Elemental Concentrations of Natural Graphite and Steelmaking Slag: Development of Microwave-Assisted Acid Digestion

Abstract To develop more sustainable purification and recovery processes, it is critical to accurately determine inorganic impurities of natural graphite (NG) and valuable elements present in low concentrations in steelmaking slag. This study applied three microwave-assisted acid digestion methods for basic oxygen furnace (BOF) slag, NG, and NG ash. Nitric acid (HNO3), hydrochloric acid (HCl), and hydrogen fluoride (HF) were used in method 1. In method 2, method 1 was followed by microwave-assisted boric acid (H3BO3) digestion. In method 3, acids (HNO3, HCl, HF, and H3BO3) were added simultaneously. Concentrations were measured using inductively coupled plasma–optical emission spectrometry. Analytical performance was evaluated using spiking recovery tests (NG) and certified reference (slag) material (CRM). Spiking recoveries for Al, Ca, Fe, K, Mg, Na, and Si were between 5% and 116% (method 1) and 94%–116% (methods 2 and 3). CRM recoveries for Al, Ca, Fe, Mg, Mn, P, Si, Ti, and V were between 2% and 102% (method 1) and 92%–103% (methods 2 and 3). The unsuitability of method 1 was evident due to the low recoveries. Methods 2 and 3 digested Ca- and Si-containing BOF slag completely without visible residue. The sufficient recoveries verified the adequacy of the methods. The suitability of method 3 probably depended on the phases of samples, as quartz was not completely digested from the NG. Method 2 completely digested the inorganic material of NG, and the spiking test showed no loss of volatile silicon compounds. These methods have not been published for NG or NG ash and validated for slag before.


