Introduction

Although chrysotile fibers have the largest industrial applications due to desirable properties, chrysotile fibers have carcinogenic (Poland and Duffin 2019; Harington 1991; Kamp 2009). Many methods had been developed to reduce the toxicity of chrysotile, and one of these methods was thermal treatment (Fedorocková et al. 2012; Dai et al. 2021; Zaremba et al. 2010). Thermal behavior of chrysotile has been widely studied in the past years (Jolicoeur and Duchesne 1981; Khorami et al. 1984; Datta,Samantaray, and Bhattacherjee 1986; Suquet 1989; MacKenzie and Meinhold 1994; Gualtieri and Tartaglia 2000; Cattaneo, Gualtieri, and Artioli 2003; Candela et al. 2007; Gualtieri et al. 2008; Gualtier, Gualtieri, and Tonelli 2008; Zaremba et al. 2010; Kusiorowski et al. 2012; Huang et al. 2013; Bloise et al. 2016, 2018; Goryainov et al. 2021; Balpanova and Baikenov 2023). Concerning the dehydroxylation of chrysotile, it has been demonstrated that its structure collapse at around 650 °C, water is released according to Eq. (1), thus causing the double-layer structure of chrysotile converted into a disordered intermediate (Zulumyan et al. 2007). The intermediate Mg3Si2O7 in Eq. (1) is defined as dehydroxylate I (MacKenzie and Meinhold 1994). In Eq. (2), dehydroxylate I early recrystallized at about 800 °C into anhydrous silicates such as forsterite (Bloise et al. 2009) and amorphous silica (Cattaneo et al. 2003; Zaremba et al. 2010; Zulumyan et al. 2014).For dehydroxylation of chrysotile, the simplified reaction is Eq. 3.

$${\text{Mg}}_{{3}} {\text{Si}}_{{2}} {\text{O}}_{{5}} \left( {{\text{OH}}} \right)_{{4}} \left( {\text{s}} \right) \to {\text{Mg}}_{{3}} {\text{Si}}_{{2}} {\text{O}}_{{7}} \left( {\text{s}} \right)\, + \,{\text{2H}}_{{2}} {\text{O}}\left( {\text{g}} \right),$$
(1)
$${\text{2Mg}}_{{3}} {\text{Si}}_{{2}} {\text{O}}_{{7}} \left( {\text{s}} \right) \to {\text{3Mg}}_{{2}} {\text{SiO}}_{{4}} \, + \,{\text{SiO}}_{{2}} \left( {\text{s}} \right),$$
(2)
$${\text{2Mg}}_{{3}} {\text{Si}}_{{2}} {\text{O}}_{{5}} \left( {{\text{OH}}} \right)_{{4}} \left( {\text{s}} \right) \to {\text{3Mg}}_{{2}} {\text{SiO}}_{{4}} \left( {\text{s}} \right)\, + \,{\text{SiO}}_{{2}} \left( {\text{s}} \right)\, + \,{\text{4H}}_{{2}} {\text{O}}\left( {\text{g}} \right).$$
(3)

Detailed model for chrysotile dehydration and dehydroxylation mechanisms and high-temperature crystallization are also reported (Khorami et al. 1984; Datta et al. 1986; MacKenzie and Meinhold 1994; Gualtieri et al. 2012; Dlugogorski and Balucan 2014; Alizadehhesari et al. 2012; Trittschack et al. 2014). For example, (Cattaneo et al. 2003) calculated apparent activation energy of the dehydroxylation of chrysotile, the calculated kinetic parameters indicate that the rate-limiting step of the reaction is one-dimensional diffusion with an instantaneous nucleation (zero order) or a deceleratory rate of nucleation, which implies a one-dimensional diffusion of the water molecules formed in the interlayer region by direct condensation of two hydroxyls. (Alizadehhesari et al. 2012) indicated the dehydroxylation of serpentine may best be interpreted as a two-step reaction, with the first step being the nucleation reaction, followed by the three-dimensional outward diffusion of the water molecule. (Trittschack et al. 2014) found that the dehydroxylation kinetics of chrysotile can be subdivided into three different stages and interpreted as interface reaction, one-dimensional diffusion and two-and three-domensional diffusion models, respectively.

However, to the best of our knowledge, the dehydroxylation process of chrysotile fiber membrane (CFM) has not been reported. Here, we discussed the difference between raw chrysotile fiber (CF) and self-made chrysotile fiber membrane (CFM) in the calcination process, the thermal behavior and thermodynamic mechanism of the dehydroxylation of chrysotile have been studied. In the future, self-made chrysotile fiber membrane (CFM) could be calcined within the range of temperature from 600 to 700 °C at an industrial level, and the obtained amorphous Mg–Si–O fiber membrane without the toxicity would be applied in many fields such as dye wastewater treatment membrane.

