Chemical Composition, Crystal Structure, and Microstructure of Slags on the Korean Peninsula from the First Copper Production Remains of the 9th Century

Abstract

The scarcity of excavated early-stage smelting sites related to copper production presents significant challenges in gaining a comprehensive understanding of the copper production process. However, the archaeological site discovered in 2018 in Daeryang-ri, Jinan-gun, Jeollabuk-do, boasts a substantial number of copper smelting remains and related slags, marking it as the first copper manufacturing production site identified on the Korean Peninsula. Consequently, this study selected 10 slag samples, chosen based on surface color and characteristics indicative of a connection to copper smelting, for scientific analysis to accurately ascertain the site’s nature. The primary component analysis of the slags indicated that CuO content ranged from 0.30 to 3.29 wt%, which, although not high, reveals significant quantities of FeO and SiO₂. X-ray diffraction analysis revealed the presence of minerals such as cristobalite, along with fayalite and wüstite, commonly found in slags, varying by sample. Furthermore, microstructural observation revealed circular copper particles containing sulfur and iron, indicating the presence of copper particles in a matte state that have not been refined. This analysis suggests that the slags recovered from Jinan Daeryang-ri bear evidence of iron smelting at the site, with the slag being produced as an intermediate by-product during copper production.

1. Introduction

The development of bronze casting culture on the Korean Peninsula spans from 300 B.C. during the Early Iron Age through the Proto-Three Kingdoms and Three Kingdoms periods. Notably, Chungcheongnam-do and Jeolla-do were the epicenters of the slender bronze dagger culture, where a large volume of copper tools was produced between the 4th and 2nd centuries B.C. [1].

Historically, copper has been used extensively in alloys consisting of copper, tin, and lead, that is, bronze. The properties of bronze differ based on the proportions of these constituent elements. Additionally, even with identical compositions, the microstructure can vary due to factors such as cooling rate, heat treatment, and machining [2]. Since metals are prone to corrosion, reconstructing the historical development of metal culture from excavated metal artifacts poses challenges. However, insights can be gained through the analysis of slags produced during smelting and artifact fabrication [3]. Generally, slags are composed of oxides such as SiO₂ and Al₂O₃, and their composition varies depending on the purpose and methods of smelting.

The primary composition of slags is governed by various factors, including raw ores, smelting furnaces, additives, etc., among which raw ores exhibit the strongest influence [4,5]. Thus, the characteristics of slags and additives in the smelting process can be determined. Furthermore, during the cooling of slags from the melting state, various microcrystals form, which aid in defining the nature of the smelting process.

Understanding the metallurgical properties of slag at various stages of copper production offers valuable insights into the production techniques of the era. Such information can aid in identifying the source ore, copper refining conditions, and specific steps involved in the production process. Hence, it is critical to take a closer look at the copper smelting process and identify the distinguishing features of the slag at each stage [1].

While many copper-related artifacts have been excavated on the Korean Peninsula, sites specifically related to the smelting and production of copper ore are relatively scarce. Notable exceptions include the Gyeongju Hwangnam-dong, Dongcheon-dong, Noseo-dong, Buyeo Gwanbuk-ri, Ssangbuk-ri, and Iksan Wanggung-ri sites (Figure 1). These sites primarily feature evidence of copper and alloy refining, making it challenging to ascertain the origins of the copper and to definitively deduce the primary smelting processes for extracting copper from ore. Furthermore, due to the lack of direct copper smelting ruins, it is somewhat difficult to understand the step-by-step process of mining, dressing, and smelting on the Korean Peninsula in detail.

Nevertheless, evidence of copper smelting using native ore was discovered in 2015 at Daeryang-ri, Jeollabuk-do Jinan-gun, by the Kunsan National University Museum. This evidence, corroborated by an academic surface survey, suggests the presence of copper production sites. A detailed excavation carried out in 2018 identified two copper production furnaces, a building site, and a large waste disposal area, marking the first substantial archaeological findings in Korea to shed light on early copper production processes.

The objective of this study is to infer the copper production process through a metallurgical analysis of various slags excavated from the Jeollabuk-do Jinan-gun Donghyang-myeon Daeryang-ri site [6]. This site is the earliest known copper production location on the Korean Peninsula, and this study aims to elucidate its nature scientifically.

The Jinan Daeryang-ri copper production site (35°50′00.6″ N, 127°34′16.4″ E) has historical significance as a special administrative district named Donghyangso. This is referenced in ancient texts such as Sinjeung Dongguk Yeoji Seungram and Yeojidoseo. The current name, Donghyang-myeon, is derived from the historical existence of Donghyangso in this area [6].

The Donghyang Mine, located at the foot of Mount Munpilbong (elevation 598.4 m), had operated until the 1980s for copper mining but was closed down due to the vanishing of economic profits. However, the presence of the Donghyang Mine suggests that copper ores obtained from the surrounding area were likely used as the basis for copper production in this area during that time [6].