Introduction
The industry has many difficult sample matrices, such as graphite and slag samples.To utilize these in their applications and to develop more sustainable processes, it is essential to determine accurately the concentrations of impurities and the chemical composition.
NG is used in various applications, such as lithium-ion batteries, refractories, electric vehicles, and lubricants (Jara et al. 2019;Chelgani et al. 2016).The demand for high-purity graphite has grown in the last decade and will continue to do so.The beneficiation of NG is usually done via froth flotation.However, by flotation, the carbon (C) content can be upgraded to 95 wt%.Chemical purification is the most common method to obtain high-purity graphite after flotation (Chelgani et al. 2016).Yet, the most commonly used methods (i.e., hydrogen fluoride (HF), acid-base methods) cause environmental challenges, and there is a growing need to develop alternative purification approaches.
In the steel industry, steel slag is formed as a side-stream.In 2021 the industry was estimated to produce steel slags, mainly electric arc furnace (EAF) and basic oxygen furnace (BOF) slag, between 190 and 280 million tons worldwide (Geological Survey U.S. 2022).Slag formation is an important stage in the steelmaking process because in this stage most impurities are separated from the molten steel (Z.Li, Guo, et al. 2022).The raw material and additives, the process conditions, and the steel quality affect the slag's chemical composition and mineralogy (Z.Li, Guo, et al. 2022;Y. Li, Guo, et al. 2022;Schollbach, Ahmed, and van der Laan 2021).These slags contain several critical and valuable metals to recover.Utilizing these secondary raw materials can save natural resources and decrease the amount of waste in industry.
Commonly the main impurities in NG after flotation are silicate minerals containing elements such as aluminum (Al), calcium (Ca), iron (Fe), potassium (K), magnesium (Mg), sodium (Na) sulfur (S), and silicon (Si) (Shen et al. 2021;Wang et al. 2016;Chelgani et al. 2016).The major components in steel slag are Ca, Fe, and Si.In addition to these, the slag contains varying amounts of several elements such as Al, chromium (Cr), Mg, manganese (Mn), phosphorous (P), titanium (Ti), and vanadium (V) depending on the used steelmaking process and the raw materials (Ragipani, Bhattacharya, and Akkihebbal 2020;Mombelli et al. 2016).Currently, the chemical composition of slags is often determined using X-ray fluorescence (XRF) (Hobson et al. 2017;Yadav and Mehra 2017;Stewart et al. 2018;Ragipani, Bhattacharya, and Akkihebbal 2020).In some studies, XRF has been used for NG samples (Wang et al. 2016(Wang et al. , 2018;;Chen et al. 2022;Jara and Kim 2020;Ri et al. 2022;Zhao et al. 2022).XRF is a fast and widely used method for analyzing the chemical composition of solid samples.However, elemental concentrations are estimated relative to each other, and XRF does not detect light elements (Margu� ı and van Griekenn 2013).
Inductively coupled plasma-optical emission spectrometry (ICP-OES) can be used to measure small concentrations, even at ppb levels and wide concentration ranges, but it is only suitable for liquid samples.As a pretreatment procedure, microwave-assisted acid digestion is commonly used for solid materials.However, it is extremely difficult to digest NG to determine its impurity levels due to graphite's stable structure (Cruz et al. 2015).In addition, both graphite and slag samples contain Si, making it difficult to fully digest the inorganic materials without using HF.The problem with HF is the formation of insoluble compounds with elements such as Al, Ca, Mg, and barium (Ba) (Muratli et al. 2012;Han et al. 2021).Boric acid (H 3 BO 3 ) (Wu et al. 1996;Wilson, Burt, and Lee 2006;Al-Harahsheh et al. 2009), perchloric acid (HClO 4 ) (Han et al. 2021), and phosphoric acid (H 3 PO 4 ) (Prasad et al. 2023) have been studied for different Si-containing materials to prevent the formation of insoluble compounds.
The traditional microwave-assisted acid digestion method US EPA method 3052 (United States Environmental Protection Agency 1996) for Si-containing samples uses nitric acid (HNO 3 ), hydrochloric acid (HCl), and HF.In addition, in the European Standard SFS-EN 13656:2020(Finnish Standards Association SFS 2020) tetrafluoroboric acid (HBF 4 ) was directly used by microwave-assisted digestion with HCl and HNO 3 for several sample types.In the standard, steel slag was used for the robustness validation tests, but recoveries were not calculated.Furthermore, the described methods have not been validated either for NG samples.Literature offers only a few studies in which microwave-assisted acid digestion is used for steel slag and other types of graphite than NG (Coedo and Dorado 1995;Teir et al. 2007;Simoes et al. 2016;Yang et al. 2021;Ambrosi et al. 2012;Watanabe and Narukawa 2000;Cruz et al. 2015).
Based on the arguments presented above, it is obvious that there is a need for an investigation of the functionality of microwave-assisted acid digestion for NG and slag samples.This research aims to find a suitable digestion method for determining the inorganic impurity concentrations of NG and for determining the elemental concentrations of high Ca-and Si-containing BOF slag accurately by ICP-OES.The challenge for using the traditional decomposition method (US EPA 3052) comes from the Al, Ca, Fe, and Mg, which form insoluble fluoride compounds during digestion.H 3 BO 3 is assumed to solve this challenge.
In this study, three different microwave-assisted acid digestion methods are applied to digest the inorganic material of NG and BOF slag.Method 1 is a traditional method using HNO 3 , HCl, and HF.In methods 2 and 3, H 3 BO 3 is used to complex excess fluoride ions causing the dissolution of insoluble fluoride compounds.Method 2 is done in two stages, and in method 3, all acids are added simultaneously.The graphite's inorganic material (ash) is also decomposed using the same methods to visually observe the precipitation formation and to compare the results to NG. Certified reference (slag) material (CRM) and spiking recovery tests are used to validate the methods used.Statistical evaluation between methods is applied through one-way analysis of variance (ANOVA).The elemental concentrations of 15 elements are measured using ICP-OES and BOF slag concentrations are compared with XRF results.C contents of NG residues are measured using an elemental (CHNS/O) analyzer.The remaining precipitations are characterized using X-ray diffraction (XRD) and field emission scanning electron microscopy with an energy dispersive X-ray Spectrometer (FESEM-EDS).

