Scientific analysis of han dynasty iron artifacts excavated from the Luojiaba Site in Southwest China

Abstract

Seventeen Han Dynasty iron artifacts from the Luojiaba site, including materials from unit H235 and a burial area, were analyzed using metallography and SEM–EDS. The results indicate the concurrent use of cast iron and wrought iron/steel technologies. Cast iron artifacts—comprising white iron, mottled iron, and malleable cast iron—were mainly employed for containers and plates, whereas wrought iron and steel objects, such as knives and awls, were produced through forging, carburization, and possibly ‘CHAO’-fining processes. The constitution of inclusion demonstrates the coexistence of bloomery iron and fining steel, with bloomery iron predominating. Comparisons between slag from H235 and artifact inclusions reveal close compositional correlations, suggesting local production of many bloomery iron artifacts, while some items, including a fining steel bar and an iron sword, likely originated elsewhere. These results provide the scientific evidence for a complex, regionally adaptive metallurgical system at that region during the Han Dynasty.

Introduction

The Iron Age in China began in the early 8th century BCE, marking the transition from the Western Zhou to the Eastern Zhou Dynasties, or the onset of the Spring and Autumn Period. This era saw the emergence and initial development of iron metallurgy in ancient China. During the Spring and Autumn and Warring States Periods, the iron industry expanded rapidly, achieving substantial progress by the late Warring States. By the Qin and Han Dynasties, China had fully entered the Iron Age¹. However, regional variations in cultural and technological development led to distinct trajectories in the adoption and evolution of iron implements.

In the pre-Qin period, Southwest China—distant from the Central Plains—was home to ethnic groups, such as the Ba (巴), Shu (蜀), Dian (滇), and Yelang (夜郎). Although these societies were influenced by Central Plains culture and technology, their development remained comparatively limited. The Qin conquest of Ba and Shu in 316 BCE brought transformative changes to the Bashu region. The establishment of new production relations stimulated economic growth and cultural advancement². Under the Han Dynasty, which inherited and consolidated Qin administrative systems, the Bashu area was further integrated into imperial governance. Policies, such as the establishment of prefectures and counties, organized migration, and water conservancy projects strengthened state control while facilitating cultural and technological exchange between the Central Plains and the southwest. These developments accelerated scientific and technological progress in the region. Northern migrants, including figures, such as the Zhuo clan and Cheng Zheng, made significant contributions to the growth of the iron industry in Southwest China.

The Chinese Iron Age was characterized by two major metallurgical systems: bloomery iron and cast iron–based production. While the Central Plains have been the primary focus of research, the technological history of Southwest China remains comparatively understudied. Situated at the crossroads of cultural interaction and technological exchange during the Han Dynasty, Sichuan represents a key region for exploring the diffusion and adaptation of ironworking traditions.

Recent archeological investigations in Sichuan Province have identified 75 sites associated with iron metallurgy, five of which have undergone systematic excavation³. Nevertheless, metallographic studies of iron artifacts from Southwest China remain limited. Li’s analysis of artifacts from Yunnan revealed two major types—wrought and cast iron—from the Ancient Dian Kingdom. By the mid- to late Western Han Dynasty, advanced forms, such as refined steel, steel-faced implements, quenched steel, and cast iron objects appeared in Dian tombs, indicating rapid technological development and widespread application of iron metallurgy in tools and weapons⁴. Chen et al. examined 11 iron artifacts from the Kele (可乐) Tomb Site in Guizhou, dating from the late Warring States to the early Western Han. Their results demonstrated that both casting and forging were used, employing materials, such as decarburized wrought iron, malleable cast iron, bloomery iron, and refined steel. The findings suggest that ironworking in the Kele area shared a technological foundation with the Central Plains, centered on cast iron smelting and steelmaking⁵.

Li et al. analyzed a cast-iron bridge pier unearthed in Guanghan (广汉) dating to 96 BCE. The 1.38-ton artifact, composed of gray cast iron and associated with casting molds, confirms that large-scale iron casting was practiced in Sichuan by the late 1st century BCE. This discovery demonstrates that the region’s metallurgical technology was on par with that of the Central Plains and served as a production hub for Southwest China⁶. Additional studies by Li et al. on artifacts from the Qiaogoutou (桥沟头) Site, dating from the Warring States to the early Western Han, identified bloomery carburized steel products and discussed their technological origins⁷. Zhang’s subsequent analysis of inclusions from the same site provided further confirmation of bloomery steel manufacture⁸. Further research by Zhang and colleagues on artifacts from the Han Dynasty Muyi (牡宜) Site in Yunnan and the Baozi (宝资) Mountain Tomb in the Chengdu Plain yielded new insights into local iron production technologies^9,10.

