New Approach for Direct Determination of Manganese Valence State in Ferromanganese Nodules by X-ray Fluorescence Spectrometry

: A new X-ray fluorescence technique is proposed as an alternative to the labor-consuming volumetric method for the estimation of manganese valence state in ferromanganese nodules. The approach is based on the measurement of the relative intensities of some X-ray fluorescence spectrum characteristic spectral lines and satellites (MnKβ 5 and MnKβ′) preconditioned by electron transfer from the valence shell. Calibration curves were created using manganese oxide samples (MnO, Mn 2 O 3 , MnO 2 ) and 12 certified reference materials of ferromanganese nodules, cobalt-bearing ferromanganese crusts, and manganese ores with certified (or determined by approved meth-ods) total and tetravalent manganese content. The presence of high iron content was taken into account. A set of oceanic ferromanganese nodules samples collected in the Magellan Seamounts (Pacific Ocean) were analyzed. Differences between the results of the X-ray fluorescence method and volumetric techniques for tetravalent manganese content were 4.9 rel.%, which is comparable with the accuracy of the volumetric technique (3.6 rel.%).


Introduction
In geological studies, the determination of the valence state of transition elements is the basis for assessing the redox conditions of rock formation [1] and, from a practical point of view, it is important for assessing the prospects for mining ores.Of all transition metals, there is only a routine method for determining the valence state of iron in geological objects.In mineral objects, iron has two main valence states, except metallic iron.Having information about at least one of them (divalent iron Fe II or trivalent iron Fe III ) and about total iron (Fe total ) content, it is possible to characterize iron valence state, for example as a ratio of ferrous (or ferric) and total iron.The determination of ferrous iron is widespread for the analysis of rocks using volumetric or spectrophotometric techniques [2][3][4][5].
In contrast to iron determination, the estimation of manganese valence state can be difficult.Manganese has three valence states in mineral objects (divalent manganese Mn II , trivalent manganese Mn III , and tetravalent manganese Mn IV ), and for the full characterization of the valence state besides total manganese (Mn total ) content, data on two of them are required.Determination of Mn total is a routine procedure.However, techniques for the diversification of manganese by valence states are rarely used [5][6][7][8].These techniques are not as common as ones for ferrous iron and are used only in the analysis of manganese ores, since the rocks contain relatively low manganese.One of the types of specific manganese ores are ferromanganese nodules and crusts, which are a promising source of valuable metals (in addition to Mn and Fe, also Co, Ni, Cu, Zn, Pb, and rare earth and precious metals), despite the relatively difficult mining conditions [9,10].When studying nodules and crusts, the valence state of ore elements is important for fundamental and application research and for assessing the ability of nodules to absorb valuable trace elements [11].Despite the dominance of oxidized Mn IV (as asbolane, vernadite, todorokite, birnessite, buserite) and Fe 3+ (as oxides and hydroxides: goethite, limonite, ferroxigite, ferrihydrite, etc.), nodules can also contain other valence states of iron (Fe II as ilmenite or impurities in amphibole or pyroxene) and manganese (Mn II as impurities in todorokite and buserite and sometimes Mn III as impurities in birnessite) [12][13][14].The X-ray fluorescence (XRF) method is widely used for the analysis of geological objects, including ferromanganese nodules and crusts [15][16][17].Currently, the XRF method allows us not only to determine the total content of an element, but also to evaluate the forms of transition elements, including iron and manganese, based on the dependence of spectral lines intensities on the valence state of an element [18][19][20].Parameters such as energy, intensity, and shape of lines preconditioned by the transitions of electrons from valence levels are the most affected by chemical bonding.There are Kβ5 line and the most intense satellite Kβ′ for iron and manganese [21][22][23][24][25][26][27][28][29][30][31][32].
Despite our rather successful experience in the development of methods for the XRF determination of Fe II in different types of igneous rocks [32], this approach seems to be not applicable to ferromanganese nodules.There is no information on the content of ferrous iron (FeO) in certified reference materials (CRMs) of ferromanganese nodules.Besides, routine volumetric determination of ferrous iron in samples of ferromanganese nodules (as well as of ferromanganese ores) is impossible due to interference caused by high manganese content.Thus, the valence state of iron in ferromanganese nodules can only be assessed qualitatively as "close to Fe III " [31].However, in the case of manganese, the XRF method looks more promising, despite the relatively small (compared to rocks) number of CRMs of ferromanganese nodules [33].The content of MnO2 was certified (according to the certificate, by single visual titrimetry) in some CRMs of the SDO series [34]; moreover, the determination of MnO2 in ferromanganese nodules can easily be carried out by reverse titration.In this work, the possibility to use the relative intensities of the MnKβ5 line and MnKβ′ satellite of the XRF spectrum to determine manganese valence state in oceanic ferromanganese nodules is considered.