Research into copper smelting in Korea was initiated in 1994, following the discovery of a crucible and copper slag at the Hwangnam-dong 376 site in Gyeongju. These findings enabled researchers to speculate about the smelting technologies of the period, and subsequent research on copper production has continued. However, most identified sites have primarily revealed refining furnaces (melting furnaces) used for improving metal quality and workshops for the fabrication of finished products. Consequently, details about the system of copper production, particularly in relation to mining and primary smelting, remain elusive [7,8].

The Jinan Daeryang-ri cooper production site was investigated in 2015 and excavated from 2018 to 2019, revealing copper production remains such as smelting furnaces and waste disposal areas (Figure 2). In the western area, characterized by high and flat terrain, auxiliary facilities related to copper production, such as first and second copper furnaces, are present, indicating a working space. In the eastern area with lower terrain, there is a large waste area, suggesting a disposal space. In other words, it was inferred that the copper production site was designed by considering the terrain to arrange auxiliary facilities and set up workspaces for efficient production [6].

From the Jinan Daeryang-ri copper smelting remnants, furnace walls, earthenware, outflow slag, white porcelain plates, and tiles were excavated. Among them, the earthenware and tiles recovered from the disposal site could be used to estimate the operational period of the site. Six large jars with wavy patterns on their necks were excavated. Such earthenware is known to have appeared on the Korean peninsula in the late-8th century and was used from the mid-9th century to the early Goryeo Dynasty [6].

Furthermore, the five charcoal samples, collected from the bottom of the second smelting furnace and from each layer of the disposal site, were radiocarbon-dated from the late-7th century to the late-10th century. These findings suggested that the site was in operation around the middle of the 9th century [6].

The Korean Peninsula has a small number of sites where ancient copper smelting remains or facilities have been discovered, making it difficult to establish the copper production system in this region clearly. However, the Jinan Daeryang-ri site demonstrates that copper production from the Three Kingdoms period to the early Joseon Dynasty was largely conducted under state control and reveals the existence of Donghyangso, a special administrative zone established for copper production [6].

2. Samples for Analysis and Analytical Methods

2.1. Samples for Analysis

The 10 slags chosen for this study are by-products of copper smelting, retrieved from both the no. 2 copper furnace (smelting furnace, Figure 3) and the associated waste site.

Slags showing the characteristics of green corrosion products or smooth surfaces with flowing traces were selected as target samples after consultation with an archaeologist. These samples are itemized in

Table 1. List of samples excavated from Jinan Daeryang-ri copper production ruins.

Sample No.	Find Spot
JA-1	Smelting furnace	Discharge port	Copper slag
JA-2	Slag disposal site	2nd floor
JA-3		5th floor
JA-4		6th floor
JA-5		9th floor
JA-6		13th floor
JA-7		15th floor
JA-8		Sedimentary deposit
JA-9		-	Slag
JA-10	Estimated to be a sedimentary deposit		Copper product

and Figure 4. Notable features include various pores as well as green and/or reddish–brown corrosion products. Specifically, sample JA-8 presents a fully vitrified state, akin to obsidian.

2.2. Analytical Methods

Parts of the selected slags were collected using a rubber mallet and then analyzed. After removing debris (such as soil) from the surface, the samples were immersed in 99.5% ethanol, cleaned with an ultrasonic cleaner, and dried. The dry samples were then ground (<25 μm) using a mortar. Prior to grinding each sample, the contaminated mortar was cleaned using powdered seasand (850–1400 μm) and wiped with ethanol (99.5%) before use.

The chemical compositions of samples were analyzed using wavelength dispersive X-ray fluorescence spectroscopy (WD-XRF; S8 Tiger, Bruker, Germany). Following sample pretreatment, the specimens were placed in liquid cups for semi-quantitative evaluation. X-ray diffraction (XRD; D8 ADVANCE with DAVINCI, Bruker, Germany) was performed to identify crystalline phases. The analytical parameters included a 2θ range of 3°–70°, a scan speed of 0.5 s/step, a step size of 0.02°, a current of 40 mA, and a voltage of 40 kV. Copper served as the target material for this analysis.

To assess the microstructure of the slag—including cross-sectional properties, pore distribution, and particle size—each sample was embedded in epoxy resin and polished sequentially, ranging from 100 to 4000 mesh sizes. Final polishing was executed using an abrasive (DP-Spray 3 and 1 µm, Struers, Denmark) until the samples were free of scratches. Subsequently, the microstructure was examined under a metallographic microscope (DM 2500M, Leica, Germany).

