Volatilized and Condensed Sb- and As-Bearing Phases Produced During Roasting of Cu-Rich Complex Concentrate in Nitrogen Atmosphere with Oxygen in Traces

The Rockliden S7574 concentrate was tested in a vertical tube furnace setup described previously by Prasad et al.[24] The sample was introduced in the furnace and held in a cold chamber while the furnace heating was started. After the attainment of the desired furnace temperature, the sample was introduced in the hot zone. The furnace temperature was set higher than the aimed sample temperature, in accordance with the wall and center temperature relationship of this setup presented in a previous study by Prasad et al.[27] For instance, the furnace temperature set-point was 404 °C when the aimed sample temperature was 350 °C in the Experiment ID 350-10. A controlled gas atmosphere was maintained from the beginning by introducing Nitrogen (Chemical Nitrogen 4.6 from AGA-Linde, Purity ≥ 99.996 pct, H₂O ≤ 5 ppm, O₂ ≤ 5 ppm) as a carrier gas. Details of the experiments performed in this setup are listed in Table II.

Figure 1 presents a schematic of the vertical tube furnace used in this study. An optional water-cooled condensation plate was used in this apparatus for the Experiment IDs 700-1 and 350-10 to facilitate the partial condensation (at room temperature) of volatiles in the gas phase before leaving through the outlet. During the Experiment ID 350-10, the temperature was held at 350 °C for 10 hours, and a higher sample mass of 50 g was used to obtain sufficient amount of the condensates from this experiment for proper characterization.

The concentrate sample was also tested in a Netzsch STA 409 instrument (sensitivity ±1 μg) to determine the mass loss and phase transformations during heating through Thermo-Gravimetric Analysis (TGA) and Differential Thermal Analysis (DTA), respectively. A Quadrupole Mass Spectrometer (QMS) connected to the thermo-gravimetric setup performed in situ analysis of the gas phase emanating from the sample. Quadstar 32-bit program was used to operate the mass spectrometer. Details of this equipment assembly are mentioned in Ahmed et al.[30] The experiments performed in this setup are listed in Table III.

The carrier gas used in these thermo-gravimetric experiments was again Chemical Nitrogen 4.6 (additionally containing H₂O(g) ≤ 5 ppm, O₂(g) ≤ 5 ppm). During the Experiment ID 1000-TG, the QMS was operated in Scanning Bargraph mode for the detection of gaseous species with mass-to-charge ratio (m/e) in a specified range between 1 and 300. In the Experiment ID 350-TG, the mode of QMS was Multiple Ion Detection (MID). In this mode, the ion current measurements in QMS were performed only for a set of preassigned m/e values.

Table IV presents the list of gaseous compounds intended for detection in the MID-mode of QMS. In case of the heteronuclear compounds of Sb and As, the m/e values were assigned for both molecular and ionic species (for instance, in case of As₃SbO₆(g), m/e = 442-443 for its molecular form, and m/e = 426-428 for its ionic form—\({\text{As}}_{3}{\text{Sb}}{\text{O}}_{5}^{+}\)[2]). This was to increase the possibility of the detection of these complex gaseous species during the in situ measurements in QMS. The Sb- and As-bearing compounds listed in Table IV were those which were predicted to form during the volatilization process by a thermodynamic model, presented later in this article (Section V). Additionally, the two complex sulfide species—As₃SbS₆(g) and As₂Sb₂S₆(g), were included in this list based on the experimental evidences of their existence in the gas phase in the work of Mauser et al.[2] These two gaseous species were not estimated in the thermodynamic calculations due to unavailability of their thermodynamic data.

An important limitation of the QMS setup used in this work is that it cannot perform detection of species with m/e >510. Therefore, m/e values for the compounds with high molecular weights—AsSb₃O₆, As₂Sb₂S₆, AsSb₃S₆, Sb₄S₃, and Sb₄O₆ could not be assigned for detection in the QMS (in the MID-mode) during the run of Experiment ID 350-TG (Table III).