Despite these advances, research on iron artifacts from Southwest China remains limited. Existing studies have concentrated primarily on metallographic features, with little analysis of inclusions. Current data are insufficient to fully characterize the development of iron metallurgy in Sichuan and the broader southwest. Scholarly debate continues over whether wrought iron and steel artifacts were produced through bloomery or ‘CHAO’-fining processes, underscoring the need for comparative analytical methods and interdisciplinary approaches. The Luojiaba Site, a key representative of Ba culture in eastern Sichuan, dates from the Warring States to the Han Dynasty (c. 4th–1st centuries BCE). Although metallurgical remains from H235 have been previously examined^11,12, the iron artifacts from both H235 and the associated burial area have not yet undergone systematic scientific analysis. This study presents the first comprehensive investigation of iron artifacts from the Luojiaba Site, aiming to reconstruct the iron production technologies of eastern Sichuan during the Han period and to assess their broader significance in the evolution of early Chinese metallurgical traditions.

Methods

Archeological Background of the Luojiaba Site

The Luojiaba Site is situated on the secondary terrace of the Hou River, a tributary of the Qujiang River, in Luojiaba Village, Jinhua Town, Xuanhan (宣汉) County, Dazhou (达州) City, Sichuan Province. Approximately 46 km from Xuanhan County and 60 km from Dazhou City, the site faces Puguang (普光) Town across the Zhong River. It occupies the confluence of the Zhong and Hou Rivers and comprises three adjoining areas: Luojiaba Outer Dam, Zhangjiaba, and Luojiaba Inner Dam. Among these, the Luojiaba Outer Dam and Zhangjiaba form the core zone, covering roughly 103.33 ha. Set within a hilly and mountainous landscape, the site is bordered by water on three sides and backed by mountains on the fourth. These natural barriers have contributed to its exceptional preservation. Archeological evidence indicates that the site dates from the Warring States to the Qin–Han periods (Fig. 1a). Iron artifacts unearthed at Luojiaba include production tools, domestic utensils, weapons, and other items. Many are heavily corroded or damaged, complicating typological identification¹³. To determine their raw materials and manufacturing techniques, and to clarify the iron-metallurgical traditions characteristic of eastern Sichuan, this study analyzed samples from iron objects excavated in Luojiaba Site H235 and the associated burial area (Fig. 1b). Systematic scientific analyses were carried out to provide new data supporting research on early iron-metallurgy technology in the Sichuan region.

Fig. 1: Location and general information map of the Luojiaba site.

a Location map of Luojiaba site. b Location map of Luojiaba site H235 and burial area.

Samples

Based on the types and preservation conditions of the excavated artifacts, iron artifacts samples were selected from H235 and part of the burial area randomly. Containers, such as iron pots and iron cauldrons were sampled from fragments of the body, while tools were from the blade. Except for some samples with severe rusting, a total of 17 samples suitable for analysis were obtained. The types of samples included daily utensils, production tools, weapons, and iron plates (Figs. 1 and 2).

Fig. 2: Iron artifacts excavated from H235 at Luojiaba site.

1. Scrap iron tool LJBF01 2. Iron knife LJBF02 3. Iron awl LJBF03 4. Iron awl LJBF04 5. Scrap iron tool LJBF05 6. Scrap iron tool LJBF06 7. Scrap iron tool LJBF07 8. Iron bar LJBF08 9. Scrap iron tool LJBF09 10. Iron knife LJBF10.

10 iron artifacts were unearthed from the H235 at Luojiaba Site in Xuanhan County. These were daily-use utensils, numbered LJBF01-LJBF10, all of which were severely rusted. The identifiable shapes included iron awls (LJBF03) and iron knives (LJBF02, LJBF10), while the shapes of the others were unclear (Fig. 2, Table 1). These iron artifacts are likely to be production tools used in workshops. Some of them are unidentified iron artifacts, which may also be forging artifacts. The age of H235 has been dated to the late Western Han Dynasty to the Eastern Han Dynasty (ca. 100 BC-AD 220)¹².

Table 1 Macroscopic Observation of Iron Artifacts Excavated from H235 at Luojiaba Site

Six iron artifacts, numbered LJBF11-LJBF16, were recovered from the burial area of the Luojiaba Site. These included three iron pots for daily use, two iron cauldrons, and one iron sword. In addition, one iron plate was extracted from furnace 2 in the burial area (Fig. 3).

Fig. 3: Sample diagram of iron artifacts excavated from the burial area of Luojiaba site.

Iron pot LJBF11 (M209:11) 2. Iron pot LJBF12 (M228:5) 3. Iron pot LJBF13 (M239:4) 4. Iron cauldron LJBF14 (M250:8) 5. Iron cauldron LJBF15 (M251:1) 6. Scrap iron sword LJBF16 (M243:4) 7. Iron plate LJBF17 (L2:1).

