*2.1. Reactive Systems*

A recent work focused on radiometric optoelectronic sensors that consider a broad spectrum, which includes the visible range up the near infrared range [7–9]. This approach is different from available commercialized sensors, which only analyze a limited selection of wavelengths. The idea here is to analyze both the continuous and discontinuous parts of the spectrum, as well as its dynamics in different time scales. The purpose is to correlate the measured spectral radiation to the operational conditions of the reactors following the emission of radiation from the oxidation reactions. These sensors also measure temperature with better precision than the Noranda Pyrometer by using

more sophisticated techniques to select the appropriate (two or more) wavelengths. The sensor thus solves the complex problem of simultaneously measuring emissivity and temperature.

Results from an experiment using a drop tube setup for the flash oxidation of copper concentrates (Figure 2) validate the formation of wustite (FeO), magnetite (Fe3O4), and copper oxides (CuO/Cu2O) as indicators of the concentrate smelting/oxidation/combustion process [7–9]. This information is of special interest to track the physicochemical dynamics of the process in real-time and establish operating criteria for a smelting reactor. Examples of such criteria could include adjustments to concentration/oxygen ratios, oxygen enrichment for the incoming blast, and the quantity of cold charge that will be required to maintain the heat balance.

**Figure 2.** Radiometric measurement scheme and associated radiative processes: (**a**) single-heated particle radiative emission with its surroundings, in which the intensity I of the emitted radiation is a function of wavelength λ, particle temperature *T*<sup>p</sup> and particle emissivity ε<sup>p</sup> and (**b**) sensing scheme depicting the different particle states as they fall through the reaction zone (adapted from [8]).

The spectral acquisition system consists of a multicore optical fiber with its own cooling process (Figure 2b), which measures the combustion flame radiation. Figure 2b shows a simple combustion scheme for sulfide particles covering the physical phase changes that a particle can experience inside the reaction zone. However, measuring the spectrum is a complex task. As shown in Figure 2a, there is an ensemble of physical and chemical processes that complicate this measurement. One example is drop tube radiation caused by increased electrical resistance due to the higher temperatures; processing these spectral signals can, however, mitigate the effect of the unwanted radiation [8,9].

Applying specialized algorithms in the treatment of the spectral signals obtained from the experiment, coupled with multivariate data analysis methodologies, allows for the identification and classification of copper and iron sulfide minerals present in the blend [10]. These results are particularly important as they demonstrate that spectral data obtained from the oxidation process can be used to identify the type of charge being treated within a molten bath. The controlled bench-scale laboratory study was carried out on several different types of concentrate. An exploratory analysis of the results using principal component analysis (PCA) applied to the spectral data depicted high correlation features among species with different mineral characteristics but similar elemental compositions. Classification algorithms were tested on the spectral data, and a classification accuracy of 95.3% was achieved using a support vector machine (SVM) classifier with a Gaussian kernel. Initial industrial-scale trials with a prototype have confirmed these results [10].

#### *2.2. Nonreactive Systems*

Despite tremendous advances in the development of passive and active photonic sensors, such as hyperspectral imaging (HI) and laser-induced breakdown spectroscopy (LIBS), real-time analytical sensors do not exist at present for the conditions of pyrometallurgical copper processing. To date, there is no commercial instrument capable of online quantification of copper content (% Cu) without contact during tapping operations. Nor can the available sensors discriminate between the phases that are of particular interest to smelter operations. The distinction between matte and slag during the tapping of a smelting furnace can significantly impact copper recovery, yet this function is heavily reliant on the experience of operators involved in tapping.

However, in the last decade, the spectral behavior of pig iron and slag in the ferrous industry has been studied to estimate different variables that allow for improved control of the tapping process [11,12]. These models describe the parameters contained in the iron–slag mixture during blast furnace tapping, such as iron emissivity, casting depth, slag layer thickness and absorption coefficients, and radiometric parameters (e.g., reflectance at the iron-slag interface). The methodology begins by determining a spectral range in which the radiation of the molten phases is comparatively high. This facilitates detection by a silicon charge coupled device (CCD) camera, which is sensitive in the visible spectral range and part of the near-infrared. An optical filter centered at 650 nm was used together with the optics, such that the radiation emitted by the wash was partially filtered in the indicated spectral range.

The results confirm that the difference in emissivity of iron and slag at 650 nm allows for the spatial distinction of these phases. Furthermore, it was identified that the radiation intensity of the molten iron remains practically constant during the process, while that of the slag fluctuates. This fluctuation is due to differences in the thickness of the slag layer, as it absorbs and transmits the radiation coming from the steel to varying degrees. Additionally, the optical system was calibrated with a high-temperature black-body radiator, allowing the temperatures to be estimated at 1500–1600 ◦C, which is considerably higher than the copper smelter temperatures (1200–1350 ◦C). The results confirm that, by using sensitive optoelectronic systems in the molten iron spectral emission band, coupled with appropriate spectral models and processing software, it is possible to develop reliable and robust systems at both the laboratory and industrial scales. The authors believe that this approach could be adapted for copper smelter processes and represents a natural pathway for future work.
