Mineralogical Study of Iron Sand with Different Metallurgical Characteristic to Smelting with Use of Japanese Classic Ironmaking

, pp. 1044 1050 Mineralogical Study of Iron Sand with Different Metallurgical Characteristic to Smelting with Use of Japanese Classic Ironmaking Furnace Tatara Hiroshi TANII, 1) Tadahiro INAZUMI 2) * and Keiichi TERASHIMA 3) 1) Volunteer of Hiroshima Prefectural History and Folklore Museum, 122 Kodakoh-Tyo, Miyoshi City, Japan. 2) Supporting Stuff of Tokyo University Museum, 3-16-1 Kimitsudai, Kimitsu City, 299-1143 Japan. 3) Professor of Chiba Institute of Technology, 2-17-1 Tsudanuma, Narasino City, Japan. (Received on November 30, 2013; accepted on March 3, 2014) This study is to investigate and clarify the details about mineralogical characteristics of iron sand used for Japanese classical iron-making furnace Tatara, that produces different kinds of iron product, natural steel and cast iron. Though the product difference is known experientially to be attributed to the difference of iron sand kind Masa and Akome, scientific reason is vague and their characterization is necessary. Specimen was prepared by separating the particles with the similar character by sieving and magnetic separation. Chemical analysis, XRD, mineral texture, EPMA, chromaticity and diffusive reflectance, Raman spectroscopy and specific gravity clarified that main difference between Masa and Akome is in the composition of particles mixed with different mineral characters in term of solid solution TiO 2% in iron oxide and quantity of hematite that is weathered magnetite. CO gas reduction test showed that such mineral characters affect the initial temperature of reduction and mineral kinds formed on the way of reduction. KEY WORDS: iron sand; mineralogical characteristic; TiO 2; weathering; Tatara. 1. Introduction * Corresponding author: E-mail: inazumi@kyouzai.co.jp DOI: http://dx.doi.org/10.2355/isijinternational.54.1044 Traditional operations of Tatara furnace in Japan found that iron sand used as starting raw material of Tatara is classified into two kinds, named Masa and Akome, in term of metallurgical characteristic, and Masa and Akome produce different kinds of iron product that are natural steel Kera and cast iron Zuku respectively. The object of this study is to investigate the main reason why the different kind of iron sand produces different product especially from the stand point of mineralogical characteristics of iron sand. It is well known that main mineral of iron sand is magnetite and that the chemical difference between Masa and Akome is distinct at TiO 2 content. And many researchers have studied the reason why TiO 2 affects metallurgical reaction and kinds of product. Those previous studies 1,2) claimed that TiO 2 content lowers melting point of slag that enhances the difference of carburization, resulting in the difference of iron product. But metallurgical process in Tatara furnace is composed of complex reactions such as metallic iron formation, sintering and carburization to form bloom texture those occur under solid state or with partial melt at relatively low temperature and not under homogeneous molten condition. Such reactions proceed heterogeneously and so solid characters such as the difference of mineral constitution and texture of iron oxide are considered to enhance heterogeneous reaction, resulting in formation of heterogeneous bloom texture. But until now the precise details of mineral characters are not known because mineral constitution and texture of iron sand is very complex. Unknown details of mineralogical differences between iron oxide of Masa and Akome and also the effect of the mineral character on metallurgy, that will be essential basic item for clarifying the different mechanism of producing either Kera or Zuku, were studied in this study. 2. Experimental Method 2.1. Iron Sand Used for This Study Raw sand (run of mine) was taken at the district of famous old mine sites where used to produce representative Masa and Akome in Shimane prefecture, that location is Hanaidani for Masa and Zakka for Akome, respectively. 3,4) The concentrates were produced by preliminary treatment of the raw sand. In this study according to the traditional technique Kanna-nagashi, water dressing method 4) similar to popular panning method for gold, was carried out. By the operation most of gangue minerals were washed out and iron oxide was concentrated. 2.2. Pre-treatment of Concentrate The complexity of mineral composition is caused to the geological origin, weathering degree of mother rocks and so force. The conventional result such as chemical analysis 2014 ISIJ 1044

