Introduction

Chromium slag is a hazardous industrial waste with high content of hexavalent chromium Cr(VI), which is a widespread contaminant owing to extensive industrial applications of chromium such as electroplating, leather tanning and dyeing industry (Wu et al. 2010; Li et al. 2015). The improper disposal of chromium slag causes environmental pollution, ultimately endangering public health (Gao and Xia 2011; Gunkel-Grillon et al. 2014). In order to meet the regulations and restrict detrimental effects, chromium slag should be treated to decrease its toxicity.

The process of transforming Cr(VI) to less toxic, insoluble Cr(III) is known as detoxification, which can be achieved through a variety of physical, chemical and biological methods (Jobby et al. 2018). Dry detoxification is the most facile approach among the reported technologies in industrial practice, due to the suitability for bulk processing of slags and for ex situ treatment. In order to reduce costs, increasing attention has been paid to the applications of other types of industrial wastes as reductants, such as municipal sludge and yellow phosphorus tail gas (Shi et al. 2007). As the excellent reducing properties and heat value, the widely available and reusable industrial carbon monoxide (CO) off-gas is put forward as a good substitute in the industrial scale dry-detoxification method (He et al. 2018).

Kinetic analysis plays a central role to understand a reaction. It has been used to understand the mechanism for Cr(VI) detoxification in the hydro-based reactions, including microbial detoxification, electrodeposition, phytoextraction, chemical immobilization and sorption (Gallios and Vaclavikova 2008; Chen et al. 2016; Guan et al. 2017; Wu et al. 2018). Nevertheless, very few studies are available on the dry-detoxification reaction under high temperature, due to the difficulties to analyze the thermally active gas–solid phase reaction.

Thermogravimetric analysis (TGA) is a method used to study thermally active reactions by thermogravimetry (TG) and derivative thermogravimetry (DTG) curves under programmed temperature control (Jaroenkhasemmeesuk and Tippayawong 2016; Liu et al. 2016). Many mathematical approaches were applied to determine kinetic parameters, but the model-free/iso-conversional approach is the most reliable method for calculating the apparent activation energy (E), as there is no need to assume reaction mechanism functions (Oladokun et al. 2016; Müsellim et al. 2018). Among the diverse selection of iso-conversional methods, the Flynn–Wall–Ozawa (FWO) method is extensively used for its accuracy and simplicity (Kissinger 1957; Ozawa 1992; Starink 2003). Thus, we studied here the mineralogical variations and apparent activation energy in dry detoxification of chromium slag by CO.

Experimental

Sample analysis and detoxification procedure

The chromium slag samples were acquired from the storage site of an idle chromium salt factory located in an industrial park in the Yunnan Province of China. The untreated slag was air-dried at room temperature and milled to less than 0.125 mm particle size.

The reduced slag was prepared through the detoxification process conducted as follows: The air-dried and milled original chromium slag was placed into a vacuum-tube electric furnace (SK-1200, Zhonghuan, Tianjin, China), followed by passing 99.9% CO cylinder gas through the furnace and heating to 450 °C for 30 min. All the cylinder gases were purchased from Messer (Messer Group GmbH, Bad Soden, Germany), and all chemicals and solvents used were of analytical grade.

The Cr(VI) leaching was analyzed by hydrazine spectrophotometry (SP-752 UV–Vis Spectrometer, Spectrum, China), and after H2SO4–HNO3 digestion, leaching procedure was conducted according to the Chinese Environmental Protection Standard HJ/T 299-2007 (MEPC 2007). The total Cr(VI) content was analyzed by atomic absorption spectrometry (AA-7000, Shimadzu, Japan) after alkaline digestion according to the Chinese Environmental Protection Standard HJ 687-2014 (MEPC 2014). The Cr(VI) leaching concentration and total Cr(VI) content in each sample were determined in triplicate.

Characterization

Chemical content in the samples was determined by X-ray fluorescence (XRF) (AxiosmAX, PANalytical, Netherlands). The mineral phases were analyzed by powder X-ray diffraction (XRD) with Cu Kα radiation and Ni filter in the 2θ ranging from 5° to 90°, and a scanning rate of 2°/min at 40 mA and 40 kV (Bruker D8 Advance diffractometer system). The Jade 6.0 software was applied for the analysis based on the mineralogical database of PDF-2004.

The electron backscattered diffraction (EBSD) analysis was conducted as follows: First, powder samples were mixed with epoxy resin and cured for 24 h; then, the cured blocks were polished using a polishing machine, and the smooth cross sections were sputtered with gold; finally, backscattered electron micrographs of morphological characterization were acquired through field-emission scanning electron microscopy (FESEM) (Gemini 300, Zeiss, Germany). The elementary composition was investigated by combining use of energy-dispersive spectrometer (EDS) (XFlash 6-30, Bruker, Germany).

