The Color Formation of “Lumu Stone” in the Weathering Processes: The Role of Secondary Hematite and Goethite

Abstract

Rocks and minerals buried in the earth’s surface usually undergo weathering processes and change color in the burying environment. A kind of yellow Chinese stamp stone named “Lumu stone”, which is buried in a yellowish weathering crust (yellowish soil), was selected to investigate its color changes in the weathering processes. In this study, the appearance features, mineral components, micromorphology, spectroscopy characteristics, and color causation of the “Lumu stone” were studied by using X-ray powder diffraction (XRD), scanning electron microscopy (SEM), an electron probe microanalyzer (EPMA), a laser ablation inductively coupled plasma mass spectrometer (LA-ICP-MS), and a UV-Visible (UV-Vis) spectrum. The “Lumu stone” usually exhibits a darker yellow outer layer and a lighter yellow core, suggesting that yellow color permeated into the stone from the surface to the core gradually and the color is secondary forming. The results from XRD and SEM show the studied samples are mainly composed of dickite and illite. The individual particles of the dickite and illite are about 2–5 μm, randomly distributing in the three-dimensional space and constituting voids among the particles. The acid pickling experiments using HCl coupled with KSCN confirmed that the mineral phases that caused the yellow color of the matrix are iron oxide and hydroxide. On the other hand, goethite and hematite were observed gathering in the yellow and brown-red cracks on the “Lumu stone” by SEM study. However, iron oxide and hydroxide in the matrix were difficult to observe and detect among the dickite and illite aggregates by SEM and XRD methods. It indicates that they may be nanoscale in size and very low in content. According to the calculation of the second derivative of Kubelka-Munk (K-M) transformed diffuse reflection spectroscopy (DRS) curves obtained from UV-Vis, the characteristic peaks of goethite and hematite were found in the yellow matrix, and their contributions to the color were confirmed. The concentrations of goethite and hematite were calculated to be 0.32 to 1.87 g/kg and 0.22 to 0.93 g/kg in the studied samples, respectively. In this study, a series of methods were employed to detect very low levels of goethite and hematite in the samples undergoing weathering processes. Additionally, nanoscale goethite and hematite were considered newly formed minerals when buried in the weathering processes and may gradually move into the voids among phyllosilicate particles. Therefore, they turned the “Lumu stone” yellow.

1. Introduction

“Lumu stone” is a type of seal stone formed in the Shoushan area (Figure 1). It has a low hardness, making it suitable for carving into artwork. It is usually excavated in the soil and has a similar formation environment, color, and appearance to the most famous and expensive seal stone, Tianhuang stone [1,2]. Yellow is considered the most valuable color of seal stone. Therefore, “Lumu stone” is a good sample to investigate whether the shallow burial after its formation will have an effect on the color changes.

For a long time, it was thought that secondary hematite or goethite caused some seal stones to turn yellow. However, very few experiments or pieces of evidence have been reported. The color-forming mechanism of Tianhuang stone was considered to be through hydrothermal alternation and water-rock reaction processes [3]. It was found that iron is free to exist in the Tianhuang stone, and it was deduced that the iron oxides/hydroxides are adsorbed at the surface of phyllosilicate mineral grains, which causes the Tianhuang stone to turn yellow. In a recent study, Tian et al. (2023) investigated the relationship between color and iron oxides in stones from Laos, whose main mineral component is dickite [4]. The other hypothesis is that isomorphism substitution in the phyllosilicate crystals, such as Fe irons replacing Al irons, causes the yellow color of the seal stones.

The study conducted X-ray powder diffraction (XRD), Raman, and scanning electron microscopy (SEM) analyses on other stones that are colored by iron oxides and/or hydroxides, such as the weathering crust of basalt and quartzite [5,6,7]. Because the concentration of iron oxides and hydroxides in them is usually high, they have good crystallization and a large grain size, making it easy to identify them as hematite and goethite. However, the detection and experimentation of secondary hematite and goethite with opposite features could be challenging.

