Cluster Analysis of Soluble Organic Fractions in Two Low-Rank Coals

Abstract

A brief overview of the composition and structure of soluble organic compounds as well as organic macromolecules in coal is sufficient to support the development of the coal chemical industry and realize the clean, effective, and high value-added exploitation of coal. In this study, 14 groups of extracts were analyzed by a gas chromatograph-mass spectrometer; seven stages of solvent-graded extraction were carried out on two typical low-rank coals, and the extracted substance spectrum data were processed by principal component analysis and cluster analysis to investigate the molecular makeup and structure of the soluble organic compounds in the two coals. CS₂ had a better extraction effect on aromatic hydrocarbons, and methanol was better for the enrichment and separation of oxygenated compounds. Hexane and petroleum ethers have a good extraction effect for aliphatic hydrocarbons; benzene has a good effect for the extraction of nitrogenous compounds; acetone has a greater extraction rate for alcohols; tetrahydrofuran has a more obvious extraction effect for ketones; the mixed solvent of tetrahydrofuran and methanol has a greater extraction rate for esters.

1. Introduction

Low- and medium-grade coals have a high organic matter content and are rich in aromatic rings, especially heteroatomic aromatic compounds [1], which are fine organic chemicals with a high added value [2]; therefore, low- and medium-grade coals are uniquely utilized as raw materials for obtaining high value-added chemicals [3,4]. China has a large proportion of low-quality bituminous coal, mainly in Shaanxi, western Inner Mongolia, and Xinjiang, and it is the goal and requirement of “double carbon” to realize its clean and efficient utilization [5]. Extracting soluble organic matter from coal by physical or chemical methods and converting it into chemicals with a high added value is an effective way to realize the clean [6] and efficient utilization of coal [7].

Gas chromatography–mass spectrometry (GC/MS) is one of the important analytical detection methods in the petrochemical and coal chemical industries [8,9]. GC has an extremely strong separation ability, and MS has a unique ability to identify unknown compounds with high sensitivity; therefore, GC/MS is one of the most powerful tools for the separation and detection of complex compounds [10] and can be used to identify coal with low molecular weight and volatility and under gasification or high-temperature substances that do not decompose [11], thus obtaining more accurate structural information. Ju et al. [12] extracted Yanzhou coal with tetrahydrofuran and analyzed the extracts by gas chromatography/mass spectrometry (GC/MS). The results show that the extracts were mainly composed of aromatic and aliphatic hydrocarbons, and the aromatic hydrocarbons were mainly naphthalene compounds. Zhang et al. [13] conducted NMP extraction of Xinjiang Zhundong coal while dissolving the extracts using organic solvents, such as methanol; the dissolved products as well as the solid residues were analyzed by GC/MS and FTIR. It was found that many nitrogen-containing compounds as well as nine sulfur-containing compounds, four chlorine-containing compounds, and two bromine-containing compounds were detected in the extracted solute of the coal. Ping et al. [14] extracted the coking coal with THF and analyzed the compound composition of the extracts by GC/MS. The results show that the THF extracts of coking coal contained the smallest relative content of aromatic hydrocarbons and the largest relative content of heteroatom-containing organic compounds, and most of them were oxygenated compounds.

In recent years, new instruments that allow the high-throughput detection of large numbers of samples and also provide multivariate data and more comprehensive chemical information, among others, have led to more powerful analytical methods for the knowledge of coal structure [15]. However, finding the required data from multiple mass spectra is difficult for traditional GC/MS data processing. Along with the rapid development of computer technology, we can use more and more data analysis methods to process the data. The dimensionality reduction idea of the principal component analysis [16] and the processing method of cluster analysis can help us to semiautomate the analysis of the obtained data, thus greatly reducing the data processing time. Zhang et al. [17] performed a graded extraction of six medium- and low-order coal samples, analyzed the extracts by GC/MS and Electron Spray Ionization-Time of Flight Mass Spectrometer, and performed a systematic cluster analysis of the extract data based on statistical methods in R language to explore the composition and distribution characteristics of molecules in coal soluble matter. Yu [15] selected eight coal samples for graded thermal solubilization, characterized the soluble organic fractions by GC/MS and Orbitrap MS, and then used chemometric methods to find the molecular composition and structural information of the organic matter from the analytical results. Zhang et al. [18] performed a graded extraction of four low-rank coals. The extracts were analyzed by Orbitrap mass spectrometry coupled with an atmospheric pressure chemical ionization ion source, and the effective information was mined from the large amount of mass spectrometry data by two statistical methods: principal component analysis and hierarchical cluster analysis.