A scanning electron microscope (SEM; MIRA3, Tescan, Czech Republic) was used for high-magnification observation and detailed chemical compositional analyses of the samples. The elemental compositions were further evaluated using an energy-dispersive spectrometer (EDS; QUANTAX200, Bruker, Germany). EDS mapping techniques were also applied to identify the distribution of sulfur within the copper particles. The analyzed samples were coated with platinum to increase their conductivity.

In addition, Raman micro-spectroscopy was used to accurately identify the microstructures (Raman Micro-Spectroscope, LabRAM Soleil, Horiba Jobin Yvon, Longjumeau, France). Raman spectra were acquired with a LabRAM Soleil Raman microscope (Horiba Jobin Yvon) using a 532 nm excitation laser. Only the platinum coating used in the SEM-EDS analysis was removed, and no additional pretreatment was performed for the Raman micro-spectroscopy analysis. Further, the Raman micro-spectroscopy results were compared with the data from the RRUFF™ Project, without calibration.

3. Results

3.1. XRF Analysis Results

Table 2. Results of WD-XRF analysis of copper slags excavated from the Jinan Daeryang-ri copper production ruins.

Sample	Chemical Composition (wt%)
	FeO	SiO2	CaO	Al2O3	CuO	K2O	MgO	P2O5	MnO	TiO2	ZnO	SrO	ZrO2	SO3	Rb2O	Na2O
JA-1	44.36	29.82	16.10	2.61	0.38	0.33	2.41	0.59	2.45	0.09	0.06	0.01	-	0.79	-	-
JA-2	26.92	44.79	10.62	8.65	2.47	1.74	2.22	0.46	0.57	0.38	0.03	0.05	0.02	0.78	0.01	0.29
JA-3	23.69	47.49	5.28	12.19	0.35	3.41	4.36	0.50	1.28	0.27	0.02	0.02	0.02	0.68	0.02	0.43
JA-4	15.87	61.21	6.20	8.83	0.72	2.29	2.17	0.48	0.50	0.43	0.03	0.01	0.02	0.86	0.01	0.36
JA-5	39.21	38.37	10.19	5.79	0.47	0.88	2.03	0.55	1.55	0.18	0.06	0.01	0.01	0.70	-	-
JA-6	43.34	33.09	13.27	3.01	0.73	0.55	2.85	0.52	2.24	0.07	0.07	0.01	-	0.25	-	-
JA-7	41.35	36.68	10.38	5.62	0.99	0.67	1.93	0.57	1.37	0.19	0.08	0.01	-	0.17	-	-
JA-8	31.64	43.04	9.91	6.78	0.33	1.73	3.64	0.48	1.13	0.25	0.03	0.01	0.01	0.59	0.01	0.41
JA-9	7.08	67.53	3.29	12.41	0.30	4.61	1.05	0.46	0.27	0.36	0.01	0.07	-	0.43	0.02	2.10
JA-10	59.74	20.93	6.23	4.69	3.29	2.16	0.90	0.65	0.26	0.25	0.02	0.02	0.01	0.83	0.01	-

presents the results of the compositional analyses of the 10 selected slags. Predominantly, FeO–SiO₂ showed the highest concentration. Additionally, components such as CaO, Al₂O₃, CuO, K₂O, MgO, P₂O₅, MnO, TiO₂, ZnO, SrO, ZrO₂, SO₃, Rb₂O, and Na₂O were identified. Notably, the copper product labeled JA-10, presumed to be a sedimentary deposit, exhibited a high FeO content (59.71 wt%) and contained the most copper oxides. Iron slags retrieved from ironworking and iron production sites across the Korean Peninsula scarcely contain PbO and ZnO, although these elements can occasionally be detected in iron ore as trace elements (Pb 0.223.5 PPM, Zn 2.6295.0 PPM) [9]. In the slags collected from the Jinan Daeryang-ri site, PbO was not detected, underscoring the need for a more comprehensive XRD analysis and microstructure examination. Furthermore, the overall concentration of CuO was low, while that of FeO and SiO₂ was high. During smelting, copper is primarily extracted from the ore in the form of matte or copper ingot, explaining the limited presence of CuO in the smelting slag [10].

To improve slag fluidity and facilitate component separation during copper production, various slag elements (e.g., SiO₂, Al₂O₃, CaO, and MgO) are incorporated. A higher slag volume indicates more efficient metal separation [9]. Moreover, Al₂O₃ concentrations exceeding 5 wt% tend to elevate the slag’s melting point and viscosity [11]. The SiO₂ content in the recovered copper slag was notably high, ranging from 20.92 to 67.47 wt%. This elevated content could result from the vitrification of clay, which forms the furnace walls, under the high temperatures reached during smelting [12]. Similarly, the CaO content, varying between 3.29 and 16.10 wt%, could either be calcareous material added during the smelting process or originate from the copper ore used as fuel.

3.2. XRD Analysis Results

X-ray diffraction (XRD) analysis was conducted to identify the primary crystalline features in the slag. The results are detailed in

Table 3. XRD analysis results.

