Influence of TiO₂% in Iron Sand on Cast Iron Production by Tatara Iron Making

Abstract

The aim of this project was to learn roles of titanium oxide (TiO₂), an impurity contained in iron sand, in the products resulting from traditional iron making processes, tatara operations. For this purpose, iron sand was collected using two different mineral processing methods from four different locations in the Chugoku area of Japan, and these samples were used to run small-scale tatara experiments. Iron sand collected with traditional gravity separation method contained 8 to 12% TiO₂, while iron sand collected with modern magnetic separation method contained less than 5% TiO₂. When gravity-separated iron sand was used in a tatara under strong reducing conditions, zuku (cast iron) flowed out of the tatara. In contrast, magnetically collected iron sand failed to produce zuku, but did produce raw steel at the bottom of the furnace. Further, even magnetically isolated iron sand could produce zuku when it was supplemented with ilmenite, a titanium-iron oxide containing mineral. The results show that TiO₂ plays a key role in producing cast iron in tatara operations, and the fact that Akome iron sand is known to produce cast iron as it contains higher levels of TiO₂. In contrast, Masa iron sand which is known to produce steel (tamahagane) contains much less TiO₂ and hence is not suitable to produce cast iron. These observations agree with historical descriptions stating that pre-modern tatara operators knew to add iron sand from a specific locality (which is rich in TiO₂) to Masa-type iron sand to produce cast iron.

1. Introduction

Japanese iron production is thought to have started in the 6th century in the Chugoku area of Japan using technology imported from mainland China. Initially, the starting material was magnetite, but soon became iron sand due to a paucity of good ore. Iron sand was obtained in rivers or seashores where it had naturally precipitated. Later, large amounts of iron sand-containing soil were subjected to a forced flow of river water to collect the iron sand (a gravity separation method called the Kanna-nagashi method). The combined use of iron sand and charcoal to make iron is called the Tatara method, and as the Chugoku area had an abundance of the two resources, it remained a center of iron production in Japan. In the 19th century (the Edo period), it is estimated that nearly 80% of Japan’s total iron production occurred in the Chugoku area.¹⁾ However, after the late 19th century (after the Meiji era), imported iron which was produced with modern technology was less expensive, and this led to a decrease in traditional Japanese iron making, and finally to the cessation of commercial Tatara iron production in 1923. Tatara iron production was temporarily revived during WWII and this revival was a national policy decision intended to help produce large numbers of traditional Japanese swords for military personnel, but such production ended following the Japanese defeat in WWII. In 1977, a paucity of tamahagane, the traditional material for making Japanese swords, became a serious problem and it was decided to reactivate the Tatara smelter in Yokota Cho in Nita-gun, Shimane Prefecture (currently Oku-Izumo Cho). This was done by the Nippon Bijutsu Token Hozon Kyokai (the NBTHK or the Society for the Preservation of Japanese Art Swords), and this Tatara is still active today.

Two iron making methods are known: one was designed to produce Kera steel (raw steel) from an iron sand called Masa, and the other method was designed to produce cast iron from an iron sand called Akome (with a carbon content of 0.1 to 2.0 mass% for Kera steel, and about 3 mass% for cast iron). In the Oku-Izumo district which is in the northern part of the Chugoku area, Masa type iron sand was the major product, and this was suitable for making Kera steel, while the southern part of the Chugoku area produced predominantly Akome type iron sand. Presently, only the NBTHK Tatara which is in Yokota Cho in the Oku-Izumo district runs a traditional iron making process using Masa type iron sand which is magnetically separated from weathered granite to produce Kera steel. Because there is no basic difference in the structure or operation of a Tatara to make Kera steel or cast iron, and the Kera-steel making method can use Akome-type iron sand (called Komori iron sand) to produce cast iron during the early half of the Tatara operation cycle, it was thought that the different outcomes (steel or cast iron) were due to differences in the chemical composition of the iron sand ores which were used.²⁾ Indeed, Masa iron sand has a larger particle size and a black color with fewer impurities. In contrast, Akome iron sand has a smaller particle size, a reddish color, and contains substantial impurities such as TiO₂. Such apparent differences were thought to indicate that Masa iron sand is composed of magnetite with a larger grain size, which made it difficult to produce cast iron (and it was therefore suited for making Kera steel), while Akome iron sand is composed of smaller sized particles and is in a more advanced oxidation state, which makes it easier to produce cast iron. However, because the traditional Tatara operations ceased after the end of WWII, the secret of cast iron production was lost, and this, in turn, led to a loss of opportunities to conduct experiments designed to understand the differences between the Akome and Masa iron sands, and why Akome iron sand can produce cast iron.