In this study, GC/MS was used to analyze 14 groups of extracts from Shenfu and Huolinguole coals. The MS data were first classified by family composition, and then the MS data were processed by origin principal component analysis and systematic cluster analysis to achieve an efficient classification and processing of complex mass spectrometry data. The content and distribution of the aromatic compounds and heteroatomic compounds in the two low-rank coal samples were visualized to determine the type, content, and distribution information of the compounds in the soluble fraction of the organic matter in the coal, thus deepening the knowledge of the composition and structure of coal organic matter at the molecular level.

2. Materials and Methods

2.1. Instruments and Reagents

Coal sample crusher; precision electronic analytical balance; Soxhlet extractor, Rotary Evaporation-3000 rotary evaporator; vacuum drying oven; and HP6980/5973 gas chromatograph/mass spectrometer (GC/MS) (Agilent Technologies, Santa Clara, CA, USA).

CS₂, n-hexane, petroleum ether, benzene, methanol, acetone, and THF were commercially available analytically pure reagents, which were refined using a rotary evaporator before use. Solvents were purchased from Sinopharm (Beijing, China).

2.2. Experimental Raw Materials

Two coal samples, Shenfu Dongsheng coal (SF) and Inner Mongolia Huolinguole coal (HLGL), were used in this study. The raw coal was crushed to below 200 mesh (74 μm), dried in a vacuum drying oven at 80 °C for 24 h, kept under vacuum and cooled to room temperature, and taken out and stored in a nitrogen-filled desiccator for backup. The coal quality analysis results of the coal samples were as follows: Shenfu coal: M_ad 5.33%, A_ad 6.32%, V_daf 30.74%, FC_ad 57.61%, C_daf 79.82%, and H_daf 4.73%; and Huolinguole coal: M_ad 3.56%, A_ad 21.65%, V_daf 45.93%, FC_ad 28.86%, C_daf 69.87%, and H_daf 5.04%.

M_ad: Air-dry basis moisture; A_ad: Air-dry basis ash; V_daf: Dry and ash-free basis volatile fraction; FCad: Air-dry basis fixed carbon; C_daf: Dry and ash-free basis carbon; H_daf: Dry and ash-free basis hydrogen.

2.3. Graded Extraction of Coal Samples

Weighed 10 g of coal sample and extracted with modified Soxhlet extractor; the solvents used were CS₂, n-hexane (petroleum ether), benzene, methanol, acetone, tetrahydrofuran, and THF/methanol mixture in that order. The extraction temperature was not higher than the boiling point of the solvent, and the extraction time was 90 h. The extracted residue of each level was dried under vacuum at 80 °C for 2 h, and then the next level of extraction was carried out until all seven levels of extraction were completed, as shown in Figure 1. The extracts of each stage were obtained after removing the solvent, and a total of 14 groups of extracts were obtained for the two coal samples, which were recorded as SF-CS₂, HLGL-CS₂, SF-HEX, HLGL-PE, SF-BEN, HLGL-BEN, SF-ME, HLGL-ME, SF-ACE, HLGL-ACE, SF-THF, HLGL-THF, SF-THF/ME, and HLGL-THF/ME.

SF-CS₂: CS₂ extract of Shenfu coal; HLGL-CS₂: CS₂ extract of Huolinguole coal; SF-HEX: n-hexane extract of Shenfu coal; HLGL-PE: petroleum extract of Huolinguole coal; SF-BEN: benzene extract of Shenfu coal; HLGL-BEN: benzene extract of Huolinguole coal; SF-ME: methanol extract of Shenfu coal; HLGL-ME: methanol extract of Huolinguole coal; SF-ACE: acetone extract of Shenfu coal; HLGL-ACE: acetone extract of Huolinguole coal; SF-THF: THF extract of Shenfu coal; HLGL-THF: THF extract of Huolinguole coal; SF-THF/ME: THF and methanol mixture extract of Shenfu coal; and HLGL-THF/ME: THF and methanol mixture extract of Huolinguole coal.