shows average bulk property, so that it is difficult to distinguish the details of individual particle character. Then the authors tried to separate particles with the similar mineral characters by sieving and magnetic separation, because particle size changes with weathering and magnetism of iron oxide differs by mineral kinds and TiO 2% in crystal. 5) The size of sieve opening is 1, 0.5, 0.25 and 0.125 mm and the method of separation by magnetic force in Table 1. 2.3. Analytical Methods of Specimen Separated by Sieving and Magnetic Force Firstly chemical composition of each specimen separated by sieving and magnetic force was analysed in order to confirm how degree the specimen is separated differently. Next, mineral constitution of the specimen was analyzed with XRD. And after the specimen was buried in resin and polished, mineral texture was observed under microscope and some individual particles are analysed with EPMA. Furthermore, in order to analyze TiO 2 bonding sate of iron oxide, Raman analysis was carried out. And in order to detect the existence of minor mineral kind such as limonite, diffusive reflection spectrum was measured. For color analysis chromaticity of the specimen as natural size and the specimen ground with agate bowl were measured. Magnetization curve with use of VSM (Vibrating Sample Magnetometer) was measured more details of magnetization strength difference among the specimens with the same kind ones by intensity of magnetism, those was classified by the method shown in Table 1. 2.4. Reduction Method of Specimen Separated by Sieving and Magnetic Force Before reduction test under actual condition of previous reports 6) will slate in future, preliminary reduction test with use of TG/DTA (Thermo-gravimetric analysis/differential thermal analysis) was carried out with simple CO gas. The atmosphere coordinates to carbon deposition range. This reduction condition is not to fit directly Tatara process, but may give basic information how particle with different mineral character behaves under reduction condition. Specimen for the test was selected to be able to evaluate the effect of representative character on reduction process. 3. Results 3.1. Distribution of Particle Size and Particle Magnetization-intensity in Concentrate Most particles (92%) of Masa concentrate belong to strong magnetism as well as main particles (71%) of Akome (Tables 2, 3). Though the specimens with medium and weak Table 1. Classification method by magnetization strength (123magnet: 1.300 Gauss). 1 Strong magnetism: Jump up and bonding to permanent magnet 20 mm above 2 Medium magnetism: Jump up and bonding to permanent magnet 5 mm above 3 Weak magnetism I: Direct contact bonding to permanent magnet 4 Weak magnetism II: Direct contact bonding to strong permanent magnet 4 100 Gauss Table 2. Distribution of particle size and particle magnetization-intensity in Masa concentrate. Magnetism intensity Particle size Strong (g) (%) Medium (g) (%) Weak1 (g) (%) Weak2 (g) (%) Total (g) Total (%) Coarse 0.5 1 mm 46 6.1% 2 7.1% 1 7.1% 1 5.6% 50 6.2% Mean 0.25 0.5 mm 38.4 51.2% 14 50.0% 6 42.9% 3 17.0% 407 50.2% Fine 0.125 0.25 mm 253 33.7% 9 32.1% 5 35.7% 1 6.0% 268 33.1% Extremefine<0.125 mm 67 8.9% 3 10.7% 2 14.3% 13 72.0% 85 10.5% Total 750 100% 28 100% 14 100% 18 100% 810 100% Magnetism (wt.%) 92.6% 3.5% 1.7% 2.2% 100% 100% Specific gravity 5.0 3.7 3.6 3.2 Table 3. Distribution of particle size and particle magnetization-intensity in Akome concentrate. Magnetism intensity Particle size Strong (g) (%) Medium (g) (%) Weak1 (g) (%) Weak2 (g) (%) Total (g) Total (%) Coarse 0.5 1 mm 13 4.4% 3 4.4% 3 6.1% nil 19 4.5% Mean 0.25 0.5 mm 67 22.5% 26 38.2% 24 49.0% 2 50.0% 119 28.4% Fine 0.125 0.25 mm 102 34.2% 25 36.8% 17 34.7% 2 50.0% 146 34.8% Extremefine 0.125 mm 116 39.0% 14 20.6% 5 10.2% nil 135 32.8% Total 298 100% 68 100% 49 100% 4 100% 419 100% Magnetism (wt.%) 71.1% 16.2% 11.7% 1.0% 100% Specific gravity 4.8 4.5 3.8 1045 2014 ISIJ