Figure 2 shows the images of the condensate collected from the roasting Experiment ID 700-1 (Table II) in the vertical tube furnace, illustrating the physical appearance of the collected condensates from the roasting experiments in the vertical tube furnace. In both the Experiment IDs 700-1 and 350-10, the collected condensates had soft and polymeric texture. Each condensate was subsequently crushed manually in a mortar under liquid Nitrogen into a fine powder. A portion of the crushed powder was mixed with liquid epoxy, and the solidified epoxy mount was polished and carbon coated. These epoxy mounts were analyzed in a Zeiss Merlin FEG SEM (Scanning Electron Microscope) together with an Oxford Instruments X-Max EDS (Energy-Dispersive X-ray Spectroscopy) detector. The EDS was operated by AZtec software. A working distance of 8.5 mm, probe current of 500 pA, and accelerating voltage of 20 kV were employed for operating the SEM-EDS. Beam calibration in EDS was done using pure copper. The elemental composition obtained from EDS measurements is reported in mol pct after normalization, ensuring the summation of the concentrations for all identified elements to be 100 mol pct.

The left over amount of powdered condensates after epoxy-molding were analyzed with XRD. These measurements were performed in a PANalytical Empyrean X-ray diffractometer in θ–θ geometry using Cu Kα radiation (λ = 0.154184 nm), beam current of 40 mA, 20 deg ≤ 2θ ≤ 90 deg  with a step size of 0.026°/s, and beam voltage of 45 mV. A PIXcel 3D detector was used. A shallow sample holder with zero/negligible background was used to hold the condensate powders during the measurement, as the obtained quantities of the condensate powders were smaller than the normal sample size for the XRD-measurement. The obtained XRD patterns were evaluated in HighScore Plus (v4.7, a software from PANalytical B.V., Almelo, The Netherlands) using the FIZ-NIST ICSD (Inorganic Crystal Structure Database) database, version 2015-1.

Results of the Experiment ID 1000-TG (Table III) in the TGA-DTA-QMS setup are presented in Figure 3. The highest ion current in QMS was measured for m/e = 32, representing O₂(g). The nearly constant ion current for this m/e value over the entire heating cycle is also suggestive of it being the trace oxygen in the carrier Nitrogen gas. QMS detected a gaseous species with m/e = 64 as the predominant volatilized compound above 400 °C. m/e = 64 may represent either S₂(g) or SO₂(g), owing to their similarity in molecular weight (≈ 64). SO₂(g) would evolve through the exothermic oxidation of sulfide minerals like chalcopyrite and pyrite. However, the peaks in the DTA curve corresponding to the QMS peaks for m/e = 64 are endothermic. Therefore, S₂(g) with m/e = 64 is the predominant volatilized species, which is expected to be emanating from the pyrolysis (endothermic process) of chalcopyrite[31] and pyrite[23]. Simultaneous detection of SO(g) corresponding to m/e = 48 further suggests that the trace oxygen in the carrier Nitrogen partially oxidizes the S₂(g). However, the QMS does not detect any Sb- or As-bearing gaseous species during this experimental run.

Figure 4 presents the molar quantities ΔS(mol), ΔSb(mol), and ΔAs(mol) of the elements volatilized during the roasting experiments on Rockliden S7574 concentrate in the vertical tube furnace at different temperatures between 200 °C and 700 °C (experimental details in Table II). Data in this plot are obtained from the chemical analyses and weights of the initial concentrate and the roasted calcines, presented in an earlier work by Prasad et al.[24] Evidently, at every temperature, the moles of S volatilized are approximately two order of magnitude higher than the quantities of Sb and As volatilized. This implies that the concentration of the volatilized Sb- and As-bearing gaseous species in the gas phase would be comparatively much lower than that of S₂(g). Additionally, there will be a significant dilution owing to a constant flow of carrier gas (N₂), thereby explaining the non-detection of Sb- and As- volatiles by the QMS (Figure 3) during the run of Experiment ID 1000-TG in the TGA-DTA-QMS setup.