There are two methods used to collect iron sand: one is gravity separation, and the other is magnetic separation. Traditionally, gravity separation (the Kanna-nagashi method) was used but was later recognized as hazardous for workers, and as leading to pollution of the environment due to the extensive amounts of soil washed downstream from the work sites, leading to a diminished use of the method. Today, the magnetic separation method has replaced the Kanna-nagashi method, and is the only method being used. Because the two methods differ in the principle used to isolate iron sand, it was thought that there was a possibility that the separation method itself affected the products.³⁾ However, because the difference was not thought to be influential enough to affect the quality of the resulting product, no experimental studies were conducted to examine the effect of the type of iron sand ore used on the characteristics of the final Tatara products. Furthermore, professional skills are required to conduct reproducible and stable operations of a Tatara smelter, which is another reason for the lack of such studies.

While it is possible to collect iron sand from different areas around the Chugoku region, it is understood that, except for iron sand collected in the Oku-Izumo district in Shimane prefecture (a Masa ore production area), iron sand often contains a magnetically non-adherent fraction. This means that there was a possibility that the difference in separation methods could have contributed to the difference in the quality of the iron sand, and also to the quality of the products. In the present study, iron sand was collected from various places around the Chugoku area by using both gravity and magnetic separation methods, and these samples were used to run small-scale Tatara operations, and to compare the different separation methods on the quality of the iron sand used and its products. In addition, a special iron sand called Kusuri-kogane which was used in the past to aid in the successful production of cast iron from iron sand which was not the best suited for producing cast iron, was collected and analyzed.

2. Materials and Methods

2.1. Iron Sand

Iron sand was collected from four places in the Chugoku district by using gravity or magnetic separation methods. These localities included the Nariwa river, the Gonokawa river, Gotsu beach, and Yonago beach. The Nariwa river is in Tojo-cho of Shobara City, Hiroshima Prefecture (on the Sanyo side), and the Gonokawa river is in Miyoshi city, and both places are known to have been areas capable of producing Akome-type iron sand. In contrast, Gotsu Beach is in Shimane Prefecture facing the Japan sea and is known to produce iron sand used for making cast iron at the Ataidani Tatara, which was described by Dr. Kuni-ichi Tawara in a book entitled “Ancient Smelting Methods using Iron Sand”.²⁾ Finally, Yonago Beach is located on the Kaike coastline in Yonago City, Tottori prefecture, which also faces the Japan sea.

To obtain iron sand for Tatara operations (5 to 60 kg was used in each operation), two methods were used. One method was magnetic separation using hand-held magnets to collect iron sand from riverbanks or from beaches where iron sand had accumulated as a result of natural gravity separation (the first separation). The collected iron sand was then washed with clean water, dried under the sun, and then magnetically purified again (the second separation). The other method was to use gravity separation. For this purpose, naturally accumulated iron sands in a river or along a beach were scooped up, and then placed in a wooden trough (40 cm wide × 360 cm long) and washed with fresh water so that the iron sand could be separated from sand by gravity. The isolated iron sand was further separated into magnetically adherent and non-adherent (residue) fractions and subjected to chemical analysis. The chemical composition of the gravity separated, and then magnetically purified adherent fraction was used as a reference material for the chemical composition of magnetically isolated iron sand used in the Tatara operations.

2.2. Structure of a Small Furnace or Tatara, Temperature Measurements, and Operating Conditions

The structure of the furnace was reported previously.⁴⁾ Temperature measurements were conducted at four positions, at the “Haguchi” (tuyeres), and 26 cm, 61 cm, and 101 cm above the tuyeres. At the tuyeres, where the temperature is sufficiently high and capable of leading to damage to the thermocouples, an optical pyrometer was used for temperature measurements. At the other three positions, thermocouples were used.

Experiments in iron production were conducted 8 times in experiment 1 (operations I to VIII), and once in experiment 2 (operation IX), for a total of 9 times. The operating conditions are shown in Table 1 and were somewhat different from the ordinary operating conditions that the author uses to produce Kera steel for sword making. Specifically, in ordinary operations using Masa iron sand, the loading ratio of iron sand to charcoal was 1.0, and the air velocity was 6.5 m/s. In the present study, however, the loading ratio was reduced to 0.8 and the air velocity was increased to 10.0 m/s to realize higher reducing conditions. The air velocity was measured using a Pitot tube inserted into the air duct.

Table 1. Operation condition.

Experiment No.	1	2	Reference*
Operation No.	I–VIII	IX
Loading ratio
(Ironsand/Charcoal)	0.8	0.8	1.0
Wind velocity (m/s)	10	10	6.5
Iron sand**	①–⑧	⑩	Masa

*  Operation to produce “Kera steel” for the “Katana” making

**  These numbers correspond to those in Table 3.

In the first series of experiments, a total of eight samples of iron sand collected at 4 different locations using the two different methods (gravity and magnetic separation) were used. In the second experiment, magnetically isolated iron sand from the Nariwa river (the TiO₂ content was 3.6%) which was supplemented with 15% New Zealand ilmenite (with a TiO₂ content of 49.4%) was used (the final TiO₂ content was 10%).