Sample No.	Mineral Composition
JA-1	Augite, Mullite, Fayalite, Magnetite
JA-2	Quartz, Augite, Cristobalite
JA-3	Quartz, Cristobalite, Fayalite
JA-4	Quartz, Cristobalite
JA-5	Augite, Quartz, Cristobalite, Fayalite, Magnetite
JA-6	Augite, Fayalite, Magnetite
JA-7	Augite, Quartz, Fayalite, Magnetite
JA-8	Amorphous
JA-9	Quartz, Cristobalite
JA-10	Augite, Malachite, Wüstite, Copper

and Figure 5. For the copper material from the waste site (JA-8), the analysis revealed no peaks, indicating a fully vitrified, amorphous slag. Generally, XRD analysis of copper slag demonstrated the presence of quartz (JCPDS card no. 46-1045) and cristobalite (JCPDS card no. 39-1425). Cristobalite serves as an indicator of high-temperature conditions and forms when quartz recrystallizes under elevated temperatures. The presence of cristobalite suggests that the material was smelted at temperatures exceeding 1200 °C, given that cristobalite forms within the 1200–1400 °C temperature range. Fayalite (Fe₂SiO₄) (JCPDS card no. 9-484 or 11-262) was detected in several samples (JA-1, JA-2, JA-3, JA-5, JA-6, and JA-7). As the only low-melting compound in the FeO–SiO₂ binary system, fayalite forms at temperatures between 1150 and 1200 °C. It is a prevalent mineral in iron-rich slag and contributes to enhanced copper recovery during slag production [1].

Augite ((Ca, Na)(Mg, Fe, Al, Ti)(Si, Al)₂O₆) (JCPDS card no. 24-201) was identified in the majority of the slag samples. This calcium-rich mineral, part of the pyroxene–clinopyroxene series, is characterized by its lower calcium content compared to diopside and hedenbergite but includes elements such as Al, Ti, and Na [13,14,15]. The high CaO content in the detected augite suggests that it acts as one of the gangue components in the slag, potentially serving as a slag-forming agent during smelting. Specifically, JA-10 displayed visible greenish corrosion, which XRD analysis identified as malachite [Cu₂(CO₃)(OH)₂] (JCPDS card no. 10-399), a type of copper oxide. Augite and wüstite (JCPDS card no. 46-1312) were also detected in JA-10, making it the only sample in the analysis to contain copper (JCPDS card no. 4-836).

3.3. Metallographic Analysis Results

The slag, which was collected at the Jinan Daeryang-ri site, was categorized into two types of slag: tap slag with traces of flow on the surface (JA-1, JA-5, and JA-7) and glassy matrix (JA-3, JA-4, JA-6, JA-8, and JA-9) with glittering and smooth surfaces as well as many pores. Additionally, green corrosion products have been observed in some samples (JA-2 and JA-10). The microstructure of these slags was observed with a metallurgical microscope, and the slag having a glassy matrix (JA-4, JA-8, and JA-9) and others having fine dendritic and columnar structures on the glassy matrix were identified as they are smelted at high temperatures. All of these samples are characterized by a wide distribution of copper and quartz particles of various sizes. JA-3 has many small and large pores, and JA-6 has white polygonal particles on the matrix (see Figure 6).

The analysis results of detailed microstructures and components of each microcrystal from JA-1 to JA-5 using SEM-EDS are presented in Figure 7 and

Table 4. EDS results for samples JA-1 through JA-5.

Position		Composition (wt%)
		SO4	CuO	SiO2	CaO	FeO	Al2O3	K2O	MgO	MnO
JA-1	ⓐ	45.91	43.20	-	-	10.88	-	-	-	-
JA-2	ⓐ	50.64	40.11	-	-	9.25	-	-	-	-
	ⓑ	-	-	44.54	14.76	22.56	13.78	3.39	0.97	-
	ⓒ	-	-	97.46	-	2.54	-	-	-	-
JA-3	ⓐ	54.54	2.20	-	-	43.26	-	-	-	-
	ⓑ	6.84	85.22	-	-	7.94	-	-	-	-
	ⓒ	-	-	48.85	5.74	19.51	17.00	3.38	5.52	-
JA-4	ⓐ	53.45	33.26	-	-	13.29	-	-	-	-
	ⓑ	-	-	100.00	-	-	-	-	-	-
	ⓒ	-	-	65.34	3.50	10.81	15.35	2.13	2.87	-
JA-5	ⓐ	55.96	29.09	-	-	14.95	-	-	-	-

. The samples consist of large and small spherical particles and irregular mineral particles evenly distributed on a glassy matrix or a glassy matrix with a high FeO content. When spherical particles of red and white colors were analyzed with a metallographic microscope using EDS (part ⓐ in Figure 7 and Table 4), they were found to be copper particles containing a large amount of sulfur. This seems to be due to the insufficient removal of sulfur (S) during calcination and smelting. Furthermore, varying FeO contents were observed in copper particles, indicating the characteristics of slag generated during the production of matte (Cu–Fe–S). High detection of elements such as S and Fe suggests the use of chalcopyrite in the raw ores to create matte.