Figure 4 in addition shows that noticeable Sb-volatilization started above 300 °C. Also, in the QMS measurements in Figure 3, S₂(g) concentration in the gas phase was low below 400 °C. Based on these two observations, an experiment was planned in the TGA-QMS setup for heating the concentrate in Nitrogen atmosphere upto a temperature of 350 °C (midway between 300 °C and 400 °C), and holding at this temperature for 10 hours under a reduced Nitrogen flow of 50 mL/min (compared to 100 mL/min in the Experiment ID 1000-TG, Table III). The possibility of detection of Sb- and As-bearing phases in the QMS measurement was expected to increase due to likely low S₂(g) volatilization at 350 °C.

Figure 5 presents the results of TGA and QMS measurements in the experiment at 350 °C (Experiment ID 350-TG in Table III). The QMS was set to measure ion currents for the selected m/e values ≤ 510 mentioned in Table IV. The concentrate experienced a total mass loss of approximately 0.8 pct in this experiment. The ion current measurements through QMS were in a wide range comprising of both positive and negative values. Consequently, a symmetric logarithmic scale developed by Webber[32] was used in Figure 5 (the transformation involved is \({{y}} = {\text{sgn}}\left({{x}}\right). {\text{log}}_{10}{(1+ |(x/C)|),}\) ‘sgn’ is the Signum function, C = 15.7) for the plotting of these widely scattered positive and negative ion current values. The QMS measurements could again clearly detect only O₂(g) and the sulfur bearing S₂(g) and SO(g) in the gas phase at this low temperature. The ion currents measured for rest of the m/e values corresponding to different Sb- and As-bearing species were in a narrow band of negative values below − 10^-15 A. Tamura et al.[33] associated such negative ion current values to the offset error of ammeter in QMS, essentially implying non-detection. However, sporadic positive ion current values are also recorded for these m/e values in Figure 5, suggesting that the Sb- and As- bearing species did exist in the gas phase, but in very low concentrations. This experiment, therefore, demonstrated the limitations in the in situ detection of the Sb- and As-bearing volatilized species through QMS during the roasting of Cu-rich sulfide concentrates in inert atmosphere.

Figure 6 presents the XRD pattern of the condensate collected during the Experiment ID 700-1 (Table II) by roasting Rockliden S7574 concentrate at 700 °C in the vertical tube furnace. It shows that the condensate primarily comprises of elemental sulfur in the form of cyclo-octasulfur, S₈(s).

Notably, similar to the failure in detecting Sb- and As-bearing species in the gas phase during measurement by the QMS, Sb and As compounds in the solid condensate could again not be identified through XRD. This was expected based on the fact that S-volatilization was two orders of magnitude higher than the quantities of Sb- and As-volatilization as shown in Figure 4.

XRD pattern of the condensate generated during the Experiment ID 350-10 of roasting the Rockliden S7574 concentrate at 350 °C is presented in Figure 7(a). It suggests that this condensate primarily comprises of Sb- and As-bearing compounds, unlike the elemental sulfur-rich condensate (Figure 6) from the roasting experiment at 700 °C. Secondly, the XRD pattern indicates that instead of the sulfides, expected to form during the roasting of sulfide concentrates in Nitrogen atmosphere, the condensate comprises of oxides. Most of the peaks in Figure 7(a) match with both As₂O₃(cubic) and Sb₂O₃(cubic) phases. Clearly, the traces of oxygen in the Nitrogen flow have contributed to the oxidation of Sb- and As-bearing volatiles. Figure 7(b) is the stem-plot showing only the peak intensities of the measured (> 5 pct) and reference patterns. Detailed examination of this stem-plot suggests that both in terms of peak positions and peak intensities, conformity of the measured pattern is relatively better with As₂O₃(cubic) compared to Sb₂O₃(cubic). The dominance of As₂O₃(cubic) over Sb₂O₃(cubic) in the 350 °C-condensate is perplexing since the Sb-volatilization was relatively much higher than the As-volatilization up to 400 °C in the roasting experiments, as shown in Figure 4.

