1 Introduction

Disposal of oily sludge (OS) from industrial residues is raising more and more concerns, due to its rapidly increasing amount and potential risk [1]. In China, OS is included in the national hazardous waste list [2]. Generally, OS is considered to be derived from the petroleum extraction and refining industry, and it is a hazardous solid waste with recycling value generated in the process of mining, storage, transportation, and processing [3]. But there are also risks of dealing with OS in other industries, and these industries are often overlooked, such as OS from steel mills.

OS is a complex, high-viscosity, semi-fluid residue, including aged crude oil, wax, resin, asphaltene, and more sediment [4]. Undoubtedly, incineration of OS has advantages such as reduction of sludge volume, destruction of organic pollutants, and energy recovery. Nevertheless, incineration has always been restricted by exhaust emissions [5, 6]. In contrast, pyrolysis can effectively avoid secondary pollution caused by flue gas [7]. Not only that, the pyrolysis reaction is more thorough, less harmful, and economical [8,9,10]. Collectively, pyrolysis technology has great potential for resource utilization of OS.

Some scholars tried to improve the pyrolysis performance of OS by using different additives. Lin et al. [11] pointed out that the participation of rice husk was beneficial to increasing the contents of saturates and aromatics in the pyrolysis oil. Alkali/alkaline metals are also of concern [12,13,14]. It has been certified that KOH could decline the concentration of asphaltenes with more saturated hydrocarbon formation [15]. Furthermore, the catalyst is also an important factor influencing the pyrolysis behavior of OS [16,17,18]. Gong et al. [19] found ferric oxide can accelerate the pyrolysis process and shorten the reaction time. The lattice oxygen in the ferric oxide can induce the cracking reaction of heavy components to generate light oil and carbon dioxide. Currently, natural ore has been widely used to prepare catalysts due to its unique structure and composition [18, 20,21,22]. Ma et al. [23] studied the catalytic effects of dolomite and olivine on the co-gasification of coal and biomass, and they found that two natural ores had a positive influence on tar reforming. It is worth noting that loading nickel on the ore can improve its catalytic performance. Tursun et al. [24] found that NiO/olivine exhibited good in situ catalytic activity for the gasification of pine sawdust. Ni catalysts were reputed to improve hydrogen yield and minimize coke formation [25]. It must be mentioned that more researches have concentrated on the effect of catalysts on product regulation and tar reforming. The influence of ore catalysts on pyrolysis kinetic behaviors has rarely been proposed.

Thermogravimetric analysis (TGA) is a necessary technique for the study of thermal decomposition kinetic behavior [26]. According to the TGA analysis, the kinetic triplet, including the apparent activation energy (\(E_{\alpha }\)), the pre-exponential factor (\(A\)), and the reaction model (\(f(a)\)), can be further determined by different methods [27,28,29,30]. Undeniably, the whole kinetic triplet has advantages in explaining reaction rate changes or reaction control factors [31]. It can also be said that these data may help identify the role of the ore catalysts in the mass transfer behavior of OS pyrolysis. Moreover, it can also provide important information for process parameter optimization, new pyrolysis reactor design, and targeted catalyst synthesis.

Although various noteworthy studies on OS were reported, there was no data to determine the pyrolysis reaction kinetics of OS from steel mills. Trego et al. [32] reported the pyrolysis kinetic parameters of heavy crude oil that can be regarded as “precursors” of OS. But there are differences between heavy crude oil and OS from steel mills. In addition, no relevant researches on the kinetic behaviors of the catalytic pyrolysis of OS by ores were reported. In our previous works [20, 21], calcined olivine, iddingsite, and xiuyan jade (C-OL, C-ID, and C-XY) were first confirmed to be an appropriate load platform. Subsequently, the C-OL (Ni), C-ID (Ni), and C-XY (Ni) prepared by impregnation exhibited good pyrolysis catalytic activity to OS. In this paper, the effects of these mentioned catalysts on the pyrolysis properties of OS were analyzed. Afterwards, mathematical fitting methods were adapted to evaluate the kinetic behavior of the main stages of the OS pyrolysis process. The study aims to provide kinetic data for the realization of high-efficiency and low-cost catalytic pyrolysis of OS.

2 Materials and methods

2.1 Materials

The OS sample used in this experiment was taken from the Baoshan Iron & Steel Co., Ltd., Shanghai. The large impurities in the samples were manually screened. Then the samples were pre-dried in a vacuum drying oven at 105 ℃ for 12 h. Proximate and ultimate analyses were measured according to the Chinese GB/T 212–2008 and GB/T 476–2001. Saturates, aromatics, resins, and asphaltenes (SARA) analysis of OS extract was conducted using the Chinese standard SY/T 5119–2008, and ICP analysis was carried out via Agilent ICP-OES 730. All the results are shown in Table 1. It is obvious that OS from steel mills was rich in Fe. Compared with the OS in Ref. [15], this OS contains more resins and lower saturates and aromatics.

