As a major climate forcing factor, CO₂ mainly generated from fossil fuel combustion and coal-fired power stations are the largest source of CO₂ emissions (Cui et al. 2021; Leng et al. 2021; Liu et al. 2021). At present, controlling the emission of CO₂ is a key challenge for China to achieve the peak carbon in 2030. Oxy-fuel combustion technology, which can capture CO₂ from power plants, has become a hot topic in recent years (Zhang et al. 2021; Yang et al. 2021). Sun et al. (2012) proposed a system of Oxy-Coal Combustion Steam System (OCCSS) of near-zero emissions, which has high net power generation efficiency and near zero CO ₂ emission. In OCCSS system, the fuel (ash-less coal) consumed by reacting with O₂ and steam in high pressure has to be pre-processed through demineralization. However, after demineralization treatment, the organic matter and chemical structure of demineralized coal will be altered to a certain extent, making the reactivity of which are quite different from that of the raw coal.

Pyrolysis is the first step in coal conversion and utilization, to what extent the pyrolysis proceeds depend on the chemical structure of the coal and the reaction conditions. Raman, FTIR, XRD, XPS etc. were commonly applied to investigate chemical structural characteristics of coal before and after acid treatment (Lin et al. 2017; Shi et al. 2013; Song et al. 2016; Ahmed et al. 2003; Li et al. 2019; Sonibare et al. 2010; Gómez-Serrano et al. 2003). Li et al. (2019) reported that the content of oxygen-containing functional groups in raw coal reduced through acid treatment by Raman. Lin et al. (2017) found that by HF-HCl combined acid treatment, the long aliphatic chain in coal was destroyed, the content of carboxyl group and phenolic hydroxyl group increased, the content of aliphatic and aromatic hydrogen decreased by FTIR. In order to study reactivity of coal before and after demineralization, a large number of pyrolysis experiments by thermogravimetric have been carried out (Slyusarskiy et al. 2017; Mandapati and Ghodke 2021). Cheng et al. (2020) found acid treatment did not dramatically alter the main devolatilization behavior of coals in a system that coupled DSC with a thermal gravimetric analyzer and a mass spectrometer. Zhu et al. (2019) reported that acid pretreatment increased the weight loss of lower rank coal with a thermogravimetric analysis. Liu et al. (2017a, b) found that acid treatment increased the pyrolysis reactivity of demineralized coal by both TGA and fixed bed experiments. However, Song et al. (2020) found that the pyrolysis reactivity of demineralized coal decreased on the contrary. In addition, some studies (Song et al. 2016; Cheng et al. 2019) indicated that the existence of minerals in ash is the main reason for the high reactivity of coal without demineralization, while the transformation and effect of inherent minerals in coal are relatively complex and still controversial (Li et al. 2017). Liu (2004) investigated that inherent minerals in coal did not have significant effect on the pyrolysis characteristics of coal. Qiu et al. (2014) found that the presence of alkaline earth metals reduced the pyrolysis reactivity of raw coal instead.

Based on previous studies, this paper focuses on the chemical structure of LPS-R and LPS-D (especially the functional groups and carbon structure), their pyrolysis characteristics and kinetic mechanism of pyrolysis, intended to describe the association between them, to better understand the effect of demineralization on coal pyrolysis reactivity. The influence of acid treatment on functional groups and carbon structure of LPS coal was discussed in this paper by FTIR and ¹³C-NMR. Thermogravimetric analysis was used to study the pyrolysis of coal before and after demineralization at three heating rates of 20, 40 and 60 °C/min, respectively. To evaluate the role of acid treatment on coals’ structure and reactivity, the pyrolysis kinetics was analyzed by iso-conversional methods like Starink, FWO and DAEM.

2.1 Samples preparation

LPS-D was obtained from LPS-R through HCl-HF-HCl three-stage acid treatment. The detailed description of specific acid treatment procedure was given elsewhere (Zhao et al. 2018). Both LPS-R and LPS-D pulverized, grounded and sieved to a particle size smaller than 75 μm were dried in a thermostat at 105 °C for 6 h. Table 1 showed the proximate and ultimate analysis of LPS-R and LPS-D, from which it can be seen that the ash content (dry basis) was significantly reduced from 15.62% (LPS-R) to 2.46% (LPS-D) after acid treatment. The mass fractions of H, N, and S (dry ash free basis) were nearly unchanged and the mass fraction of C (dry ash free basis) increased greatly after acid pretreatment.