To further ascertain the constituting phases, the epoxy mounts of these condensates were examined in SEM-EDS. It was found through the EDS measurements that these condensates comprised of the elements—S, O, As and Sb. Figure 8 presents the EDS measurements of the elemental composition of the condensates performed at numerous spots over the polished surface of the epoxy mounts. The composition values in mol pct are arranged in the decreasing order of S-content along the x-axis. The nearly continuous nature of the plots, especially for the S-content in both the condensates indicates that the size of uniform phase fields in the condensate is much lower than the spot-size of EDS measurements. Otherwise, the measurement values for the elemental compositions would be clustered at the stoichiometric values corresponding to the specific compounds (for instance near 40 mol pct for the plot of As, if the EDS spot solely covers As₂O₃(cubic)). Therefore, the individual compounds in these condensates are fine-grained (in sub-micron sizes) and heterogeneously intermixed, evident from wide variation in the compositions over different spots.

Following are the salient observations based on the elemental compositions in Figure 8(a) for the 700 °C-condensate:

The prominent observations from the elemental compositions of the 350 °C-condensate shown in Figure 8(b) are as follows:

Figure 9 presents the same compositional data for the 350 °C-condensate of Figure 8(b), but in the form of a scatter plot between O mol pct vs. S mol pct. EDS measurements on a hypothetical mixture containing finely intermixed As₂S₃(s), Sb₂S₃(s), As₂O₃(cubic), Sb₂O₃(cubic) would follow the line AB shown in this plot. The actual measurements follow this line as a lower limit of S-content. The measured data points above this line indicate that the spots contain elemental sulfur in addition to the trisulfides and trioxides of Sb and As. The upper limits of S-content also nearly fit over a straight line CD, almost parallel to the line AB. This line having a slope of − 1, suggests that the elemental sulfur, S_n(s) got partially oxidized to form polysulfur oxides, S_n−xO_x(s) via an equal replacement of S atoms by O atoms (x-moles). Equal replacement of S by O means that, for instance, if the unoxidized elemental sulfur in the condensate was S₈(s), the polysulfur oxide compounds would be S₇O(s), S₆O₂(s) etc., and not S₈O(s), S₇O₂(s) or S₆O(s). Also, the S-content at O = 0 mol pct not reaching 100 mol pct yet again indicates that the elemental sulfur is fine-sized and much smaller than the spot-size of EDS measurements, such that no spot comprises of only the elemental sulfur.

Based on the above characterization investigations, the condensates collected at room temperature in this study during the roasting of Rockliden S7574 concentrate in Nitrogen atmosphere (bearing traces of oxygen) were found to consist of the following compounds:

The Rockliden S7574 concentrate exhibited higher Sb-volatilization compared to As-volatilization below 500 °C in Nitrogen atmosphere, as presented in Figure 4. Arsenopyrite is the only As-bearing mineral in this concentrate.[24] Figure 10 presents a BSE image of a fully liberated particle of arsenopyrite in the calcine obtained from the Experiment ID 350-10 (Table II). The BSE gray-level throughout the particle is uniform, suggesting lack of any perceptible transformationin arsenopyrite at 350 °C. The elemental compositions at the indicated spots in this figure measured via EDS are presented in Table V. As content was found to be slightly lesser than in the theoretical stoichiometry of arsenopyrite (33.3 mol pct Fe, 33.3 mol pct As, 33.3 mol pct S). Therefore, arsenopyrite experienced only minor transformation at 350 °C, consequent to which the As-volatilization was low below 500 °C (Figure 4).

Regarding substantial Sb-volatilization at low temperatures, Prasad et al.[24] in a previous study attributed this behavior to the thermal decomposition of gudmundite (FeSbS) into pyrrhotite and metallic-Sb between 300 °C and 400 °C. They also demonstrated that other Sb-bearing minerals in the concentrate did not experience any noticeable transformation at 350 °C, thereby concluding that the decomposition of gudmundite was the reason for a relatively high Sb-volatilization at low temperatures.