Materials and methods

Sample preparation and physicochemical analyses

The chrysotile fiber (CF) (single fiber length: 40–70 μm, diameter: about 4–8 μm) (shown in Fig. S1a used in this paper was produced in Qinghai, without any treatment before use. The chrysotile fiber membrane (CFM) (single fiber length: 6–40 μm, diameter: about 240 nm) (shown in Fig. S1b) used in the study was prepared according to the previous work of the research group (Cai et al. 2019). CF and CFM were calcined from room temperature to 500 °C,550 °C,600 °C,650 °C,700 °C and 800 °C for 2 h in a muffle furnace (KSL-1200X,hefei) at a heating rate of 5 °C min−1 in air atmosphere.The X-ray diffraction (XRD) patterns of the samples were collected from a DX-2700 diffractometer equipped with a Cu-Kα radiation source (λ = 0.15405 nm) at room temperature (the voltage and the current: 35 kV/30 mA; the scanning range: 5–40°; step: 0.02°; 0.2 s/step). The morphologies of the samples were observed by scanning electron microscope (SEM) (KYKY-EM2600, BeiJing) (magnification: 2000–5000 x; accelerating voltage: 20 kV; probe current: 2.47 A). The surface functional groups of the samples were analyzed by Nicolet iS50 infrared spectrometer of Thermo Fisher Scientific (wave number: 4000–400 cm−1; resolution: 4; number of scanning: 32).

TG-DSC experiment

Almost, 10 mg fiber sample was placed in an alumina crucible and heated from room temperature to 1000 °C, at five different heating rate of 2.5 °C min−1, 5 °C min−1, 7.5 °C min−1, 10 °C min−1 and 12.5 °C min−1 under air atmosphere at the flow rate of 50 mL min−1 in a simultaneous TG-DSC analyzer (STA449C, NETZSCH, Germany).

Mathematical model development

For the analyses of the TG-DSC data, a mathematical model was developed. In isoconversional method the rate of decomposition of a material is given by

$$\frac{d\alpha }{{dt}} = kf\left( \alpha \right),$$
(4)

where

$$\alpha = \left( {m_{0} - m_{t} } \right)/\left( {m_{0} - m_{\infty } } \right),$$
(5)

where, \(\alpha\) is the conversion rate, \(m_{0}\) is the initial mass, \(m_{t}\) is the change in mass and \(m_{\infty }\) is the residual mass.

Using Arrhenius (Laidler 1984) temperature dependency of K, Eq. (4) is written as

$$\frac{d\alpha }{{dt}} = A\exp \left( { - \frac{E}{RT}} \right)f\left( \alpha \right).$$
(6)

Introducing the heating rate, \(\beta\) (℃ min−1), we obtain Eq. (7)

$$\frac{d\alpha }{{dt}} = \frac{A}{\beta }\exp \left( { - \frac{E}{RT}} \right)f\left( \alpha \right),$$
(7)

where \(A\) is the pre-exponential factor, \(E\) is the activation energy (kJ mol−1), \(R\) is the universal gas constant (8.314 \({\text{J/K}}^{ \cdot }\).mol), \(T\) is the pyrolysis temperature (K).

Now, integrate Eq. (7), we obtain Eq. (8):

$$G\left( \alpha \right) = \int_{{T_{0} }}^{T} {\left( {\frac{A}{\beta }} \right)} \exp \left( { - \frac{E}{RT}} \right)dT,$$
(8)

where, G is the Gibbs free energy (kJ mol−1).

Kinetic and thermodynamics parameters calculation

Kinetic parameters of a thermal reaction are necessary for accurate prediction of pyrolytic behavior and to optimize the process for thermal degradation of CF/CFM. The kinetic and thermodynamic parameters of the sample were calculated by the FWO (Flynn–Wall–Ozawa) method (Flynn and Wall 1966; Ozawa 1965), Starink method (Starink 1996) and Freidman method (Friedman 1964), as described below.

FWO method

The Flynn–Wall–Ozawa method is based on the kinetic equation of Arrhenius non-isothermal and heterogeneous reaction in Eq. (6). The temperature is further integrated and approximately transformed to obtain the following formula:

$$\lg \beta = \lg \left( {\frac{AE}{{RG\left( \alpha \right)}}} \right) - 2.315 - 0.4567\frac{E}{RT}.$$
(9)

The LHS of Eq. (9) was plotted on the y-axis, and inverse of pyrolysis temperature was plotted on the x-axis, for selected \(\alpha\) value to calculate kinetic parameters from the value of the slope.