2.2 FTIR and solid-state ¹³C-NMR

The functional groups in samples were characterized by a German Bruker VERTEX 80 V Fourier transform infrared spectrometer, with a scanning range of 350–4000 cm⁻¹ and a scanning number of 32. In sample preparation, firstly taking 2 mg pulverized coal and KBr to mix at a mass ratio of 1:100, and then grinding the mixed sample to powder form.

The solid-state ¹³C-NMR detection were performed by AVANCE III-600 MHz NMR fully digital superconducting NMR instrument from Bruker, Germany. The experiment was performed by using a 4 mm MAS probe with a magic angle rotation of 14 kHz and a resonant frequency of 150.9 MHz. The cross-polarization contact time was 2 ms and the cycle delay time was 6 s.

Peak Fit version 4.12 software was used to divide the absorption peaks obtained by both FTIR and ¹³C-NMR. By analyzing the second derivative of spectral data, the approximation of the position and number of fitting peaks was obtained. The residual sum of squares between the original spectrum and the fitting spectrum was considered as the minimum objective function for fitting, and the fitting spectrum was continuously optimized by adjusting the parameters of the fitting peaks.

2.3 TGA experiments

Pyrolysis of the coals was conducted on a TGA/SDTA851e thermogravimetric analyzer. The pyrolysis process was carried out in 99.999% high purity Ar atmosphere with a flow rate of 80 mL/min. The coal samples were heated from 25 °C to 1000 °C at 20, 40 and 60 °C/min, respectively, and the weight of the coals used in each experiment was about 26 mg. When 1000 °C is reached, a cooling ramp was initiated at 20 °C/min in the program to restore the system to room temperature. A group of empty crucible control experiments were carried out in each group of working conditions. Do at least two parallel experiments.

3.1 FTIR analysis

Figure 1 shows the infrared spectra of LPS-R and LPS-D by FTIR, from which it shows that the kaolin absorption peak located at 3700–3600 cm⁻¹ (Fig. 1a) in LPS-R almost disappeared after acid treatment and the absorption peak intensity caused by minerals of 600–400 cm⁻¹ (Fig. 1f) was notably reduced, indicating that acid treatment can remove a large number of ash such as clay and silicate in the coal. In addition, to distinguish the difference between LPS-R and LPS-D spectra, curve fitting method was adopted for analysis, and the spectrum was divided into four parts: aliphatic functional groups of 3000–2800cm⁻¹ (Fig. 1b), oxygen-containing functional groups of 1750–1390 cm⁻¹ (Fig. 1c) and 1390–960 cm⁻¹ (Fig. 1d), and aromatic functional groups of 920–720 cm⁻¹ (Fig. 1e). The curve-fitted results of the infrared spectra of LPS-R and LPS-D are shown in Fig. 2, where the correlation coefficient R² of each region are > 0.995. The functional group distribution and corresponding content of the two coals are shown in Table 2.

FTIR spectra of LPS-R and LPS-D. a 3700 – 3600 cm⁻¹ b 300 –2800 cm⁻¹ c 1750–1390 cm⁻¹ d 1390–960 cm⁻¹ e 920–720 cm⁻¹ f 600 –400 cm⁻¹

Curve fittings of LPS-R and LPS-D sample’s infrared spectrum in different wavenumber bands

