinsek@fkkt.uni-lj.si Prejem rokopisa - received: 2011-01-16; sprejem za objavo - accepted for publication: 2011-08-23 Several quartz moulding sands were used for the preparation of foundry models. All the moulding sands were similar with respect to their sieving parameters, yet different in terms of some morphological characteristics. Foundry models were analysed with regards to their mechanical properties. It was shown that the flexural strength of the prepared foundry models varied substantially, despite the fact that the grain size distributions obtained by sieving analyses of the moulding sands were very similar. The observed dissimilarities were explained pertaining to some differences in the micro-morphological characteristics of the sands, i.e., the particle shape and the grain statistical parameters. A quantitative morphological analysis of the sands was made on photographs taken with optical and electron microscopy using Zeiss KS 300 software. Crucial morphological parameters were defined by treating the moulding sands in laboratory-scale homogenization equipment and subsequent flexural strength measurements. Key words: quartz moulding sands, foundry cores, mechanical properties, morphological analysis Različni kremenovi peski so bili uporabljeni za pripravo livarskih jeder. Pred določitvijo mehanskih lastnosti livarskih modelov so bili vsi peski analizirani s sejalno analizo. Kljub podobnim osnovnim morfološkim lastnostim različnih kremenovih peskov so bile razlike v upogibni trdnosti livarskih modelov relativno velike. Razlike v mehanskih lastnostih livarskih modelov smo razložili z mikromorfološko karakterizacijo peskov. Karakterizacijo mikrostrukture smo izvedli z optičnim ter vrstičnim elektronskim mikroskopom ter kvantitativno analizirali rezultate z uporabo programa Zeiss KS 300. Z medsebojno primerjavo rezultatov se je izkazalo, da se ključni morfološki parametri, kot so oblika ter površina zrn, ter statistični parametri velikosti zrn pri različnih peskih spreminjajo. Izvedli smo tudi obedavo peska na vrtečih se valjih, s čimer smo na preprost način simulirali obdelavo peska na večjih industrijskih napravah, ki se uporabljajo v te namene. S tem je bil demonstriran način obdelave peska z namenom spreminajanja njegovih morfoloških parametrov ter posledično mehanskih lastnosti pripravljenih livarskih jeder. Ključne besede: kremenov pesek, livarska jedra, mehanske lastnosti, morfološka analiza 1 INTRODUCTION properties of the sand cores4-8. Such chemically bonded sand cast systems are often used for the moulds in In foundry practice, sand cores are very extensively ferrous (iron and steel) and nonferrous (copper, alumi- used to form various complicated casting cavities. The nium, brass) castings processes. main ingredients of the moulding sand cores are base When producing cores and moulds, the so-called sand (i.e., a high-quality silica sand or lake sand), binder cold-box method has become increasingly widely used. (clay binders, organic binders or inorganic binders) and This method does not require any heating of the moisture, if clays such as kaolinite are used as binders1,2. moulding sand for hardening. In short, the method is The ceramic moulding cores should have sufficient based on the preparation of a mixture of particle-formed mechanical strength, resistance to erosive wear and material (base sand) and a bonding agent (i.e., a polyiso- chemical corrosion of the liquid metal, high refracto- cyanate compound and a polyhydroxy compound), then riness, low expansion coefficient and superior thermal the mixture is given the form desired and, finally, it is stability. At the same time, no chemical reactions of the hardened by means of a catalyst. The cold-box method foundry core with liquid metal at high temperatures are enables accelerated machining of the moulds in large allowed3. quantities. The raw strength of the moulds, i.e., the The mechanical strength of the moulding cores prior strength directly after machining, is high enough to to pouring the molten metal into the mould is termed minimize the risks involved in handling the cores and "green strength". The key to obtaining optimal sand moulds, so they can be used in the casting process cores with a high green strength and the desired shortly after preparation. Moulds and cores prepared by performance during moulding is a feasible core binder. the cold-box method are also characterized by excellent Therefore, producers and researchers have paid great disintegration after the casting of the metal9. attention to selecting and developing optimum binders, The mechanical properties of moulds are determined which can correspond well with the various required on the basis of strength moduli. The flexural modulus determination is often chosen as the strength parameter, because it can be easily determined experimentally in a practical way10-12. However, the mechanical properties of the moulding cores are strongly influenced by the binding mechanism in the moulds' machining process and also by the morphological properties of the sand being used. The aim of this paper was to demonstrate the relationship between the morphological characteristics of the mould sand and the flexural strength of the prepared moulds. For the first time, crucial morphological parameters were defined and subsequently tested through the tailoring of moulding sands. 