Kissinger method (Kissinger 1957; Criado and Ortega 1986)

The Kissinger method selects the peak temperature \(T_{P}\) of TG or DSC at different heating rates for calculation. Upon rearranging and taking logarithm of both sides of Eq. (5), we obtained

$$\ln \left( {\frac{\beta }{{T_{P}^{2} }}} \right) = \ln \left( \frac{AR}{E} \right) - \frac{E}{{RT_{P} }}.$$
(10)

Starink method

Starink method is based on Flynn–Wall–Ozawa method Eq. (9) and Kissinger method Eq. (10).

$$\ln \left( {\frac{\beta }{{T^{1.8} }}} \right) = C_{s} - 1.0037\frac{E}{RT},$$
(11)

where, C is a constant.

Now, the LHS of Eq. (11) was plotted on the y-axis, and inverse of pyrolysis temperature was plotted on the x-axis to calculate thekinetic parameters from the value of slope.

Friedman method

Friedman method is based on the Arrhenius equation:

$$\ln \left( {\frac{\beta d\alpha }{{dT}}} \right) = \ln \left[ {Af\left( \alpha \right)} \right] - \frac{E}{RT}.$$
(12)

Similarly, the decomposition rate dα/dT at the same conversion rate α of the curve at different heating rates β was taken, and the plot of ln[β(dα/dT)] to 1/T was plotted. The activation energy E was obtained from the slope, and the pre-exponential factor A was obtained from the intercept.

Results and discussion

XRD analysis and FTIR analysis

Figure 1 showed the XRD pattern of CF and CFM at different calcination temperatures, the marked characteristic were indexed to Mg3Si2O5(OH)4 (JCPDS card No.82-1838) (chrysotile) when the CF and CFM was uncalcined,after calcining at 500 °C and 550 °C for 2 h, the diffraction peaks of chrysotile show a downward trend. Compare with the initial CF/CFM, in the FTIR spectra (Fig. S1),vibration bands of chrysotile including Mg–OH(3685 to 3650 cm−1,606 to 591 cm−1), Si–O–Si (1080 to 950 cm−1), Mg–O (542 cm−1) gradually weakened when the calcination temperature increased from 500 to 550 °C (Kusiorowski et al. 2012).

Fig.1
figure 1

a XRD patterns for CF (0–800 ℃), b XRD patterns for CFM (0–800 ℃)

Full size image

Figure 1b exhibited the XRD pattern of CFM calcined at different temperatures, the amorphous region (600–700 °C) appeared when CFM was calcined to 600 °C, Zaremba indicated that within the range of temperature from 600 to 800 °C chrysotile completely dehydroxylates to form serpentine anhydride, a material void of structure and absent of chrysotile properties(Zaremba and Peszko 2008), it may be due to the network structure of CFM, which makes the contact area too small, resulting in slow mass transfer during the calcination process, instead of forming an observable crystalline forsterite, amorphous dehydroxylate I (Mg3Si2O7) was formed(Bloise et al. 2018; Rim et al. 2020). While there was no amorphous region in CF (Fig. 1a) after calcination. Three vibration bands (Mg–OH, Si–O–Si, Mg–O) gradually weakened or even disappeared in Fig. S1, the new characteristic band at 875 to 840 cm−1 and 606 cm−1 can be assigned to the SiO4 and MgO6 in forsterite (Mirhadi et al. 2016), these results confirmed the formation of forsterite as previously was observed in the XRD pattern (Fig. 1a).

Usually, the temperature at which forsterite begins to appear during the calcination of chrysotile is 800 °C. While, the temperature at which Mg2SiO4 (JCPDS card No.85-1346) (forsterite) appears in CF is earlier than that in CFM, and the diffraction peak of forsterite appear at 650 °C indicating the total destruction of chrysotile, this result is the same as that of Zaremba (Zaremba et al. 2010). While the diffraction peak of forsterite begins to appear in CFM after 800 °C (d’Azevedo et al. 2006), this process conforms to Eq. 2. After calcinated at 800 ℃, the bands related to the characteristic bands of forsterite appeared in the range of 1100 to 980 cm−1 related to Si–O–Si, 875 to 840 cm−1 related to SiO4 stretching, 600 to 670 cm−1 related to SiO4 bending, 475–506 cm−1 due to the MgO6 modes (Kharaziha and Fathi 2009). These results confirmed that forsterite appeared in CF and CFM when the calcination temperature was above 800 °C as show in Fig. 1.