According to the relative contents of aliphatic structure fitting by peaks in Table 2, absorption peaks caused by CH₂ stretching vibration in LPS-R and LPS-D accounted for 80.34% and 78.85%, respectively, suggesting that the lipids in LPS-R and LPS-D were mainly in the form of long chains, with relatively few branched and side chains. After acid treatment, the proportion of symmetrical and asymmetric stretching vibration of –CH₂ in LPS-D decreased by 1.49%, indicating that the long aliphatic chain was destroyed and shortened slightly. The peak at around 1695 cm⁻¹ was attributed to the stretching vibration of carboxylic acid, and the absorption strength increased by 13.04% after acid treatment, revealing that the alkali metal elements linked to COO– could be removed, forming COOH group. The peaks near 1600, 1565 and 1500 cm⁻¹ were attributed to the conjugated extension vibration of aromatic C=C. The relative content of aromatic C=C reduced by 0.01% merely, indicating that the aromatic C = C skeleton was stable enough which was not damaged easily by acid treatment. At 1100–1350 cm⁻¹, the absorption vibration peaks were mainly caused by C–O and C–OH stretching vibration of ether oxygen and phenolic hydroxyl. The corresponding contents of hydroxyl increased from 22.28% to 55.93% and the contents of ether oxygen increased from 25.5% to 42.01%, due to the removal of a large number of alkali and alkaline earth metal associated with oxygen and hydroxyl groups by acid treatment. Additionally, the absorption peaks of silica-alumina minerals at 1010 and 1038 cm⁻¹ almost disappeared after acid treatment. In LPS-R, benzene ring 3 substitution (accounting for 73.16%) was the dominant type of aromatic hydrocarbons, while benzene ring 4 substitution (accounting for 73.42%) was dominant in LPS-D after acid treatment, suggesting that the structure of aromatic hydrocarbons was altered by substitution reaction to a certain extent during acid treatment.

In order to better understand the chemical structure of LPS coal before and after acid treatment, the infrared structural parameters in Table 3 were analyzed to characterize the carbon skeleton structure of coal according to the intensity of fitting peaks. The aromatic carbon ratio f_a is calculated as follows:

In Eq. (1), C_al/C is the ratio of aliphatic carbon to total carbon, H_al/H is calculated by Eq. (2), which refers to the ratio of aliphatic hydrogen to total hydrogen content, H/C atomic ratio is calculated by ultimate analysis, and H_al/C_al takes the empirical value of 1.8.

Table 4 shows the infrared structural parameters of LPS-R and LPS-D referred in Table 3. The aliphatic hydrogen content I₁ of LPS-D decreased by 18.1%, while the structurally stable aromatic hydrogen content I₂ increased by 74% compared to that of LPS-R; The aliphatic hydrocarbon branch chain ratio (CH₃/CH₂) of LPS coal has little difference before and after demineralization, indicating that acid treatment has weak effect on the aliphatic side chain; C’ and Doc increased by 62% and 34% respectively, which suggests that the maturity of coal quality and the degree of aromatic ring polycondensation were promoted after acid treatment; The f_a increased by 11.5%, revealing that the aromatic structure increased; The oxygen enrichment degree (C=O/C–O) of LPS coal increased from 0.088 to 0.228, suggesting the C–O bond decreased and the C=O bond increased relatively; The A_ar/A_al of demineralized coal increased by 58%, revealing that acid washing treatment may lead to the formation of aromatic hydrocarbons, which were more stable. In general, the structural stability of LPS coal was enhanced to a certain extent after acid treatment on the basis of the infrared parameters.

3.2 ¹³C-NMR analysis

Different carbon skeleton structures were investigated by solid state ¹³C-NMR studies (Qian et al. 2014; Jing et al. 2019; Kawashima et al. 2000). According to the ¹³C-NMR spectra of LPS-R and LPS-D in Fig. 3, both of them contain two main peak clusters, namely, which represent aliphatic carbon (0–90 ppm) and aromatic carbon (90–165 ppm). Meanwhile, they also contain a small amount of carbonyl carbon with chemical shift of 165–220 ppm. To clearly figure out the influence of demineralization on carbon skeleton structure, ¹³C-NMR spectra were decoupled by peaks to analyze the structural parameters of LPS-R and LPS-D, which was shown in Table 5. LPS-R is mainly composed of aliphatic carbon and aromatic carbon, accounting for 38.72% and 55.84% respectively, the same as that of LPS-D, with the content of aliphatic carbon and aromatic carbon of LPS-D changed little by 1.18% and 1.47% respectively after acid treatment, suggesting that the macromolecular structure of LPS coal was not altered greatly. The total content of oxygen-containing functional group structures in LPS-R mainly consisted of oxy-aliphatic carbon f_al^O = 3.39%, oxy-aromatic carbon f_a^P = 5.76%, carboxyl and carbonyl carbon f_a^C = 6.72%, which decreased from 15.87% to 8.60% due to the protonation of oxygen-containing functional groups by acid treatment process. Based on the curve-fitted ¹³C-NMR spectra results in Table 5, six lattice parameters were calculated and the results were summarized in Table 6. Seen from it, the average ratio of bridge carbon to peripheral carbon of aromatic compounds X_BP, which reflects the polycondensation degree of aromatic compounds, that is, the size of aromatic core, increased from 0.22 to 0.31 after acid treatment, indicating that demineralization process made the aromatic carbon condense to a greater extent, so as to increase the size of aromatic clusters. The number of C_a, C_al, C_cl and C_p increased by 45.6%, 37.6%, 41.9% and 27.4%, indicating that the LPS-D has larger cluster structures than LPS-R to a certain extent.