Table 1 The proximate analysis, ultimate analysis, SARA analysis, and ICP-OES analysis of the OS sample
Full size table

In this paper, three natural ore catalysts, OL, ID, and XY, were purchased from Xinjiang Province, China. The same pre-treatment method as in Ref. [20] was adopted to obtain C-OL, C-ID, and C-XY. The catalysts were crushed and sieved to 60 mesh to attain suitable catalyst carriers. The catalysts were calcined at 900 ℃ under an air atmosphere for 4 h. Afterwards, the impregnation method was utilized to load Ni on the ore surfaces. The details were introduced in Ref. [21]. The calcined catalysts were added to the nickel chloride solution (1.96 wt.%) and stirred mechanically for 12 h, and then the obtained precipitates were dried at 105 ℃ overnight and calcined at 900 ℃. The final obtained catalysts were denoted as C-OL (Ni), C-ID (Ni), and C-XY (Ni), respectively.

2.2 Experimental methods

The pyrolysis experiments were performed by non-isothermal thermogravimetry in a thermogravimetric analyzer model STA449-F1 (TA Instruments, manufactured in Germany). Before the thermogravimetric experiments, the OS and the catalysts were blended in a ratio of 5:2. In each experiment, 10 ± 0.5 mg of sample was added into a crucible, followed by heating from room temperature up to 900 ℃ with a heating rate of 5 ℃/min in the nitrogen atmosphere. The gas flow rate was 100 mL/min.

2.3 Kinetic calculation

For variations in chemical components of OS, the entire pyrolysis reaction mechanism cannot be accurately predicted. But a popular formula [33] can be used to describe the overall reaction mechanism,

$$\mathrm{Oilysludge}\;\left(\mathrm{Solid}\;\mathrm{Waste}\right)\xrightarrow{k(T)}\mathrm{Volatile}\;\left(\mathrm{Gas}+\mathrm{Tar}\right)+\mathrm{Char}\left(\mathrm{Solid}\;\mathrm{residue}\right)$$
(1)

This formula can be adapted for different pyrolysis stages. We approximate each stage as the only single-step one. Hence, the rate constant of reaction (\(k\)) can be given by the Arrhenius equation: \(k=A exp\left(-\frac{E}{RT}\right)\) Here, \(k\) is the reaction rate constant, \(A\) is the pre-exponential factor (min−1), \({E}_{\alpha }\) is the apparent activation energy (kJ/mol), \(R\) is the gas constant (8.314 J/(mol·K)), and \(T\) is the absolute temperature (K). The reaction rate (\(d\alpha / dt\)) of pyrolysis of the OS is a linear function of rate constant and reaction kinetic model (\(f(\alpha )\)), and can be written as:

$$\frac{d\alpha}{dt}=k\cdot f(\alpha)$$
(2)

where \(\alpha\) is the conversion rate of sample, which can be calculated by the following equation:

$$\alpha =\frac{\left({m}_{0}-{m}_{t}\right)}{\left({m}_{0}-{m}_{f}\right)}$$
(3)

where \({m}_{0}\) is the initial weight of the sample, \({m}_{t}\) is the mass of sample at time \(t\), and \({m}_{f}\) is the mass of the sample at the end of the pyrolysis reaction. By substituting the Arrhenius equation into Eq. (2), we get,

$$\frac{d\alpha }{dt}=Af\left(\alpha \right)exp\left(-\frac{E}{RT}\right)$$
(4)

For non-isothermal experiments, the temperature is a linear function of time, and the slope is the heating rate, which is a constant, described as \(\beta =dT /dt\). Thus, Eq. (5) and Eq. (6) can be obtained,

$$\frac{d\alpha }{dT}=\frac{A}{\beta }f\left(\alpha \right)exp\left(-\frac{E}{RT}\right)$$
(5)
$$\mathrm{g}\left(\mathrm{\alpha }\right)={\int }_{0}^{\alpha }\frac{da}{f(a)}=\frac{A}{\beta }{\int }_{{T}_{0}}^{T}exp\left(-\frac{E}{RT}\right)dT=\frac{AE}{\beta R}p(x)$$
(6)

Equation (6) is the integral form of the reaction model and \(x=\frac{E}{RT}\) The temperature integral (\(p(x)\)) has no exact analytical solution; therefore, an approximate integral formula can be used instead [34],

$$1n\left[\frac{\mathrm{g}(\mathrm{\alpha })}{{T}^{2}}\right]=1n\left(\frac{AE}{\beta R}\right)-\frac{E}{RT}$$
(7)

The common reaction mechanism models in integral form are shown in Table 2. Kinetic plot between \(1n\left[\frac{\mathrm{g}(\mathrm{\alpha })}{{T}^{2}}\right]\) versus \(\frac{1}{T}\) will give slope and intercepts used for calculation of apparent activation energy \(E\) and pre-exponential factor \(A\).