SEM of the samples

Figure 2 showed the SEM images of CF/CFM without calcination and after calcination. As shown in Fig. 2h–l, the CFM fibers at different calcination temperatures maintain a network structure, while the CF after calcination at 600 °C (Fig. 2c–f) chrysotile fibers melt into a large bundle shape, Fig. 2f shows the SEM image of the cross section of the CF fiber after calcination at 800 °C,it can be clearly seen that the single fiber is fused into a large bundle after calcination at 800 °C. The results of SEM are more able to confirm that the XRD phase difference of CF and CFM calcined at different temperatures may be due to the accumulation form of chrysotile fibers, CF is mainly in the form of multiple fiber accumulation winding, while CFM is a single fiber arranged in a network structure.

Fig.2
figure 2

SEM image of a CF, b CF-500, c CF-600, d CF-650, e CF-700, f CF-800, g CFM, h CFM-500, i CFM-600, j CFM-650, k CFM-700, l CFM-800

Full size image

TG/DTG/DSC curve

The TG/DTG curves of CF and CFM in the range of 400–800 °C are shown in Fig. 3 when the heating rate is increased from room temperature to 1000 °C at 2.5, 5, 7.5, 10, 12.5 °C min−1 in air atmosphere. The weight loss of CF/CFM at different heating rate as shown in Table S1.

Fig.3
figure 3

a TG, b DTG curves of the pyrolysis process of CF at different heating rate 2.5 ℃ min−1, 5 ℃ min−1, 7.5 ℃ min−1, 10 ℃ min−1, and 12.5 ℃ min−1, respectively, c TG, d DTG curves of the pyrolysis process of CFM at different heating rate 2.5 ℃ min−1, 5 ℃ min−1, 7.5 ℃ min−1, 10 ℃ min−1, and 12.5 ℃ min−1, respectively

Full size image

As shown in Fig. 3a and b, the DTG curves at different heating rates indicated that there were two stages in the TG curve of CF in the range of 400–800 °C. The temperature range of the first stage was 500–650 °C, and the weight loss rate was about 8.15%. The temperature range of the second stage is 650–700 °C, and the weight loss rate is about 4.15%. DSC (Fig. S2a) also indicated that the dehydroxylation process of CF at 400 to 800 °C can be subdivided into two different stages. The decomposition of other Serpentine minerals study by other scholars(Viti 2010; Teixeira et al. 2013; Bloise et al. 2016; Kusiorowski et al. 2015), the DTG results show that there is only one weight loss stage in the decomposition process of serpentine minerals, which is different from the decomposition process of raw chrysotile (CF) studied by us. The specific mechanism is not clear, pending further in-depth study in the future.

As shown in Fig. 3c and d, the DTG curve of CFM is dominated by a single peak, and the weight loss rate is about 12.27% due the chrysotile dehydroxylation represented by Eq. (1). DSC (Fig. S2b) also showed that CFM has only one stage during pyrolysis at 400–800 °C. The wide peak at 636 ℃ in the DCS curve of CFM is due to the chrysotile dehydroxylation, which is close to the results of previous studies (Viti 2010; Bloise et al. 2016).

The calculation of activation energy (Ea)

Without obtaining the most probable mechanism function in advance, the activation energy (Ea) was obtained by using the Flynn–Wall–Ozawa equation (Eq. 9), the Starink equation (Eq. 11) and Friedman equation (Eq. 12), and the relationship between the activation energy and the reaction process (Ea ~ α) was obtained. The data at 2.5 ℃ min−1, 5 ℃ min−1, 7.5 ℃ min−1, 10 ℃ min−1, and 12.5 ℃ min−1 were selected for calculation.

The stage of CFM selected the data with the conversion rate in the range of 0.3–0.8 for calculation. The linear fit plots for determining activation energy of the CFM calculated by the FWO, Starnk and Friedman, respectively, is shown in Fig. 4a, b and c. The relationship between the activation energy (Ea) and the conversion rate (α) calculated based on the Flynn–Wall–Ozawa equation and the Starink equation is shown in Fig. S4d. The calculation results are shown in Table S4, the average activation energy calculated by Flynn–Wall–Ozawa equation, Starink equation and Friedman equation are 243.33 kJ/mol, 241.69 kJ/mol and 227.81 kJ/mol, respectively. From Fig. 4d, it can be seen that the activation energy results calculated by the FWO method and the Starink method are not much different, the calculation results of Friedman method have a relatively small change, indicating that the Ea calculated by FWO method are reliable.