2 EXPERIMENTAL PROCEDURE Six types of silica moulding sands (silica content > 98 %), similar with respect to their sieving parameters, were used in the experimental work. The particle size analysis was carried out on dry samples (110 °C, 2 h) of about 1 kg reduced to a mass of about 100 g using a Jones splitter and a +GP+ mechanical sieve shaker equipped with 12 sieves from 3.000 mm to 0.063 mm. The sieves were shaken continuously for a period of 12 min. After shaking, the sieves were taken apart and the sand left over on each of the sieves was carefully weighed and expressed as a percentage of the total mass. Prior to the testing of the mechanical properties, mixtures of various sands (5 kg) and bonding agents (50 g of di-isocyanate and benzyl-ether polyole) were prepared, shaped in a sand rammer into bar specimens (18x3x3) cm and hardened according to the cold-box method with the ethyl-di-methyl-amine catalyst (T = 105 °C, t =15 s). The flexural strength of the specimens was measured one day after their preparation using +GF+ apparatus and was expressed as an average flexural strength of 11 measurements for each sample. For the morphological analysis, the sands were first embedded into Technovit resin and polished. After polishing, the samples were analyzed with a Leitz optical microscope (Leitz Wetzlar). The quantitative morphological analysis of the sands was performed on digital images and expressed as the parameters FERET X, FERET Y, FERET MIN and FERET MAX (intercept lengths in the x or y directions and the minimum or maximum intercept lengths, respectively), FERET RATIO as a ratio of the min and max ferrets, PERIM F (perimeter of the filled analysed region), PERIM C (perimeter of the convex shell of the analysed region), D CIRCLE (diameter of the circle with equivalent area as analysed region), F CIRCLE (form factor of the analysed region - sphericity) and AREA (area of the analysed region). The images were digitized into pixels with 255 different gray values using Zeiss KS300 3.0 image-analysis software. To obtain statistically reliable data, 7-10 different images were analysed in each case. In order to tailor the morphological characteristics of the selected powder, ~ 1.5 kg of sand was put into a ball mill's grinding bottle and rotated (~0.2 r/s) for an extended time (no grinding balls were used). 3 RESULTS AND DISCUSSION The results of the sieving analysis of the six different silica moulding sands are summarized in Figure 1 and Table 1. It appears that the basic morphological characteristics of all six sands, including the granulometric interval, the mean diameter value x and the standard deviation ds, are very similar and within the range of the morphological requests of silica sands used in a foundry. However, one of the crucial parameters for distinguishing silica sands in the foundry industry is the mechanical strength of the prepared casting model. From this point of view, the six silica sands differ substantially, as indicated in Table 1. The highest values of flexural strength were measured in the case of the samples Sand 1 and Sand 2 (777 and 679, respectively), while the lowest values were obtained for the samples Sand 4 and Sand 5 (369 and 317, respectively). Since the basic morphological parameters for the sample pairs Sand 1 and Sand 5, and Sand 2 and Sand 4 are very close, and cannot explain the big difference in the measured flexural strength of the prepared models, a more in-depth morphological analysis of the sands was performed. The optical micrographs and the results of the quantitative morphological analysis are summarized in Figure 2 and in Table 2. Figure 1: Granulometric histograms of the moulding sands Slika 1: Histogrami različnih livarskih peskov Table 1: Basic morphological characteristics of the silica sands and the flexural strengths of the prepared foundry models Tabela 1: Osnovne morfološke karakteristike kremenovih peskov in upogibna trdnost pripravljenih livarskih jeder Sample x /mm ds/mm ffflex/ (N cm-2) Sand 1 0.196 0.079 777 Sand 2 0.164 0.054 679 Sand 3 0.250 0.095 567 Sand 4 0.167 0.050 369 Sand 5 0.210 0.099 317 Sand 6 0.254 0.081 382 Table 2: Morphological parameters of silica sands Tabela 2: Morfolo{ki parametri kremenovih peskov Parameter Sand sample Sand 1 Sand 2 Sand 3 Sand 4 Sand 5 Sand 6 FERET X / ^m 100.785 102.899 143.911 120.273 74.796 123.465 FERET Y / ^m 100.221 103.729 