Operating conditions (the furnace structure, furnace material, charcoal, and air velocity) were kept the same throughout the experiments. The resulting iron and slag were all recovered for chemical analysis. The chemical compositions of the furnace materials are shown in Table 2 and were used for the evaluation of furnace wall erosion.

Table 2. Chemical composition of furnace cray (mass%).

T.Fe	FeO	Fe2O3	TiO2	SiO2	Al2O3	MnO	P2O5	S
1.88	0.27	2.39	0.66	55.56	27.63	0.082	0.032	0.007

3. Results

3.1. Experiment 1: Effect of Separation Method on the Tatara Products

3.1.1. Effect of the Separation Method Used on the Quality of Iron Sand

The chemical composition of gravity-separated and magnetically separated iron sand and residues isolated from four different locations (a total 12 samples) is shown in Table 3. Iron sand separated by gravity contained more than 8 mass% of TiO₂ while magnet-separated iron sand contained less than 5 mass% of TiO₂. A sample of 100 g of gravity-separated iron sand, for example, was composed of 80 g of magnetically separated iron sand and 20 g of residue which contained about 30 mass% of TiO₂. Thus, the separation method clearly affected the composition of the iron sand, and especially the TiO₂ content.

Table 3. Chemical compositions of iron sands.

Extracted place	Seperation method	T.Fe	FeO	Fe2O3	TiO2	SiO2	Al2O3	P2O5	S	Content ratio (%)
Nariwa river	①Gravity	57.18	22.9	56.25	12.5	2.90	1.52	0.055	0.006	100
	②Magnetic	65.34	24.37	66.27	3.60	2.10	1.30	0.062	0.004	78.4
	①–②: Residue*	28.38	17.02	21.55	33.97	13.56	4.07	0.044	0.007	21.6
Gonokawa river	③Gravity	57.53	24.41	55.14	9.07	2.70	1.85	0.08		100
	④Magnetic	63.13	26.70	60.60	4.79	1.26	1.56	0.08		82.0
	③–④: Residue	29.79	14.53	26.46	29.82	12.86	5.03	0.10		18.0
Gotu beech	⑤Gravity	60.04	22.71	60.63	8.63	2.85	1.15	0.16	0.007	100
	⑥Magnetic	63.52	24.61	63.49	4.66	2.63	1.11	0.17	0.006	85.0
	⑤–⑥: Residue	38.68	10.41	43.76	27.92	8.83	1.66	0.20	0.048	15.0
Yonago beech	⑦Gravity	59.09	21.12	61.03	8.47	3.63	1.10	0.21		100
	⑧Magnetic	64.45	24.86	64.54	4.82	1.78	0.92	0.18		78.0
	⑦–⑧: Residue	38.68	12.01	41.97	26.40	9.37	1.40	0.32		22.0
New Zealand	⑨Ilmenite	31.79	13.60	30.34	49.40	1.55	0.48	0.05	0.0053
Mixed**	⑩(85%② + 15%⑨)	60.30	22.75	60.88	10.47	2.02	1.18	0.06	0.004

*  Magnetically non-sticking fraction of iron sand.

**  Chemical compositions of the Mixed ⑩ was calculated from the results of ② and ⑨.

3.1.2. Results of Tatara Operations

Tatara operations were conducted using eight different samples of iron sand collected from four different places using two different methods [Table 3, ① to ⑧] under the same operational conditions (a loading ratio of 0.8 and an air velocity of 10.0 m/s). Each operation which used iron sand from ① to ⑧ is labeled here as operation I to VIII, respectively, and Table 4 shows the carbon content of the resulting iron, and the chemical composition and the mineral texture of the slag. Although the operating conditions were the same, only gravity-separated iron sand rich in TiO₂ was found to yield cast iron (outflowing iron from the Tatara) which contained about 3 mass% carbon. Magnetically separated iron sand (with a lower TiO₂ content) yielded Kera steel containing about 1 mass% carbon at the bottom of the furnace.

Table 4. Chemical compositions of products and mineral texture of the slag produced.