¹³C-NMR spectra and the de-convoluted curves

3.3 Thermogravimetric analysis.

Figure 4 shows the TG-DTG curves of LPS-R and LPS-D pyrolysis experiment by TG at three different heating rates. The characteristic temperature of coal pyrolysis are defined as follows: the initial pyrolysis temperature T₀ is defined at the intersection point of X axis and the line of sample conversion of 0.05 and 0.50 on the TG curve(Yan et al. 2019; Wang et al. 2016), and the peak temperature corresponding to (dw/dt)_max is defined as T_p. The pyrolysis characteristic parameters of LPS-R and LPS-D are shown in Table 7. With the increase of heating rate, T_p of LPS-R coal shifted from 457.15 °C (20 °C/min) to 482.75 °C (60 °C/min), resulting in thermal hysteresis (Wang et al. 2022) in DTG curve. Based on Fig. 4 combined with Table 7, the TG curve decreases along with the increasing of temperature, and at 1000 °C, the mass weight loss of LPS-R or LPS-D at different heating rates are both close to the contents of their respective volatiles in Table 1 proximate analysis (LPS-RAW: 38.9%, LPS-D: 42.09%).

Pyrolysis characteristic curve of LPS-R and LPS-D

Taking the heating rate of 40 °C/min as an example, as can be seen from Table 7: the (dw/dt)_max of LPS-R is 2.68%/min, and the (dw/dt)_max of LPS-D is 2.54%/min. According to FTIR and ¹³C-NMR analysis, the aliphatic hydrogen content I₁ and the content of aliphatic carbon of LPS-R are 0.149 and 38.72%, higher than that of LPS-D respectively. Therefore, the covalent bond breaking between aliphatic carbon in LPS-R can generate more small molecular tar fragments quickly at T_p, contributing to the result that the (dw/dt)_max of LPS-R is 0.14%/min higher than that of LPS-D. The T_P of LPS-R and LPS-D are 471.89 °C and 473.79 °C respectively. Compared with LPS-D, LPS-R contains more oxygen-containing functional groups (15.87%) with lower bond energy and less aromatic carbon (55.84%) with higher bond energy, which is easy to decompose, so the T_p is about 2 °C lower than that of LPS-D. Furthermore, as a pyrolysis reactivity parameter (Yan et al. 2020), T_p reflects the average stability of the macromolecular structure of coal (Han et al. 2013a, b), which indicates that LPS-R is more likely to fracture in the pyrolysis process, and the pyrolysis reactivity of LPS-R is higher.

As shown in Fig. 5, the pyrolysis process of LPS-R and LPS-D at the heating rate of 40 °C/min is divided into three stages according to the second derivative method (Wang et al.2016), among which the division of stage 1 and stage 2 is where the slope of TG curve changes obviously and the division of stage 2 and stage 3 is where the d²m/dT² changed slightly. The first stage is related to water evaporation and gas desorption in coal pores, 25–222 °C for LPS-R and 25–205 °C for LPS-D. During this stage, the slope of TG curve drops little, and both DTG curve have a small water loss peak near 100 °C, contributed by the evaporation of crystal water in coal. The second stage is regarded as the main pyrolysis stage (LPS-R: 222–693 °C; LPS-D: 296–707 °C), TG curve slope gradually increases, DDTG curve is roughly sinusoidal, during this main pyrolysis stage, part of bridge bonds and side chains with weaker binding energy, oxygen-containing functional groups and macromolecular side chains undergo decomposition reaction successively making a great weight loss peak appeared in DTG curve. The third stage is related to the semi-coke condensation stage (LPS-R: 693–1000 °C; LPS-D: 707–1000 °C), in which pyrolytic coke is formed by polycondensation reaction. On the DTG curve of LPS-R at 780 °C, there is an obvious weight loss peak caused by the decomposition of minerals in ash such as carbonate (Cheng et al. 2019; Shi et al. 2013; Zhu et al. 2018), which almost disappeared on that of LPS-D after acid treatment.

