Non-isothermal kinetics of the thermal desorption of mercury from a contaminated soil

The Almadén mining district (Ciudad Real, Spain) was the largest cinnabar (mercury sulphide) mine in the world. Its soils have high levels of mercury a consequence of its natural lithology, but often made much worse by its mining history. The present work examines the thermal desorption of two contaminated soils from the Almadén area under non-isothermal conditions in a N2 atmosphere, using differential scanning calorimetry (DSC). DSC was performed at different heating rates between room temperature and 600 °C. Desorption temperatures for different mercury species were determined. The Friedman, Flynn-Wall-Ozawa and Coasts–Redfern methods were employed to determine the reaction kinetics from the DSC data. The activation energy and pre-exponential factor for mercury desorption were calculated.


INTRODUCTION
Mercury (Hg) is one of the most toxic of global pollutants (Gochfeld, 2003).Its toxicity depends on the type of exposure suffered and the chemical species involved being organic forms the most toxic (Luciana et al., 2012).Bioaccumulation and biomagnification intensify its ecological and toxicological impacts (Luciana et al., 2012 andFitzgerald andLamborg, 2003).
The mining districts of Almadén (Spain), Idrija (Slovenia) and Monte Amiata (Italy) have together produced more than half of the total Hg extracted and commercialized in the world.Although operations have now ceased, Almadén was the world's largest and oldest Hg mining operation, its production making up a full one third of the total Hg ever extracted (Hylander and Meili, 2003).This activity led to the gradual dissemination and redistribution of Hg in the surrounding area.Millán et al., (2006) and Millán et al., (2011) report Hg to be present in concentrations of up to 40,000 mg kg −1 in the soils affected by mining activities.Even areas further away from the centres of exploitation were affected by the atmospheric and hydrological transport of Hg (Higueras et al., 2003).
In others countries, small-scale gold (Au) mining activities are harmful to the environment in part because of the widespread use of Hg in the extraction process.After thoroughly grinding Au-containing ore or silt, Hg is added, creating an amalgam.Subsequent heating concentrates the gold into a pellet but releases elemental mercury into the environment (Paruchuri et al., 2010).In soil samples collected from the Tapajós (Brazil) gold mining reserve, the total mercury concentration in soil is nearly 13 times that recorded for background sites (Egler et al., 2006).
The technologies used to remove or stabilize Hg in contaminated solid waste or soil include solidification/stabilization, soil washing, thermal treatment, solar thermal desorption, vitrification and electrokinetic remediation.A comparison of the different technologies is provided by Wang et al. (2012) and López-Delgado et al. (2012a and2012b) the Resource Conservation and Recovery Act (RCRA) of the United States suggests that Hg-containing waste be treated by solidification/stabilization when the Hg content is less than 260 mg kg −1 , and that thermal desorption methods be used when the Hg values are higher than 260 mg kg −1 (USEPA, 2008).
Thermal desorption treatment usually involves an ex situ technology that converts mercurial compounds into volatile Hg.The safety and reduced emissions due to the use of suitable filters associated with these techniques have led to their becoming preferred technologies in this field (Chang and Yen, 2006).Hg thermal desorption experiments have demonstrated the feasibility of mercury removal at temperatures between 127 °C and 600 °C (Kunkel et al., 2006).
Thermal decomposition techniques have been used to identify Hg compounds in soil, sediment samples, iron-based sorbents, and even in Hg lamp waste.The Hg species thermally released from contaminated soils can be analysed in several ways, including temperature-controlled continuous heating of samples in a furnace coupled to an atomic absorption spectrophotometer (AAS) (Windmoller et al., 1996) the use of solid-phase Hg-thermodesorption techniques complemented by selective extraction of organically bound Hg (Biester et al., 2000) by temperature programmed decomposition desorption (TPDD) (Ozaki et al., 2008), or by pyrolysis and X-ray absorption fine spectroscopy (XAFS) (Liu et al., 2006).Table 1 shows the mercury speciation derivatives recorded by different researchers.
In the present work, the thermal desorption of Hg was analyzed by differential scanning calorimetry (DSC) in a N 2 atmosphere, with the aim of elucidating the associated reaction mechanisms and kinetics.