Goethite and hematite are widely spread in the weathering crust of the earth’s surface, such as in the loess and the soil [8]. They prefer to form nano-sized and poorly crystallized particles and concentrate in low amounts in the soil. Goethite and hematite are the causes of yellowish and reddish colors in soils [8,9]. Their particle size is usually between 10 and 20 nm [5,6,10,11]. Due to their low contents in soils and the detection limit of XRD, purification is necessary before conducting XRD experiments [9,12]. They can be observed by transmission electron microscopy (TEM) and detected by the Mössbauer spectrometer after purification when their content is not too low [5,6,9,10]. However, the newly formed (secondary) iron oxides/hydroxides are usually less than 1% in igneous, metamorphic, and sedimentary rocks and soils. They are hard to extract or identify using the above techniques because of their very low content and poor crystallization.

Diffuse reflection spectroscopy (DRS) is a method to test the reflectance and absorbance spectra of minerals at wavelengths from UV to near-infrared (220~1000 nm). It is a new, rapid, and effective method to determine the very low content of goethite and hematite in the loess and the soil [13,14,15]. The detection limit of DRS is as low as ~0.01 wt% [16]. Therefore, DRS is widely applied to quantify the content of hematite and goethite in soil research [13,17,18,19,20]. In addition, this method has been used in researching gemstones such as Beihong agates [21]. Based on the DRS, more data analysis methods have been proposed to improve the quantification of goethite and hematite. Calculating the derivates of the spectra was recommended using different calculation methods, such as first-order derivates [16,17,22,23] and second-order derivates [14]. At the same time, with the development of the Kubelka-Munk (K-M) theory in soil color research [24,25], the second-order derivate of K-M transformed DRS was used to identify and quantify iron oxides and hydroxides and establish accurate content formulas to calculate the content of goethite and hematite [15].

The present study aims to investigate the cause of “Lumu stone” color by employing a range of methods to confirm the formation of goethite and hematite. These methods were specifically designed to detect the very low concentrations of newly formed nanoscale iron goethite and hematite.

2. Geological Background

The studied “Lumu stone” samples of this work were excavated from the Lumu field (north latitude 26°17′04.76″, east longitude 119°17′23.15″) at the foot of Mabei mountain, Shoushan Village, Fujian Province, China. The field is a phyllosilicate assemblage generated by hydrothermal alternation and has a variety of valuable gemstones belonging to the classification of “Shoushan stone” [1,2]. “Lumu stone” breaks off from the Mabei mountain and tumbles to the hillside in small pieces. The most common ones generally weigh tens of grams. Because no stream goes through the area, the rocks are buried almost without migration in yellowish soils (the weathering crust of the earth’s surface) and generate a yellowish weathering layer. Mabei mountain is located in the “Gaoshan-Duchengkeng metallogenic belt” [26] in the southern Down Field of the Shoushan stream (Figure 2).

The Shoushan village area is located in the Shoushan-Emei volcanic eruption basin, which formed in the volcanic eruption zone of the coastal region of eastern Fujian during the Early Cretaceous [26,27]. A mass of pyroclasts accumulated around the crater and formed a volcanic cone caused by violent volcanic activity. Many lavas erupted from the volcanic chamber, leading to an underground empty chamber, and then the volcanic cone collapsed and became a caldera. The collapse produced circular, radial, and even step-like fractures and formed an interlayer fracture zone, providing genetic space for a hydrothermal solution to be filled. In lithology, the pyroclastic rock of the Xiaoxi Formation is the surrounding rock of “Shoushan stone”, which is composed of ignimbrite, rhyolite, and rhyolitic crystal tuff [26,27]. Additionally, the pyroclastic rock enriches salic minerals; the phyllosilicate could be formed when hydrothermal alternation occurs [26]. The rhyolitic crystal tuff is a primary rock in the “Gaoshan-Duchengkeng metallogenic belt”.

3. Materials and Methods

For this study, 10 samples were selected from excavations dug by local miners (Figure 3).

For specific gravity (SG), each sample was weighed three times using hydrostatic measurement, and the average value was recorded. Hardness was determined using the Mohs hardness method. External and internal features were observed using the Leica M205A/DFC 550 microsystem.