Op No.	Iron Sand*	Product C%	Slag TiO2%	Slag T.Fe	Slag M.Fe	Slag FeO%	Slag Fe2O3	Slag SiO2	Slag Al2O3	Mineral texture**
I	①	Cast iron 2.89	25.89	18.72	0.68	13.66	10.61	27.94	7.45	I, Ps
II	②	Kera Steel 1.17	8.67	30.02	0.67	37.54	0.24	30.20	9.58	U
III	③	Cast iron 3.11	29.08	19.84	2.45	19.59	3.10	21.48	8.96	I, U
IV	④	Kera Steel 1.14	12.84	36.08	0.66	45.05	0.60	25.57	6.31	U, F
V	⑤	Cast iron 3.00	27.01	20.82	3.01	20.76	2.40	22.91	8.88	I, U
VI	⑥	Kera Steel 0.97	9.16	37.01	0.56	45.82	1.22	31.69	2.25	U, F
VII	⑦	Cast iron 3.49	26.63	17.97	1.10	12.33	10.42	29.71	3.07	I, Ps
VIII	⑧	Kera Steel 1.33	13.31	37.15	0.58	45.64	1.56	25.21	4.64	U, F

*  These numbers correspond to those in Table 3.

**  Mineral texture: I = Ilmenite, Ps = Pseudobrookite, U = Ulbospinel, F = Fayalite.

3.1.3. Chemical Composition and Mineral Texture of the Slag

The chemical composition and mineral texture of the slag were largely different between the two operations which yielded cast iron (using gravity-separated iron sand) and Kera steel (using magnetically separated iron sand). The slag from the former operation contained high levels of TiO₂ and low levels of FeO, and in addition, ilmenite (I) and pseudobrookite (Ps) were observed. In contrast, the slag from the latter operations contained low levels of TiO₂, and high levels of FeO, ulvospinel (U), and fayalite (F) were observed.

Figure 1 shows scanning electron microscopy photos of the representative slag samples and the results of EDX analysis (arrows). The slag from operation I which produced cast iron (Fig. 1(a)) showed ilmenite (arrow 2) and pseudobrookite (arrow 1), while the slag from operation II which produced Kera steel (Fig. 1(b)) showed ulvospinel (arrow 1).

Fig. 1. Scanning electron microscopy images. (a) Slag from operation I (production of cast iron); arrows 1, 2, and 3 indicate ferrous pseudobrookite, ilmenite, and glassy material (SiO₃, Al, Ca etc.), respectively. (b) Slag from operation II (production of Kera steel): arrows 1 and 2 indicate ulvospinel and glassy materials, respectively.

3.1.4. Balance Sheet and Furnace Wall Erosion

The amount of clay eroded from the furnace wall can be calculated from the chemical composition of the iron sand used, the slag, the furnace wall, the amount of iron sand loaded, and the slag which has flowed out of the Tatara. Table 5 shows the balance sheet for operations I and II, and the amount the furnace wall eroded.

Table 5. Balance sheet of the two operations.

Operation	Iron sand (kg)	Product (kg)	Yield (%)	Products C (%)	Slag (kg)	Furnace erosion amount (kg)
I	48.0	21.6	45.0 (cast iron)	2.89	10.3	2.67
II	60.2	25.5	42.3 (Kera Steel)	1.17	23.8	10.67

In operation I, from 48 kg of iron sand loaded, a total 21.6 kg of cast iron was produced (11.0 kg was in the furnace and 10.6 kg had flowed out), and the yield (produced iron/loaded iron sand) was 45%. Outflowing slag amounted 10.3 kg, and the amount of the eroded furnace wall was 2.67 kg.

In operation II, from 60.2 kg of iron sand loaded, 25.5 kg of Kera steel was produced, and the yield was 42.3%. Outflowing slag amounted to 23.8 kg and the amount of the eroded furnace wall was 10.67 kg.

The slag produced from operation I consisted of a small quantity, but was spongy and light, whereas the slag produced from operation II was larger in quantity and was dense and heavy. Operation I which produced cast iron had lower amounts of slag and a lesser extent of furnace wall erosion compared to operation II which produced Kera steel, and the yield of iron was higher in operation I. Inspections of the furnace wall after breaking down the furnace showed that the extent of erosion in the furnace wall was clearly less in operation I than in operation II.

3.1.5. Temperature Measurements Inside the Furnace

Figure 2 shows temperature changes during operations I and II. In both operations, the temperature was around 1500°C at the tuyeres, 1050 to 1100°C at 26 cm above the tuyeres, 800 to 850°C at 61 cm above the tuyeres, and around 500°C at 101 cm above the tuyeres, which corresponds to the top of the furnace. No major differences were observed between the two operations.

Fig. 2. Temperature changes in the furnace for the Operation I (a) and Operation II (b). (◆) tuyere, (■) lower part of the furnace (26 cm above the tuyere), (▲) middle part of the furnace (61 cm above the tuyere), (×) top of the furnace (101 cm above the tuyere).