Metallographic and SEM - EDS analysis

To reveal the material composition and manufacturing techniques of these iron artifacts, metallographic analysis was conducted on the collected samples and measured the composition of the inclusions using a scanning electron microscope and an accompanying energy disperse spectroscopy. Metallographic structure refers to the internal crystal structure of metals, which carries important information, such as alloy composition, forming method, and subsequent heat treatment. Inclusions refer to various forms of slag produced during the smelting and reprocessing of metals. By obtaining information on the composition of inclusions, it is possible to determine the raw materials and the processing techniques used.

To perform the above analysis, the samples were first embedded in resin, then polished and ground to obtain a smooth cross-section of 0.5 microns. The samples were then etched with a 3% nitric acid alcohol solution to observe the metallographic structure. After metallographic analysis, samples were polished again to remove the etching layer and undergo spraying carbon, then analyzed for inclusion composition using a scanning electron microscope and accompanying energy disperse spectroscopy.

The instrument used for metallographic analysis was a Leica DM4000 metallographic microscope manufactured in Germany.

Scanning electron microscope analysis was used to study the composition of slag inclusions by a Tescan VEGA3 XMU scanning electron microscope equipped with a Bruker XFlash 610 M energy disperse spectroscopy. After metallographic observation, samples were polished, carbon-sprayed, and analyzed using SEM-EDS to obtain the overall elemental composition of the inclusions. The manual selection analysis function was used to select the entire inclusion as much as possible to avoid the impact of uneven composition on data reliability. During analysis, the acceleration voltage was set to 20 kV, the working distance to 15 mm, and the signal acquisition time to 60 s. Meanwhile, an SEM photo with a scale bar was taken in this field of view, and the area of the inclusion was obtained using the measurement function of ImageJ software for easy use of the weighted area (Oxide*) of the inclusion in data analysis¹⁴.

$$\% {{Oxide}}^{* }={\sum }_{i=1}^{n}\left( \% {E}_{i}\times \frac{{S}_{i}}{{S}_{T}}\right)$$

(1)

The multivariate statistical approach proposed by Disser et al. ¹⁵, was employed to determine the probability that specific inclusion components originate from different production systems. The analytical procedure consisted of the following steps. First, the purity of these inclusions was tested and preliminarily screened using the non-reduced oxide (NRC) ratio analysis method to determine whether the samples were composed of different types of iron and to remove data with significant component deviations that were not representative of the original metallurgical slag. At this stage, we should select at least 20 independent inclusions from each sample for analysis in principle. However, due to severe corrosion in some specimens, the effective analytical area—where inclusions can be clearly distinguished from corrosion products—was considerably limited, resulting in an insufficient number of analyzable inclusions. For such samples, we believe that although very few inclusions were analyzed, the consistency shown by the morphology of the inclusions and their compositional data still allows us to use this (limited) data for discussion and to make a judgment about the metallurgical process. The second step was to appropriately convert the (weighted average) content value of specific oxides in the inclusions (Eq. 2) and then normalized it again before substituting it into Eq. 3 to calculate the p-value: when p < 0.3, the sample can be judged as bloomery iron products; when p > 0.7, the sample can be judged as ‘CHAO’-fining steel products¹⁵.

$${ \% {Oxide}}^{* * }=\left({ \% {Oxide}}^{* }\times 100\right)\div\left(100-{{FeO}}^{* }\right)$$

(2)

$${Logit}\left(p\right)=\mathrm{ln}\frac{p}{1-p}={\beta }^{0}+{\beta }^{{Mg}}[{ \% {MgO}}^{* * }]+{\beta }^{{Al}}[{ \% {{Al}}_{2}{O}_{3}}^{* * }]+{\beta }^{{Si}}[{ \% {{SiO}}_{2}}^{* * }]+{\beta }^{P}[{ \% {P}_{2}{O}_{5}}^{* * }]+{\beta }^{K}[{ \% {K}_{2}O}^{* * }]+{\beta }^{{Ca}}[{ \% {CaO}}^{* * }]+{\beta }^{{Mn}}[{ \% {MnO}}^{* * }]$$

(3)

Results

Based on metallographic analysis, the samples examined in this study can be preliminarily classified into three categories: (1) cast iron, (2) annealed cast-iron products, and (3) wrought iron or steel artifacts whose manufacturing techniques remain undetermined. For the third category, scanning electron microscopy coupled with energy-dispersive spectroscopy (SEM–EDS) was employed to analyze the composition of inclusions within the metal matrix. The resulting data were processed and compared with existing reference datasets. The following section presents a detailed discussion of these analytical results.