Equipment
The investigation was carried out using the equipment of the Center for Isotopic and Geochemical Research (Vinogradov Institute of Geochemistry, Siberian Branch of the Russian Academy of Sciences, Irkutsk, Russian Federation) [35].XRF spectra were studied using a S4 Pioneer wavelength-dispersive spectrometer (Bruker AXS, Karlsruhe, Germany) equipped with a Soller optical system, scintillation counter, and an X-ray tube with an Rh anode.All measurements were performed in vacuum mode using a lithium fluoride (LiF (220), 2d = 2.848 Å) crystal combined with a 0.23° collimator.A voltage of 50 kV and an X-ray tube current of 40 mА ensured a sufficient intensity of radiation.For spectra measurements, the step size was 0.024°, and the measurement time for each point was 4 s.

Samples
To obtain the calibration, a set of CRMs was used.CRMs of ferromanganese nodules and cobalt-bearing ferromanganese crusts provided by the N. Fedorovsky All-Russian Scientific Research Institute of Mineral Raw Materials (CRM 408-10: GMK-1 and GMK-2, CRM 409-10: KMK-1 and KMK-2); the Research Institute of Applied Physics of Irkutsk State University together with the P.P. Shirshov Institute of Oceanology of the Russian Academy of Sciences (OOPE601 (SDO-4), OOPE602 (SDO-5), OOPE603 (SDO-6) and OOPE604 (SDO-7)) [34], sample certified during GeoPT proficiency testing program (FeMn-1 (GeoPT 23a)) [36]; and the Central Laboratory of Mongolia as a part of interlaboratory testing for certification (MnN) [37]; as well as three pure-grade manganese oxides (MnO, Mn2O3, MnO2).Despite the large number of certified elements, only certificates for four CRMs of the SDO series [34] as well as a certificate for CRM MnN contained semiquantitative data on the mineral composition (Table 1).For CRM 408-10 (GMK-1 and GMK-2), vernadite, todorokite and birnessite were noted as the main Mn-containing minerals; aluminosilicates, plagioclase, and quartz were noted as the rock-forming minerals.For CRM 409-10 (KMK-1, and KMK-2), vernadite and apatite were noted as the main minerals; silicates, zeolite, quartz, and calcite were noted as the rock-forming minerals.For CRM FeMn-1, no data on mineral composition were found.Certificates of only four of the listed ten CRMs contain data allowing us to estimate manganese valence state (MnO2 iscertified in CRMs of SDO series); therefore, the other 8 samples were analyzed by volumetry (the technique will be described below).The calibration set was expanded using CRMs of manganese ores 5404-90 and 5407-90 with certified content of MnO2 that contained predominantly Mn II -and Mn IV -containing minerals, including pyrolusite and psylomelane, as well as oxide and silicate manganese minerals (jacobsite, rhodonite, rhodochrosite, coronadite).CRMs of manganese ores were provided by Tsentrgeolanalit LLP (Karaganda, Kazakhstan).Other available CRMs of manganese ores contained a lot of Mn III as hausmanite, manganite, and braunite, and were not included in the calibration set.
For validation, a set of 15 samples of oceanic Co-bearing ferromanganese crusts collected on the Govorov and Kotzebue Guyots (Magellan Seamounts, Pacific Ocean) [38,39] were used.
Since the material of nodules and crusts is highly hygroscopic, all studied samples were dried at 120 °C for 48 h to constant weight to obtain air-dry samples.The content of hygroscopic moisture in the studied samples reached 15 wt.%.After drying, samples were pressed on a boric acid substrate and kept in a desiccator.