Table 4. Chemical composition of specimen by particle size and magnetism intensity. Kind Size Magnet. T.Fe FeO Fe 2O 3 SiO 2 Al 2O 3 CaO MgO MnO TiO 2 Masa Mean Strong 66.13 25.3 66.37 3.23 0.99 0.74 0.39 0.27 0.87 Masa Fine Medium 59.55 22.1 60.48 8.14 2.02 0.24 0.15 0.92 1.91 Akome Mean Strong 65.37 18.8 72.55 2.18 1.33 0.03 0.06 0.36 3.74 Akome Fine Medium 62.55 4.96 83.92 4.10 1.35 0.04 0.07 0.31 3.48 Akome Mean Weak I 32.95 13.8 31.76 9.68 1.21 0.23 0.14 3.77 38.26 Fig. 1. Microscopic photographs of Masa (a) and Akome (b). magnetism are small amount in concentrate, Akome is slightly more than those of Masa. The mean size of Masa is 0.25/0.5 mm and Akome is slightly smaller than Masa. The specific gravity of Masa with strong magnetism is a little higher than that of Akome. And the weaker the magnetism, the lower the specific gravity (Tables 2 and 3). 3.2. Mineral Characteristic of Specimen by Sieving and Magnetic Separation 3.2.1. Specimen with Strong Magnetism TiO 2% of the specimen in Akome is higher and FeO% is lower than those in Masa (Table 4). Under microscope most grains are homogeneous magnetite with partial hematite transferred from magnetite by weathering (Fig. 1). Weathering proceeded more remarkably than our conventional idea. Hematite increases with decrease of particle size. Hematite quantity in Akome is more than Masa. Small exolusion ilmenite is observed in Akome. XRD (Fig. 2) results show that main mineral kind of both Masa and Akome with strong magnetism is magnetite. Both sands include hematite and a little γ hematite. Also Akome includes a little ilmenite. TiO 2 in magnetite and hematite of Akome measured with EPMA is higher than that of Masa. Also indices of magnetite and hematite peaks of XRD shift with TiO 2% (Fig. 2). Raman spectrum difference between Masa and Akome indicates Ti bond in Akome (Fig. 3). Consequently the important difference between Masa and Akome is the difference of solid solution TiO 2% in iron oxide. 3.2.2. Color Analysis and Details of Magnetization Color indices of b (yellow axis) and a (red axis) on chromaticity coordinates are plotted into the different zones Fig. 2. XRD of Masa (a) and Akome (b) with different size and magnetism. 2014 ISIJ 1046

Fig. 3. Raman spectrum of Masa (a) and Akome (b) with strong magnetism. Fig. 4. Color indices on chromaticity. (Online version in color.) between Masa group and Akome group, except Masa with weak magnetism (Fig. 4). And the coefficient of a/b ratio of ground specimen is lower than that of non-ground specimen with natural size. The different color indices between ground and non-ground specimen is due to the difference of observation portions between surface and inside of particle. Also color indices of the specimens tend to be different by particle size. As for indices of ground specimen, the finer the specimen size is, the larger a and b value. In old time Murage, superintendent of Tatara works, has diagnosed the kind of iron sand between Masa and Akame by the difference of glance color as state and again the glance color of sand diminished by hand. 4) Such old technique like knowhow means for characterization to need estimating both color of surface and inner portion of sand. Chromaticity is effective to diagnose kinds of iron sand and the complex composition of iron oxide kinds in iron sand. These color difference is considered to be attributed to firstly TiO 2 content in iron oxide and secondarily quantity of hematite and ilmenite. To know the meaning of a and b value, specimens were heated in air, and a value changed. Nakashima showed by his experiment with use of reagent magnetite, hematite or limonite that color change by hematite increase occurs along a axis on line of low b/a ratio, and also that increase of b value means limonite increase. 7) Small amount of limonite was confirmed by calculating spectrum of diffusive reflectance with formula of Kubelka Munk model 7) (Fig. 5). Therefore increase of a & b value with decrease of particle size is attributed to the increase of hematite and limonite formed by weathering of magnetite. And the results of XRD and reflection spectrum showed that hematite and limonite increase with increase of TiO 2 and size down of particle. One of the reasons why the higher coefficient of a/b of nonground specimen than ground one is considered limonite formation on particle surface. Among specimens with strong magnetism, saturation magnetization (72 emu/gr) of magnetite in Masa is relatively stronger than that (69 emu/gr) in Akome. It is well known that TiO 2 solid solution in iron oxide weakens magnetization of magnetite. 5) Therefore the result that the finer the particle 1047 2014 ISIJ