3.2. Experiment 2: Effects of TiO₂ Supplementation

In experiment 1, it was strongly suggested that the TiO₂ contained in the residue contributed to the production of cast iron. However, the residue contained not only TiO₂ but also SiO₂ and Al₂O₃ (the gangue component), and thus gangue components other than TiO₂ could be involved in the production of cast iron. To exclude this possibility, in experiment 2, high quality ilmenite which contains little gangue other than TiO₂ was added to the magnetically isolated iron sand to see if cast iron could be produced. Specifically, magnetically isolated iron sand from the Nariwa river ② and high-quality New Zealand ilmenite ⑨ were mixed to form a mixed iron sand ⑩ as indicated in Table 3. The Tatara operation was conducted as in experiment 1 which was under strong reducing conditions (there was a loading ratio of 0.8 and an air velocity of 10.0 m/s: operation IX in Table 1).

3.2.1. Results of Tatara Operations

Following operation IX which used the mixed iron sand (⑩ in Table 3), cast iron was produced (with a carbon content of 3.43 mass%), which confirmed the role of TiO₂ in cast iron production. The carbon content of the resulting iron and the chemical composition of the slag are shown in Table 6. The slag was somewhat viscous, but the operation was smooth and stable and, as in experiment 1 which produced cast iron (operations I, III, V, and VII), the slag contained a high level of TiO₂ (28%) and a low level of FeO (14%) and contained ilmenite and pseudobrookite.

Table 6. Chemical composition of products from operation IX which used a mixture of magnetically separated iron sand prepared for the operation II and 15% ilmenite which contains 49.4% TiO₂ (mass%).

	C%	T.Fe	FeO	Fe2O3	TiO2	SiO2	Al2O3	P2O5	S
Product	3.43							0.19	0.017
Slag		21.75	14.30	15.22	27.53	24.67	8.35	0.024	0.011

4. Discussion

4.1. Effect of the Iron Sand Isolation Method on the Tatara Products

In experiment 1, cast iron was produced only when gravity-separated iron sand was used, and magnetically separated iron sand did not yield cast iron, even though the Tatara operations were conducted under the same strong reducing conditions. Because iron sand separated by gravity or by magnet have distinctive differences in their TiO₂ content, it was suspected that TiO₂ plays an important role in the production of cast iron. In the subsequent experiment 2, it was confirmed that magnetically separated iron sand which was supplemented with high quality ilmenite could produce cast iron under strong reducing conditions. Consequently, it has become clear that TiO₂ is involved in the production of cast iron.

There are archival documents which support the present findings. Kawano wrote in his book that in the northern area of Hiroshima prefecture, iron sand collected in Awaya-cho, Miyoshi city which is called Kusuri iron sand (kusuri refers to something with mystical properties), was used as a supplemental material which was added to iron sand from which it was difficult to produce cast iron.⁵⁾ The chemical composition of iron sand collected from the red clay in Awaya-cho is shown Table 7. Since it had a high content of TiO₂ (16.37 mass%) when isolated with a gravity separation method, it is clear that the Kusuri iron sand functioned to supplement the amount of TiO₂ present and facilitate cast iron production from low TiO₂-containing iron sand. This chemical composition data strongly supports the present experimental results. The reason why the technique used to produce cast iron was lost is because the method to isolate iron sand changed from the traditional gravity separation method (Kanna-nagashi) to a modern magnetic separation method after the end of WWII, which resulted in a decreased content of TiO₂. The magnetic separation method leads to a loss of magnetically non-adherent components which includes TiO₂, leading to difficulties in producing cast iron. Figure 3 schematically shows the role of the separation method on the quality of iron sand and the products produced in Tatara operations.

Table 7. Chemical compositions of Kusuri iron sand (mass%).

Separation method	T.Fe	FeO	Fe2O3	TiO2	SiO2	Al2O3	MnO	CaO	P2O5	S
Gravity	48.25	15.74	51.51	16.37	9.15	2.22	0.24	0.67	0.054	0.0094
Magnet	50.53	13.30	57.47	8.85	10.20	5.57	0.18	0.62	0.075	0.012

Fig. 3. Schematic presentation of the role of TiO₂ which facilitates the production of cast iron from magnetically separated iron sand (i.e., contains less non-sticking fraction).

The content of TiO₂ differed largely depending on the isolation method (gravity vs. magnetic) for all samples including Kusuri iron sand. Tanii studied the composition of iron sand collected from different places in the Chugoku area of Japan and reported that gravity separated iron sand contained different levels of magnetically non-adherent components.⁶⁾ In general, iron sand containing high levels of TiO₂ is more commonly found in Japan, and iron sand containing low levels of TiO₂ (Masa iron sand), which is less affected by the separation method used, is found in localized areas such as Oku-Izumo.⁷⁾ It is therefore necessary to recognize the fact that the quality of the iron sand tends to be affected by the separation method, and hence when the chemical composition of iron sand is reported, the isolation method needs to be specified.