X-ray Fluorescence Kβ Spectra of Manganese
According to its mineral composition, tetravalent manganese Mn IV , as a compound of asbolane, vernadite, and todorokite, is the main compound of ferromanganese nodules.Figure 1 shows an XRF spectrum of a Mn IV oxide (MnO2) sample in the energy range 6.42-6.57keV.The most intense peak corresponds to the doublet MnKβ1,3 line (theoretical energy is 6490 eV here and below, according to the NIST database [40]) preconditioned by transitions К-М2,3.This line has visual asymmetry in a low energy area that is related to the presence of relatively intense MnKβ′ satellite interpreted as transitions К-М2,3 (main MnKβ1,3 line) with a simultaneous ejection of an electron from level М4,5.The relatively intense peak in the energy range 6.52-6.54keV is the MnKβ5 line (theoretical value is 6535 eV) preconditioned by transitions К-М4,5.According to [21], there are some satellites near the MnKβ5 line: MnKβ″ (~6520 eV) and MnKβ′′′ (~6550 eV).Therefore, in fact, the peak in 6.52-6.54keV energy range is a superposition of several lines.However, some of these lines do not appear for some compounds, and the resolution of the spectrometer is insufficient to split these peaks visually.Hence, for further study, the peak in the energy range 6.52-6.54keV was considered as the MnKβ5 line.To find a position for intensity measurement, a decomposition procedure was performed for the MnO2 sample using the pseudo-Voigt function as a fitting function [41]: where  According to decomposition, energy positions for analytical lines measurement were found as 6529 eV (MnKβ5) and 6469 eV (MnKβ′).Because the intensity of any line of the manganese XRF spectrum depends on the gross manganese content, the ratios of selected lines to the intensity of line MnKβ1,3 were used as analytical parameters.To provide count rate statistic error at a level less than 1 rel.%, the exposition of measurement was defined as 10 s for the MnKβ1,3 line, 50 s for the MnKβ′ line, and 100 s for the MnKβ5 line.Background intensity was measured in the 6590 eV energy position, where intense analytical lines of elements contained in ferromanganese nodules and crusts are absent.
Figure 2 shows the dependence and good convergence of the ratios of the MnKβ5/MnKβ1,3 and MnKβ′/MnKβ1,3 line intensities on the manganese valence state (N Mn ) for manganese oxides.A polynomial equation was used for approximation.Unlike oxide, silicate, and carbonate ores, ferromanganese nodules have a specific phase and elemental composition; in particular, they contain high amounts of iron (in some cases exceeding the manganese content), which can significantly affect manganese line intensity due to possible spectral overlap with FeKα1,2 line.Figure 3 shows the XRF spectrum of the CRM of ferromanganese nodule SDO-4 with a total iron content of 12.05 wt.% and a total manganese content of 23.17 wt.%.The spectral overlap between the FeKα1,2 line (theoretical energies are 6.391 keV and 6.404 keV) and manganese spectrum lines under consideration seems insignificant.But in cases where iron content is comparable to or exceeding manganese content (for example, in the CRM of ferromanganese nodule SDO-7, total iron and total manganese content is quite the same: 15.49 wt.% and 15.38 wt.%, respectively), the high intensity of FeKα1,2 lines can distort the measured intensities of selected manganese spectrum analytical lines.Moreover, the FeKβ line (7058 eV) excites the secondary fluorescence of manganese (energy of manganese K edge is 6538 eV).There is a reason why, unlike silicate and carbonate manganese ores [30], the investigation of manganese valence state in ferromanganese nodules requires taking into account iron presence.The exposition of measurement of FeKα1,2 line was 10 s.