Fig.4
figure 4

The stage of CFM a Flynn–Wall–Ozawa equation lgβ ~ 1000/T plot, b Starink equation ln(β/T^1.8) ~ 1000/T plot, c Friedman equation ln[β(dα/dT)] ~ 1000/T plot at different conversion rate; d The relationship between activation energy Ea and conversion rate α in weight loss stage of CFM

Full size image

The activation energy calculation results of the two steps of the CF dehydrogenation process are shown in Figs. S3 and S4, the corresponding thermodynamic parameters are shown in Tables S2 and S3. The activation energy required for the first stage of CF dehydrogenation is less than that for the second stage. The activation energy for the dehydrogenation of CFM is between the two stages of CF.

The calculation of the mechanism function

According to the calculation results of the activation energy, the activation energy of CF and CFM tended to be stable during the pyrolysis process, so we used the invariant kinetic parameters’ method (Lesnikovich and Levchik 1983) to obtain the pre-exponential factor A. Substituting the mechanism function in Table S5 into the Coats-Redfern equation (Coats and Redfern 1964), the formula is as follows:

$$\ln \left[ {\frac{G\left( \alpha \right)}{{T^{2} }}} \right] = \ln \left[ {\frac{AR}{{\beta E}}} \right] - \frac{E}{RT}.$$
(13)

The fitting equation is obtained by linear fitting of \(\ln ({\text{G}}(\alpha ){\text{T}}^{2} ) \sim 1/{\text{T}}\) in the equation, according to the slope and intercept of the fitting equation, the corresponding activation energy Ea and pre-exponential factor function lnA are obtained respectively. The calculation results are shown in Tables S6 and S7. The most probable mechanism function can be obtained by comparing the calculated results with the activation energy of CF/CFM calculated by FWO method.

The results are shown in Table 1, the calculation results of CF are similar to the work done by previous researchers(Cattaneo et al. 2003; Alizadehhesari et al. 2012; Trittschack et al. 2014), and the dehydroxylation of chrysotile conforms to both diffusion mechanism and nucleation mechanism. The reason for the difference of the mechanism functions G(α) may be due to the different origin and composition of chrysotile. From the results, it can be seen that the stage of CFM and the second stage of CF both conform to the two-dimensional diffusion mechanism, but the mechanism function G(α) of the two is different. The stage of CFM conforms to the ‘Valensi’ model with mechanism function G(α) = (1−α)ln(1−α) + α,while the second stage of CF conforms to the ‘Jander’ model (n = 2) with mechanism function G(α) = [1−(1−α)1/2]2. It is worth noting that the second stage of CF conforms to the random nucleation and subsequent growth ‘Avrami-Erofeev’ model (n = 3/2) with mechanism function G(α) = [−ln(1−α)]2/3. This just shows that the XRD results (Fig. 1) are reasonable. The reason why CFM does not appear forsterite phase but presents amorphous phase is that the dehydroxylate process of CFM only conforms to the two-dimensional diffusion mechanism. The forsterite appearance of CF is because its pyrolysis process conforms to both two-dimensional diffusion and random nucleation and subsequent growth mechanisms.

Table 1 Reactive kinetic parameters in the stage of the pyrolysis process of CF/CFM
Full size table

Conclusions

In this paper, the different phenomena of calcination of raw chrysotile film (CF) and self-made chrysotile film membrane (CFM) in air were studied. At 500 °C–600 °C and 800 °C, the XRD and FTIR of CF/CFM have little difference, corresponding to the processes of chrysotile → dehydroxylate I and dehydroxylate I → forsterite, respectively. When the calcination temperature is 600 °–700 °C, the XRD of CF and CFM shows obvious difference. CFM is amorphous and corresponds to dehydroxylate I, while CF appears forsterite in advance. Through the DTG curve of CF/CFM, it is found that the dehydrogenation of CFM is a one-step reaction process and the activation energy of CFM dehydroxylate stage is 243.33 kJ/mol. While the dehydroxylate process of CF is divided into two stages, the activation energy of stage 1 is 316.04 kJ/mol, the activation energy of stage 2 is 316.04 kJ/mol. Finally, through the determination of the mechanism function, the CFM weightlessness platform conforms to the two-dimensional diffusion mechanism function G(α) = (1−α)ln(1−α) + α. The first stage of CF conforms to the two-dimensional diffusion mechanism function G(α) = [1−(1−α)1/2]2, the second stage of CF conforms to the random nucleation and subsequent growth mechanism function G(α) = [−ln(1−α)]2/3.