140.119 119.183 77.62 122.288 FERET MIN / ^m 78.271 80.627 106.389 89.137 56.567 91.705 FERET MAX / ^m 117.792 121.547 168.389 143.656 93.003 146.425 FERET RATIO / 0.6645 0.6633 0.6318 0.6205 0.6082 0.6263 PERIM F / ^m 368.745 379.89 569.329 467.135 340.922 485.994 PERIM C / ^m 314.482 322.784 443.152 375.305 241.109 385.725 D CIRCLE / ^m 90.636 92.619 122.214 102.79 63.158 105.489 F CIRCLE / 0.6915 0.6836 0.5402 0.5577 0.4585 0.5531 AREA / ^m2 8799.479 8703.74 15228.13 9298.84 4614.615 10431.86 Figure 2: Optical micrographs of silica moulding sands (silica particles were embedded into polymer resin and polished) Slika 2: Morfolo{ka analiza kremenovih peskov z opti~nim mikroskopom (kremenovi peski so bili zaliti v polimerno rezino in polirani) According to the results summarized in Table 2, the FERET X and FERET Y values are very close for each analysed sample. Some differences among the samples can be seen if the values of the minimum or maximum intercept lengths and their ratios (FERET MIN, FERET MAX and FERET RATIO) are compared. The highest FERET RATIO values (close to 0.665) were calculated for the samples Sand 1 and Sand 2. These three values together somehow also indicate the origin of the silica sands. If the silica sands are treated with mechanical processing in crushers, the grains of the finally prepared sands are normally irregularly shaped, meaning that such sands should have fairly different values of FERET MIN and FERET MAX and consequently relatively low FERET RATIO values. A similar deduction may also be used when the parameters PERIM F (perimeter of the filled analysed region) and PERIM C (perimeter of the convex shell of the analysed region) are compared. Both parameters should be dissimilar if the grains are irregularly shaped. Again, by calculating the ratio of PERIM C/PERIM F, a single number is obtained, which indicates in some way the grain shape. The closer the calculated PERIM C to PERIM F ratio is to 1, the more spherical and full (without cavities or large pores) are the silica grains. The highest PERIM C to PERIM F ratios were calculated for the samples Sand 1 and Sand 2 (0.85 in both cases) and the lowest perimeter ratio in the case of the sample Sand 5 (0.71). The parameters D CIRCLE and AREA are not indicative when the morphological characteristics of the sands are compared with the flexural strength of the casting models. These two values describe the size class of an average sand grain. In contrast, the F CIRCLE (sphericity) value is significant for the final mechanical strength of the prepared models. When comparing the values of the measured flexural strength and sphericity, it can be concluded that higher F CIRCLE values also result in a higher flexural strength of the foundry models. More precisely, the highest F CIRCLE values (0.69 and 0.68) were determined for the samples Sand 1 and Sand 2, respectively, which also exhibited the highest flexural strength, while a relatively low flexural strength was characteristic for the prepared foundry model from the Sand 5 sample with the lowest F CIRCLE value of 0.46. Such a relationship, in which more spherically shaped (less irregular) silica particles also resulted in a higher flexural strength of the prepared foundry models, is in accordance with the principles of the dispersion strengthening of composite materials. Foundry models can be understood as a composite of hard silica particles and softer, more ductile polyurethane, which is formed by the reaction between the di-isocyanate and the benzyl-ether polyole. In such a composite material, any sharp-edged particles locally act as stress intensifiers, reducing the critical stress needed for the mechanical degradation of the tested model. To confirm that silica particle shape is one of the crucial parameters for controlling the flexural strength of Table 3: Change of morphological parameters and flexural strength during sand treatment Tabela 3: Sprememba morfolo{kih parametrov in upogibne trdnosti med obdelavo peskov Parameter Time of sand treatment t = 0 t=3d t = 6 d t = 10 d t = 14 d FERET X / ^m 120.273 118.963 111.461 101.811 100.969 FERET Y / ^m 119.183 125.086 112.723 103.674 105.883 FERET MIN / ^m 89.137 91.304 85.124 79.106 80.856 FERET MAX / ^m 143.656 146.075 133.087 123.079 123.951 FERET RATIO / 0.6205 0.6292 0.6396 0.6427 0.6523 PERIM F / ^m 467.135 466.154 424.539 382.031 380.529 PERIM C / ^m 375.305 381.347 352.102 320.762 323.869 D CIRCLE / ^m 102.79 103.319 95.852 85.997 88.268 F CIRCLE / 0.5577 0.5757 0.5897 0.6297 0.6488 AREA / ^m2 9298.84 10462.27 9542.703 7058.766 7416.08 o flex / N . cm-2 330 / / / foundry models, silica sand (Sand 4) was rotated in a grinding bottle for several days. Such a treatment of the silica sand should not change its mean particle size considerably; however, it should tailor the shape of the sand grains. The results of the quantitative