TG, DTG and DDTG curves of LPS-R and LPS-D at heating rate of 40 °C/min

The DTG curve is mainly caused by the cleavage of different covalent bonds in coal (Li et al. 2015). However, since pyrolysis is a process where overlapping reactions take place in series or in parallel, it is necessary to decouple the DTG curve to further intuitively analyze the effect of demineralization on pyrolysis behavior. To correlate the bond cleavage behavior with the coal pyrolysis, the generalized reduction gradient method (GRG) is used to resolve multiple subpeaks of DTG curve (Shi et al. 2013). The peak temperature of each subpeak is regarded as the average reaction temperature of chemical bond breaking, and the peak intensity is a measure of the corresponding mass loss. Figure 6 shows the curve-fitted DTG results and specific parameters are shown in Table 8.

DTG curve of LPS-R and LPS-D fitted by subcurves

In Fig. 6, DTG curve of LPS-R and LPS-D were decoupled into six and five subpeaks according to GRG method (Shi et al. 2013) respectively, among which peak 1 was considered as the release of bound water and decomposition of carboxylic acid (Eskay et al. 1997), and a distinct weight loss peak appeared in DTG curve for both LPS-R and LPS-D at 250 °C. According to FTIR analysis, the proportion of carboxylic acid increased by 13.04% after acid treatment, thus the area of peak 1 for LPS-D was increased by 4.56% in Table 8. Peak 2 was related to the cleavage of covalent bonds formed by aliphatic carbon and heteroatoms on the side chain of aromatic structure, including Cal-O/N/S (Shi et al. 2013; Liu et al. 2008); peaks 3 and 4 were considered for the cleavage of chemical bonds centered on aliphatic carbon and aromatic carbon respectively, including Cal-O, Cal-Cal (Heek and Hodek 1994) and Car-O (Hodek et al. 1991); All peaks 2, 3 and 4 were the main organic devolatilization peaks, among which peak 3 for LPS-R and LPS-D occupied 65.43% and 65.59%, respectively, indicating that the cleavage of chemical bond in aliphatic carbon was the main cause for the weight loss of coal pyrolysis; Peak 5 was mainly caused by carbonate decomposition(Li et al. 2015), which was disappeared in LPS-D after acid treatment; Peak 6 was related to the polycondensation of aromatic rings (He et al. 2015). The evolution temperature of peak 6 for LPS-D was about 100 °C lower than that of LPS-R, and the area was 3.33% larger than that of LPS-R. This was mainly because acid treatment promoted the condensation of the aromatic structure, leading to the early release of volatiles, and similar conclusions were found in the study of Cheng et al (2019).

3.4 Kinetic analysis

The contents of various functional groups of LPS-R and LPS-D vary to some extent on the basis of the former analysis by FTIR and ¹³C-NMR, and the reactivity of coal pyrolysis related to the thermal stability of functional groups can be reflected in the pyrolysis kinetics. According to the recommendations of Kinetics Committee of the International Confederation for Thermal Analysis and Calorimetry (Vyazovkin et al. 2011), as the crucial composition of coal pyrolysis, pyrolysis kinetics of iso-conversional method not only avoid the error brought by the mechanism function selection, making the calculated kinetic parameters more reliable, but also reveal the complex nature of the solid state reaction. Therefore, three iso-conversional methods, Starink (Khan et al. 2015), Flinn-Wall-Ozawa (FWO) (Flynn and Wall 1996; Ozawa 1965; Han et al. 2013a, b) and Distribution Activation Energy model (DAEM)(Miura and Maki 1998) were adopted in this study. The specific formulas are listed in Table 9 and the Arrhenius plots of the left side of the equations versus 1/T under different heating rates at specific conversion were presented in Fig. 7.