Solid samples and reagents
NG extracted in Finland (Kaukkala, P€ alk€ ane) was obtained by flotation in the Oulu Mining School pilot process.Inorganic material, namely the ash of the NG sample was obtained by burning 10 g of NG in a muffle furnace (Nabertherm GmbH, L5/11/P320, Germany) at 950 � C.
BOF slag (12.5 microns, D50) was obtained from a steel mill in northern Finland.ECRM 805-1 (Institut de Recherches de la Siderurgie, IRSID, France) was used as a CRM (basic slag) for the BOF slag.

Microwave-assisted acid digestion
The samples were decomposed with three different methods via microwave-assisted (Milestone microwave, ETHOS UP) acid digestion adapted from US EPA method 3052 standard procedure (United States Environmental Protection Agency 1996) described in Table 1.HF-resistant components and labware were used.The main purpose of H 3 BO 3 was to dissolve and prevent the formation of insoluble fluoride compounds by complexing excess fluoride ions.
Replicate samples (n ¼ 2-5) were prepared by weighing 300 mg (NG, BOF, CRM) or 150 mg (NG ash) into PTFE digestion vessels.Different sample types were processed separately with the microwave to ensure a constant temperature of 180 ± 5 � C. Digestion blanks were made for each microwave-assisted digestion run.After the microwave treatment, the vessels were allowed to cool to room temperature for at least an additional 90 min.In methods 1 and 3, all acids were added to the vessels before the microwave treatment.In method 2, microwave-assisted acid digestion was completed in two stages.First, the samples were digested using HNO 3 , HCl, and HF (stage 1).After cooling, 4.5 wt.% H 3 BO 3 was added, and the digestion program was driven again (stage 2).The solutions were filtered through filter paper (Whatman, 589/2 ashless/White ribbon, / 110 mm) into volumetric flasks (100 mL, polypropylene) and diluted with ultrapure water.

ICP-OES analysis and characterization
The ICP-OES measurements were performed by Agilent 5110 VDV (Agilent Technologies, USA).HF-resistant components were used in the ICP-OES measurements.The analyzed elements were Al (237.312nm (NG, ash), 396.152 nm (slags)), Ca  ; 213.857 nm).External calibration was done using matrix-matched standard solutions.Calibration was accepted with a correlation coefficient �0.999.Matrixmatched blanks and quality control samples were used in all measurements.Yttrium (Y; 1 mg/L, 371.029 nm) was used for internal standardization.
To compare the ICP-OES results of the NG and ash samples, the elemental concentrations [mg/g] of the ash samples (c NG ash ) were calculated to correspond to the original NG concentrations using Equation ( 1), as follows: where c ash [mg/g] is the concentration of the measured elements in the ash sample, m NG [g] is the mass of the NG sample before heat treatment, and m ash [g] is the mass of the obtained ash.
The instrumental limits of detection (LOD) and quantitation (LOQ) were determined with ICP-OES.The LOD was calculated by using Equation ( 2) and LOQ by Equation (3).
where SD is the standard deviation of the blank (n ¼ 5) and S is the slope of the calibration curve.One-way ANOVA was used to statistically compare the means of elemental concentrations of the three methods at a 95% confidence level.
The mineralogy and phases of the samples were characterized using a PANalytical X'Pert Pro XRD (Malvern Panalytical, Almelo, Netherlands).The XRD patterns of the NG sample were collected at 40 mA and 45 kV, with 2h values from 6 � to 90 � by Cu K a1 radiation k ¼ 1.54060 � Å Similarly, the XRD patterns of the residues were collected at 10 mA and 40 kV.The XRF was measured by a PANalytical Axios mAX 4 kW wavelength dispersive XRF spectrometer (Malvern Panalytical, Almelo, Netherlands) using loose powders through mylar film in He-atmosphere.
The morphology and composition of the formed insoluble residues of the NG ash and BOF samples were characterized by FESEM-EDS using a Zeiss Ultra Plus (Carl Zeiss Microscopy GmbH, Jena, Germany).The elemental mapping and composition analysis was operated at 15 kV and a working distance of 8.5 mm using AZtec software from Oxford Instruments.The C, nitrogen (N), hydrogen (H), sulfur (S), and oxygen (O) contents of the untreated NG and the C contents of NG residues were measured using a CHNS/O analyzer (Thermo Scientific FlashSmart Elemental Analyzer, Italy).