Fig. 5. Limonite absorption of spectral. (Online version in color.) Fig. 6. TG/DTA reduction test with CO gas (Masa and Akome with strong magnetism). size of specimen, the lower magnetization of the specimen is attributed to increase of hematite and TiO 2. And so as TiO 2 and hematite increase with decrease of particle size, magnetization become lower. But it seems strange that the magnetization intensity does not drop so much despite increase of TiO 2 and hematite. Formation of γ hematite and γ titano-hematite by weathering is thought to be one of the reasons, but the precise study is future item. As the result weathering of magnetite in iron sand is found to have proceeded more remarkably than our conventional idea. And weathering of Akome proceeds more easily than that of Masa. Most of fine particle is considered to be the degraded one from weathered potion. 3.2.3. Specimen with Medium and Weak Magnetism Quantity of the specimen with medium magnetism in concentrate is small 3.5% in Masa, but in Akame is 16.2%. Specimen with weak magnetism in concentrate is 3.9% in Masa and 12.7% in Akame. TiO 2% become higher and FeO% lower in the specimens with medium and weak magnetism than those with strong magnetism (Table 4). Akome specimens with medium and weak magnetism is more than those of Masa. Under microscope, periphery and cracks of magnetite particles changed remarkably to hematite or titano-hematite in Masa and Akome respectively, also singleedged of iron oxide and gangue minerals were observed in the particle with weak magnetism. XRD results showed that though some magnetite is left in both sand, ilmenite and titano-hematite increase in Akome and hematite in Masa comparing to the specimen with strong magnetism. And γ hematite in Masa and γ titano-hematite in Akome were detected. EPMA result showed that TiO 2 % in hematite of Akome is high, and so it is titano-hematite, but magnetite of Masa transformed to ordinal hematite, those solid solution difference are detected by magnetite peak shift of XRD. Result of chromaticity showed that a and b of the specimen with 2014 ISIJ 1048

medium and weak magnetism is higher than that with stronger magnetism (Fig. 4), and showed that the smaller the size of specimen are, the higher a and b value (Fig. 4). It is remarkable that color index of Masa specimen with medium magnetism is located in the zone of Akome group. Masa particles with medium magnetism has one with the same character of Akome, that means that Masa includes partly the same kind particles of Akome and that iron sand is considered to be ordinarily more or less mixed with representative Masa and Akome character. The specimen with medium and weak magnetism is composed of mainly weathered particles as well as gangue minerals and single-edged. 3.3. Result of Reduction Test One of the remarkable mineral characters of Akome is that iron oxide is titano-magnetite and titano-hematite. On the other hand iron oxide of Masa is relatively pure magnetite and hematite. In this study the effect of different TiO 2 solid solution % in magnetite and quantity of weathered minerals on the reduction with simple CO gas was analyzed with use of TG/ DTA.TG result shows small decrease of weight until peak of around 500 C and after the peak weight increase remarkably (Fig. 6). DTA change shows exothermal reaction from hematite to magnetite starts around 200 C, increases until around 500 C peak and after the peak endothermic reactions including reactions of magnetite to wustite or metallic iron and also carbon deposition occur. The weight change shows balance of those reactions. Reduction result of Masa with strong magnetism and Akome with medium magnetism showed that TiO 2% in magnetite depresses reduction reaction, but adversely titano-hematite tends to lower the starting temperature of exothermal reduction reaction from hematite to magnetite and also lower the peak temperature around 500 C. In addition reduction test of Masa with weak magnetism, that has the same mineral characters of representative Akome, showed that the reduction behavior is similar to representative Akome specimen. XRD results showed that mineral kind after reduction is wustite for Masa and ilmenite for Akome respectively. And cementite Fe 3C is found to be formed in both sands (Fig. 7). 4. Discussions The results about the intrinsic differences between Masa and Akome is that TiO 2 contains in iron oxide of both magnetite and hematite as solid solution, and another kinds of minerals containing TiO 2 is not included except a little ilmenite. Main representative iron oxide of Masa is relatively pure magnetite and hematite. On the other hand, main representative iron oxide of Akome is titano-magnetite and titano-hematite. Magnetite in Masa transforms to hematite via γ hematite by weathering. Titano-magnetite in Akome transforms to titano-hematite via γ titano-hematite. This study showed that weathering of such magnetite in iron sand proceeded more remarkably than our conventional idea. The previous reports 8,9) are summarized that reducibility is high by the following order of hematite, titano-hematite, magnetite and titano magnetite. Our results accorded to the results by previous researchers that titano-magnetite is the most difficult to reduce, 8) and reduction of titano-hematite as well as hematite is easier than that of pure magnetite. 