Manganese Valence State
Because quantitative data on different valence state of manganese content in nodules are difficult to obtain, the only source for manganese valence state estimation is mineral composition, usually obtained by X-ray powder diffractometry.However, it is hard to find quantitative data on the mineral composition of nodules: the results of X-ray powder diffractometry analysis are often only qualitative, in particular due to the impossibility to identify amorphous iron and manganese minerals.Even CRMs often contain only semiquantitative data (Table 1).
Taking into account the simultaneous presence of three main manganese valence states (Mn II , Mn III , and Mn IV ), the average manganese valence state of manganese can be calculated as: where С(Mn II ), С(Mn III ), and C(Mn IV )-content of divalent, trivalent, and tetravalent manganese, accordingly; 2, 3, and 4-manganese valence states.According to data in the literature, nodules contain mainly Mn IV -containing minerals with low quantities of Mn II -containing minerals, usually in a mixture with Mn IV [12][13][14]; therefore, N Mn can be calculated only from the content of Mn IV and Mn II (as the difference between the content of Mn total and Mn IV ), accepting that content of Mn III is negligible.
Nonetheless, some minerals contain Mn III combined with Mn II and Mn IV .For example, CRMs SDO-5 and SDO-6 contained up to 8% of buserite and up to 5% of birnessite (which together can contain up to 12% of Mn III ).This fact may have affected the accuracy of N Mn value calculation in Formula (2) if only Mn II and Mn IV content were used.For these two CRMs, N Mn values were calculated using Formula (2), assuming that both CRMs samples contained the maximal possible (taking into account phase composition, Mn total , and Mn IV content) or no content of Mn III .Due to the fact that the CRM SDO-5 has certified MnO2 (31.01 ± 0.4 wt.%) and MnO total (25.16 ± 0.28 wt.%) content, recalculating to Mn total and Mn IV (due to the relatively wide confidence interval for MnO2) leads to the conclusion that its content of Mn IV exceeds its content of Mn total (which is obviously impossible).Therefore, for the calculation, the highest and the lowest values in the confidence interval for Mn total and Mn IV content, respectively, were used.The calculated N Mn values were 3.95 and 3.90 (relative deviation is 1.3%) for CRM SDO-5 and 3.98 and 3.97 (relative deviation is 0.4%) for CRM SDO-6.These discrepancies will be discussed further.
To use all listed CRMs of ferromanganese nodules and crusts as a calibration set (and to analyze real samples of Co-bearing ferromanganese crusts), a certified volumetric technique was applied for the determination of MnO2 content.The technique used is based on the reduction of MnO2 by ammonium iron (II) sulfate Fe(NH4)2(SO4)2 and further titrating by potassium dichromate K2Cr2O7 with sodium diphenylamine-4-sulfonate as the redox indicator.
To validate the volumetric technique, four CRMs of the SDO series (SDO-4, SDO-5, SDO-6, and SDO-7) with certified MnO2 content were analyzed (Table 2).Repeatability was estimated by two independent measurements of CRMs SDO-4 and SDO-5; the deviations between results were 0.06 and 0.13 wt.%, respectively (less than 0.5 rel.%).A systematical underestimation from −0.7 to −1.6 wt.% (from 2.5 to 4.8 rel.%) of MnO2 results obtained by the volumetric technique is observed; the standard deviation value for analyzed CRMs is 1.2 wt.% (3.6 rel.%), whereas the confidence interval of certified values is 0.4-0.5 wt.% (1.1-1.7 rel.%).Table 2 also contains the N Mn value calculated using Formula (2) from MnO2 content obtained by volumetry or certified MnO2 content, assuming that the content of trivalent manganese C(Mn III ) is insignificant.The relative standard de-viation (RSD) for N Mn determination is 1.7%.In comparison with the case where N Mn values were calculated for CRMs SDO-5 and SDO-6, taking into account the possible presence of trivalent manganese Mn III (1.3 and 0.4 rel.%), errors in volumetric analysis were significantly higher (1.9 and 2.0 rel.%).Therefore, for further studies, it was accepted that ferromanganese nodules contain only divalent and tetravalent manganese (Mn II and Mn IV ), which corresponds to previous studies [12][13][14].Table 3 shows the results of MnO2 determination in 6 CRMs of ferromanganese nodules and Co-bearing ferromanganese crusts.

Calibration
To build calibration curves, 12 CRMs of ferromanganese nodules, Co-bearing ferromanganese crusts, and ores as well as three samples of manganese oxides were used.Ratios of MnKβ5/MnKβ1,3 and MnKβ′/MnKβ1,3 line intensities were used as analytical parameters.The dependence of manganese valence state N Mn on analytical parameter R I was approximated by different variants of the equation: where N Mn -average manganese valence state, R I -analytical parameter (ratio of analytical lines intensities), I(FeKα1,2)-the intensity of the FeKα1,2 line included to take into account the presence of high iron content, a0, a1, a2-coefficients of regression.
To discover the necessity of taking into account the presence of high iron content (as a possible spectral overlap of the FeKα1,2 line and the exciting of secondary fluorescence of manganese by the FeKβ line), Equation (3) with and without including the intensity of the FeKα1,2 line was also considered.The parameters of calibration curves (standard deviation (SD) and correlation coefficient (rxy)) are shown in Table 4.Despite the fact that dependencies for three manganese oxides were approximated by a polynomial equation (Figure 2) for both the MnKβ5/Mnβ1,3 and MnKβ′/Mnβ1,3 analytical parameters, for CRMs, polynomial equations did not lead to a significant decrease in SD value or significant increase in rxy value.As can be seen in Table 4, without taking into account the influence of iron presence, both MnKβ5/Mnβ1,3 and MnKβ′/Mnβ1,3 provide the same poor accuracy (SD > 0.3, rxy < 0.70), whereas taking into account iron presence allows us to reduce SD value up to 2.5 times.The minimal SD value (and sufficient rxy value more than 0.96) is observed when both analytical parameters are used.However, in this case, the added intensity of the FeKα1,2 line as an additional member to Equation (3) does not allow us to significantly reduce SD value.Probably, it is the close influence of iron presence for both the MnKβ5 and MnKβ′ lines that compensates for this influence.Thus, four analytical parameters providing minimal SD values were used for further validation: ratios of MnKβ5/Mnβ1,3 and MnKβ′/Mnβ1,3 line intensities (with the intensity of the FeKα1,2 line as additional member), and the combination of these ratios (equations are shown in Supplementary Table S1).All selected equations have the same degrees of freedom (12).Figure 4 shows the correlations between N Mn values obtained from XRF analysis using different analytical parameters and ones calculated using Formula (2) for the calibration set.  4 and Table S2: Equation (S2) (a), Equation (S4) (b), Equation (S5) (c), Equation (S6) (d) and calculated using Formula (2) for the calibration set.Black squares are manganese oxides, blue dots are CRMs of ferromanganese nodules, Co-bearing ferromanganese crusts, and manganese ores.