XRD was performed using a PANalytical B.V. X’ Pert PRO Dy2198 X-ray diffractometer fitted with Ni-filtered Cu-Kα radiation. Powdered samples (200 mesh) were mounted on a flat holder, and diffraction patterns were collected in the range of 2θ: 3~65° at a scanning speed of 0.4°/s, a scanning step of 0.0167°, and operating at 40 kV and 40 mA. Mineral components were identified by comparing them with the standards in the ICDD Powder Diffraction File using the software X-ray Run 2018, with reference to the standard files of dickite (PDF 76-0632), illite (PDF 43-0685), pyrophyllite (PDF 24-0011), chlorite (PDF 13-0003), and svanbergite (PDF 76-0630).

UV-Vis measurements of all samples were performed using a Skyray GEM-UV 100 UV-Visible spectrometer fitted with a total reflectance integrating sphere using the reflectance method. The spectra were collected under the conditions of an integration time of 110 ms, an average of five times, ranging from 220 to 1000 nm. All spectra were processed to improve the signal-to-noise ratio. The wavelength ranges of the original DRS were selected from 300–900 nm to avoid the impact of UV light instability and the strong hydroxy absorption of phyllosilicates. The original and first-order derivate spectra were smoothed using the Savitzky-Golay method with 30 points [28]. The second-order derivative spectra of DRS and the K-M spectra were smoothed with 40 points. All spectra were not processed by baseline calibration.

Samples were coated with carbon to enhance conductivity, and surface morphology was investigated using an SEM from Thermo Fisher Helios G4. Various accelerating voltages (20 kV, 5 kV), currents (1.4 nA, 0.34 nA), and magnifications were used to show the most evident range of interests. Chemical analyses were detected using an EDS from Oxford AZtec Ultim Max 100, which incorporated AZtec 5.0 software.

Mineral chemical components were also analyzed using an EPMA from JEOL JXA-8230 equipped with five wavelength dispersive spectrometers (WDS). The instrument was operated under an accelerating voltage of 15 kV, a beam current of 20 nA, and a spot size of 1 μm. The data were automatically corrected using the ZAF correction. The concentrations of Si, Al, K, and Fe were quantified using diopside (Si), almandine (Al), orthoclase (K), and olivine (Fe) standards.

In-situ trace element analysis of minerals by laser ablation inductively coupled plasma mass spectrometer (LA-ICP-MS) was conducted at the Wuhan SampleSolution Analytical Technology Co., Ltd., Wuhan, China. Laser sampling was performed using a GeolasPro laser ablation system that consists of a COMPexPro 102 ArF excimer laser (wavelength of 193 nm and maximum energy of 200 mJ) and a MicroLas optical system. An Agilent 7700e ICP-MS instrument was used to acquire ion-signal intensities. The spot size of the laser was set to 44 µm. The contents of the samples were calibrated against multiple-reference materials (NIST 610) combined with internal standardization. The method of internal standardization utilized the mean value of Al₂O₃ (weight percentage, standard deviation: 0.159) obtained from the EPMA test results.

SG, hardness, microsystem, and UV-Vis measurements were performed at the Gemological Institute of the China University of Geosciences (Wuhan) (CUG), while XRD, SEM, EDS, and EPMA analyses were performed at the State Key Laboratory of Geological Processes and Mineral Resources (GPMR), CUG.

4. Results

4.1. Internal and External Features

4.1.1. Conventional Appearance Observation and Tests

The “Lumu stone” has a yellowish weathering layer and many white pits on the surface. The raw stone is angular or subangular with an earthy luster. The weathering layer is the outer layer, which showed much lower transparency (caused by the looser texture) and a stronger yellow tone than the internal fresh matrix. Usually, the interior matrix has a lighter yellow than the outer layer, while some samples even have a white interior, which can be referred to as the fresh part. Red-colored regions often exist in the yellow matrix. The values of SG and Mohs hardness are 2.51~2.68 and 2~3, respectively. The low hardness provides the opportunity to be carved easily by the burin, the calligrapher, or the artist. To discover the features of the weathering layer, microscope observation is necessary.