Spiking recovery tests
Due to a lack of suitable CRM, spiking recovery tests were employed to validate the methods used for the NG sample.Elements whose concentration in the NG was >1 mg/g were selected for investigation.Spiking solutions of Al (47.0 g/L), Ca (9.9 g/L), Fe (157.6 g/L), Mg (18.8 g/L), Na (12.7 g/L), K (10.6 g/L), and Si (24.5 g/L) were prepared by diluting salts with ultrapure water.Microwave digestion (Table 1) was performed for all samples at the same time.Spike recoveries (R [%]) were calculated by:

Characterization of the samples
The BOF slag characterization was presented in a previous publication.

Comparison of different digestion methods for natural graphite
There was no suitable CRM available for the NG sample, and spike recovery tests were therefore conducted to validate the studied methods.The spiking recovery tests help to detect whether there are losses of the studied elements during the investigated methods.The blank samples (n ¼ 4) and NG samples (n ¼ 5) were spiked.Spiking recoveries of the spiked NG samples were calculated using Equation (4).The recoveries are shown in Table 2. Method 1 was not suitable for Al, Ca, Mg, and Na determinations, since spiked NG samples had losses of those elements.Yang et al. (2021) decomposed spent graphite after calcination using a mixture of HNO 3 , HF, and hydrogen peroxide (H 2 O 2 ) (Yang et al. 2021).However, the possible precipitation of AlF 3 was not considered.Recoveries of the spiked NG samples of methods 2 and 3 imply that there were no losses of the studied elements when H 3 BO 3 was added.
The NG sample was not fully decomposed during microwave digestion due to the stable structure of graphite (Cruz et al. 2015).Due to that, ashing was performed to observe the decomposition of the inorganic impurities.Upon filtration of the NG ash sample, it was observed that a white precipitate had been formed during digestion method 1. White residue was also observed after method 3, although the quantity was smaller.The inorganic material in a sample must be completely digested to determine the total amount of impurities.Method 2 left no visible residue.
ICP-OES measurements of all filtered solutions and a comparison of different methods by ANOVA are shown in Table 3.Even though residue was left when digestion   methods 1 and 3 were used, it seems that Cr, Cu, Mn, Mo, P, and Zn can be measured by using any of the digestion methods based on the ANOVA.By comparing the different methods (Tables 2 and 3), it can be observed that the precipitate of method 1 contained at least Al, Ca, Mg, and Na (i.e., elements that form insoluble compounds with HF) (Muratli et al. 2012;Han et al. 2021).The characterization of the precipitates is presented in detail above.
In methods 2 and 3, excess H 3 BO 3 was added to dissolve the insoluble fluoride compounds.Method 3 did not dissolve Si completely, based on the NG and NG ash ICP-OES results (Table 3) and the visible residue of the NG ash sample.In method 3, H 3 BO 3 was added before HF.It seems that HF was consumed by reacting directly with H 3 BO 3 before the dissolution of quartz by HF was complete.A likely explanation for this observation lies in the fact that the latter reaction is a two-phase reaction, whereas the former takes place in the solution.The mass transfer through the solid-liquid interface was probably much slower than the reaction with HF and H 3 BO 3 in the neighboring solution (Gr� enman, Salmi, and Yu.Murzin 2011).In the spiking recovery test (Table 2), there was no loss of spiked Si.Si was spiked as a solution, while in the NG sample, it was in the quartz phase.Spiking tests may prove that Si is not lost as volatile silicon tetrafluoride (SiF 4 ), which may be a problem in open-vessel systems (Sawhney and Stilwell 1994;Jones and Dreher 1996).Open microwave digestion could be a suitable method for graphite materials that do not contain Si as an impurity, as Watanabe and Narukawa ( 2000) pointed out in their study.The impurities of high-purity graphite were determined by decomposing the graphite sample completely with a mixture of HNO 3 and H 2 SO 4 (Watanabe and Narukawa 2000).Ambrosi et al. (2012) mentioned in their study that it is possible that some of the impurities may stay trapped in the graphite crystals when microwave digestion is used.However, they used only concentrated HNO 3 for the digestion of commercial graphite (Ambrosi et al. 2012).It seems that this was not the case in our study, as the elemental concentrations of NG and NG ash for the most part corresponded with each other (method 2).The most significant differences were in the Ca, Cr, and Na concentrations.However, the concentrations of Cr (0.08-0.11 mg/g) were very low.The heterogeneity of the samples should also be considered.Otherwise, it would seem that the ashing step is not required to determine the impurities in NG.At least, the determination of all the main impurities directly from the NG seemed to be possible with method 2. Direct digestion of the sample is preferable because ashing requires a much larger sample size.The heterogeneity of the samples is a likely reason for the rather poor repeatability obtained for some elements (high RSD values).There are no similar microwave-assisted acid digestion studies of NG in the literature.Nonetheless, method 2 has previously been confirmed to be suitable for the digestion of different materials, such as soils (Wilson, Burt, and Lee 2006).Based on the results in Tables 2 and 3, method 2 is the most suitable digestion method for impurity determination in NG.