The solvent extraction rate η per stage was calculated by the following equation:

$$\mathsf{\eta }=\frac{100\times { \mathrm{W}}_1}{\left(100-{ \mathrm{A}}_{\mathrm{ad}}-{ \mathrm{M}}_{\mathrm{ad}}\right)\times \mathrm{W}/100}$$

(1)

where η denotes extraction rate, %; W₁ is the weight of the extract, g; W is the weight of the original coal sample, g; A_ad denotes the air-dry basis ash of coal, %; M_ad denotes air-dry basis moisture of coal, %.

2.4. Analytical Methods

GC analysis: HP 6890 gas chromatograph, HP-1 (Methyl Siloxane) capillary column (30 m × 250 m × 0.25 m); FID detector; nitrogen as carrier gas, flow rate 1.0 mL/min, split ratio 20:1; ramp-up procedure: 100 °C (2 min), 10 °C/min to 300 °C (15 min). Quantification by external standard method.

GC/MS analysis: HP 6890/5973 gas chromatograph/mass spectrometer, HP 190915-433 capillary column (30 m × 250 m × 0.25 m); nitrogen as carrier gas, flow rate 1.0 mL/min, splitting ratio 20:1; inlet temperature 300 °C; EI source, ionization voltage 70 eV, ion source temperature 230 °C. The mass scan range was 30~500 amu.

Compound identification: The structure of the identified compounds was determined by computer search against the mass spectral data of the compounds in the NIST library according to the PBM method and by confidence or similarity; for compounds difficult to identify in the library, the structure of the compounds was determined by GC retention time, major and characteristic ion peaks, molecular weight, etc., compared with the literature chromatography and mass spectrometry data, and supplemented by GC/FTIR data. The structures of the compounds were determined by GC/FTIR data.

3. Results and Discussion

3.1. Coal Extraction Rate

Table 1. Extraction rate and extractant composition of Shenfu coal in graded extraction.

Extractant	δ/(cal·cm−3)1/2	Polarity Parameters	Extraction Rate/%	Aliphatic Hydrocarbon Content/10−4 g	Aromatic Content/10−4 g	Heterocyclic Substance Content/10−4 g
CS2	10	0.15	0.7160	32.12	105.51	13.63
n-hexane	7.3	0.06	0.2830	2.4141	0.1318	0.076
Benzene	9.2	3	0.3067	1.795	0.91	0.889
Methanol	14.5	6.6	0.6247	0.259	0.054	4.838
Acetone	9.9	5.4	2.7197	7.275	8.166	14.176
THF	9.5	4.2	7.3968	1.485	0.583	0.22
THF and methanol	12	5.4	0.5142	0.7409	0.5815	0.1452

shows the graded extraction rates of Shenfu coal. The extraction rates of the coal at all levels were affected by the solvent polarity size and solubility parameters, and the extraction rates of polar solvents were higher than those of nonpolar solvents, mainly because the more similar the polarity of the solute and solvent, the stronger the intermolecular gravitational force was relatively and the more favorable the dispersion of the solute in the solvent and thus of the separation. For Shenfu coal, CS₂, acetone, and THF extracted better. Among them, CS₂ extracted about 70% of the aliphatic hydrocarbons, 91% of the aromatic hydrocarbons, and 40% of the heteroatomic compounds in the coal, which is because CS₂ is the first stage of extraction and it could not effectively separate the different components in the coal. Acetone extracted 42% of the heteroatomic compounds.

The solubility parameter δ can be an important parameter to measure the mutual solubility between two substances and is widely used in solvent extraction [19]. The closer the δ of the selected solvent and solute, the higher the mutual solubility of the two. The extractants with different δ were selected for the extraction of Shenfu coal, and the trend of the extraction rate with δ is shown in Figure 2. From Figure 2 the extraction rate of Shenfu coal in the range of the solubility parameter of 9–10 could be maintained above 7%, and the maximum extraction rate was reached at δ = 9.5, while the extraction rate of the solubility parameter in the ranges of 7–9 and 10–15 was only about 0.4%. This indicates that not the higher solubility parameter is better, beyond the optimal solubility parameter, and the extraction rate will drop rapidly.

Table 2. Extraction rates and composition of extractables in graded extraction of Huolinguole coal.