Cast Iron

Metallographic observations indicate that the cast-iron artifacts can be further divided into two subtypes: unaltered cast iron and decarburized (annealed) cast iron. The unaltered cast-iron products comprise both white and gray iron, with a total of five specimens. Among these, LJBF15 is white iron, while LJBF11, LJBF12, LJBF13, and LJBF14 are gray iron. One annealed pig iron product has been identified—specimen LJBF17 is malleable cast iron. The metallographic characteristics and compositional data are summarized in Table 2.

Table 2 Sample Information for Mottled Iron/White Iron Materials

The metallographic structure of white iron is characterized primarily by ledeburite. Depending on carbon content, white iron can be subdivided into hypereutectic and hypoeutectic types. Hypereutectic white iron displays abundant strip-like primary cementite within a ledeburitic matrix (Fig. 4a, b), whereas hypoeutectic white iron features pearlite—transformed from primary austenite—dispersed within ledeburite. The examined white iron samples were mainly used for domestic utensils and iron plates.

Fig. 4: Micrographs showing the microstructure of white iron and malleable cast iron.

a, b Micrographs of the microstructure of LJBF15 iron cauldron, both magnified 200 times, ledeburite structure, white iron; c, d metallographic structure photos of the LJBF17 iron plate sample, all magnified 200 times, show flocculent graphite distributed on a ledeburite matrix. Rusting has occurred near the flocculent graphite areas. The material is malleable cast iron.

Ancient metalworkers learned from experience that heating brittle white cast-iron objects could reduce their brittleness and improve usability. Research shows that when the annealing temperature reaches about 900 °C or slightly higher, and the cast iron is held at that temperature for a sufficiently long duration, the cementite in white cast iron decomposes completely and graphite forms. The graphite aggregates into flocculent clusters, which enhances the mechanical properties of the cast iron and produces malleable cast iron. Depending on the annealing temperature, atmosphere, and duration, malleable cast iron with different matrix structures can be obtained. In sample LJBF17, the iron plate did not undergo complete softening; flocculent graphite is distributed on a ledeburitic matrix (Fig.4c, d).

Mottled iron represents an intermediate type between white and gray iron. Its matrix consists of a mixture of ledeburite, flake graphite, and cementite. During the casting process, variations in the cooling rate of molten iron within the mold lead to distinct microstructures. Near the mold walls, heat dissipates rapidly, producing a higher cooling rate and favoring ledeburite crystallization. In contrast, the slower cooling at the core allows the formation of flake graphite (Fig. 5a–d). Similar microstructural differences can occur within different sections of the same artifact. These observations suggest that gray and white iron are fundamentally similar in composition, with their structural differences primarily resulting from variations in cooling rate during solidification.

Fig. 5: Metallographic Structure of Mottled Iron.

a LJBF11 iron pot metallographic microstructure photo, magnified 200 times, mottled iron. b LJBF12 iron pot metallographic microstructure photo, magnified 200 times, mottled iron. c LJBF13 iron pot metallographic microstructure photo, magnified 200 times, mottled iron. d LJBF14 iron cauldron metallographic microstructure photo, magnified 200 times, mottled iron.

Wrought Iron and Steel Products Containing Inclusions

Excluding the corroded samples and those previously discussed, the remaining specimens analyzed in this study consist of wrought iron or steel as the base material. Their metallic matrices contain numerous non-metallic inclusions (Figs. 6, 7). These samples may represent either cast-iron or decarburized cast-iron products; therefore, compositional analysis of the inclusions is essential to determine their technological origin. Detailed information on these specimens is provided in Table 3.

Fig. 6: Metallographic Structure Photos.

a LJBF01 scrap iron metallographic structure photo, magnified 100 times, ferrite structure, inclusions deformed and elongated along the processing direction. b LJBF03 iron nail metallographic structure photo, magnified 100 times, ferrite, pearlite, carbide structure in some areas, single-phase inclusions, folded forging, ferrite grains deformed and elongated along the processing direction. c LJBF04 iron awl metallographic structure photo, magnified 200 times, ferrite structure, inclusions deformed along the processing direction. d LJBF08 iron bar metallographic structure photo, magnified 100 times, ferrite structure, elongated inclusions along the processing direction.

Fig. 7: Metallographic and SEM Photos of LJBF10 and LJBF16.

a LJBF10 iron knife metallographic structure photo, magnified 100 times, with higher carbon content in the upper part, network-like cementite, ferrite structure in the lower part, and single-phase and sub-complex inclusions in the middle, with the inclusions deformed along the processing direction. b SEM photo of LJBF10 iron knife inclusions, complex-phase inclusions; c Metallographic microstructure photo of LJBF16 iron sword, magnified 200 times, ferritic Widmannstätten structure, inclusions deformed and elongated along the processing direction, hypoeutectoid steel; d SEM photo of LJBF16 iron sword inclusions, single-phase inclusions.