4.1.2. Microscopic Features Observation

Figure 4 shows the magnified regions of “Lumu stone” samples. There are many pits with irregular shapes and different sizes on the surface of the weathering layer (Figure 4a). All samples show a lighter yellow interior matrix, except one sample, which has a white interior matrix (Figure 4b). It indicates that the interior part of some samples did not undergo complete weathering processes, and the interior is fresher than the exterior. The yellow color is filamentous and distributed along with phyllosilicate particles (Figure 4). Additionally, some yellow spots exist in the interior matrix (Figure 4c). Figure 4d shows the feature of the brown-red crack, as indicated by the white arrow, which extends from the surface of the yellowish weathering layer into the white interior. In addition, in the interior matrix, the yellow color diffuses from the crack into the white matrix (as indicated by the yellow arrow in Figure 4d).

4.2. Mineral Components

XRD analysis has been widely used to investigate mineral components because of its outstanding ability to analyze mineral phases, especially to distinguish polytypes of phyllosilicate. The XRD data of “Lumu stone” samples are shown in Figure 5. “Lumu stone” can be divided into dickite-type and illite-type according to the main mineral component.

(1)

In dickite-type samples (Figure 5a), XRD patterns showed a feature of the kaolin group that was characterized by (002) reflection near 7.17 Å. Kaolin-group minerals include kaolinite, dickite, nacrite, and so on. The six diffraction peaks split distinctly in the range of 18–24° (2θ). The phase that has peaks at the d values of 3.96 Å and 3.79 Å is attributed to dickite. The intensity and resolution of the shape of the six peaks are directly related to the order degree of the dickite, especially the shape of the 4.28 Å peak. The patterns of the samples, which had clearer and sharper peaks, indicated a higher-order degree of the dickite [29]. In addition, a small amount of illite had been identified in LM-4 samples according to the (002) reflection at 10.02 Å. LM-3 and LM-5 samples were relatively composed of pure dickite.

(2)

In illite-type samples (Figure 5b), LM-1, LM-6, LM-9, LM-17, and LM-21 were composed of pure illite. Sample LM-7 contained pyrophyllite, identified by the (001) reflection at 9.22 Å. Sample LM-2 had relatively complex mineral components of pyrophyllite, dickite, and chlorite that were identified as coexisting with illite.

XRD did not detect the existence of any iron oxides or hydroxides. This does not mean they do not exist in the samples, but due to their lower contents, which might be below the detection limit of the XRD instrument, and/or their poor crystallization [22], they remained undetected. Further experiments are needed to explore whether they exist.

4.3. Micromorphology Features and Micro-Chemical Analysis

4.3.1. Micromorphology Features

SEM can be used to study the micromorphology of minerals and provide chemical analysis from EDS, which is employed in the SEM. In the fresh fracture of dickite-type samples, the subhedral- and anhedral-shaped particles of dickite can be observed as flakes with a grain size of about 2–5 μm and a thickness of no more than 300 nm. The dickite particles crystallized randomly in three-dimensional space (Figure 6a). Additionally, in illite-type samples, most illite particles had the same size and shape as the dickite, but their thickness was no more than 100 nm. Some of the illite particles spread randomly, but a few of them were stacked similar to books (Figure 6b).

A brown-red region (as shown in the inserted optical micrograph in Figure 6c) showed high contrast in backscattered electron (BSE) images. EDS showed that they contained Fe and O elements, so they were iron oxide or hydroxide particles. Different varieties of iron oxides and hydroxides have different morphologies [24], so their morphology under secondary electron (SE) images is significant for phase identification.

Figure 6d–h shows the various morphologies. Figure 6d shows “iron roses”, which were composed of mica-like hematite and specularite (slabby hematite). Because mica-like hematite is more compact, it appeared to be more enriched in Fe than specularite (as shown in the inserted optical micrograph in Figure 6d). Figure 5e shows reniform hematite with a radiated structure. Figure 6f shows acicular aggregations composed of goethite. The brown-red crack surface (as shown in the inserted optical micrograph in Figure 6g) was adhered by amounts of tiny granular hematite (Figure 6g) and formed oolitic hematite with a grain size of 300~500 nm (Figure 6h).