Table 2 The common kinetic models in integral form for solid-state reactions [35]
Full size table

3 Results and discussion

3.1 Pyrolysis behavior of OS

Figure 1 represents the TG and DTG profiles of the OS pyrolysis at 5 ℃/min from the room temperature to 900 ℃. There are five stages in the pyrolysis process, which is similar to the point in Ref. [19], and the peaks representing each stage have been marked on the DTG curve. The first stage is water evaporation, which occurred up to the temperature of 140 ℃. The mass loss recorded at this stage is 0.956 wt.%. The maximum weight loss rate in this stage was 0.0422%/min, which appeared at 64 ℃. The weight loss in this stage is not obvious, which was attributed to the pre-drying of the samples. The second and third stages were inseparable. In the range of 140 to 345 ℃, the slope of the TG curve is stable. However, the DTG curve showed that this cannot be seen as a single stage. There was a shoulder between 140 and 270 ℃, determined as the second stage. The weight loss in the second stage was 8.161 wt.%. This stage was represented by the DTG peak that reached the maximum at 254.29 ℃, where the highest weight loss rate of this stage was 0.55%/min. The shoulder was related to the priority of light components escape, such as saturates with small molecular mass. The third stage was observed between 270 and 345 ℃ where the weight loss is 8.854 wt.%. The weight loss rate of the third stage of OS pyrolysis was even higher than that of the second stage, indicating the continuation of the light components escape. The maximum DTG peak was recorded at 299 ℃ with a weight loss rate of 0.745%/min, which can be attributed predominantly to the simultaneous crack of heavy components. Specifically, the resin was related to this stage because it contained a large number of labile alkyl chains, which were cracked from 200 ℃ [32]. This phenomenon would continue until the resins were completely converted to coke. The temperature interval of the fourth stage started at 345 ℃ and ended at 510 ℃, where a weight loss of 9.516 wt.% took place. The maximum weight loss rate of this stage was 0.402%/min at 434 ℃. The weight loss of the fourth stage was mainly linked to partial carbon decomposition. In detail, the cracking and polycondensation of asphaltenes were the main theme of this stage. Additionally, the resin was also converted into coke through asphaltenes. The significant weight loss rates at a temperature above 730 ℃ indicated the beginning of a new stage, defined as the fifth stage. The fifth stage continued up to 900 ℃ with a weight loss of 9.745 wt.%. The maximum weight loss rate in this stage was read as 0.426%/min at 799 ℃. The weight loss of this stage can be associated with the decomposition of inorganic minerals. The final residue amount at the end of pyrolysis was 58.653 wt.%, which is significantly greater than the ash content determined by the industrial analysis, which means that there was still more coke remaining in the residue.

Fig. 1
figure 1

TG and DTG curves of OS

Full size image

3.2 Pyrolysis behavior of OS with catalysts

Figure 2a and b illustrate the TG and DTG curves of OS with C-OL and C-OL (Ni) recorded from room temperature to 900 ℃ at a heating rate of 5 ℃/min. Figure 2a presents that the catalysts could effectively inhibit the production of residue. The addition of C-OL and C-OL (Ni) dropped the final residue to 56.79 wt.% and 47.49 wt.%, respectively. Compared with C-OL, the nickel-based catalyst had better performance. It can be seen from Fig. 2b that the maximum weight loss rate of pyrolysis blending with catalysts appeared in advance. The DTG curves demonstrated that the maximum weight loss of pyrolysis with catalysts at the second and third stages was higher than without. However, in the fourth stage, the catalysts reversely promoted carbon decomposition represented by the lower weight loss rates of 0.347%/min and 0.361%/min, respectively. In the last stage, C-OL (Ni) performed significantly for the weight loss rate of 0.575%/min, indicating that nickel could disrupt the process of asphaltene polycondensation to provide more small molecules.