Table 3 Basic Information on Wrought Iron/Steel Products Containing Inclusions in Metal Matrix in This Study

Inclusion Analysis and Data Processing

Metallurgical archeologists have conducted extensive studies on iron production through the analysis of non-metallic inclusions. Dillmann et al. established correlations between inclusion compositions and smelting processes by examining the proportional relationships among oxides, such as Al₂O₃, MgO, K₂O, CaO, and SiO₂, and subsequently proposed a data-processing method for their interpretation. They also developed a model to differentiate between bloomery iron and ‘CHAO’-fining steel products, identifying the relative concentrations of P₂O₅ and CaO as key diagnostic indicators.

Building on this framework, Dillmann and L’Héritier introduced a method based on the ratio of non-reduced compounds (NRCs) within inclusions¹⁴. This approach assumes that certain oxides—such as MgO, Al₂O₃, SiO₂, K₂O, and CaO—exhibit chemically stable behavior during smelting and refining. Because these oxides are either difficult to reduce or readily reoxidized, their relative proportions remain largely constant throughout processing. If new inclusion-forming processes occur during later production stages, however, the NRC ratios will deviate from those of the original inclusions. Accordingly, ratios, such as Al₂O₃/SiO₂, K₂O/CaO, and MgO/Al₂O₃ can be used to assess whether the inclusions within a given sample share a uniform origin (Fig. 8). In Dillmann and L’Héritier’s original protocol, at least 40 inclusion-composition data points were recommended per sample to ensure statistical reliability. Due to the preservation state and limited availability of material, the present study adopts a minimum threshold of 20 data points per sample (Inclusion data are shown in Supplementary Table 1).

Fig. 8

K₂O/CaO, Al₂O₃/SiO₂, and MgO/Al₂O₃ Ratio Diagram of Inclusions in Wrought Iron/Steel Samples from the Luojiaba Site.

As mentioned earlier, based on the findings of Disser et al., the composition characteristics of inclusions in smelted iron were used to determine whether the iron artifacts were bloomery iron. The results for each sample, including the sub-component ratio of weighted oxides in inclusions (Oxide**), the number of inclusions, the total area of inclusions, and the p-value, are shown in Table 4.

Table 4 Sub-component Ratios of Weighted Oxides in Inclusions (Oxide**) in the Nine Samples of This Study Number of Analyzed Inclusions, Total Area of Inclusions, and p-values

Figure 9 shows that sample LJB08 falls within the ‘CHAO’-fining steel range, while LJB02, LJBF03, LJBF04, LJBF09, LJBF10, and LJB16 plot within the bloomery-iron range. Samples LJB01 and LJB07 lie between these two fields, rendering their classification uncertain. To verify these results, a logistic regression method proposed by Disser and Dillmann¹⁵ was applied. In this model, values of p < 0.3 indicate bloomery iron, whereas p > 0.7 denote fining steel. The p-value of LJBF08 exceeds 0.7, confirming it as ‘CHAO’-fining steel; the p-values of LJB01, 02, 03, 04, 07, 09, and 16 are below 0.01, and that of LJBF10 is below 0.14 (Table 4). Accordingly, LJBF08 can be preliminarily identified as a ‘CHAO’-fining steel product, while LJBF01, 02, 03, 04, 07, 09, 10, and 16 represent bloomery-iron products.

Fig. 9

Chart Distinguishing Direct and Indirect Processes by Considering the Weighted Content of Several Compounds¹⁴ (Supplementary Table 2).

Comparison of H235 Slag Composition Data and Iron Inclusion Data

The slag-composition data for H235 were published previously (Supplementary Table 3)¹¹. In this study, those data were compared with the inclusion compositions of iron artifacts excavated from H235 and the associated burial area. Prior to comparison, the datasets were divided into five groups:

• (1) Type A slag from H235;

• (2) Type B slag from H235;

• (3) inclusions from samples LJBF01, 02, 03, 04, 07, 09, and 10 (bloomery iron) from H235;

• (4) inclusions from LJBF08 (fining steel) from H235; and

• (5) inclusions from the iron sword LJBF16 from the burial area, preliminarily identified as bloomery iron.