So, the iron oxides and hydroxides crystallized and gathered in some larger voids and cracks.

4.3.2. Micro-Chemical Analysis

LA-ICP-MS was employed to examine both the weathering layer and the interior matrix. With higher sensitivity to trace elements, LA-ICP-MS could detect lower concentrations of trace elements that EPMA had failed to detect due to its higher detection limit. The results (Table S1 in Supplementary Materials) showed that, other than Fe, most of the trace elements, such as Cr, V, Mn, Co, and Ni, have too low a content to relate to colors. The weight percent of total FeO was higher than that of other trace elements (0.3–1.2 wt%). Due to the large beam size (44 μm) of the laser beam, it is inevitable to detect several particles at once. In Figure 7, the contents of total FeO (TFeO) and Na₂O in the weathering layer and interior matrix are compared. It is evident that most of the weathering layers exhibit higher levels of TFeO and Na₂O compared to the interior matrix. Consequently, further focus should be placed on the position of the Fe element.

In addition to the cracks, the weathering layer and the interior also showed a yellow color, but iron crystals were not found in these parts by SEM. Two assumptions were proposed: one is that the Fe ion enters the crystal structure and replaces Al^Ⅵ in dickite and illite by isomorphism substitution; the other is that iron oxides and hydroxides are too small to be observed by SEM. The EPMA experiment was used to detect the chemical composition of dickite and illite and to find whether Fe content is higher in the yellowish weathering layer than that in the white interior matrix. The chemical composition was tested in the center area of the dickite or illite particles to avoid the influence of iron minerals at the boundary of the dickite or illite particles.

Table 1. The mineral chemical composition of dickite- and illite-type samples.

Sample No.	LM-5-R-1	LM-5-R-2	LM-5-R-3	LM-5-P-1	LM-5-P-2	LM-5-P-3
Oxides (wt.%)
SiO2	46.09	46.08	45.54	45.99	44.45	46.34
Al2O3	42.14	41.75	42.09	42.18	42.04	41.93
K2O	0.02	0.02	0.01	-	0.02	0.02
TFeO	0.01	0.01	0.02	0.10	0.11	0.07
Total	88.26	87.86	87.65	88.27	86.63	88.36
Ions based on 14 oxygens
Si	3.87	3.89	3.85	3.86	3.62	3.89
Al	4.17	4.15	4.20	4.18	4.51	4.15
K	<0.01	<0.01	<0.01	-	<0.01	<0.01
Fe	<0.01	<0.01	<0.01	<0.01	<0.01	<0.01
Sample No.	LM-6-R-1	LM-6-R-2	LM-6-R-3	LM-6-P-1	LM-6-P-2	LM-6-P-3
Oxides (wt.%)
SiO2	48.79	48.19	47.69	49.96	48.16	47.66
Al2O3	41.37	40.44	40.59	40.31	41.18	40.29
K2O	9.48	9.43	9.47	9.03	9.56	9.57
TFeO	0.05	0.06	0.06	0.09	0.01	0.25
Total	99.69	98.13	97.82	99.39	99.00	97.76
Ions based on 11 oxygens
Si	3.04	3.05	3.03	3.1	3.02	3.03
Al	3.03	3.01	3.04	2.95	3.05	3.02
K	0.75	0.76	0.77	0.72	0.77	0.78
Fe	<0.01	<0.01	<0.01	<0.01	<0.01	0.01

shows the chemical composition of the dickite (LM-5) and illite (LM-6) types. The particles had similar iron content and were always in the weathering layer or white interiors of a sample. Thus, Fe does not take the place of Al^Ⅵ in the crystal structure of illite or dickite.

EDS was used to conduct element mapping in order to analyze the distribution of major and trace elements within the weathering layer and interior matrix. Figure 8a,f are the BSE images of the weathering layer and interior matrix of LM-5 (dickite-type); others are the corresponding element mappings. Figure 9a,g are the BSE images of the weathering layer and interior matrix of LM-6 (illite-type); others are the corresponding element mappings. Figure 8 and Figure 9 demonstrate the distribution of elements, where major elements are enriched at dickite and illite particles while lacking at boundaries among the particles. The distribution of Fe is complementary with that of the major elements and was distributed among particle boundaries in the weathering layers (Figure 8e and Figure 9f). The interior matrix is lighter than the weathering layers, especially the LM-5 interior, which is pure white, so the distribution of Fe is not obvious.