SEM-EDS analysis for samples JA-6 through JA-10 is presented in Figure 8 and

Table 5. EDS results for samples JA-6 through JA-1.

Position		Composition (wt%)
		SO4	CuO	SiO2	CaO	FeO	Al2O3	K2O	MgO	MnO
JA-6	ⓐ	39.24	57.56	-	-	3.20	-	-	-	-
	ⓑ	-	-	37.91	16.39	35.14	4.10	-	4.79	1.67
JA-7	ⓐ	51.66	45.14	-	-	3.20	-	-	-	-
JA-8	ⓐ	56.20	17.46	-	-	26.34	-	-	-	-
	ⓑ	-	-	48.53	11.36	22.43	10.01	2.15	5.52	-
JA-9	ⓐ	41.67	26.62	6.95	0.50	22.11	2.14	-	-	-
	ⓑ	-	-	45.40	7.88	31.18	10.54	2.63	2.37	-
JA-10	ⓐ	-	98.29	-	-	1.71	-	-	-	-
	ⓑ	39.66	52.72	-	-	7.62	-	-	-	-
	ⓒ	-	-	34.89	18.29	33.32	8.38	5.12	-	-

. All samples displayed spherical copper particles set against a high-iron-content matrix. A high sulfur content was identified within the copper particles(part ⓐ in Figure 8 and Table 5), with FeO detected as a minor impurity. Specifically, JA-10 displayed textures with elevated sulfur content surrounding high-copper-content particles, which were encased by wüstite.

EDS mapping (Figure 9) revealed concentrations of Cu, S, and Fe on a high-Si matrix. The secondary electron images indicated significant residual sulfur (S) and iron (Fe) within the copper particles.

Figure 10 and

Table 6. EDS analysis results of samples JA-1, JA-3, JA-5, JA-6, JA-7, and JA-10.

Position.		Composition (wt%)
		SO4	CuO	SiO2	CaO	FeO	Al2O3	K2O	MgO	MnO
JA-1	ⓐ	-	-	38.04	29.27	26.77	4.13	-	1.78	-
	ⓑ	-	-	25.69	16.75	49.76	0.98	-	4.10	2.72
JA-3	ⓐ	-	-	47.55	6.93	19.71	18.41	3.17	3.21	1.02
	ⓑ	-	-	36.91	3.05	34.43	6.21	1.59	15.90	1.90
JA-5	ⓐ	-	-	28.90	6.72	56.93	2.55	-	2.57	2.33
	ⓑ	-	-	49.72	13.20	23.99	8.75	0.64	3.69	-
	ⓒ	-	-	39.38	12.37	39.01	5.01	1.02	3.21	-
JA-6	ⓐ	-	-	30.24	14.96	45.93	3.02	0.44	3.08	2.32
JA-7	ⓐ	-	-	-	-	98.05	1.95	-	-	-
	ⓑ	-	-	18.99	5.32	68.66	-	-	4.32	2.71
JA-10	ⓐ	-	-	33.78	15.44	31.34	11.00	8.44	-	-
	ⓑ	-	-	29.83	17.80	40.53	5.36	5.38	1.10	-
	ⓒ	-	-	-	-	100.00	-	-	-	-

present the observed microstructures of areas other than the glassy matrix and copper particles when examined under a metallographic microscope. Fayalite, characterized by a high FeO content, was present within the matrix, while wüstite manifested in polygonal or irregular shapes (Table 6: JA-7-ⓐ, JA-10-ⓒ).

The slags analyzed in this study exhibited sulfur and iron components that were not entirely removed during the matte production process. This suggests that slags generated during matte production in the copper production process, utilizing chalcopyrite ore, can be identified through microstructural analysis.

Furthermore, Al₂O₃ (4.10–17.00 wt%) and MgO (0.97–5.52 wt%) were also detected in the background. Tissues that were challenging to discern under a microscope were identified as bright gray, resin-like tissues in SEM images. Consequently, Raman micro-spectroscopy was employed to accurately determine the structures present within the matrix.

3.4. Raman Micro-Spectroscopy Results

Raman micro-spectroscopy was utilized to definitively identify the micro-crystals observed, based on the results of XRF, XRD, and microscopic analysis of the fine structure. The samples selected for this analysis were JA-1, JA-2, JA-5, and JA-10, representing various crystals within the structure.