Validation
For the validation of the proposed XRF technique, 15 samples of oceanic Co-bearing ferromanganese crusts collected on the Govorov and Kotzebue Guyots (Magellan Seamounts, Pacific Ocean) were studied.The content of Mn total was determined by the XRF method [17] and varied from 16.5 to 26.0 wt.%.Average manganese valence state was calculated from the volumetric results of MnO2 determination and XRF results of Mn total determination using Formula (2), as well as directly determined by the proposed XRF technique using selected analytical parameters.Results are shown in Supplementary Table S2.
rxy values between the obtained N Mn values were not more than 0.47, but this result is not informative for such a thin range of values (3.61-4.00).It can be concluded that most samples contained predominantly Mn IV ; however, in sample 3, N Mn was minimal in both XRF (3.61 using MnKβ5/Mnβ1,3 as an analytical parameter) and volumetric (3.59) results.
Table 5 shows a comparison of average valence state (N Mn ) values and recalculated MnO2 content obtained by volumetric and proposed XRF techniques.As was concluded for manganese ores [41], the ratio of MnKβ′/Mnβ1,3 line intensities is influenced by manganese speciation, but for samples with relatively close mineral composition, this parameter is optimal.The RSD value between N Mn results was 2.3%, which is comparable with the volumetry error (1.7 rel.%).When recalculating to MnO2 content, SD was 0.8 wt.% (4.9 rel.%), which is comparable with a confidential interval of certified values (0.4-0.5 wt.%, or 1.3 rel.%).

Conclusions
The XRF method has been presented as a fast and cost-efficient method for estimating the valence state of manganese in ferromanganese nodules, where manganese occurs mainly as Mn II and Mn IV .The accuracy of the proposed method was 4.9 rel.%, which is comparable with the accuracy of volumetric analysis (3.6 rel.%).The XRF method can be used to determine the valence state of manganese (simultaneously with the elemental XRF analysis of nodules) as an alternative to the labor-and time-consuming volumetric method.

Citation:
Chubarov, V.M.New Approach for Direct Determination of Manganese Valence State in Ferromanganese Nodules by X-ray Fluorescence Spectrometry.Minerals 2023, 13, 1329.https://doi.org/102023 by the author.Licensee MDPI, Basel, Switzerland.This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/license s/by/4.0/).
is Lorentzian; A is the amplitude of the distribution centered at energy Ep; σg and σL are parameters defining the widths of the Gaussian and Lorentzian functions; Cg determines the portion of Gaussian and Lorentzian and varies in the interval (0.1).

Figure 1 .
Figure 1.X-ray fluorescence spectrum in the field of MnKβ1,3 line for manganese (IV) oxide sample (MnO2).Dots are measured spectrum; lines are results of decomposition using the pseudo-Voigt function.

Figure 3 .
Figure 3. X-fluorescence spectrum in the field of MnKβ1,3 and FeKα1,2 lines of the CRM of ferromanganese nodule SDO-4).Dots are measured spectrum; lines are results of decomposition using the pseudo-Voigt function.

Table 1 .
Mineral composition of certified reference materials of ferromanganese nodules, %.

Table 2 .
Results of MnO2 determination by volumetric technique (MnO2 vt ) in comparison with certified values (MnO2 cert ).

Table 3 .
MnO2 content and average manganese valence state (N Mn ) determined by volumetric technique for CRMs of ferromanganese nodules.

Table 4 .
Parameters of calibration curves.