In the previous blog, we discussed the importance of minerals analysis for efficient iron ore mining and beneficiation. In this blog we will discuss the added value of mineralogy monitoring at the next step of the ore-to-metal process, iron ore sintering.

Sintering is one of the iron ore post-processing steps to prepare iron ore fines for a blast furnace (another common process of iron fines agglomeration is pelletizing, but this will be part of another blog).

Feed for a sinter plant consists of iron fines, coke, and flux (eg limestone). The feed is placed on a sintering bed, where thermal agglomeration (1300-1480 °C) takes place to produce clustered lumps, aka iron sinter (5-20 mm in size).

At a common sinter plant, the following sinter quality parameters need to be controlled:

  • Basicity, CaO/SiO2 – FeO (Fe2+)
  • Sinter strength index, SSI
  • Tumbler index, TI
  • Reducibility index, RI
  • Reduction degradation index, RDI
  • Low-temperature degradation, LTD

All the above parameters are linked to the properties of the mineral phases, comprising iron sinter. The main sinter phases can be divided into iron oxides and silicates, which are gluing the iron oxides together:

Iron oxides:Silicates of SFCAs:
Hematite: Fe3+2O3Larnite: Ca2SiO4
Magnetite: Fe3+2+3O4SFCA-a: Silica-ferrites of calcium aluminium, Fe2+ only
Wuestite: Fe2+OSFCA-b: Silica-ferrites of calcium aluminium, Fe3+, some Fe2+

In our first blog we established that X-ray diffraction (XRD) is a fast, versatile, and accurate mineralogy probe, which can be easily implemented in the process flow at mine operation and processing plant. Since all sinter quality parameters are linked to its mineralogy, XRD is a unique tool, providing in a minimum of time a comprehensive assessment of the sinter quality parameters. In the following case study, we discuss the added value of XRD for the process control at a sinter plant.

Accurate analysis of sinter mineralogy using XRD

For this case study, we used 49 sinter samples from a producing sinter plant. All samples were prepared as pressed pellets and were measured on Aeris Metals benchtop diffractometer with a scan time of 5 minutes, followed by an automatic quantitative phase analysis.

Figure 1 shows an example of full-pattern XRD analysis of one of the sinter samples, used in the study.

Figure 1. Typical result of XRD analyses of iron sinter using Aeris Metals. 5 minutes scans followed by the automatic phase quantification.

Any XRD pattern is a set of diffraction peaks of different intensities, located at certain diffraction angles (2q), specific to a certain mineralogical phase. Peak positions enable identification of present phases. The relative intensities of each mineral contribution to the XRD pattern allows us to quantify the relative amount of each present mineral using full-pattern Rietveld refinement [1].

In the example in Figure 1, the sinter sample primarily consists of hematite and magnetite with 23% of crystalline calcio-silicates (so-called SFCA phases) and 21% of the amorphous phase.

Comparing the XRD result of the sample in Figure 1 with the rest of the samples in the set (Figure 2), we see that mineralogy doesn’t change; however, the relative phase amounts differ from sample to sample.

Figure 2. Combined result of quantitative phase analysis of the entire iron sinter sample set.

Using the stoichiometry of the present mineral phases, the sinter FeO content, the most important sinter property since it is connected to the energy consumption in the blast furnace, can directly be extracted from the mineralogical composition (Figure 1, 2).

FeO quantification is done simultaneously with the quantitative phase analysis and is reported along with the phase composition.

A comparison of FeO values, as extracted from the XRD data, with the reference values, obtained by wet chemistry, is shown in Figure 3. We see a very good agreement between the XRD results and the given reference values. Every red data point on this graph takes 10 minutes on average, including automatic sample preparation, 5 minutes scan using Aeris Metals followed by an automatic quantitative analysis. Compared to a few hours of manual sample analyses using hazardous chemicals, the XRD is a fast and safe alternative. Would you still want to use wet chemistry for routine assessment of sinter FeO content?

Figure 3. Comparison of sinter FeO content as obtained from XRD (red circles) with reference values, obtained by wet chemistry (black diamonds).

Getting more from the same XRD data set

In the previous section we established that XRD is a fast and accurate tool for quantitative analysis of sinter mineralogy and determination of sinter FeO content. However, there are other sinter process parameters, which are dependent on the sinter mineralogy (e.g. strength, degradation index, etc.). Can we extract them from the same XRD data set?

The answer is “Yes”, we can by using a modern statistical method, Partial Least Square Regression, (PLSR) [2]. In short, a statistical model is built using a set of reference samples, for which the value of a process parameter (e.g. RDI, SSI, LTD) is known. Afterward, this model is used to predict the process parameter(s) directly from an XRD pattern. This eliminates the need for additional time-intense physical tests and increases the frequency of monitoring.

We applied the PLSR approach to the sinter sample set used in the case study. Part of the set, for which we had reference values of process parameters, was used to build (calibrate) the corresponding PLSR models. Afterward, the obtained PLSR models were used to predict the process parameters directly from the XRD data. Figure 4 summarizes the results for the studied sinter sample set.

Figure 4. results for the studied sinter sample set

Using PLSR approach we obtained sinter strength index SSI, reduction degradation index, RDI, and basicity. The rest of the sinter process parameters, e.g. low-temperature degradation index, tumbler index, reducibility index) can be obtained using the same approach.

Note that both, automatic full-pattern mineral quantification using the Rietveld method and PLSR analysis, can be simultaneously performed on the same XRD pattern. Thus, a single, 5-minute XRD measurement on the Aeris Metals benchtop diffractometer provides the full sinter phase composition along with all-important process parameters.

To summarize, most of the iron sinter quality parameters (aka process parameters) are determined by the properties of the mineral phases. XRD is an indispensable tool for fast, accurate, and tailored mineralogical analysis, which can be easily implemented into the process flow. XRD can be used not only for the fast-quantitative assessment of the full mineralogical composition of iron sinter and its FeO content. New statistical methods (e.g. partial least square regression) opened up new possibilities and enabled the extraction of relevant process parameters directly from the same diffraction data set, eliminating the need for additional time-consuming, costly tests.

To learn more about the benefits of sinter process control by XRD, watch our webinar on-demand discussing PLSR approach to the process control in general with the example on sinter processing added value of XRD and review the application report discussing X-ray diffraction for iron ores sinter analysis.

Additional information can be found in the following papers:

  • König, U., Degen, T. & Norberg, N. (2014): PLSR as a new XRD method for downstream processing of ores – Case study: Fe2+ determination in iron ore sinter. Powder Diffraction, Vol. 29, No. S1, December 2014, pp. S7-S83, DOI: https://doi.org/10.1017/S0885715614001109.
  • König, U. & Norberg, N. (2017): Iron sinter process control using X-ray diffraction – Part 2: Monitoring of low-temperature degradation. Iron Ore Conference / Perth, Australia, AUSIMM 3/2017, 75-78, ISBN 978-1-925100-58-7.

References:

  • [1] H.M. Rietveld, A profile refinement method for nuclear and magnetic structures, J. Appl. Cryst. (1969), 2, 65 – 71.
  • [2] S. de Jong, Simpls: An alternative approach to partial least square regression, Chemometrics Intell. Lab. Syst., (1993), 18(3), 251-263.