Analyses of the copper particles observed through a metallographic microscope revealed the compositions of JA-1 and JA-2 as chalcopyrite (CuFeS₂) and cuprite (Cu₂O), respectively; this JA-1, JA-5 composition was confirmed by the observed Raman shift of 293, 354 cm⁻¹, while that of JA-2 was corroborated by the detected Raman shifts of 107, 149, 218, and 627 cm⁻¹. This result indicated that smelting at the Jinan Daeryang-ri site was conducted using chalcopyrite (Figure 11).

The gray tissues distributed throughout the fine structure were identified as augite, a task that was challenging with a metallographic microscope but successfully achieved through Raman micro-spectroscopy (Figure 12, Figure 13 and Figure 14). The Raman spectroscopy analysis of slag JA-1 showed that the gray area (a) distributed in the microstructure was identified as augite by detecting Raman shifts of 315, 381, 661, 855, and 1011 cm⁻¹. It was difficult to identify them with a microscope, so it was accurately identified by Raman spectroscopy analysis. Moreover, a white polygon (b) has been identified as magnetite by detecting a Raman shift of 680 cm⁻¹. The columnar structure was identified as fayalite by detecting a Raman shift of 815 cm⁻¹ (Figure 12).

With a significant presence of fayalite detected alongside augite (Figure 13 and Figure 14), shifts at 664, 814, 839, and 1012 cm⁻¹ were detected for JA-2 in both positions (a) and (b). Similarly, for JA-5, shifts at 318, 348, 383, 665, 817, 836, and 1010 cm⁻¹ were also detected.

Samples JA-5 (a) and JA-5 (b) were observed under a metallographic microscope to be brighter than the matrix tissue, with (b) presenting as a lighter gray crystal compared to (a). This suggests a composition of fayalite and augite within the matrix tissue, as supported by the EDS analysis results (Table 6), where colors are distinguished by composition.

The presence of wüstite distributed around copper particles was observed in JA-10, and the bright gray long-columnar tissue in the structure was identified as fayalite by the detected Raman shifts of 814 and 842 cm⁻¹ (Figure 15: (a), (b)). Furthermore, leucite (K(AlSi₂O₆)) was identified in the matrix surrounding fayalite by the detected Raman shifts of 151, 495, and 527 cm⁻¹. It was also detected in EDS results as SiO₂ (40.42 wt%), Al₂O₃ (18.29 wt%), and K₂O (18.01 wt%) (Figure 15: (c)).

4. Discussion

4.1. The Technology of Smelting

In ancient times, copper smelting involved mixing copper ore with charcoal in a crucible or a dugout mud furnace, with air supplied through bellows. Since only a fraction of the smelted material was converted into copper particles, these were manually separated and remelted to produce ingots [16].

The standard procedure for refining copper from copper ore encompasses several steps: calcination and smelting to create matte, followed by additional smelting to produce crude copper, and, finally, refining to obtain purified copper. Elements such as tin or lead are introduced to alloy the matte copper.

Calcination serves as the initial phase, oxidizing and eliminating sulfur from the ore to facilitate its subsequent reduction. Post-calcination, the ore is heated to yield an intermediate product called matte. This is a metallic mass featuring a Cu–Fe–S bonded structure that settles at the furnace’s bottom, while FeO converts into slag and rises to the top. The matte is further smelted to remove its iron content, generating copper sulfide (Cu₂S), while sulfur is oxidized to produce crude copper.

The crude copper then undergoes refining to remove trace impurities such as arsenic, iron, and residual sulfur, yielding refined copper [1]. Slags excavated from the Jinan Daeryang-ri site exhibited a greenish surface corrosion, indicative of copper smelting activity. Compositional analyses revealed these slags to be rich in SiO₂ and FeO, with CuO content ranging from 0.30 to 3.29 wt% and SO₃ ranging from 0.17 to 0.86 wt%. Such components are rarely found in iron slags on the Korean Peninsula, thus confirming the slags’ association with copper production. The composition also included quartz, cristobalite, mullite, fayalite, and wüstite. Notably, some slags contained augite, a calcium-rich mineral. High-temperature indicator minerals, such as cristobalite, suggest that the smelting process occurred at temperatures exceeding 1200 °C. Augite, with its high calcium content, likely acted as a flux, improving fluidity and aiding in the desulfurization process during smelting.

Microstructural analyses of the slags revealed the presence of spherical copper particles embedded in predominantly glassy matrices, with substantial sulfur and iron content. And the presence of wüstite and fayalite in the microstructure provides critical insights into the copper production methods employed at this site.

According to the analysis of these slags, the Jinan Daeryang-ri site is characterized by slag generated during the smelting process of chalcopyrite, specifically at the matte (Cu–Fe–S) stage (Figure 16).