morphological analysis of the treated silica sand and the flexural strength of the prepared foundry models are summarized in Table 3. The main characteristic of silica-sand treatment is the substantial increase in the flexural strength value if foundry models machined from non-treated and maximally treated sands are compared. However, sand treatment tailors not only the flexural strength of the foundry models but also alters the morphological characteristics of the sand. The parameters that describe the size class of an average sand grain (FERET X, Figure 3: Optical micrographs of the silica sand (Sand 4) before (left image) and after (right image) the sand treatment Slika 3: Slike opti~nega mikroskopa vzorca kremenovega peska (Sand 4) pred obdelavo (levo) in po njej (desno) FERET Y, FERET MIN, FERET MAX and D CIRCLE) diminish over time. In this respect, the diameter of the average sand grain (D CIRCLE) was reduced by 14 %. However, strictly from the aspect of grain size, silica sand (Sand 4) is within the range of morphological requirements to be used in the foundry. In view of the fact that the parameters determining morphological size (if silica grain size is within the range of requests) are not decisive for the foundry model flexural strength, as discussed previously, the increased flexural strength of the foundry model was ascribed to the shape change of the average sand grain. The sand treatment increased the ratio PERIM C to PERIM F from 0.80 to the final value of 0.85. Simultaneously, the F CIRCLE value also increased from 0.56 to 0.65, meaning that the sand grains became increasingly spherical. This change in the sphericity of the particles is also evident from the optical micrographs of the non-treated and treated silica sand (Figure 3). 4 CONCLUSION Foundry models prepared from various silica sands may express quite different values of flexural strength. These differences may be interpreted by the morphological characteristics of silica sands. It was shown that the basic morphological parameters, i.e., the mean particle diameter and the standard deviation, were not sufficient to explain the measured differences in flexural strength. Instead, an in-depth morphological analysis revealed that when the silica grain size is within the range of requests, particle shape is one of the crucial parameters for controlling the flexural strength of the foundry models. More spherically shaped silica particles also resulted in a higher flexural strength of the prepared foundry models. The crucial morphological parameters of the silica sands to predict the flexural strength of prepared foundry models are F CIRCLE (sphericity), the PERIM C to PERIM F ratios (ratio between the perimeters of the convex shell of the analysed region and the filled analysed region, respectively) and the FERET RATIO as a ratio of the ferret min to ferret max. 5 REFERENCES 1 D. M. Gilson, The role of different core binder systems in iron casting production: Effect of porosity defects and casting properties, AFS Trans., 101 (1993), 491-496 2M. Stancliffe, J. Kroker, X. Wang, Focusing core binder needs, Modern Casting, 97 (2007), 40-48 3 F. Jorge Lino, T. Pereira Duarte, Ceramic components for foundry industry, Journal of Materials Processing Technology, (2003), 628-633 4 A. Ferrero, M. Badiali, R. Schreck, J. Siak, W. Whited, New binder for casting cores: An industrial application to safety suspension parts, J. Mater. Manufact., 107 (1998), 894-899 5 Y. Kato, T. Zenpo, N. Asano, New core binder system for aluminium casting based on polysaccharide, AFS Trans., 113 (2005), 327-332 6 Z. P. Xie, J. L. Yang, Y. Huang, The effect of silane contents on fluidity and green strength for ceramic injection moulding, Journal of Materials Science Letters, 16 (1997), 1286-1287 7 W. Yu, H. He, N. Cheng, B. Gan, X. Li, Preparation and experiments for a novel kind of foundry core binder made from modified potato starch, Materials and Design, 30 (2009), 210-213 8 N. A. Ademoh, A. T. Abdullahi, Assessment of foundry properties of steel casting sand moulds bonded with the grade 4 Nigerian acacia species (gum arabic), International Journal of Physical Sciences, 4 (2009), 238-241 9 J. Jakubski, S. M. Dobosz, The thermal deformation of core and moulding sands according to the hot distortion parameter investigations, Arch. Metall. Mater., 52 (2007), 421-428 10 J. O. Aweda, Y. A. Jimoh, Assessment of Properties of Natural Moulding Sands in Ilorin and Ilesha, Nigeria, Journal of Research Information in Civil Engineering, 6 (2009), 68-77 11 J. Petrfk, P. Gengel, The capability of green sand mould strength and mould permeability measurement process, Acta Metallurgica Slovaca, 15 (2009), 86-92 12F. Peters, R. Voigt, S. Z. Ou, C. Beckermann, Effect of mould expansion on pattern allowances in sand casting of steel, International Journal of Cast Metals Research, 20 (2007), 275-287