Because FeO values in inclusions may derive from the surrounding metal matrix, FeO was excluded from subsequent data processing (Supplementary Table 3). Principal component analysis (PCA) and oxide-ratio plots (Al₂O₃/SiO₂, K₂O/CaO, and MgO/Al₂O₃) reveal that the Type A and Type B slag from H235 align closely with the bloomery-iron inclusion data (Fig. 10a–d). This correspondence suggests that the slag recovered from H235 represents forging and bloomery slag^11,12. During smithing, slag particles could have become embedded in the metal, forming inclusions within the finished tools. These results indicate that the seven bloomery-iron artifacts from H235 were likely produced locally and include small tools and utilitarian ironware. In contrast, the inclusions in LJBF08 (‘CHAO’-fining steel bar) are more compositionally homogeneous and distinct from the bloomery-iron data, while those in LJBF16 (iron sword from the burial area) show only partial overlap. This pattern suggests that the iron artifacts from the Luojiaba Site were not produced from a single source of raw material but rather originated from multiple metallurgical sources.

Fig. 10: Comparison of slag composition from pit H235 at Luojiaba with the composition of inclusions in iron artefacts recovered from the graves at Luojiaba and in iron artefacts from pit H235.

a PCA diagram of the composition of furnace slag excavated from H235 at the Luojiaba Site and the inclusions in iron artifacts excavated from H235 and the burial area; b Diagram of the Al₂O₃/SiO₂ ratio of the composition of furnace slag excavated from H235 at the Luojiaba Site and the inclusions in iron artifacts excavated from H235 and the burial area; c Diagram of the K₂O/CaO ratio of slag components excavated from H235 at the Luojiaba Site and iron artifacts excavated from H235 and the burial area; d Diagram of the MgO/Al₂O₃ ratio of slag components excavated from H235 at the Luojiaba Site and iron artifacts excavated from H235 and the burial area.

Discussion

Metallographic and inclusion analyses demonstrate that both cast iron casting and bloomery iron forging technologies were employed concurrently at the Luojiaba site during the Han Dynasty. Cast iron artifacts—comprising primarily white iron and mottled iron—were mainly used for containers and domestic utensils. In contrast, wrought iron and steel implements, including knives, awls, and other tools, were produced through forging, carburization, and possibly ‘CHAO’-fining processes.

Inclusion chemistry further supports the coexistence of these technological systems. Most wrought iron and steel samples (e.g., LJBF01, 02, 03, 04, 07, 09, 10, 16) exhibit compositional signatures characteristic of bloomery iron, notably low P₂O₅ and elevated SiO₂, Al₂O₃, and MnO contents. In contrast, LJBF08 displays high P₂O₅ and CaO (p > 0.7), indicating a fining steel origin. These findings collectively suggest that both technological traditions were practiced locally, reflecting a complex and versatile metallurgical system.

Early iron artifacts in China can be divided into two major systems: the northwest system, mainly characterized by bloomery iron metallurgy, and the Central Plains system, characterized by the coexistence of bloomery iron metallurgy and liquid cast iron metallurgy, with cast iron metallurgy predominating¹. This indicates that during the origin and development of iron metallurgy techniques, two metallurgical systems coexisted in China: cast iron metallurgy and bloomery iron metallurgy. These two technologies are related yet distinct. One of the main differences is the smelting temperature: cast iron metallurgy requires approximately 1300 °C, while bloomery iron metallurgy operates at 1100–1200 °C. As a result, cast iron produced via cast iron metallurgy is liquid iron, whereas iron obtained through bloomery technology is solid bloomery iron. Additionally, cast iron metallurgy was generally conducted on a larger scale with higher production efficiency but required greater inputs of raw materials, such as ore and fuel. In contrast, bloomery ironmaking was smaller in scale and less efficient, yet required considerably less resource investment.

The primary materials used in cast iron artifacts are white iron, gray iron, and mottled iron, with material type largely determined by the solidification rate and silicon content. Gray iron is the most widely used type of cast iron in modern industry, valued for its low cost, excellent casting performance, and superior mechanical properties¹⁶. In modern production, silicon content typically ranges from 1.5–3%, improving fluidity and facilitating solidification^16,17. By contrast, cast iron in ancient China contained very low silicon, making precise control of cooling rate during solidification essential to produce gray iron. In ancient times, craftsmen typically preheated molds and placed them in furnaces for heat preservation after pouring to reduce the cooling rate¹⁸. Similar techniques persisted in the 19th century in Huizhou, Guangdong, for producing low-silicon gray iron¹⁹.

Gray iron has lower hardness than white iron but is less brittle, with better wear resistance and lubricity, as well as good fluidity, thermal stability, and low shrinkage. These properties made gray iron suitable for casting molds and other functional objects during the Qin and Han Dynasties. The solidification process of gray iron requires precise operation; insufficient solidification time or excessively rapid cooling leads to mottled iron, which exhibits microstructural features and mechanical properties intermediate between white iron and gray iron. By the mid-Western Han Dynasty, China had already produced cast iron artifacts with rough surfaces, such as the iron plowshare excavated from the Han tomb in Mancheng (满城) (M2:01)²⁰, the iron block from Tieshengou (铁生沟) in Gong County (T5:42)²¹, the iron adze from Mianchi (渑池) Site (MJC:458), and the hexagonal crucible (MJC:32)²².