Although magnetically isolated iron sand did not produce cast iron in the present study, it should be theoretically possible if the loading ratio is further reduced, and the air velocity is increased. However, it should be noted that a further reduction in the loading ratio will cause a severe increase in the viscosity of the slag and lead to a failure of the operation. This is because under such conditions, the iron component in the iron sand is sufficiently reduced, leading to a sharp decline in the FeO content in the slag, and results in relative increase of SiO₂ levels and formation of tridymite which has a high melting temperature. Also, a further increase in the air velocity would lead to a scattering of iron sand and result in a reduction of the actual loading ratio. In this context, although it was suggested that the invention of the highly efficient balance-type Tenbin-fuigo bellows in the Edo Period had enabled an increase in the air velocity (and generated strong reducing conditions) thereby made it possible to produce cast iron from Masa-type iron sand,⁸⁾ this notion needs to be reconsidered even after taking into account the fact that the thermal efficiency is better in larger furnaces than in smaller ones. Note that operations with Masa-type iron sand under strong reducing conditions may occasionally produce a small outflow of iron along with the discharged slag.⁴⁾ However, this out-flowing iron is viscous, and its carbon content is lower (about 2 mass%). Furthermore, the slag contains a high content of FeO and exhibits an ulvospinel structure. This Kera-associated out-flowing iron is likely to be produced as a result of physical impacts delivered to the Kera mass when iron bars are inserted through the slag holes in order to break up obstructing solidified slag (this is done to allow the melted slag to flow out of the Tatara).

In short, to avoid confusion in understanding the technology of a cast-iron producing Tatara, it is important to recognize that the smooth cast iron out-flows (with about a 3 mass% carbon content) which occur during operations with Akome iron sand (traditionally called the Zuku Oshi method) is different from the more viscous (but present in a smaller amount) out-flowing low-carbon cast iron (about 2 mass% carbon) which is occasionally observed when using Masa iron sand (traditionally called the Kera Oshi method).

4.2. Classification of Akome versus Masa Iron Sands

Traditionally, iron sand was classified on the quality of the products (i.e. the carbon content) after a Tatara run. Specifically, when the Tatara produces cast iron with a carbon content of 3 mass% or higher, it is called an Akome type iron sand, and when the product is Kera steel with a carbon content below 2 mass%, the iron sand is called a Masa type. Furthermore, because TiO₂ was regarded as undesired impurity which causes problems in operating the Tatara due to the formation of high-melting-point slag, Akome iron sand which contains a larger amount of TiO₂ was regarded as being inferior to Masa iron sand. However, the present study shows that Akome iron sand contains a high TiO₂ content and TiO₂ is essential for the production of cast iron. Since cast iron is produced only from iron sand which contains TiO₂ higher than 8 mass%, and only Kera steel is produced from iron sand which contains TiO₂ below 5 mass%, there is a classification boundary with a TiO₂ content between 5 to 8 mass% which separates Masa iron sand from Akome iron sand. Kusuri iron sand belongs to the Akome iron sand group, but its TiO₂ level is distinctively higher than that of ordinary Akome iron sand, and thus it was used as a supplement to low-TiO₂ containing iron sand in order to facilitate the production of cast iron.

4.3. Different Slag Compositions in Tatara Operations Aimed at Producing Cast Iron or Kera Steel

During Tatara iron production, the iron sand added to the top of the furnace slowly descends while being heated by hot gas and is converted to metallic iron following direct contact with the high-temperature charcoal and forms slag. Because the chemical compositions of the slag and the produced iron are affected by the quality of the iron sand used and by the operating conditions, it is important to analyze the slag to fully understand the processes which occur in the furnace. For this purpose, various studies were conducted on the slag.^9,10,11,12) However, due to difficulties in conducting stable Tatara operations which consist of complex procedures, there have not been many reports describing the operating conditions, chemical compositions of the iron sand, furnace materials, the resulting iron products, and the slag. The present study provides information on the materials used, operating conditions, and the resulting products, which is suitable for an analysis of the slag and the thermodynamics involved.

Figure 4 shows the relationship between FeO and TiO₂ in the slag. Among the 9 operations I to IX, when the product was cast iron, the slag always contained high levels of TiO₂ and low levels of FeO (group A), whereas when the product was Kera steel, the slag always contained low levels of TiO₂ and high levels of FeO (group B), and thus the two groups were clearly separated. From the compositions of the slag, two different mechanisms for carburization can be hypothesized one possibility is that TiO₂ accelerates the penetration, and the other is that FeO suppresses carburization, and decreased FeO concentrations stimulate carburization. However, it was reported previously that when low-TiO₂ Masa iron sand was used to produce Kera steel, the increased loading ratio (decreased reducing conditions) resulted in increased FeO in the slag and decreased carbon content in the Kera steel.⁴⁾ The results indicate that it is FeO rather than TiO₂ which affected the carbon content in the resulting steel. It seems important to lower the FeO content in the slag for better production of cast iron.