Thermogravimetric and differential thermogravimetric analysis

Thermogravimetric and differential thermogravimetric (TG–DTG) analysis was carried out to obtain the thermogravimetric data (TG and DTG curves) under diverse heating rates with a thermal analyzer (Labsys Evolution, setaram, France). Approximately, 7.5 mg of the untreated slag was placed in a platinum crucible and put in the thermal analyzer, and the temperature was gradually increased from 25 to 800 °C in an inert atmosphere of 90% CO and 10% argon purified gases at the heating rates of 5, 10, 15 and 20 °C min−1, respectively, while the flow rate of mixed purified gas was maintained constant at 50 mL min−1.

Results and discussion

Mineral analysis and variation

Results show that Cr(VI) leaching was 373 mg L−1 in untreated slag and 0.01 mg L−1 in reduced slag. Total Cr(VI) was 7040 mg kg−1 in untreated slag and 11 mg kg−1 in reduced slag. Also, 99.84% of Cr(VI) was reduced during detoxification.

Minerals were analyzed in untreated slag and reduced slag by XRD analysis (Fig. 1), which indicated that the mineral composition of chromium slag was complex (Földi et al. 2013). According to Hillier et al. (2003), the mineral compounds can be classified into three groups: The first group was unreacted feedstock ore, only included chromite; the second group was “parent” minerals produced in the high-temperature roasting process, mainly included brownmillerite, periclase and larnite; the third group was weathering products and minerals formed in the leaching stage, such as brucite, calcite, ettringite and hydrogarnet.

Fig. 1
figure 1

X-ray diffraction (XRD) patterns of untreated and reduced slag. The Na2CrO4 in the untreated slag disappeared after reduction, and eskolaite was detected in the reduced slag, which indicated that Cr(VI) was reduced to Cr(III) by carbon monoxide. The variation of mineral compound before and after detoxification indicated the decomposition of the third-group minerals

Full size image

The first-group minerals have not been detected. Minerals detected in untreated slag were mostly consistent with the previously published mineral assembly in other researches (Deakin et al. 2001). They mainly belonged to the second and third-group minerals. The second-group minerals included periclase, brownmillerite, srebrodolskite (a subclass of brownmillerite) and larnite. The third-group minerals included gyrolite, hydrogarnet, lizardite-1T and lizardite-1M. Hydrogarnet is the hydration product of brownmillerite, which has been demonstrated to be the main Cr(VI)-containing minerals (Geelhoed et al. 2002). Gyrolite, lizardite-1T and lizardite-1M were layered double-hydroxide minerals of high anion exchange capacity which could hold chromate anions in the interlayer (Földi et al. 2013).

Minerals in the reduced slag mainly include: periclase, brownmillerite, calcite and quartz. As the hydration and weathering products were unstable under high temperature, calcite and part of periclase are the decomposition products of the third minerals under heating. Quartz occurred only in trace amounts, and its sources are complicated (Hillier et al. 2003).

From the perspective of mineral composition, the third-group minerals completely disappeared after reduction. Weak intensity XRD peaks of Na2CrO4 and eskolaite were detected in untreated and reduced slag, respectively, which is in agreement with the equation of the main reducing reaction (Eq. 1) and with the chemical analysis results. Our findings thus suggest that the decomposition of the third-group minerals is responsible for the release of Cr(VI), which was then reduced by CO.

For in-depth determination of the minerals and elements, XRF and EBSD–EDS were then conducted. XRF analysis showed that six elements, O, Ca, Mg, Fe, Al and Cr, accounted for over of 90% of samples mass, their proportions being 50.9%, 15.0%, 13.3%, 8.31%, 3.2% and 3.2% in the untreated slag, and 31.8%, 21.9%, 14.1%, 13.7%, 5.0% and 5.0% in the reduced slag. The elements determined by the EDS were similar than those from XRF results. Electron micrograph of particles and element distribution are presented in Fig. 2. The particle sizes in untreated slag were larger, and the size difference between particles was also greater than that of reduced slag, as shown in Fig. 2a, b. Element distribution maps illustrate that Ca, O, Fe, Al and Cr are uniformly distributed throughout the particles, and their distribution contours were overlapped. Though these minerals could not be clearly separated, complementary analysis using mineral formulas provides evidence that brownmillerite and its converted products were the main phases in the samples, and that Cr(VI) was bound to the mineral matrix, which also led to the difficulty of mitigation of Cr(VI) (Tinjum et al. 2008). Overall, results show that detoxification relies in the decomposition of the third-group minerals.

Fig. 2
figure 2

Backscattered diffraction scanning electron microscopy images (polished surface) of untreated slag (a, c) and reduced slag (b, l). Elemental distribution map of Ca, O, Mg, Al, Fe, Si, Cr and Na in the untreated slag (c–k). Elemental distribution map of Ca, O, Mg, Al, Fe, Si, Cr and Na in the reduced slag (m–t). Lighter colors indicate higher concentration of elements

Full size image

Calculation of apparent activation energy E by Flynn–Wall–Ozawa method

The main reaction equation of chromium slag reduction by CO is:

$$3{\text{CO}}\left( {\text{g}} \right) + 2{\text{Na}}_{2} {\text{CrO}}_{4} \left( {\text{s}} \right) = 2{\text{Na}}_{2} {\text{O}}\left( {\text{s}} \right) + {\text{Cr}}_{2} {\text{O}}_{3} \left( {\text{s}} \right) + 3{\text{CO}}_{2} \left( {\text{g}} \right)$$
(1)

Thermal analysis kinetic research was conducted through thermogravimetric and differential thermogravimetric (TG–DTG) analysis (Fig. 3). The thermogravimetric (TG) curve showed very rapid mass loss, in combination with the X-ray diffraction analysis (XRD) and Eq. 1, and the mass loss was concluded to be largely attributed to the decomposition of minerals in the slag and the reduction in Na2CrO4.