Comparison of digestion methods for steelmaking slags
For the BOF slag, there was a suitable CRM available.Table 4 shows the elemental concentrations of CRM measured by ICP-OES and the statistical comparison of different methods by one-way ANOVA.Based on ANOVA there are significant differences in the elemental concentrations between the three methods except for P. A white precipitate was formed during digestion method 1, as in the NG ash sample.The ICP-OES results (Table 4) show that there were losses of Al, Ca, Fe, and Mg.Recovery of the reference material was between 92% and 103% in method 2 and between 93% and 103% in method 3.In addition, the measured concentrations of both methods corresponded with each other.The standard deviations (s) of the analyzed elements with methods 2 and 3 varied between 0.01 and 1.72 mg/g, indicating that the repeatability of digestions was good.There have been no studies in which method 3 was applied to slag or any other material.Al-Harahsheh et al. (2009) studied a similar two-stage microwaveassisted acid digestion method for certificated reference materials of ores (Al-Harahsheh et al. 2009).The results of methods 2 and 3 regarding the recoveries and standard deviation levels are consistent with their results.
In addition to the CRM, the BOF slag was also digested using all methods.The ICP-OES results of all the filtered solutions are shown in Table 5.Similar to the CRM, there were losses of Al, Ca, Fe, and Mg.It can be stated, based on Tables 4 and 5, that method 1 is not suitable for the elemental determination of Ca-, Al-, and Fe-containing steelmaking slags.The characterization of the BOF slag residue is discussed below.
Methods 2 and 3 did not form any visible residue.As mentioned in the section above", method 3 was not able to completely dissolve quartz from the NG sample.The reason why method 3 was successful for BOF but not for NG might lie in the different phases of Si.Some support for this statement can be obtained by comparing the standard Gibbs free energy changes for the formation of ortho-silicic acid (D r G 0 ) from quartz and calcium-silicate phases, such as larnite.In real solutions, many other species of Si exist in the solution in addition to ortho-silicic acid.However, comparing the D r G 0 for these simple reactions is a quick way to obtain some insight into the relative stability of the two solid phases in an acidic environment.The standard Gibbs free energy changes at 180 � C were estimated using the reaction equations module of the HSC Chemistry 10 software (Roine 2021).
In the NG sample, quartz was the major phase of Si (Figure 1), and its D r G 0 was about þ14 kJ/mol at 180 � C. In the BOF slag, Si was mainly in calcium-silicate phases, such as larnite, whose D r G 0 was much lower, at around −220 kJ/mol at 180 � C.These values suggest that larnite is more readily dissolved in acidic solution than quartz.Due to that it should be noted that if the methods are used for new sample types, validation should be done.Methods 2 and 3 seemed to digest all the phases of the BOF slag.The concentrations of 15 elements (Table 5) corresponded with each other.As mentioned, Wilson, Burt, and Lee (2006) studied a procedure similar to method 2 but for soil samples.The results of their publication support our findings regarding reference material recoveries (Wilson, Burt, and Lee 2006).In the literature, only a few studies were presented related to the use of H 3 BO 3 for slag sample digestion.Coedo and Dorado (1995) used a similar microwave-assisted acid digestion method for CRM (slag) to determine major and minor elements by ICP-MS.However, the acid combinations (HF, HCl) were different, and the H 3 BO 3 stage was not processed in the microwave.They also mentioned that HNO 3 was not suitable to use in combination with other acids because the combination did not digest the samples completely (Coedo and Dorado 1995).As mentioned in the introduction, European Standard SFS-EN 13656:2020 presented a method where HBF 4 was directly used with HCl and HNO 3 .In this study, the two-stage addition of H 3 BO 3 and direct addition of HF and H 3 BO 3 were validated using slag CRM.Digestion of Si was also validated.Even though it seems based on the standard that HBF 4 can be directly used for many sample types, it was not studied for the NG samples.It should be noted that at least in this study, the direct addition of HF and H 3 BO 3 was not a suitable method for the NG sample.
As mentioned in the introduction, slag concentrations are typically determined using XRF.For a comparison, the XRF results of the BOF slag are shown in the same table (Table 5).XRF did not detect Cu, Na, or Zn, which were present in low concentrations in the slag.Considering the heterogeneity of the sample, the XRF and ICP-OES results were generally in reasonable agreement.Teir et al. (2007) also used both ICP-OES and XRF to determine the initial concentrations of steelmaking slags (Teir et al. 2007), and their results are consistent with ours.However, the dissolution method was not accurately described or validated.As can be seen from Table 5, significant concentration differences (>24%) occurred at low concentrations (Cr, K, Mg, P) when ICP-OES and XRF were compared.When comparing low concentrations in ICP-OES, there were no significant differences between methods 2 and 3 for the BOF slag and CRM.Good recoveries of small concentrations (Table 4, 92%-98%) indicate that ICP-OES is also accurate at low concentrations.Instrumental LOD and LOQ obtained with ICP-OES are shown in Table 6.The LODs and LOQs were calculated by Equations ( 2) and (3) for concentrated solutions of CRM and spiked NG samples.The dilution factors (CRM (1:20) and NG (1:10)) should be taken into account for the diluted samples.Small concentrations can be measured because obtained LOD and LOQ values are good and suitable for this application.Based on the results (Tables 4 and 5) both methods 2 and method 3 can be assumed to be suitable for the determination of the elemental concentrations of BOF slag.