Therefore, the concentrations of Fe within the dickite and illite particles in the weathering layer and interior matrix have not changed. Fe isomorphism substitution did not take place on dickite and illite crystals in the weathering process but formed as iron oxides and/or hydroxides and crystallized on the surface of dickite or illite particles and surrounding the particles. This process must be what makes the part turn yellow.

4.4. Identifying Goethite and Hematite

4.4.1. Acid Pickling Experiment

To strengthen the evidence from element mapping analysis, a sample acid pickling experiment was conducted. The iron oxides and hydroxides could be dissolved by acid, while Fe in dickite or illite could not be dissolved by acid. The iron oxides and hydroxides do react with HCl, such as hematite (Fe₂O₃) and goethite (FeOOH), and generate Fe³⁺. KSCN is an important chromogenic reagent with Fe³⁺ that appears red. The chemical reaction equations are as below:

Fe₂O₃ + 6 HCl = 2 FeCl₃ +3 H₂O

(1)

FeOOH + 3 HCl = FeCl₃ + 2 H₂O

(2)

FeCl₃ + 3 KSCN = Fe(SCN)₃ + 3 KCl

(3)

Three groups of samples were set up: a blank control group (Figure 10a), a control group (Figure 10b), and experimental groups (Figure 10c–e). Nothing was initially added in the blank control group, while hematite was initially added in the control group. In the experimental groups, “Lumu stone” was divided into three groups: two powders (about mesh 200) and one small chunk (1 × 1 × 0.5 cm in length, width, and thickness). Excessive diluted hydrochloric acid (concentration: 1 mol/L) was added to every group and then stood for 30 min to ensure the reaction sufficiently. The chunk group (Figure 10e) stood for 5 days because it was harder for it to react sufficiently compared to the powder groups (Figure 10c,d). The powder group (Figure 10c) shows a troubled yellow solution, so the powder group (Figure 10d) was filtered by two filter papers, and a clear, colorless solution was obtained. The FeCl₃ solution should appear yellow, but all the solutions of this step were colorless except the powder group (Figure 10c) due to the low content of Fe³⁺. Excessive KSCN was added to every group except the chunk group (Figure 10e), and the solution colors of the control group and the two experimental powder groups (Figure 10b–d) changed. The blank control group was still colorless because nothing had reacted with HCl or KSCN. The control group appeared to have a light rosy color, implying it contained Fe³⁺. The solution of the powder group (Figure 10c) had changed color from yellow to orange, and the other powder group (Figure 10d) had changed from colorless to light red. The chunk group stood for 5 days, then the solutions changed color from colorless to yellow.

The different reddish color changed by KSCN and yellow changed by HCl, indicating iron oxides and/or hydroxides existed in the samples.

4.4.2. The Spectra of DRS

The UV-Visible spectrometer is particularly useful in studying the causes of color minerals. Now DRS has been widely used in identifying and quantifying the hematite and goethite in soil [13,14,15]. It has achieved a remarkable effect in the correlation of color and iron oxides in the yellow seal stones [4].

To identify and quantify the variety of iron oxides and hydroxides, the results of DRS were analyzed by four groups: original DRS, first-order derivates of DRS, second-order derivates of DRS, and second-order derivates of K-M transformed DRS curves.

The original DRS reflectance spectra are shown in Figure 11a. Low resolution caused many peaks to combine, so it was hard to distinguish different peaks from different minerals. As the order of derivates arose, the resolution obviously improved, and more peaks could be distinguished. The first-order derivate spectra (Figure 11b) represented the slope changes of the original spectra. The highest peak center, located from 551 to 571 nm, belonged to hematite. The peaks with weak shoulders are located at approximately 510 nm, and the weak secondary peaks nearly approach 434 nm, attributed to goethite [22,23,30]. Their centers could be experiencing a redshift accompanied by increasing concentrations [22]. Therefore, the spectra indicated that they are composed of different proportions of mixed hematite and goethite.