Fig. 4. Compositions of FeO versus TiO₂ in the slag produced in each experiment. (A) Results for the operations I, III, V, and VII to produce cast iron from gravity-separated iron sand. (B) Results for the operations II, IV, VI, and VIII to produce Kera steel from magnetically separated iron sand. (★) Results for the operation IX which used a mixture of magnetically separated iron sand used for the operation II (production of Kera steel) supplemented with 15% ilmenite which contains 49% of TiO₂. See Table 4 for the individual data.

TiO₂ forms various compounds with FeO such as ulvospinel (Fe₂TiO₄), ilmenite (FeTiO₃), and pseudobrookite (FeTi₂O₅). These compounds are produced when high TiO₂-containing iron sand was used because the TiO₂ content in the slag increases. However, under the conditions when a strongly reducing atmosphere is used to produce cast iron, the FeO content in the slag decreases, and more ilmenite and pseudobrookite are produced (their melting temperature is higher than other compounds). Then, the question is whether slag containing ilmenite and pseudobrookite can flow or not. The answer is likely to be yes because in the remains of historic iron making sites, high TiO₂-containing iron sand and ilmenite- and pseudobrookite-containing slag with high TiO₂ content have been discovered.¹³⁾ Also, in the present experiments which used gravity-separated iron sand and produced cast iron, the slag contained ilmenite and pseudobrookite. Sasabe et al. created a Fe₂SiO₄–TiO₂ binary phase diagram and showed that there is a low-melting-temperature eutectic point in the TiO₂-containing slag.⁹⁾ Also, Itaya et al. constructed a FeO-TiO₂-SiO₂-5%Al₂O₃ state diagram, and identified many eutectic points, and showed that there exists a wide low melting temperature zone below 1250°C in the TiO₂-containing slag.¹¹⁾ This means that while it is generally thought that Tatara iron making is a method to produce Kera steel, which was accompanied a low temperature slag containing a high FeO content represented by fayalite (Fe₂SiO₄), Tatara operations were also conducted to produce cast iron, which was accompanied by a high-TiO₂ containing slag such as ilmenite and pseudobrookite.

One may next ask how TiO₂ can contribute to the production of cast iron. In experiment I, Tatara operations were conducted using iron sand collected from 4 different places in the Chugoku area using gravity as well as magnetic separation methods under strong reducing atmosphere conditions (with a loading ratio of 0.8, and an air velocity of 10.0 m/s). Table 8 shows the mean values of total Fe (T.Fe), TiO₂, and SiO₂ components (taken from Table 3) in the iron sand, and the FeO, TiO₂, and SiO₂ components in the slag (taken from Table 4) according to the separation method; namely, gravity [①, ③, ⑤, ⑦] or magnetic [②, ④, ⑥, ⑧] separation.

Table 8. Chemical compositions of iron sand and the slag. Concentrations of T.Fe, TiO₂ and SiO₂ in iron sand that were collected at four different places in Chugoku area of Japan by gravity (①③⑤⑦) or magnetic (②④⑥⑧) separation method, and of FeO, TiO₂ and SiO₂ in the slag obtained after tatara operations using the corresponding iron sand. Means of mass% for four different iron sands are shown.

Separation method	Iron sand (%)			Slag (%)
	T.Fe	TiO2	SiO2	FeO	TiO2	SiO2
Gravity ①③⑤⑦	58.5	9.7	3.0	16.6	27.2	25.5
Magnetic ②④⑥⑧	64.1	4.5	1.9	43.5	10.9	28.2

In a thermodynamic analysis, the Fe/FeO oxidizing/reducing power is expressed by an equilibrium of partial pressure of oxygen, while in the reaction of 2Fe + O₂ = 2FeO, the partial pressure of oxygen can be regarded as nearly proportional to (FeO)². In other words, when low TiO₂-containing iron sand is used under the same loading ratio and air velocity, the equilibrium of partial oxygen pressure which leads to equilibration of the slag becomes (43.5/16.6)² = 6.9 times higher, and cast iron is not produced.

A further increase in reducing conditions would realize a decreased FeO content in the slag below 20%, but which leads to a SiO₂ content higher than 40% and causes the slag to be highly viscous due to crystallization of high-melting-temperature tridymite, and thereby continued operation becomes impossible.