Characterization of the digestion residues
The C content of the NG sample digestion residues was 94.55 wt.-% ± 0.36% in method 1, 98.34 wt.-% ± 0.92% in method 2, and 95.96 wt.-% ± 0.16% in method 3. The C content of the method 2 residues was comparable to commercial graphite, whose C content was measured as reference (98.41 wt.-% ± 0.73%).As expected, the residues of methods 1 and 3 had lower C content, which was due to the formation of insoluble compounds and quartz.These results support the suitability of method 2 for the determination of NG impurities.
The residue of the NG ash sample after method 3 could not be characterized due to the smaller amount of residue.The white residue from method 1 of the NG ash sample was characterized using XRD (Figure 2a), and as expected, the residues contained mostly fluoride precipitates.This result was supported by SEM-EDS (Figure 2b).The residue consisted mostly of the F-containing phase, ralstonite (Na 3 Mg 3 Al 13 (OH) 17 F 31 (H 2 O) 7 ).In addition, aluminum fluoride (AlF 3 ) was precipitated.Akermanite (Ca 2 Mg(Si 2 O 7 ) may have been present due to coprecipitation.The results are consistent with those from ICP-OES (Table 3).Ralstonite and other fluoride compounds (CaAlF 5 , CaMg 2 Al 2 F 12 , MgF 2 ) have been also reported in the literature to precipitate during HF digestion (Yokoyama, Makishima, and Nakamura 1999).The SEM-EDS results indicate that the quartz phase remained in the residue.However, according to XRD, Si was in the previously mentioned phases (Ca 2 Mg(Si 2 O 7 ).The possible quartz phase could not be detected by XRD.Based on these results, it cannot be concluded whether Si coprecipitated or some quartz remained undissolved in method 1.
According to the XRD and SEM-EDS results (Figures 2c,d), the BOF slag precipitate consisted of calcium fluoride (CaF 2 ) and iron aluminum fluoride (FeAlF 5 ).The main identified phase was CaF 2 , which is supported by the ICP-OES results (Table 5).The literature presents similar results for igneous rock, shale, and soil samples, where formed precipitate is mainly composed of fluorides and main elements of the starting material (Kemp and Brown 1990;Zhang et al. 2012;Hu et al. 2013).In this respect, the results support the composition of the precipitate formed from BOF slag.In addition, from the ICP-OES results, it can be observed that Al was almost completely precipitated, and some of the FESEM-EDS and XRD also supported these results.