Samples
The material analysed consisted of two Hgcontaminated soils (S1 and S2) from the Almadén area (Fig. 1).Soil S1 (34 mg kg −1 of total Hg) was taken from Dehesa de Castilseras (northeast of the El Entredicho mine).The area is an open Mediterranean forest where livestock graze.The Hg concentration of this soil is lower than the other soil S2 (Table 2) and is mainly of natural origin although this site has also been affected by mining activity.Soil S2 (10,497 mg kg −1 of total Hg) was from an abandoned metallurgical site in Almadenejos.This site dates from the 18th century and lies 12 km from Almadén.It occupies an area of some 30,000 m 2 and is surrounded by a wall built between 1756 and 1759.The area contains six pairs of abandoned roasting furnaces once used to obtain primary Hg from cinnabar.The soil in this area has a high Hg concentration (Table 2) -partly of natural origin and partly the result of mining activity, Hg waste storage, and the use of these cinnabar roasting furnaces (Millán et al., 2011).As sampling strategy, a composite sample of each study soil was collected from a 5 m 2 by obtaining 10 cm of the surface soil at five points.Afterwards, both soil samples were air dried and sieved to obtain the fraction less than 2 mm ready for the different analysis and trials.Furthermore, reference materials with known concentrations of Hg were used: CRM 051 from an Hg-contaminated area (29.9±5.96mg kg −1 of total Hg) in the western USA, and SRM 2709a, an agricultural soil (1.40±0.08 mg kg −1 of total Hg) from San Joaquín (Querétaro, Mexico).

Physico-chemical analysis of soils
The pH (H 2 O, 1:2.5) and electrical conductivity (EC) (H 2 O, 1:5) of the soils were measured according to the official methods of the Spanish Ministry of Agriculture, Fisheries and Food (MAPA, 1994).The EC of the saturated extract (EC) ES was determined by multiplying EC by 6.4 according to the method proposed by Loveday et al., (1972).The organic matter (OM) content of the soils was determined using the method of Walkley-Black and the soil textures recorded using the Bouyoucos method, according to standard procedures (Page et al., 1987).

Determination of soil Hg concentration
The Hg concentration of the soil samples was directly measured using an atomic absorption spectrophotometer specifically designed for Hg determination (Advanced Mercury Analyser -AMA254 -LECO Company).The certified reference material CRM 051 (soil from USA contaminated area, 29.90±5.96mg kg −1 of total Hg) was used as a standard to determine the accuracy and precision of the measurements and to validate the applied method.The mean value of total Hg determined for 10 measurements of the certified material using the AMA254 equipment was 29.90±2.89mg kg −1 of total Hg.At a 95% confidence level, no significant differences were detected between the certified value and the experimental one, this method was therefore considered to be reliable for the determination of total Hg.
When samples showed a high Hg content, out of equipment range limit (>600 ng), they were preprocessed by an acidic digestion using a MARS5 microwave oven (VERTEX Technics) following EPA Method 3052 (USEPA, 1996) in order to dilute for measuring by the above analyser.Recovery percentage for certified reference material (CRM 051) was 99%.
Table 2 shows the physico-chemical characteristics of the contaminated and reference soils.

Thermal desorption study soil constituents
S1, S2, CRM 051 and SRM 2709a were subjected to DSC analysis in a N 2 atmosphere (flow rate 20 mL min −1 ).About 60 mg of each sample were placed in a 175 µL sealed aluminium crucible and heated at a rate of 10 °C min −1 between 25 and 600 °C or 650 °C in a Setaram Model 3D-EVO analyser, which also recorded the temperature peaks associated with Hg desorption.

Differential scanning calorimetry kinetics: Mathematical models for determining Hg desorption kinetics
The kinetics of Hg desorption were then studied by DSC using the apparatus mentioned above.DSC experiments were performed at four different heating rates (5, 10, 15, and 20 °C min −1 ) between room temperature and 600 °C.Temperature calibration was achieved using ICTAC-recommended DSC standards.The accuracy of the reported temperatures was estimated to be ±2 °C.The sample mass used was again about 60 mg, and all experiments were performed in a N 2 atmosphere (flow rate 20 mL min −1 ).
For mercury desorption, it is generally assumed that the rate of conversion is proportional to the concentration of reacted material.The rate of conversion can be expressed by the following basic rate equation (Eq.1): where α is the degree of advance of reaction, f(α), β the heating rate (°C min −1 ) and k(T) are functions of conversion and temperature.For the DSC experiments, equation (2) can be expressed as: where dH dt is the heat flow above baseline and ∆Htotal the peak area of the reaction, expressed in mJ.By combining equations ( 1) and ( 2), the rate of conversion can be written in the form Eq. ( 3): where k(T) is the temperature dependent on the rate of heat flow; this is often modelled successfully by the Arrhenius Eq. ( 4): where E (kJ/mol) is the activation energy, A (min −1 ) the pre-exponential factor, and R (8.314 J mol −1 K −1 ) the gas constant.By combining equations ( 3) and ( 4), the reaction rate can be written as Eq. ( 5):

Friedman method
Friedman analysis (Friedman, 1964), which is based on the Arrhenius equation ( 4), takes into account the logarithm of the conversion rate (dα/dt) as a function of the reciprocal of the temperature (i.e., 1/T) at different degrees of conversion α, according to Eq. ( 6): where i is the index of conversion, j is the curve index, E a the apparent activation energy at α i , and f(α i,j ) the function dependent on the reaction model (assumed to be constant for a given reaction progress α i,j for all curves j).Since f(α) is constant at each degree of conversion αi, the curve for the logarithm of the reaction rate vs. 1/T is linear, with a slope of E a /R and an intercept A.