The second-order derivate spectra represent the inflection points of the original spectra; in other words, the real peaks were fitting out. The strongest absorbance in iron oxides and hydroxides is 2(⁶A₁)→2(⁴T₁), which belongs to the electron pair transition (EPT). As Figure 8c shows, there are strong absorbance bands at 520~565 nm (red area) and 470~520 nm (yellow area), attributed to hematite and goethite, respectively [15,24]. It indicates that the newly formed goethite and hematite coexisted in the samples. Other peaks at 430 to 450 nm (green area) are assigned to ⁶A₁→⁴E⁴A₁ of hematite and goethite, and at 350 to 380 nm (blue area) they are assigned to ⁶A₁→⁴E [24]. Using the K-M theory to process the original DRS can transform the reflectance into absorbance [25]. Calculating the second derivate K-M transformed DRS can yield a more credible result. The spectra (Figure 11d) show the strongest peaks at around 485 nm, 420 nm, and 340~360 nm belonging to the EPT (yellow area), ⁶A₁→⁴E⁴A₁ (blue area), and ⁶A₁→⁴E of goethite, respectively. The weak peaks at around 525 nm and 380 nm were assigned to the EPT (red area) and ⁶A₁→⁴E of hematite, respectively.

Based on the calculation of second derivate spectra and the K-M theory, Scheinost et al. (1998) proposed a measurement to quantify the concentrations of hematite and goethite [15]. This method has been widely used in soil research [31,32,33]. The amplitudes between ~415 nm and ~445 nm for goethite (Y₁, blue area) and between ~535 nm and ~580 nm for hematite (Y₂, red area) were selected to calculate the contents with the formula:

w(Gth)/(g/kg) = −0.06 + 268Y₁ (R² = 0.86; n = 40; p < 0.001;)

(4)

w(Hem)/(g/kg) = −0.09 + 402Y₂ (R² = 0.85; n = 40; p < 0.001;)

(5)

The results obtained from the quantified formula are presented in

Table 2. The concentrations of goethite and hematite.

Sample Number	Y1 (×103)	Y2 (×103)	w(Gth) (g/kg)	w(Hem) (g/kg)	Total (g/kg)
LM-1	2.31	0.76	0.56	0.22	0.78
LM-2	1.43	1.27	0.32	0.42	0.75
LM-3	7.19	0.95	1.87	0.29	2.16
LM-4	6.62	1.16	1.71	0.38	2.09
LM-5	4.70	1.13	1.20	0.37	1.56
LM-6	1.45	1.42	0.33	0.48	0.81
LM-7	2.04	2.55	0.49	0.93	1.42
LM-9	2.38	1.87	0.58	0.66	1.24
LM-17	1.57	1.19	0.36	0.39	0.75
LM-21	1.53	1.56	0.35	0.54	0.89

. The concentrations of goethite were from 0.32 to 1.87 g/kg, the hematite was from 0.22 to 0.93 g/kg, and the total iron oxides/hydroxide were from 0.75 to 2.16 g/kg. These results indicate that the contents of goethite in most of the samples are higher than those of hematite, which causes the color to be yellow. In contrast, the brown hue increases when hematite is higher in concentration than goethite, although the total contents are below 0.90 g/kg. Such low concentrations of goethite and hematite in sample powders cannot be detected by XRD but can be detected by the second derivative of K-M transformed DRS.

5. Discussion

The surface of the rock usually undergoes weathering processes when exposed to the air or buried in the soil. During these processes, some of the components dissolve, while others migrate into the rock and crystallize as secondary minerals, eventually forming a weathering layer on the exterior [7,34].

The raw “Lumu stones”, or some white parts of the stones, had not undergone weathering processes (Figure 4b). The experiments of SEM, LA-ICP-MS, EPMA, EDS, acid pickling, and DRS were designed to find what caused the yellowish color and how it exists in the stone.