4.2. Comparison of Copper Production Sites on the Korean Peninsula

Copper production remains that have been found so far in the Korean Peninsula include the Gyeongju Hwangnam-dong 376 site of the mid-6th century [17], Buyeo Neungsan-ri temple site of the 6th century [18], Gyeongju Dongcheon-dong site of the 7th-to-8th centuries [7,19], Gyeongju Noseo-dong site of the 7th-to-10th centuries [20], Naju Bokam-ri site of the 7th century [21], Buyeo Ssangbuk-ri site of the 7th century [22], Buyeo Gwanbuk-ri site of the 6th-to-7th centuries [8], Wanju Ungyo site of the 4th-to-7th centuries [23], and Wanju Sinpung site of the 10th-to-14th centuries [24].

Among these, the sites where the remains of the copper production process were identified through scientific analysis were reviewed. The Gyeongju Hwangnam-dong remains were identified as refined copper materials, and the crucible excavated was interpreted to have been a crucible used in the process of making refined copper [17]. Gyeongju Dongcheon-dong was found to be a site where alloys were produced by adding galena and tin to copper ore to make bronze, as well as casting in the process of refining crude copper to make pure copper [7,20]. The molten material of the crucible excavated from the Gyeongju Noseo-dong site was either pure copper or a bronze alloy with varying tin content, and the crucible was used to make pure copper or a bronze alloy [20]. In other words, cassiterite was added to the smelted copper to make a bronze alloy, or copper ore and cassiterite were melted together to make a bronze alloy, and arsenic bronze with a content similar to the bronze from the Dongcheon-dong site was identified. Therefore, it is likely that cassiterite was added after the copper was refined to make the bronze alloy or that copper ore and cassiterite were melted together. The bronzes found at the Dongcheon-dong site were found to be arsenic bronzes with similar content.

The remains from Buyeo Gwanbuk-ri [8] were identified as refining and alloying intermediate products of Cu–Sn or Cu–Sn–Pb. The remains from Buyeo Ssangbuk-ri were characterized as copper, tin, and lead, identified as a bronze melt resulting from the melting and alloying of copper in its refined state. The remains from Buyeo Neungsan-ri consisted of four bronze lumps, analyzed as two crude coppers from smelting matte, one Cu–Sn bronze alloy produced by adding tin to refined copper, and one Cu–Sn–Pb bronze alloy created by adding tin and lead to refined copper.

The Wanju Sinpung sites [3] were confirmed to be slags generated during the alloying stage, while the Wanju Ungyo remains were identified as smelting slags with relatively effective calcination.

Thus, the copper production sites on the Korean Peninsula identified so far involve by-products of refined copper or alloys produced through processes subsequent to matte. However, for the Jeollabuk-do Jinan Daeryang-ri site, the process of smelting chalcopyrite to produce matte has not been excavated, leaving the subsequent processes unknown.

These results were compared with copper by-products from the geographically proximate Wanju Ungyo site, Sinpung site, and Buyeo Gwanbuk-ri site (refer to Figure 17 and

Table 7. EDS analysis results by archaeological site [3,8,23].

Excavate	Analysis Position		Composition (wt%)
			CuO	SnO2	PbO	SiO2	CaO	FeO	Al2O3	K2O	MgO	ZnO
Wanju Ungyo Site (Copper Refining Slag)	A	1	2.27	-	-	-	0.24	88.40	6.38	-	2.71	-
		2	6.91	-	2.12	48.92	13.27	10.54	9.95	1.14	5.02	2.12
		3	97.33	-	-	-	-	2.67	-	-	-	-
Wanju Sinpung Site (Alloy Slag)	B	1	7.06	3.32	27.04	33.76	0.58	8.77	16.47	2.02	0.98	-
		2	100	-	-	-	-	-	-	-	-	-
		3	1.93	74.67	5.59	10.33	-	2.19	5.30	-	-	-
Gwanbu-ri Site (Alloy Slag)	C	1	22.63	-	-	40.63	15.56	0.81	17.24	3.13	-	-
		2	53.13	4.59	-	0.98	0.97	40.33	-	-	-	-
		3	5.78	0.73	-	0.57	-	92.92	-	-	-	-

). The findings indicate that the slags from all four sites contained copper particles dispersed within a glassy matrix. However, at the Ungyo site and Daeryang-ri site, microstructural analysis revealed that fayalite was produced during the process of removing iron and gangue materials present in chalcopyrite.

The slags excavated from the Wanju Ungyo site [17] exhibited fayalite, magnetite, augite, and copper particles in the microstructure analysis, with SO₄ detected in some copper particles by SEM-EDS analysis. However, the relatively low sulfur content (0.35–18.97 wt%) confirmed that it was refined copper, indicative of effective calcination. Moreover, the high iron content (A-1) suggests that the copper was extracted from raw ores of the copper sulfide ore type, with spherical copper oxides widely distributed (see Figure 17, Table 7: A-3). This indicates that the copper refining slag contains a higher proportion of gangue and impurities than the alloy slags from the Wanju Sinpung site.