Flynn-Wall-Ozawa method
The Flynn-Wall-Ozawa method (Flynn andWall, 1996 andOzawa, 1965) is derived from the integral isoconversional method.Using Doyle's approximation (Doyle, 1961) the reaction rate in logarithmic form can be expressed as Eq. ( 7): where β is the heating rate and g(α) is the function of conversion.Thus, for any constant α value, the plot ln β vs. 1/T recorded at different heating rates should be a straight line.The E a can be then determined from its slope.

Coats-Redfern method
The Coats-Redfern method (Coats and Redfern, 1964) provides the thermal decomposition mechanism from the mass loss.An asymptotic approximation of 2RT/E a <1 for the resolution of Eq. ( 8): Non-isothermal kinetics of the thermal desorption of mercury from a contaminated soil • 5 Revista de Metalurgia 50(1), January-March 2014, e001.ISSN-L: 0034-8570 doi: http://dx.doi.org/10.3989/revmetalm.001allows equation ( 9) to be obtained: The Coats-Redfern method is one of the most widely used procedures for the determination of reaction processes.The E a at a constant heating rate for any of the g(α) functions listed in Table 3 can be obtained from equation (9).

Thermal decomposition of contaminated and reference soils
Figure 2 shows the DSC curves by heating the soils at a rate of 10 °C min −1 from room temperature to 650 °C.Two or three endothermic peaks can be seen.Table 4 shows the temperature data for these peaks.
Three peaks can be seen for S1, with maximum temperatures (T m ) of 109.5 °C, 304.8 °C and 533.8 °C.S2 has only two peaks of T m 121.8 °C and 305.7 °C.
CRM 051 has two peaks of T m 93.4 °C and 546.4 °C, while SRM 2079a has three of T m 122.5 °C, 279.2 °C and 528.5 °C.
Using the data in Table 1 as a reference, Hg is released from HgCl 2 between 72 and 182 °C, according to the reaction HgCl 2 (s) → Hg(g) +2Cl(g) (peak 1 in the DSC curves; see Fig. 2).In a well-aerated soil as is the case of the study soils, the conditions are normally appropriate for the formation of inorganic Hg 2+ as HgCl 2 .The mercurous cation is rarely stable under ambient conditions and, together with Hg 0 is easily oxidized to Hg 2+ .Therefore, Hg form mainly present in the soil will be HgCl 2 (Adriano, 2001 andGaona, 2004).
The peak of T m =93.4 °C in the DSC curve for CRM 051 could be due to the desorption of
The T m values for desorption of Hg from the different phases of the contaminated soils follow the order Hg 0 <HgCl 2 <HgS<HgO.These temperatures agree quite well with those reported by other authors using other techniques (Windmoller, 1996 andLópez et al., 2010).This order suggests that the thermal release of mercury species is related to its vapour pressure (Stein et al., 1996).When the vapour pressure decreases, the Hg release temperature increases.
Hg is subject to a wide array of chemical and biological transformation processes, such as Hg 0 oxidation, Hg 2+ reduction, and methylation, depending on the soil pH, temperature, and humus content.The formation of organic Hg 2+ complexes is known to be a dominant process, largely due to the affinity of Hg 2+ and its inorganic compounds for sulphur-containing functional groups (Skyllberg et al., 2006).Besides the above mentioned peaks, all samples show a sharp endothermic peak at 575 °C, that could be attributed to the polymorphic transformation of hypothermic quartz to hyperthermic quartz which starts at this temperature.It is due to the phase transition in quartz (α-Quartz trigonal to β-Quartz hexagonal) (Karathanasis et al., 1994;Plante et al., 2009;Salgado et al., 1995 andSalgado et al., 2004).X-ray diffraction of soil samples shows that the percentage of quartz in soil S1 (82%) is higher than the soil S2 (60%).This result is in agreement with the different depth of the S1 and S2 peaks.Thereby, the peak of S1 is deeper than the peak of S2.
Figure 3 show the DSC curves for S1 and S2 at different heating rates (5, 10, 15 and 20 K min −1 ).Table 5 shows the T m values for the peaks obtained.A shift towards higher T m values was seen as the heating rate increased (Fig. 3).This displacement of DSC curves with heating rate has been described by other researchers (Aboulkas et al., 2010).
Both soils present the same two firs peaks showing similar chemical bonds.However soil S1 has a third peak which does not appear in the case of S2.Therefore, S1 seems to have significantly more HgO than S2 has.