At the Luojiaba Site, no gray iron castings dating to earlier periods have been found, but castings made of mottled iron and white iron were already in use during the Han Dynasty. The samples analyzed in this study—LJBF11, LJBF12, LJBF13 (iron pots), and LJBF14 (iron cauldron)—are made of mottled iron (Fig. 5), whereas LJBF15 (iron cauldron) is made of white iron (Fig. 4a, b). These objects are all identified as vessels.

Metallographic observations reveal that cast iron and solid decarburized cast iron artifacts undergo a liquid smelting process, resulting in a relatively pure matrix with few inclusions. They typically feature a ledeburite structure and various forms of graphite, which are completely different from bloomery iron products. However, for some artifacts with steel or wrought iron matrices containing abundant non-metallic inclusions, it is necessary to analyze the composition of the inclusions to determine their smelting technology. Such artifacts may originate from bloomery iron production or from fining of cast iron, in which liquid-state cast iron is decarburized.

Since the steel ‘CHAO’-fining process is mainly an oxidation reaction, high P₂O₅ content in inclusions is generally considered a key characteristic of ‘CHAO’-fining steel products²³. In bloomery smelting, the weak reducing atmosphere allows only partial reduction of phosphorus into the metallic iron. In contrast, in cast iron smelting, the stronger reducing atmosphere and higher temperature allow phosphorus to reduce and enter the cast iron, which is then re-oxidized during ‘CHAO’-fining and incorporated into inclusions.

Steel could also be produced through carburization of bloomery iron. Wrought iron or low-carbon steel produced via bloomery iron technology was heated in hot charcoal at 900 °C–950 °C for carburization, sometimes accompanied by forging. Carburization may have initially been performed in a forge and later in a carburizing box²⁴. Carburized steel exhibits significantly improved hardness. Samples LJBF03 (iron awls) (Fig. 6b) and LJBF10 (iron knives) (Fig. 7a) from Luojiaba show microstructures consistent with carburized steel.

Alternatively, steel could be produced via ‘CHAO’-fining of cast iron, involving heating cast iron to a liquid or semi-liquid state and decarburizing it in a strongly oxidizing atmosphere to produce steel or wrought iron. ‘CHAO’-fining furnaces have been discovered at sites, such as Tieshengou (Gong County) and Wafangzhuang (Nanyang). Compared with carburization of bloomery iron, ‘CHAO’-fining represented a more efficient production model, satisfying the growing social demand for workable iron. After the invention of ‘CHAO’-fining steel technology, bloomery carburized steel became rare. However, ‘CHAO’-fining steel technology did not dominate before the Northern and Southern Dynasties, during which cast iron annealing was dominant; it became widely adopted only afterward²⁵.

From a regional perspective, the Luojiaba assemblage aligns closely with the metallurgical traditions of the Central Plains, where cast iron metallurgy predominated, although bloomery iron continued to be widely utilized. Comparable coexistence of casting and forging technologies has been identified at other sites, such as Qiaogoutou^7,8 and Kele⁵, where inclusion analyses indicate shared technological ancestry. At Luojiaba, the predominance of bloomery iron among smaller tools may reflect local adaptation to resource availability or specific functional requirements. The absence of gray iron and the prevalence of mottled and white iron suggest that casting control, particularly regulation of cooling, was still in developmental stages. This observation may indicate a technological transmission lag or regional experimentation in Eastern Sichuan during the Han Dynasty.

Comparative compositional analyses of slag from H235 and inclusions within iron artifacts reveal strong correlations, particularly in the ratios of Al₂O₃/SiO₂, K₂O/CaO, and MgO/Al₂O₃. These correspondences suggest that most bloomery iron artifacts were likely produced locally near H235, using similar ore sources and smelting parameters. In contrast, samples LJBF08 and the iron sword LJBF16 deviate chemically from the main cluster, implying alternative raw material sources or distinct production locales. This divergence supports the notion of a non-centralized production and supply system, in which iron artifacts circulated through interregional trade networks, population movements, or military provisioning, rather than being confined to a single manufacturing center.