Fig. 2
figure 2

TG and DTG curves of OS with different catalysts. (a) and (b) C-OL/C-OL (Ni); (c) and (d) C-ID/C-ID (Ni); (e) and (f) C-XY/C-XY (Ni)

Full size image

TG and DTG curves of the OS mixed with C-ID and C-ID (Ni) are displayed in Fig. 2c and d. It can be observed from Fig. 2c that C-ID and C-ID (Ni) can also reduce the amount of pyrolysis residues. The pyrolysis residues involved in C-ID and C-ID (Ni) were 56.79 wt.% and 49.52 wt.%, respectively. Figure 2d shows that the DTG curve of C-ID in the second and third stages shifted to low temperature and had a higher rate of weight loss, revealing that it preferentially promoted the escape of light components and the crack of heavy components. But in the later stages, C-ID did not perform a good catalytic effect as a lower maximum weight loss rate in the DTG curve. C-ID (Ni) had a clear advantage in the final stage for the fast-declining TG curve and the high maximum weight loss rate of 0.55%/min. However, in the range of 160 to 510 ℃, the maximum weight loss rate of pyrolysis with C-ID (Ni) is relatively low, only 0.665%/min and 0.31%/min at 263 ℃ and 403 ℃, respectively.

Figure 2e and f have displayed the TG and DTG curves of OS blending with C-XY and C-XY (Ni). XY-derived catalysts presented a similar performance to the others mentioned in this study in reducing pyrolysis residues. With the participation of C-XY (Ni), there was 48.603 wt.% of pyrolysis residues. The DTG curves viewed in Fig. 2f demonstrated that the temperature corresponding to the maximum weight loss rate shifted to left in all stages. In the range of 160 to 345 ℃, the maximum weightless peak of OS with C-XY was recorded as 0.91%/min, which is higher than that of OS with C-XY (Ni) observed as 0.8%/min. The maximum weight loss rates of C-XY and C-XY (Ni) were 0.347%/min and 0.316%/min at 394.19 ℃ and 403.27℃, respectively. In the final area, the weightless peak of pyrolysis of the OS with C-XY (Ni) arrived at 0.56%/min, while that of pyrolysis of the OS with C-XY was 0.362%/min, which is below 0.426%/min represented by without additives.

In general, the catalysts have a significant impact on the pyrolysis process of OS, whether it is loaded with Ni or not, and the Ni-loaded ones are mainly responsible for the changes in the pyrolysis characteristics of the OS in the high-temperature section.

3.3 Kinetic analysis of OS pyrolysis with different catalysts

The main reactions of OS pyrolysis appeared at 140–510 ℃ and 700–900 ℃, which corresponded to stage 2, stage 3, stage 4, and stage 5. The stage was a rough division of the reaction mechanism, which was not suitable to represent the kinetic model. In the OS pyrolysis, the escape of light components and the crack of heavy components were simultaneous behaviors according to the DTG curve. Therefore, from the perspective of the kinetic model, an attempt to build a model that includes stage 2 and stage 3 was appropriate. A simplified kinetic model is a necessary method for kinetic analysis. Figure 3 presents the kinetic curves of the main reaction stages of OS pyrolysis calculated by fitting models. The kinetic parameters of the apparent activation energies (Ea), correlation coefficients (R2), and frequency factors (A) of OS pyrolysis are illustrated in Table 3. Stage 2/3 was defined as the combination of stage 2 and stage 3. It is clear that the reaction mechanisms of the three stages all follow the diffusion model (the stage 2/3 and stage 4 follow D2, stage 5 follows D3), and the correlation coefficients of the stages are all above 0.90, denoting a good fitting effect. The fitting apparent activation energies of all stages were 61.03, 18.07, and 62.92 kJ/mol, respectively. The apparent energy is usually used to estimate the difficulty of the reaction, which represents the lowest effective energy required to break down the chemical bonds. The E of the stage 2/3 was recorded as a higher value. This may be attributed to the cleavage of heavy components consuming more energy. The apparent activation energy of the stage 4 was much lower than the others. In general, resins disrupted the π–π and polar bonding interactions between asphaltene monomers, which results in asphaltenes being broken with low energy [36, 37]. The stage 5 expanded the most energy, which was associated with the complex interaction of coke and ash residues [38]. Iron and its oxides were the most abundant components in the ash residues. In the OS, the iron content was observed as high as 129.8 mg/g, and iron ions were more easily formed, which subsequently attracted nitrogen, sulfur, and polar oxygen functional groups in asphaltenes through coordination bonds, causing the formation of coke, were adsorbed on the ash residues through coordination. The break of this bond costs more energy. To deduce an insight into the reaction chemistry of the pyrolysis of OS, the order magnitude of the pre-exponential factor is a useful reference [39]. The pre-exponential factors of the all stages were 543.78, 0.029, and 10.07 min−1, respectively. In the case of an order of magnitude lower than 109 min−1, the surface reaction is the working pathway [35, 39, 40].