Further results from the evaluation of gudmundite transformation at 350 °C (Experiment ID 350-10, Table II) will now be presented. Interestingly, minor concentrations of As were found at the sites of partially transformed gudmundite as shown in Figure 11. Gudmundite in Rockliden S7574 did not contain any As from the beginning before roasting.[24] Figure 11 presents the EDS spectra of composition measurements at two different sites in the roasted calcine—one inside a fully liberated particle of gudmundite, and another in a particle hosting gudmundite-chalcopyrite intergrowth. As was detected in these spots of gudmundite in the roasted calcine at statistically significant levels, confirmed from low values of the statistical error, σ (corresponding to the As-content) in the displayed composition tables in Figure 11. (Composition for the site at the gudmundite-chalcopyrite intergrowth comprises additional elements—Cu, Zn, owing to small size of the uniform phase field in the complex intergrowth). Presence of As at these sites points towards a mechanism of Sb-volatilization which involves interaction of As-bearing species in the gas phase with the metallic-Sb generated from the thermal decomposition of gudmundite.[24] This observation is in line with the earlier evidences of formation of complex Sb-As compounds during volatilization when Sb and As coexist in the material undergoing roasting.[1‐3,9,10,15]

The calculations were primarily based on the findings from Prasad et al.[24] that Sb-volatilization at 350 °C occurred through the thermal decomposition of gudmundite (FeSbS) into pyrrhotite (Fe_1−xS) and metallic-Sb. The thermodynamic framework, therefore, involved a set of simplified initial conditions of Sb(s) interacting with N₂(g), O₂(g), H₂O(g) (components of the carrier gas), and S₂(g) (labile sulfur volatilized from the concentrate) at 350 °C. An As-bearing sulfide vapor phase As₄S₄(g) was also included for consideration in the gas mixture. This was based on the detection of As at the sites of the transformed gudmundite, presented in Figure 11. The quantity of As₄S₄(g) was set as a study parameter and varied in a range of relatively insignificant values compared to the other gaseous components. The small quantities of As₄S₄(g) in the gas mixture were considered based on the observation of imperceptible transformation in the As-bearing mineral, arsenopyrite, at 350 °C (Figure 10, Table V). The model inputs of initial quantities of the reactants were calculated on the basis of 1 mole of Sb(s). Determination of the quantities of the gaseous compounds commensurate with 1 mole of Sb(s) is presented in Appendix 1, Table A1.2.

Figure 12 presents the quantities of Sb- and As-bearing oxide and sulfide gaseous species (with available thermodynamic data) equilibrating at 350 °C corresponding to different initial quantities of As-sulfide in the gas phase.

The results suggest that the concentration of individual oxide and sulfide compounds of As and Sb will be much higher than those of the complex heteronuclear Sb-As compounds—As_nSb_4−nO₆(g) and As_nSb_4−nS₆(g). However, the equilibrium calculation does not include the compounds—As₃SbS₆(g) and As₂Sb₂S₆(g) due to unavailability of their thermodynamic data. These two compounds are expected to be present in much higher concentration than AsSb₃S₆(g) at lower temperatures, for instance at 300 °C as measured by Mauser.[2] Therefore, the vapor-phase complexation of sulfides of Sb and As might be prominent in reality, contrary to the indications from thermodynamic calculation.

Secondly, in Figure 12, the concentration of all the three heteronuclear oxide compounds—As₃SbO₆(g), As₂Sb₂O₆(g), and AsSb₃O₆(g), are much lower than for the complex sulfide—AsSb₃S₆(g). Therefore, the Sb-As complex oxides are likely to be relatively insignificant in the gas phase compared to the Sb-As complex sulfides

The input parameters of the stochastic model are explained below:

A detailed description of the mathematical formulation of the stochastic model is presented in Appendix 2.

Appropriate values of the input parameters—S^max, H^max, Mode, and \({{R}}_{\text{O}}\) are obtained by performing parametric analyses of the model and comparing the results with the EDS measurements of elemental composition for the 350 °C-condensate (already presented in Figures 8(b) and 9).