Conclusions
Three different microwave-assisted acid digestion methods for the elemental determination of NG and BOF slag were compared.Method 1 was the traditional digestion method following the US EPA method 3052 (HNO 3 , HCl, HF).In method 2, traditional microwave-assisted acid digestion (method 1) was followed by a second treatment with H 3 BO 3 .In method 3, all acids (HNO 3 , HCl, HF, and H 3 BO 3 ) were added once.Without the addition of H 3 BO 3, insoluble fluoride compounds precipitated (method 1).The results showed that the addition of H 3 BO 3 was necessary for the complete digestion of Si-containing inorganic material.
The impurities of the NG were determined using two approaches.First, the NG was directly digested using the microwave-assisted methods.Second, it was burned, and the same methods were applied to the ash of the NG.The literature does not present validated microwave-assisted acid digestion methods for NG or NG ash.In this study methods 2 and 3 were successfully validated for NG using spiking recovery test.It was stated that there were no losses of studied elements as the recoveries were between 95% and 114% in method 2 and 94%-116% in method 3.However, method 3 was not capable of digesting quartz from the NG sample.Therefore, it seems that method 3 suitability depends on phases of the samples.This study indicates that method 2, is suitable for NG and NG ash samples.The results show that the main impurities can be determined directly from NG without the ashing step.The direct digestion of impurities without the ashing step is faster and requires a much smaller sample size.
In this study, new validated methods without visible precipitation for BOF slag was presented.A CRM was used to validate digestion methods and the recoveries of method 2 and 3 varied between 92% and 103%.Repeatability was also good with low concentrations (CRM �2 mg/g).The ICP-OES results of BOF slag were compared with the XRF results.As expected, the results for large concentrations were consistent between both techniques, but significant differences were obtained for the low concentrations.Based on the results, methods 2 and 3 are suitable for determining the elemental concentrations of BOF slag.A significant benefit of method 3 is faster implementation.

Figure 1 .
Figure 1.XRD analysis of the NG sample.

Figure 2 .
Figure 2. Insoluble residue formed during method 1 (a) XRD of the NG ash residue, (b) SEM-EDS analysis of the NG ash residue, (c) XRD of the BOF residue, and (d) SEM-EDS analysis of the BOF residue.
a n ¼ 4. b Fe and Si.c Al, Fe, and Si.d Al, Fe, K, Na, and Si; were measured from diluted solutions.

Table 6 .
Instrumental LOD and LOQ obtained with ICP-OES [mg/L] for the CRM and spiked NG samples.