Tin oxide was observed in the Wanju Sinpung site alloy slag, and the lead content was uniformly distributed in the tissues (Table 7: B). Cassiterite was also detected, and when this was combined with the component analysis result of EDS, the Wanju Sinpung site appears to have produced copper (Cu)–tin (Sn)–lead (Pb) ternary alloys (Table 7: B).

In addition, tin was detected in the alloy slags obtained from the Gwanbuk-ri site (see Figure 17 and Table 7: C-1, C-3), and the presence of cassiterite was confirmed based on the XRD results [8]. The Gwanbuk-ri site is characterized by Cu–Sn binary alloying, and needle-shaped delafossite (CuFeO₂) was observed. This iron and copper alloy suggests that iron was introduced to remove impurities from the alloy structure.

Therefore, the analysis of slags related to copper production excavated on the Korean Peninsula reveals that the Wanju Ungyo and Sinpung sites exhibit characteristics indicative of copper refining or alloy slag. In contrast, the Jinan Daeryang-ri site is distinguished from the others by speculation that it represents the first site on the Korean Peninsula to have produced matte through the smelting of raw ore, subsequently yielding refined copper. The discovery of various by-products of copper production in future excavations would establish it as a pivotal site for elucidating the copper production process on the Korean Peninsula.

5. Conclusions

In this study, a comprehensive analysis of the primary constituents, crystalline phases, and microstructures in slags unearthed from the Jinan Daeryang-ri site was conducted, leading to several key conclusions. First, XRF analyses indicated that SiO₂ and FeO were the most abundant elements in the slag. CuO concentrations ranged from 0.30 to 3.29 wt% and SO₃ concentrations varied from 0.17 to 0.86 wt%. Additionally, ZnO was present by more than 0.1 wt%. Given that CuO and ZnO are seldom found in ferrous slags on the Korean Peninsula, this confirms that the slags from Jinan Daeryang-ri were generated through the smelting of non-ferrous copper ore.

Second, the XRD analyses further revealed that quartz was the dominant phase, accompanied by cristobalite, mullite, fayalite, and wüstite, depending on the sample. The presence of high-temperature minerals such as cristobalite and mullite suggests smelting operations at the site occurred at temperatures exceeding 1200 °C. Moreover, augite was detected in some samples; its high calcium content implies that it may have acted as a flux during smelting.

Third, microstructural assessments unveiled circular copper particles with primarily glassy structures and significant internal sulfur and iron content. Notably, EDS mapping elucidated the sulfur distribution surrounding the copper particles. Crystal phases such as fayalite and wüstite were also identified, and likely formed during the removal of iron (Fe) from the copper ore gangue through the addition of quartz and sand (SiO₂) in iron-rich samples.

Fourth, Raman micro-spectroscopy analyses confirmed that they were refined with chalcopyrite. Furthermore, in addition to fayalite, magnetite, wüstite, and augite, the presence of leucite was confirmed through XRD and EDS analyses. However, augite and diopside were difficult to distinguish owing to their similar XRD and Raman peaks. This is also related to the slag heterogeneity, so it is necessary to perform an accurate analysis to distinguish the exact type of pyroxene. Additionally, it is assumed that the high Ca content of pyroxene minerals worked as a flux that facilitated the separation of slag from copper during smelting.

Fifth, in the Korean Peninsula, remains with refined copper or alloy have been excavated and studied. However, no cases of excavation of smelting remain that produced copper, so studies related to copper production were not conducted. Hence, while this study is meaningful in that it studied the first remains that produced copper, discovered in the Korean Peninsula, it was challenging to generalize the copper production technology due to the limited number of samples. Therefore, excavating various remains related to copper production in the Korean Peninsula by period and region is essential. If detailed scientific analysis is conducted, it will greatly contribute to furthering our understanding of copper production technology in the Korean Peninsula.

Drawing on these conclusions, the slags from Jinan Daeryang-ri conclusively demonstrate that copper smelting was practiced at this site. Furthermore, both archaeological and metallurgical analyses suggest that the slag originated from copper sulfide and underwent calcination and smelting to produce an intermediate product, namely, matte. Future excavations that uncover additional copper production artifacts, such as crucibles used for smelting and alloying copper and tin, will enhance our understanding of ancient metallurgical practices on the Korean Peninsula.

The excavation of the Jeollabuk-do Jinan Daeryang-ri copper production site revealed the remains of smelting, the primary production of copper from ores, and its by-products (copper slag and furnace walls). This provided an important opportunity to review the entire copper production technology system, from mining to refining. Using the scientific data with sufficient metallurgical analysis and research will allow us to subdivide the later stages of copper smelting and characterize them, shedding light on the development of copper metallurgy in ancient times.