Extractant	δ/(cal·cm−3)1/2	Polarity Parameters	Extraction Rate/%	Aliphatic Hydrocarbon Content/10−4 g	Aromatic Content/10−4 g	Heterocyclic Substance Content/10−4 g
CS2	10	0.15	1.0988	23.78	53.889	13.63
Petroleum ether	16.1	0.01	0.0253	0.0466	0	0.014
Benzene	9.2	3	0.0440	0.016	0.101	1.187
Methanol	14.5	6.6	2.1682	0.126	0	1.348
Acetone	9.9	5.4	0.5201	0	0	144.952
THF	9.5	4.2	0.9214	0	0	32.571
THF and methanol	12	5.4	1.6829	0	0	6.5115

shows the graded extraction rates of Huolinguole coal. The comparison shows that the total extraction rate of Shenfu coal was greater than that of Huolinguole coal, which was 12.5611%. CS₂, methanol, and THF/methanol mixture had higher extraction rates for Huolinguole coal. The better extraction of both coals by CS₂ may be due to the fact that both coals contained more small molecules [20], and CS₂ was the first stage of extraction. By observing the composition of the extracts, it was found that CS₂ extracted about 99% of the aliphatic hydrocarbons and 99.8% of the aromatic hydrocarbons from the coals. Acetone extracted 72% of the heteroatomic compounds, and THF extracted 16.3% of the heteroatomic compounds. At the third stage of extraction, the aromatic hydrocarbons were already extracted completely, while the aliphatic hydrocarbons were extracted completely at the fourth stage. Meanwhile, when the aromatic and aliphatic hydrocarbons were extracted completely, the content of the heteroatomic compounds increased rapidly.

Figure 3 shows the variation trend of the extraction rate with δ. From 0, the extraction rate of Huolinguole coal could be maintained above 2% in the range of a solubility parameter of 14–15, and the maximum extraction rate was reached at δ = 14.5. When the extractant polarity was 5.4, a large amount of heteroatomic compounds with medium polarity were extracted for medium polarity substances. Among the different extractants of the two coals, the polarity of CS₂ was not consistent with the content of the components it extracted and other solvent results, and its polarity parameter was only 0.15, but many compounds could be extracted. The reason is that polarity was not the main factor affecting the extraction rate of CS₂; its own structure and composition were the key factors determining its extraction [21].

3.2. Analysis of the Family Composition of GC/MS Results of Coal Extracts

As shown in Figure 4, the family composition analysis was performed on the GC/MS results of 14 groups of coal extracts to obtain information on aliphatic hydrocarbons, aromatic hydrocarbons, oxygenated compounds (alcohols, aldehydes, esters, phenols, ketones, carboxylic acids, ethers, furans, and pyrans), nitrogenous compounds (amides and carbazoles), sulfur-containing compounds (thiophenes), and other compounds.

Figure 4 shows that both Shenfu coal and Huolinguole coal had no aliphatic hydrocarbons in the THF and THF/methanol extracts. Furthermore, Huolinguole coal also had no aliphatic hydrocarbons in acetone. The content of olefins, chain alkanes, and cycloalkanes in the CS₂ extracts of both was relatively high, and the content of aliphatic hydrocarbons in the other solvent extracts was small in comparison, which proved that a large number of aliphatic hydrocarbons were extracted in the first stage of extraction. The aromatic compounds of Shenfu coal were higher in the CS₂ and acetone extracts, while in the extracts of Huolinguole coal, they were only present in the CS₂ and benzene extracts, and the content of the CS₂ extracts was extremely high. The aromatic compounds in the solvent extracts were dominated by dicyclic aromatic hydrocarbons and tricyclic aromatic hydrocarbons, except for the CS₂ extracts, which contained more polycyclic aromatic hydrocarbons. Shenfu coal oxygenated compounds were present in each type of solvent extracts, but were relatively low in the n-hexane, THF, and THF/methanol extracts. Among them, the phenols were relatively high in the CS₂ extracts, which was due to the strong nucleophilicity of CS₂ and the easy solubilization of the phenols. In contrast, Huolinguole coal had less oxygenated compounds in the petroleum ether, benzene, and methanol extracts. Since diacetone alcohol is like acetone in nature and has better solubility, the acetone extracts contained more than 99% of the two alcohols diacetone alcohol, which could extract the alcohols better. The content of nitrogen- and sulfur-containing compounds in Shenfu and Huolinguole coals was low, and the structure of the elemental analysis was approximately the same as in the previous paper.