Arrhenius plots of LPS-R and LPS-D with Starink, FWO and DAEM method

Figure 8 shows the pyrolysis conversion curves of LPS-R and LPS-D with temperature under different heating rates. It can be seen that when the temperature > 700 °C (corresponding to α ≥ 0.7), the coal pyrolysis is in the stage of semi-coke condensation. As the clusters of aromatic ring structure in coal char gradually become larger, the difficulty of polycondensation reaction also increases, so that the weight loss rate monotonically decreases, resulting in the curve of α-T gradually overlaps and finally shifts to low temperature with the increase of heating rate. At this stage, the linear correlation between ln (β/T^1.92), ln β, ln (β/T²) and 1/T becomes poor due to the overlap of α-T curve, and the description of experimental data by kinetic fitting is likely to have a great deviation. Therefore, the specific parameters of apparent activation energy of LPS-R and LPS-D obtained by the three iso-conversional models at selected conversion from 0.05 to 0.65 are shown in Tables 10 and 11 and the apparent activation energy of LPS-R and LPS-D pyrolysis under different conversion is shown in Fig. 9.

Pyrolysis conversion curve of LPS-R and LPS-D with temperature at different heating rates

Apparent activation energy of LPS-R and LPS-D pyrolysis at different conversion

As seen in Fig. 9, when α < 0.3, since the aliphatic group of LPS-D contains more methyl carbon and methylene carbon, the chemical bonds with lower bond energy are easy to decompose at low temperature, the apparent activation energy of LPS-D is low, which indicates that acid treatment is conducive to promoting the pyrolysis reactivity of demineralized coal, resulting to the initial pyrolysis temperature T₀ of LPS-D lower than that of LPS-R (in Table 7, T₀ of LPS-D at different heating rates is 60–70 °C lower than that of LPS-R); When 0.3 ≤ α ≤ 0.4, the apparent activation energy of LPS coal before and after demineralization is the same; When α > 0.4, it is in the main pyrolysis stage. The analysis of pyrolysis experiment shows that the stability of LPS coal increases after acid treatment, and a higher apparent activation energy is required for the cleavage of chemical bonds during LPS-D pyrolysis; When α = 0.65, the activation energies of LPS-R and LPS-D increased significantly, indicating that stable structures such as aromatic bridgehead carbon (f_a^B) and protonated aromatic carbons (f_a^H) with high bond energy begin to destroy at high temperature. Since the content of f_a^B and f_a^H increased by 5.40% and 5.87% after acid treatment, the apparent activation energy of LPS-D is about 300 kJ/mol higher than that of LPS-R. The arithmetic mean activation energy values of LPS-R and LPS-D are 76 ± 4–463 ± 5 kJ/mol and 84 ± 2–758 ± 12 kJ/mol, respectively.

The fundamental understanding of the pyrolysis characteristics of demineralized coal is considerable for the stable and efficient operation of the Oxy-Coal Combustion Steam System (OCCSS). In present study, the impact of acid treatment on pyrolysis characteristics of coal dependent on its chemical structure were investigated. FTIR and ¹³C-NMR were used to characterize the chemical structure of LPS-R and LPS-D and the pyrolysis experiments were carried out by temperatures-programmed TG pyrolysis. The pyrolysis kinetics of LPS-R and LPS-D were analyzed by using three iso-conversional models. The main conclusions are as follows:

1. (1)

   FTIR shows that demineralization reduces the content of aliphatic hydrogen and C–O, increases the content of aromatic hydrogen and C=O, and improves the maturity of coal, making the chemical structure more stable.

2. (2)

   ¹³C-NMR analysis shows that acid treatment has little effect on the proportion and distribution of aliphatic carbon and aromatic carbon, while LPS-D has a larger cluster structure and polycondensation degree of aromatic compounds due to the demineralization.

3. (3)

   TGA experiments indicate that the weight loss, which is mainly caused by the cleavage of aliphatic covalent bonds and the formation of small molecular fragments according to the curve-fitted DTG results, of LPS-D increased slightly by demineralization treatment, but the pyrolytic reaction at the (dw/dt)_max decreased and the pyrolysis reactivity of LPS-R is higher.

4. (4)

   Kinetic analysis shows that acid treatment is conducive to promoting the pyrolysis reaction when α < 0.3, whereas higher apparent activation energy is required for LPS-D in the semicoke polycondensation stages owing to the more stable chemical structure.