On the Detection of Spectral Emissions of Iron Oxides in Combustion Experiments of Pyrite Concentrates

Abstract

In this paper, we report on the spectral detection of wustite, Fe(II) oxide (FeO), and magnetite, Fe(II, III) oxide (Fe₃O₄), molecular emissions during the combustion of pyrite (FeS₂), in a laboratory-scale furnace operating at high temperatures. These species are typically generated by reactions occurring during the combustion (oxidation) of this iron sulfide mineral. Two detection schemes are addressed: the first consisting of measurements with a built-in developed spectrometer with a high sensitivity and a high spectral resolution. The second one consisting of spectra measured with a low spectral resolution and a low sensitivity commercial spectrometer, but enhanced and analyzed with post signal processing and multivariate data analysis such as principal component analysis (PCA) and a multivariate curve resolution - the alternating least squares method (MCR-ALS). A non-linear model is also proposed to reconstruct spectral signals measured during pyrite combustion. Different combustion conditions were studied to evaluate the capacity of the detection schemes to follow the spectral emissions of iron oxides. The results show a direct correlation between FeO and Fe₃O₄ spectral features intensity, and non-linear relations with key combustion variables such as flame temperature, and the combusted sulfide mineral particle size.

Keywords: combustion; multivariate data analysis; optical sensors; principal component analysis; signal detection; signal processing; signal reconstruction; spectral measurements.

Figure 1

Experimental setup [6].

Figure 2

Radiometric measurement scheme and associated radiative processes: (a) Single heated particle radiative emission with its surroundings; (b) sensing scheme depicting the different particle states during their passing through the reaction zone.

Figure 3

Data processing and analysis schemes implemented in this work: (a) Spectral signals acquisition and pre-processing, (b) PCA decomposition, (c) MCR-ALS implementation, (d) temperature estimation.

Figure 4

Average spectral emission from pyrite minerals in combustion experiments by considering different particle sizes.

Figure 5

Spectral FeO emission measurements and reference spectral features.

Figure 6

Extraction of spectral features with PCA from pyrite combustion for different particle sizes, P.V. 1 (loadings vectors) are depicted.

Figure 7

Extraction of spectral features with MCR-ALS from PyE pyrite sample combustion, each spectrum is normalized to its maximum: (a,b) are associated to a sodium doublet, named as I₁ and I₂ for subsequent analysis, (c) is associated with FeO emission, I₃ and (d) emission of other iron oxides emission, I₄. Obs. Spectral features are normalized to their maximum and offsets to each spectrum are included for comparison purposes.

Figure 8

Predominance diagram for the Fe-O system at high temperatures constructed with HSC^® software. The sections delimited by the continuous lines, represent the conditions at which the described species are thermodynamically stable.

Figure 9

Non-linear curve fitting results for a spectrum from combustion of PyE sample, RMSE (root mean square error) = 0.4393, θ₀ = 0.0249, T_s = 220.3 K, θ₁ (related to I₁ or Na) = 3.4583, θ₂ (related to I₂ or Na) = 15.7373, θ₃ (related to I₃ or FeO) = 6.2484, θ₄ (related to I₄ or Fe₃O₄) = 7.3093 and θ₅ (related to an offset) = 9.6 × 10⁻⁸.

Figure 10

Relative concentrations (median values for each data set) of FeO and Fe₃O₄ band emissions estimated from spectra of pyrite combustion. Obs. Sample particle sizes decrease from left to right.