The application of inclusion analysis to distinguish bloomery iron from ‘CHAO’-fining steel has proven effective, yet methodological limitations remain. Logistic regression models (e.g., Disser et al.) provide a robust quantitative framework, but overlaps in inclusion chemistry—especially in corroded or altered samples—can obscure precise classification. Additionally, post-depositional corrosion may modify or destroy original glassy phases, potentially biasing compositional interpretations. And the distinction between ‘CHAO’-fining steel and bloomery carburized steel remains subject to debate. Chen and Han argued that the volume of eutectic slag in bloomery carburized steel is relatively large and minimally deformed, with a silicate matrix containing spherical iron oxide particles rich in iron but low in silicon. Elemental distribution is inhomogeneous, showing significant fluctuations in phosphorus, calcium, and manganese. These inclusions contain little to no potassium, aluminum, or magnesium, though trace amounts of copper and sulfur may occur²⁶.

By contrast, slag inclusions in ‘CHAO’-fining steel are typically deformed and elongated along the processing direction, featuring single-phase silicate inclusions as well as sub-complex ferrous oxide–silicate phases. The matrix is generally a single-phase slag inclusion containing a secondary structure, with no clear boundary between matrix and second phase, and elemental composition varies considerably. Other studies provide complementary perspectives. South Korean scholars Jang-Sik Park and Li have noted that evidence is currently insufficient to conclusively distinguish bloomery iron from ‘CHAO’-fining steel based solely on slag morphology and chemistry^7,27. Yang et al. proposed that phosphorus content in non-metallic inclusions can serve as an indicator of fining steel production²⁸. Guo et al. identified bloomery iron based on the presence of large ferrous oxide particles within a silicate eutectic inclusion²⁹. Li and Huang considered single-phase silicate inclusions in steel products as calcium-silicon slag residues, whereas bloomery iron inclusions corresponded to iron olivine–iron oxide eutectic³⁰. Liu et al. suggested that inclusions in bloomery iron are primarily composed of iron olivine (Fe₂SiO₄), with high FeO content and minor MgO, Al₂O₃, and CaO³¹.

High phosphorus oxide content in the inclusions of ‘CHAO’-fining steel supports using CaO and P₂O₅ content in slag as indicators for distinguishing between the two technologies, though this method is not universally applicable. Yang further proposed that steel inclusions are mainly iron-rich, low-silicon sub-complex inclusions consisting of ferrous oxide and a glass-phase matrix, with minor complex inclusions containing ferrous oxide crystals (iron olivine + wüstite + glass-phase matrix), small amounts of aluminum, magnesium, and potassium, and a highly fluctuating phosphorus-to-calcium ratio. In contrast, bloomery iron exhibits high Si, Mn, Al, and Ca content, virtually no phosphorus, and primarily consists of metal oxides, such as iron olivine, manganese oxide, aluminum oxide, and calcium oxide³². Chen and Zhang highlighted remaining uncertainties in identifying slag in steel and proposed calcium phosphate as a potential marker to differentiate ‘CHAO’-fining steel from bloomery carburized steel. Zhang additionally observed that cast iron products differ from ‘CHAO’-fining steel by their Fe-Mn-Si-P-O inclusions: exogenous inclusions are small (<10 μm) and primarily calcium-silicate, while endogenous inclusions typically include MnS, MnS-FeS, SiO₂, and Fe-Mn-Si-P-O, with type and proportion correlated to the carbon content of the iron matrix³³. There remains considerable work to fully define criteria for distinguishing bloomery iron from ‘CHAO’-fining steel based on morphology and chemical composition of inclusions. Future studies could focus on (1) developing China-specific inclusion composition criteria and (2) examining the glass phase in inclusions to mitigate the effects of corrosion. Using methods proposed by Dillmann and Disser, preliminary judgments on wrought iron/steel artifacts from Luojiaba indicate the coexistence of bloomery iron and ‘CHAO’-fining steel technologies, with bloomery iron inclusions exhibiting variable characteristics, reflecting diverse ore sources. These results suggest that Luojiaba iron artifacts originate from two distinct technological systems, highlighting a complex network of production, circulation, and supply.

This study provides the first definitive scientific evidence for the coexistence of bloomery iron and ‘CHAO’-fining steel at Luojiaba, revealing a sophisticated and diversified metallurgical system during the Han Dynasty. The findings emphasize technological complexity and regional variability in Eastern Sichuan, where casting and forging traditions coexisted and evolved in response to local resources and functional requirements.

Future research should expand the analytical framework by incorporating larger, geographically diverse datasets to better capture regional variability. Integrating trace element and lead isotope analyses with metallographic and inclusion studies will enhance provenance determination. Equally important is correlating laboratory data with archeological field investigations, including furnaces, slag heaps, and production remains, to reconstruct the organization of ironworking activities. Establishing China-specific inclusion chemistry reference databases will further improve diagnostic precision and facilitate inter-site comparisons. Overall, results from Luojiaba contribute to a deeper understanding of iron technology diffusion, adaptation, and innovation in ancient Southwest China, highlighting the dynamic interplay between technological transmission and local experimentation during the Han period.