Fig. 3
figure 3

The kinetic curves of the main reaction stages of OS pyrolysis: (a) stage 2/3, (b) stage 4, (c) stage 5

Full size image
Table 3 Fitting parameters of the main reaction stages of OS pyrolysis
Full size table

Table 4 shows the fitting parameters of the main reaction stages of OS pyrolysis with catalysts. The kinetic curves of all samples are illustrated in the Appendix. It is demonstrated that some curves fitted well with the reaction order models (R2 was higher than 0.9). But in fact, these curves also had a similar degree of fit to diffusion models. It is virtually impossible to conclude that the catalysts had changed the reaction mechanism of OS pyrolysis based on the value of R2. In general, the catalysts could lower the total apparent activation energy of OS, except for C-ID (Ni). The E of stages 2/3, 4, and 5 decreased from 61.03, 18.07, and 62.92 kJ/mol to 29.69, 12.33, and 18.46 kJ/mol with the addition of C-OL, respectively. Its performance was conspicuous. But when nickel was loaded on C-OL, the E change was not obvious, especially in stage 2/3, which only dropped by 4.21 kJ/mol. The same performance also appeared in C-ID (Ni) and C-XY (Ni). In the case of C-ID (Ni) participation, the apparent activation energy increased to 65.83 kJ/mol instead. The change of apparent activation energy is closely related to the reaction model. For the diffusion model, the apparent activation energy is positively correlated with the change of the heat and mass diffusion. In other words, for a reaction dominated by a diffusion model, the reaction rate is determined by the heat and mass diffusion rather than the chemical reaction. The ore catalysts had a well-developed pore structure, which was beneficial to the diffusion of light components during the reaction. Nonetheless, when nickel chloride was loaded, part of the pore structure may be blocked, causing the catalyst’s mass transfer performance to be weakened. Therefore, it can be seen that nickel chloride has the greatest restriction on C-ID. It can also be explained that the apparent activation energy of stage 5 loaded with nickel chloride is higher than that without loading. Consequently, it implied that the catalyst did not change the pyrolysis reaction model of OS, and it was still a diffusion model. The addition of C-OL reduced the apparent activation energy of stages 2/3, 4, and 5 to 32.15, 16.67, and 24.84 kJ/mol, respectively. Furthermore, C-XY dropped the apparent activation energy of the three stages to 26.83, 16.21, and 20.81 kJ/mol, respectively. From the perspective of apparent activation energy, C-OL, C-ID, and C-XY had a similar catalytic performance for OS; however, after loading nickel, C-XY(Ni) was prior to the others, which was consistent with experimental results [21]. The addition of catalysts did not significantly change the order of magnitude of the pre-exponential factor of OS pyrolysis (all A were lower than 109 min−1), which indicated that the reaction still took the surface reaction as the path.

Table 4 Fitting parameters of the main reaction stages of OS pyrolysis with catalysts
Full size table

4 Conclusion

In this work, the non-catalytic and catalytic pyrolysis using natural ore catalysts of OS were investigated by thermogravimetric analysis. The results showed that the OS pyrolysis can be divided into five stages: water evaporation, light hydrocarbon escape, the crack of resins (heavy components), asphaltene decomposition, and inorganic mineral decomposition.

On the TG/DTG curves, the selected natural ores did not show obvious differences. The addition of catalysts reduced the mass fraction of OS pyrolysis residue and caused the DTG curves to shift to the left. The ore carriers had a positive effect on the cracking of heavy components in the OS, which was manifested in the prominent DTG peak. The participation of nickel had a significant effect on the formation and decomposition of coke during the pyrolysis of OS.

From the perspective of kinetic analysis, OS pyrolysis mainly included stage 2/3, stage 4, and stage 5, and their apparent activation energies were 61.03, 18.07, and 62.92 kJ/mol, respectively. The magnitude of the pre-exponential factor indicated that OS pyrolysis was a surface reaction. The catalysts could effectively reduce the apparent activation energy of OS pyrolysis. C-OL had the best performance, with apparent activation energies reduced to 29.69, 12.33, and 18.46 kJ/mol, respectively. Consistently, the pyrolysis reaction path of OS was not changed by the catalysts. The study unveils for the first time the kinetic parameters of the pyrolysis of OS from steel mills under nickel-ore–based catalysts, which provides a new perspective for the treatment of OS from steel mills.