Figure 14 presents the parametric study results obtained by changing the values of H^max and S^max in succession. The compositional limits in a S vs. O plot form domain boundaries in the shape of trapezium. Figure 14(a) shows that the value of H^max decides the orientation of the line AD, one of the non-parallel sides in this trapezium (Lines AD₁ to AD₆ corresponding to increasing values of H^max from 0 to 100 mol. pct). Superimposition of the EDS measurements suggests that the appropriate value of H^max is 60 mol pct, corresponding to the line AD₄, which offers the closest confinement of the data points from EDS measurements. Therefore, the most oxidized polysulfur oxide in the condensate should be approximately S₃O₅(s), owing to x ≈ 5 in the chemical formula, S_n−xO_x(s) (rounded off from the exact value of x = 4.8, calculated by substituting H = H^max = 60 mol pct and n = 8 corresponding to cyclo-octasulfur in Eq. [4]). Similarly, as shown in Figure 14(b), the location of the line CD is decided by the value of S^max and S^max = 75 mol pct leads to an apt validation with the EDS measurements. Consequently, the sulfur content representing elemental sulfur, S_n(s) (excluding sulfur in the As, Sb trisulfides), is below 75 mol pct at any site in the condensate.

After determining the suitable values of H^max and S^max, subsequently, the model is tested with changing input values of Mode and \({{R}}_{\text{O}}\). The effect of these changes are studied in the As vs. Sb plot owing to the higher relevance of Mode and \({{R}}_{\text{O}}\) in the modeled Sb- and As-contents. Figure 15(a) shows that changing the value of \({{R}}_{\text{O}}\) has no effect on the As-Sb compositional domain when Mode = 2 (Table VI, \(\left(\frac{{\text{Sb}}_{2}{{\text{S}}}_{3}\text{(s)}}{{\text{Sb}}_{2}{{\text{S}}}_{3}\text{(s) + }{\text{As}}_{2}{{\text{S}}}_{3}\text{(s)}}\right) \, \in \, \left[{0, 1}\right]\)). The same domain boundary-ABCDA is obtained irrespective of the different input values of \({{R}}_{\text{O}}\) = 0, 5, 10, 15, 20, 25, 40, 50, 75, 90, and 100 mol pct. Also, the superimposed EDS measurements suggest that the modeled compositional limits obtained with Mode = 2 are much wider than the scattered variations in the actual measurements. Interestingly, the modeled limits become responsive to the changes in \({{R}}_{\text{O}}\) when Mode = 1, as shown in Figure 15(b). The domain boundary obtained in each set of calculations is A_pB_pC_qD_qA_p, where

Evident from Figure 15(b), the EDS measurements are well confined by the modeled domain boundary A₀B₀C₂D₂A₀ (p = 0 and q = 2), corresponding to \({{R}}_{\text{O}}\) = 10 mol pct and Mode = 1. The agreement between the modeled composition limit and the EDS- in Figure 15(b) when Mode = 1 points towards the existence of the complex sulfide species, As_4−nSb_nS₆(g) in the gas phase during volatilization of Sb and As from Rockliden S7574 concentrate at 350 °C. Additionally, it suggests \(\frac{{\text{Sb}}_{2}{{\text{O}}}_{3}\text{(s)}}{{\text{As}}_{2}{{\text{O}}}_{3}\text{(s)}} \approx \frac{1}{{9}}\) (ratio calculated by putting \({{R}}_{\text{O}}\) = 10 mol pct in Eq. [2]) in the 350 °C-condensate.

Table VII lists the values of the input variables of the stochastic model (described in Section VI–A) obtained through parametric analyses and subsequent comparison with the EDS measurements.

Until now, the S vs. O and As vs. Sb graphical representations have been used to demonstrate the close agreements of the modeled composition limits and the EDS measurements. Figure 16 presents further validation of the modeled composition limits in the other four possible binary scatter plot representations—Sb vs. S, As vs. S, Sb vs. O, and As vs. O. These illustrations prove the consistency of the modeled results generated from the parametrized inputs in all the possible graphical representations for the condensate composed of S, O, As, and Sb.