The TiO₂ content in the slag depends on the TiO₂ content in the iron sand and on the operating conditions (i.e., on the strength of the reducing conditions), and only when high levels of TiO₂-containing iron sand is used under strong reducing conditions, cast iron can be produced, while in the slag, the TiO₂ content becomes high, and FeO content becomes low. The presence of TiO₂ can allow both, the maintenance of fluidity and a decreased FeO content in the slag, which makes it possible to produce cast iron in a Tatara at a low temperature with charcoal.

From ancient times, producing Kera steel in a Tatara was described as a three-day operation, which is shorter by one day than cast iron making which was described as a four-day operation. Because constructing the furnace requires a substantial amount of labor and materials, including clay and charcoal to desiccate the furnace, and a number of additional processes, it is desirable to run the Tatara as long as possible. However, the Kera steel making method yields more FeO which rapidly penetrates the furnace wall and reacts with SiO₂ to produce fayalite (Fe₂SiO₄) and erodes the wall, thereby making a long-term operation impossible. Furthermore, since the Kera steel making method yields slag which contains a higher content of Fe, the yield of iron is lower than that in the cast-iron making method.

5. Conclusion

It has been thought that TiO₂ which is contained in Akome iron sand is an undesirable impurity which reduces the fluidity of the slag. The present study showed, however, that TiO₂ is critically important in the production of cast iron, increases the yield of iron production, and enables a longer-term Tatara operation.

There is no report which describes the differential production of cast iron and Kera steel by using iron ore which contains different levels of TiO₂. Although the Japanese traditional Tatara iron making method lasted over one thousand years without major technical modifications, it is not a simple antique or historical technique, but can now be seen to be a worldwide highly specialized and sophisticated iron making technique.

Acknowledgements

The author is indebted to the Metallurgical Research Laboratory of Hitachi Metal Ltd. (currently the Proterial Co. Ltd.) and the Kiguchi Technology Co. Ltd. for their kind help in analyzing the chemical compositions of the specimens. In addition, the author greatly appreciates the support of Dr. Hiroshi Arai for his kind advice during the preparation of this manuscript, and the assistance of Dr. Leon Kapp and Dr. Nori Nakamura for their help in preparation of the English manuscript.

References

• 1)   M.  Sasabe,  M.  Tate and  K.  Terashima: Kinsei Tatara Seitetu no Rekishi (History of Early Modern Iron making), Maruzen-planet, Tokyo, (2003), 50 (in Japanese).
• 2)   K.  Tawara: Fukkoku Kaisetu-Ban Korai no Satetu Seirenhou, Tatara Seitetsuhou (Study Group of Iron Sand Refining Method from Ancient Times), ed. by T. Tate, Keiyuu-Sha, Tokyo, (2007), 42 (in Japanese).
• 3)   S.  Hirai: Research Group Report on Analytical Evaluation of Iron and Its Related Remainders, Forum of Technology and History of Iron, ISIJ, Tokyo, (2002), 49 (in Japanese).
• 4)   Y.  Kubo,  Y.  Sato,  Y.  Murakawa and  K.  Kubota: Tetsu-to-Hagané, 91 (2005), 83 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.91.1_83
• 5)   H.  Kawano: Miyoshi-City Awaya-Chou ni okeru Kanna-Iseki (Kanna-Ruins in Awaya-Cho, Miyoshi City), General Information of Miyoshi-City Futami-Gun, Miyoshi, (1956), 3 (in Japanese).
• 6)   H.  Tanii,  T.  Inazumi and  K.  Terashima: ISIJ Int., 54 (2014), 1044. https://doi.org/10.2355/isijinternational.54.1044
• 7)   K.  Hasegawa: Wagakuni Satekko to sono Riyo (Japanese Iron Sand Ore and its Utilization), Gijyutu-Shoin, Tokyo, (1963), 34 (in Japanese).
• 8)   I.  Takahashi: Bull. Iron Steel Inst. Jpn, 1 (1996), 937 (in Japanese).
• 9)   M.  Sasabe,  T.  Yamashita,  S.-j.  Utsuno and  M.  Tate: Tetsu-to-Hagané, 91 (2005), 33 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.91.1_33
• 10)   M.  Haba: Tetsu-to-Hagané, 91 (2005), 23 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.91.1_23
• 11)   H.  Itaya,  T.  Watanabe,  M.  Hayashi and  K.  Nagata: ISIJ Int., 54 (2014), 1067. https://doi.org/10.2355/isijinternational.54.1067
• 12)   K.  Ito and  K.  Matsuzaki: Tetsu-to-Hagané, 91 (2005), 28 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.91.1_28
• 13)   M.  Osawa: Tatara-Kenkyu, 29 (1988), 327 (in Japanese).