9) The reducibility difference is attributed mainly to the mineral formation under way of reduction, that is, the mineral after reduction is wustite for pure magnetite of Masa and ilmenite for titano-magnetite of Akome. On the other hand the smaller size of iron ore, the easier the reduction. As particle size of Akome is smaller that that of Masa, Akome has advantage property at this point, but adversely magnetite with high TiO 2 of Akome depresses reduction. In addition it is remarkable that titano-hematite affects reduction though titano-hematite in Akome is minor content. TG/DTA results showed in this study that Akome with high TiO 2 affect starting temperature of reduction, bring forward exothermal reaction from hematite to magnetite. Though total reduction of Akome is worse than Masa, Akome with high TiO 2 has titano-hematite that has such characteristic reducibility. Iwase 8) and Park 9) pointed out that Fig. 7. XRD pattern of Akome after 500 C reduction. 1049 2014 ISIJ

reducibility of titano-hematite by roasted titano-magnetite is better than pure magnetite. Weathering of Akome proceeds more than Masa, that indicates weathering of titano-magnetite is easier than that of pure magnetite. How solid metal is formed at early stage of reduction is the important theme for Tatara metallurgy. As metallic iron formation is dependent on reduction starting and the metal formation at early stage by depress of slag formation, these details of mineral characters are important as affecting factor. And in this study cementite Fe 3C is found. Though the reduction condition is different from Tatara process, Park pointed out that cementite is formed more easily from γ titanohematite. 9) Park considers that diffusion of solid solution Ti in iron oxide affects cementite formation. In this study both Masa and Akome formed cementite under the reduction condition of carbon deposition. The reason why even in Masa with low TiO 2 cementite was found is inspected that a small amount titano-hematite in Masa formed cementite, becauses iron sand is composed of mixed particles with either representative characteristic of Masa or Akome. In fact some of Masa particles with medium and weak magnetism have the same characteristic of Akome (Fig. 4), and the reduction test of such Masa with similar characters of Akome showed the same reduction behavior to that of representative Akome. Though the reason weather cementite is formed in Tatara furnace or not is future item, but if cementite is formed in Tatara furnace, cast iron Zuku is easy to form. As such mechanism has not been known as yet, future study will be expected. This study elucidated that ordinary iron sand is composed of mixed particles of Masa or Akome with various ratio, resulting in mixed structure of steel and iron. To clarify the details of such mechanism further experiment under wide reduction conditions and slag formation is necessary. 5. Conclusion (1) Intrinsic difference of mineral characters between Masa and Akome is in TiO 2% difference in iron oxide as solid solution and quantity of hematite that is weathered magnetite. The mineral composition reflects its color, and the delicate color difference is able to distinguish quantitatively and effectively with use of chromaticity. (2) Weathering of magnetite in iron sand is found to have proceeded more remarkably than conventional idea, and the weathering degree affects metallurgical properties of iron sand. Such mineral characters change reducibility, reduction start temperature and formation of mineral kind under reduction. As iron sand is ordinarily the mixed composition of representative Masa and Akome, the complexity is considered to affect to form complex bloom of mixed steel and iron. REFERENCES 1) Y. Kubo and K. Kubota: Tatara Kenkyu, No.50 (2010), 29. 2) M. Amatatsu: 25th Forum of Technology and History of Iron, ISIJ, Tokyo, (2013). 3) S. Ishihara: Geological News, 243 (1974). 4) K. Tawara : Japanese Classical Smelting Method, Maruzen, Tokyo, (1933), Reprint with commentary by M. Tate, Keiyousha, Tokyo, (2007). 5) O. Kawano: Introduction to Rock Magnetism, Tokyo Univ. Press, Tokyo, (1982). 6) K. Nagata: Kinzoku, No.7 (2005) No.8 (2008), No.5 (2010) No.5 (2011). 7) S. Nakashima: Color Change on the Earth, Kinmiraisha, Nagoya, (1994). 8) K. Iwase: Studies of Sand Iron, Kagakukoughousha, Tokyo, (1945). 9) E. Park and O. Ostrovski: ISIJ Int., 43 (2003), No.9, 1316, 44 (2004), No. 1, 7. 2014 ISIJ 1050