Fig. 3
figure 3

Thermogravimetric (TG) curve at the heating rate of 5 °C min−1 (a) and differential thermogravimetric curves (DTG) at different heating rates (5, 10, 15 and 20 °C min−1) (b). The mass losing procedure during the detoxification of chromium slag was proved by (a). The data for the calculation of the apparent activation energy (E) were obtained from (b)

Full size image

The basic kinetic equation of the non-isothermal heterogeneous reaction could be expressed as follows (Eq. 2):

$$\frac{{{\text{d}}a}}{{{\text{d}}t}} = A\exp \left( { - \frac{E}{RT}} \right)f(a)$$
(2)

where a is the reaction conversion degree defined by weight, A is the pre-exponential factor, E is the apparent activation energy, and f(a) is the simulated mathematical function of the reaction.

In general, the conversion degree a could be expressed by the weight loss as follows (Eq. 3):

$$a = (w_{0} - w_{t} )/(w_{0} - w_{\text{f}} )$$
(3)

Herein, w0, wt and wf denote the mass of the sample initially, at time t and finally, respectively.

In the non-isothermal reaction, the term β was defined as different heating rates (Eq. 4):

$$\beta = {\text{d}}T / {\text{d}}t$$
(4)

By substituting Eq. (4) to Eq. (2), the expression for the detoxification at different heating rates is provided by Eq. (5):

$$\frac{{{\text{d}}a}}{{{\text{d}}t}} = \frac{A}{\beta }{ \exp }\left( {\frac{ - E}{RT}} \right)f\left( a \right)$$
(5)

Subsequently, Eq. 5 was integrated to obtain the conversion function g(α), which describes the kinetics of the detoxification reaction at diverse heating rates (Eq. 6):

$$\mathop \smallint \limits_{0}^{a} \frac{{{\text{d}}a}}{f\left( a \right)} = g\left( a \right) = \frac{A}{\beta }\mathop \smallint \limits_{{T_{0} }}^{T} {\text{e}}^{{\left( { - E/RT} \right)}} {\text{d}}T$$
(6)

The solution to Eq. 6 could be obtained by the Doyle’s approximation: P(E/RT)  =  − 5.331–1.052E/RT, which forms the theoretical basis for the iso-conversional kinetic models of Flynn–Wall–Ozawa (FWO) method (Doyle 1965). The final expression was described as follows (Eq. 7):

$${ \lg }\beta = \lg \left( {\frac{AE}{Rg\left( a \right)}} \right) - 2.315 - 0.4567\frac{E}{RT}$$
(7)

The maximum mass loss (wMax) at a specific temperature was obtained at the interaction of tangent to the peak of the differential thermogravimetric (DTG) curve, known as peak temperature (Tp). For \({ \lg }\left( {AE/Rg\left( a \right) } \right)\) is a constant, R is universal gas constant (8.314 J mol−1 K−1), the plot of lgβ against 1/T could be built at diverse heating rates, and E value is calculated from the slope − 0.4567 E/R.

Apparent activation energy (E) is one of the kinetic three factors, which is thought to be the minimum energy required for a chemical reaction, and it could reflect the reaction rate of chemical reactions. The corresponding data of FWO methods for solving E value are presented in Table 1. The wMax values were very close, and the conversion degrees at Tp were approximately equal. The slope of the fitting equation was − 2271.89, and E was calculated to be 41.36 kJ mol−1. The liner correlation coefficient of the fitting equation was 0.9426, indicated that the plot was well fitted. Therefore, thermogravimetric analysis (TGA) was proved to be a reliable instrumental method in exploring the kinetic parameters of detoxification of chromium slags by CO.

Table 1 Thermogravimetric parameters related to the calculation of activation energy by the Flynn–Wall–Ozawa method
Full size table

Conclusion

XRD and EBSD–EDS analysis illustrates visually the variation of mineral phases in the chromium slag. Combined with the results of the leached and total Cr(VI) before and after detoxification, it is concluded that the third-group minerals were the major Cr(VI)-containing minerals, which amplified the difficulties in the detoxification of chromium slag. Their decomposition was responsible for the release of Cr(VI) and was the crucial premise for the reducing reaction. Furthermore, a detailed instrumental thermal analysis kinetics research by TG–DTG was performed, and apparent activation energy (E) was calculated by FWO method, which made up for the insufficient instrumental analysis kinetic research of dry detoxification of chromium slag.