Goethite and hematite particles, which crystallized at the yellow cracks and brown-red regions, were easy to observe by SEM. They were recognized based on their micromorphology and EDS. A yellowish weathering layer was observed in the studied samples. They are composed of many tiny dickite and illite particles (2~5 μm and thousands of nanometers thick) with voids and a loose texture, so the special structure provides space to accommodate newly formed nanosize particles (Figure 6a,b). According to the results obtained from LA-ICP-MS analysis, the weathering layers are found to contain relatively high concentrations of TFeO and Na₂O. The EDS element mapping was used to detect whether there was an additional chemical composition or new crystals formed around the particles of the matrix. The Fe element was found surrounding the original particles in the weathering layer of “Lumu stone” (Figure 8e and Figure 9f).

Meanwhile, EPMA experiments were used to detect the chemical composition of the dickite and illite crystals. Almost no chemical composition changes were found in the dickite and illite crystals, whether they were in the yellow outer layer or the white core of the stones (Table 1). Therefore, the Fe element did not enter the crystal structures of dickite and illite. That is to say, Fe does not replace Al in dickite and illite by isomorphism substitution. Combined with acid pickling and DRS experiments, the Fe material around the dickite and illite crystals was found to be iron oxides and hydroxides, and further to be goethite and hematite (Figure 10 and Figure 11). The newly formed (secondary) goethite and hematite content in “Lumu stone” is not greater than 2.16 g/kg.

The iron in soils is mostly found in primary minerals such as ferric and silicate minerals that contain iron. Once exposed to the surface of the Earth, these primary minerals undergo a weathering process and decompose into various minerals composed of soil [24]. During pedogenic and weathering processes, the iron is released from these minerals and transformed into various secondary common iron minerals in soils [7,24,35]. If there is sufficient water present, iron may also precipitate directly from the hydrolysis of iron-rich fluids [36]. For rocks, weathering processes begin at exposed surfaces and cracks, as they offer easier access to the atmosphere and facilitate various reactions that generate iron oxides or hydroxides [7]. It is considered that the color changes of soils and weathering crusts can indicate different proportions of goethite and hematite and infer climate in various ways [37]. The former reports suggested that relatively high temperatures and low humidity are suitable formation conditions that are beneficial to form hematite, while goethite is present in cool, humid climates [24,37,38,39]. They can transform into each other when the climate changes [37,40]. Therefore, it is possible that the “Lumu stone” was buried in the yellow soil and reacted with an aqueous solution containing Na⁺ and Fe²⁺/Fe³⁺. As a result of this interaction, the surface of the stone could have undergone supersaturated precipitation, leading to the production of hematite and goethite. In this study, nanoscale goethite and hematite were considered to form in the stone as secondary minerals under the weathering processes. They existed around the dickite and illite particles and made the stone appear yellowish. Additionally, they crystallized in larger sizes at cracks, which can be directly observed by SEM. In the weathering processes, newly formed goethite and hematite are primarily attached to the exposed surfaces and cracks, as they offer easier access to the atmosphere and facilitate various reactions that generate iron oxides or hydroxides [7]. The secondary goethite and hematite adhered to the surface of the original particles; they were concentrated externally first and formed a weathering layer. As the weathering process continues, they diffuse gradually into the interior from the outer layer and the cracks (Figure 4d).

6. Conclusions

The “Lumu stones” were mainly composed of dickite and illite. The particle size of dickite and illite was about 2–5 μm. They were randomly distributed in the three-dimensional space, so voids occurred among the particles. Acid pickling experiments confirmed that iron oxides and hydroxides caused “Lumu stones” to turn yellow. The second derivative K-M transformed DRS method identified the mineral phases of iron oxides and hydroxides as goethite and hematite and quantified their concentrations at 0.32 to 1.87 g/kg and 0.22 to 0.93 g/kg, respectively. Such low contents and poor crystallinity make their existence undetectable by XRD and SEM. So, the secondary goethite and hematite form at the nanoscale and come into the voids or attach to the dickite and illite crystals under the weathering processes. They concentrated at the surface of the stone and permeated into the interior gradually, at last forming a darker yellow weathering layer and a lighter yellow-colored core.