Calculation of the activation energy
The Friedman (FR) and the Flynn-Wall-Ozawa (FWO) methods were first used to calculate the E a for all heating rates.
With the FR method, the E a was calculated for different conversion values by plotting ln(dα/dT) against 1/T for a constant α value.Table 6 shows the results for S1 and S2.
Equation ( 7) was used with the FWO method.The results supplied are independent of the degradation mechanism.The E a for S1 and S2 were obtained from the plot of log (β) vs. 1/T at a fixed conversion rate (the slope being 0.4567E/R).Table 6 shows the E a values for S1 and S2.
The E a values obtained with the FWO method were higher than those obtained by the FR method.This is to be expected since the FWO method involves a systematic error that does not affect the FR method (Vyazovkin, 2001).The E a values obtained by the FR method are therefore deemed more reliable.
For all the decomposition reactions studied, the variation in E a with the degree of conversion over the interval 0.1<α<0.9was practically constant (Figure 4).This suggests that these Hg reactions occur in a single stage.
The variation in E a , independent of the calculation method used, shows that E a HgCl 2 <E a HgS<E a HgO.The vapour pressure and decomposition T m values for each Hg compound follow the same order.The E a of each reaction was greater for S2 than for S1, in agreement with the quantity of Hg in each soil.
Employing equation ( 9), the Coast-Redfern method was also used to determine the E a values for all the f(α) functions listed in Table 2 at a constant heating rate.Table 7 shows the E a and A values for S1 and S2 at a constant heating rate of 10 °C min −1 .
Table 8 summarises E a and A values and the thermal degradation mechanisms determined by the Coats-Redfern method.The thermal desorption of HgCl 2 is likely to be of the F n type since this mechanism has an E a (44.4 kJ mol −1 for S1 and 53.2 kJ mol −1 for S2), similar to that obtained by the FR method (44.6 kJ mol −1 for S1 and 49.7 kJ mol −1 for S2).The thermal degradation mechanism of HgS is probably A n -type (n=1.1) for S1 and F n (n=1.06) for S2.The thermal degradation mechanism of HgCl 2 is F n -type (n=1.38 for S1 and n=1.31 for S2).Finally, the thermal degradation mechanism of HgO is likely A n -type (n=1.5).
L 'vov (1999) reported an E a for the decomposition of pure HgO of 201.7±3 kJ mol −1 in an O 2 atmosphere, and of 93±8 kJ mol −1 in a vacuum.Later, L'vov et al. (2004) reported a value of 186±1 kJ mol −1 in O 2 .The literature appears to have no information on the decomposition energies of HgCl 2 and HgS.

CONCLUSIONS
The E a was found to be practically constant over the 0.1-0.9α interval for S1 and S2, suggesting that the decomposition of Hg compounds   is a single-step process with an E a of 44-53.2kJ mol −1 for HgCl 2 , 139.6-187.8kJ mol −1 for HgS, and 231 kJ mol −1 for HgO.The corresponding E a and A values were calculated in order to interpret the relationships between the soils.The Coats-Redfern method was successfully used to predict the reaction mechanism of thermal degradation of Hg compounds.The decomposition reaction model for HgCl 2 is first order, whereas that of HgO is described by the Avarani-Erofe've model, and that of HgS is first order and also described by the Avarani-Erofe've model, depending on the soil type.Knowledge of these E a values could be important when attempts are made to remediate Hg-contaminated soils by thermal decomposition.

Figure 3 .
Figure 3. DSC curves for the decomposition of contaminated soils at different heating rates.a) for S1 and b) for S2.

Figure 4 .
Figure 4. Apparent activation energy, as determined by the Friedman method for a) Soil S1 and b) Soil S2.

Table 2 .
Physico-chemical characteristics of contaminated soils Figure 1.Location of soil collection sites.

Table 3 .
Algebraic expressions of functions of the most common reaction mechanisms

Table 4 .
DSC results for the thermal desorption of Hg from the contaminated and references soils (heating rate: 10 °C min −1 ) o : Start of peak temperature; T m : maximun temperature; T e : End of peak temperature)

Table 5 .
DSC data for the thermal decomposition of the soils at different heating rates

Table 6 .
Activation energies for S1 and S2, as determined by the Friedman and the Flynn-Wall-Ozawa methods

Table 7 .
Activation energies and pre-exponential factors for S1 and S2 as determined by the Coasts-Redfern method for different f(α) functions

Table 8 .
Activation energies of S1 and S2 as obtained by the Coats-Redfern method