3.3. Principal Component Analysis of GC/MS Results of Coal Extracts

The principal component analysis method (PCA) is one of the most widely used algorithms for data dimensionality reduction. The PCA method is used to analyze GC/MS data to determine how to cluster and find outlier samples. After doing z-score normalization of the data, 14 extracts were analyzed using the PCA method, and the first principal component was determined to be olefins, the second principal component to be chain alkanes, and the third principal component to be aromatics, which included monocyclic aromatics, bicyclic aromatics, and polycyclic aromatics. The first principal component explained 48.2% of the variance; the second principal component explained 27.0% of the variance, and the third principal component explained 9.3% of the variance. The principal component scores were plotted with the first, second, and third principal components as the axes, as shown in Figure 5. The figure shows that CS₂ and acetone could be better distinguished from the extracts of the other five solvents, and the differences between the extracts were large.

3.4. Clustering Analysis of GC/MS Results of Coal Extracts

Systematic cluster analysis (HCA), also known as hierarchical cluster analysis, is one of the most commonly used methods of cluster analysis. HCA is calculated by starting with one data point as a cluster, and then for each cluster, merging the clusters based on the same criteria. This calculation is performed until finally only one cluster is left, and the hierarchy of clusters is found. Because the principal component analysis only explains 84.5% of the overall variable information and does not fully reflect its classification, it is necessary to perform a systematic cluster analysis. Using the class averaging method in systematic cluster analysis, the GC/MS results of the two coal extracts were analyzed by clustering tree diagram using the Euclidean distance to obtain the clustering of the family components based on their contents. As shown in Figure 6, the 190 compounds detected were classified into 14 groups, from left to right, aliphatic hydrocarbons, aromatic hydrocarbons, phenols, aldehydes, alcohols, esters, ketones, carboxylic acids, ethers, furans, pyrans, sulfur-containing compounds, other compounds, and nitrogen-containing compounds.

The GC/MS results of the two coal extracts were analyzed by clustering heat map using the Euclidean distance using the sum of squares of deviations method in a systematic cluster analysis to obtain the clustering of the family components based on their contents. As shown in Figure 7, the different colors in the clustering heat map represent the content of each family component, and the color from light to dark indicates that the content of the family component gradually increases. It can be seen from the Figure that the CS₂ extracts of the two coal samples contained the most aromatic hydrocarbons, mainly polycyclic aromatic hydrocarbons, such as naphthalene, anthracene, pyrene, phenanthrene, and their derivatives and a small number of thick-ringed aromatic hydrocarbons, among which SF-CS₂ contained more bicyclic, tricyclic, and tetracyclic aromatic hydrocarbons, while HLGL-CS₂ contained more bicyclic and tricyclic aromatic hydrocarbons. The tree diagram above the thermogram shows that the 14 groups of coal extracts were clustered into six groups by the similarity in the types and contents of the soluble components of organic matter contained, and the five groups SF-HEX, HLGL-PE, SF-BEN, SF-THF, and SF-THF/ME were clustered into one group because they contained a large number of aliphatic hydrocarbons in addition to small numbers of aromatic hydrocarbons. The three groups SF-ME, HLGL-THF, and SF-ACE were grouped together because they contained many ketones. HLGL-ME and HLGL-THF/ME were grouped together because they contained many esters. HLGL-ACE was grouped separately because it contained many alcohols. The above results reflect that polar solvents were more effective in the extraction of oxygenated compounds [22], which was because solvents, such as methanol and acetone, could break stronger hydrogen bonds along with van der Waals forces and weaker hydrogen bonds, so the content of oxygenated compounds in their extracts will be relatively high. HLDL-BEN was clustered into a separate group because of the large number of nitrogenous compounds contained in it.

4. Conclusions

The GC/MS analysis of 14 extracts from Shenfu and Huolinguole coals was performed, and the results were classified by family composition to determine that the composition of compounds in the extracts was mainly three major groups: aliphatic, aromatic, and heteroatomic compounds. The clustering analysis of GC/MS data using Origin showed that CS₂ had a better extraction effect on the aromatic hydrocarbons, and most of them were two- to three-ring polycyclic aromatic compounds; n-hexane and petroleum ethers had a better extraction effect on aliphatic hydrocarbons; benzene had a good extraction effect on nitrogen-containing compounds; acetone had a greater extraction rate on alcohols; THF had a more obvious extraction effect on ketones; tetrahydrofuran and methanol mixed the extraction rate of tetrahydrofuran, and methanol was larger for esters; methanol was more effective for oxygenated compounds, which could be attributed to the difference of solvent polarity. The comparative analysis of PCA and HCA can clearly show the enrichment and separation effects of different extractants on soluble organic components in coal, which provides some reference for the further selection of target compound families in coal for separation.