An improved microwave assisted sequential extraction method followed by spectrometric analysis for metal distribution determination in South African coal samples

Some metal pollutants are corrosive in nature, are associated with fouling and slagging challenges of the coal boilers, are highly volatile and might cause air pollution and are catalyst poisoners during Fischer–Tropsch catalytic reaction. Therefore, this work describes an improved microwave-assisted sequential extraction (MW-ASE) method followed by ICP-OES/MS analysis for metal distribution determination in South African coal samples. The multivariate optimum conditions for each sequential step were 0.1 g, 200 °C and 5 min for sample amount, microwave temperature and extraction time, respectively. Under the optimum conditions, Ga, Sr and Ba were the only metals that showed solubility towards water, therefore, these metals are classified as highly mobile and eco-toxic under wet environmental conditions. Additionally, all the investigated metals showed solubility towards acidic conditions (HCl and HNO3). These results suggest that, most metal ions are predominantly bonded to sulphate, sulphide, and carbonate coal minerals. Alternatively, Ce, Cr and Y showed total extraction recoveries of ≤ 90%, confirming their strong affinity towards quartz coal minerals. In overall, the proposed MW-ASE method reported short extraction time (0.34 h), environmentally friendly reagents (H2O and diluted H2O2) and rapid multivariate optimization with acceptable extraction efficiencies (79–98%) and reproducibility (RSD ≤ 5%).


Methods
Reagents, materials and solutions. All chemical reagents were of high-purity and were used without any further processing. The ultrapure water with resistivity of 18 MΩ cm from a Milli-Q system (Millipore, Bedford, MA, USA) was utilised for all the preparations of calibration standards, sample solutions and rinsing of the glassware. The suprapure inorganic acids such as hydrochloric acid (30%, vv −1 ), nitric acid (65%, vv −1 ) and hydrogen peroxide (30%, vv −1 ) were purchased from Sigma-Aldrich, South Africa and were used as extracting reagents during MW-ASE method. The proposed MWASE method was validated by using three coal certified reference materials ( SARMs 18,19 and 20) and were all supplied by Council for Mineral Technology (MINTEK) in the Republic of South Africa. It has to be noted that, SARMs 20, 19 and 18 were all having an average particle size of ≤ 106 μm and were sampled from Sasolburg, Orange Free State and Witbank, respectively. In addition, the three real coal samples assigned as coal sample A (CSA), coal sample B (CSB) and coal sample C (CSC) were collected from one of the South African coal mines and were sieved to ensure that they match the same particle size as the CRMs. All glassware were pre-soaked for 24 h in diluted nitric acid (5%, vv −1 ), extensively rinsed with Milli-Q water, and dried in an oven (Digital Scientific series 2000 oven, Scientific Engineering (Pty) Ltd, South Africa) for overnight. It is worth to indicate that, the oven was also used for dehydration (50 °C) of coal samples in order to achieve constant weight amount.
Commercially available single stock solutions (1,000 mg L −1 ) of Be, Sc, V, Cr, Co, Ga, Sr, Y, Ba, Ce, Pb, and Th in 1% (vv -1 ) HNO 3 acid were purchased from Merck, South Africa. These stock solutions were diluted in order to make three sets of external calibration standards (0.2-1, 5-80 and 100-800 μg L −1 ) to ensure that all the concentration levels of various metal ions in both samples and CRMs were accommodated. Furthermore, Indium solution of 20 µg L −1 was prepared from 1,000 mg L −1 in 1% (vv −1 ) HNO 3 , used as an internal standard for both ICP-OES and ICP-MS measurements and was also purchased from Merck, South Africa. instrumentation. The proposed sequential extraction method was conducted by using the same MARS 6 One Touch Technology Microwave lab station (CEM Microwave Technology Ltd., North Caroline, USA) that was described in our research group 27 .
The sequential step 2 (5 M HCl) extracts were analysed by using Spectro ARCOS 165 ICP-OES (SPECTRO Analytical Instrument GmbH, Germany) equipped with Cetac ASX-520 auto-sampler. The other extracts from sequential steps 1 (H 2 O), 3 ( www.nature.com/scientificreports/ Perkin Elmer NexION 300 ICP-MS Spectrometer (Perkin Elmer, Waltham, MA, USA) with a triple cone interface (thereby allowing less spread of ions, photons and neutrals as were transferred to ion optics), two modes of operation (collision and standard modes) and a single quadrupole The measurement conditions of the ICP-MS were optimized daily by following supplier's recommendations and this technique was attached with Perkin Elmer S10 autosampler. Therefore, Table 1 shows optimum operating conditions for both ICP-OES and ICPMS. It has to be noted that, argon with purity of 99.996% was purchase from Afrox, South Africa and was used for all the plasma based measurements.
Microwave-assisted sequential extraction (MW-ASe) procedure. The proposed MW-ASE was conducted by using the same procedure that was previously report in our research group 28 . However, the only difference is that, a mixture of 2 M HNO 3 and 7 M H 2 O 2 was used in the last sequential extraction step (see Table 2).
Multivariate optimization for the proposed MW-ASE procedure. The multivariate optimization of extraction time (min), microwave temperature (°C) and coal amount (g) was carried out by using full factorial design and response surface method, as explained in the previous report 28 . Additionally, it is worthy to indicate that, the concentration levels of HCl, NHO 3 and H 2 O 2 were already optimised in the previous studies 5, 24 , therefore, these concentration levels were fixed at 5, 2 and 3 M, respectively. Lastly, the volumes of H 2 O, HCl, HNO 3 and HNO 3 -H 2 O 2 were fixed at 12 mL as previously reported [24][25][26] . It is worthy to indicate that, the minimum (−), central (0) and maximum (+) levels of the investigated factors are presented in Table 3 and the first order experimental design resulted in 11 experiments, which were later visualise by using Pareto charts. The significant factors were then further optimised by using response surface methodology (RSM), as illustrated in Table 4.    www.nature.com/scientificreports/ illustrated in Figs. S1,S2 (Appendix A), respectively. It has to be noted that, the bar length of the Pareto chart is directly proportional to the absolute value of the estimated effects. The ANOVA results observed in Fig. 1, show that the other two parameters (coal amount and temperature) exceeded the p-value (i.e. their bar length exceeded the red reference line) for Sr and Ba, demonstrating that coal amount and temperature are significant at 95% confidence level for the extraction of these two metals. Similar results were observed for (5 M) HCl ( Fig. S1) and (2 M) HNO 3 ( Fig. S2) extracting reagents for extraction of most metals, hence time (C) was fixed at 5 min, since it was the only insignificant factor. Therefore, the significant parameters were further optimised by using RSM optimisation tool.

Results and discussion
Second order multivariate optimization. In order to obtain optimum conditions for the other two significant factors (microwave temperature and coal amount), second order multivariate optimization strategy was conducted, which involved the application of response surface methodology based on the central composite design. The matrix of the central composite design contained 14 experiments with responses (% recovery) correlating to each and every experimental run (see Tables S4-S6). The ANOVA data acquired and the quadratic equations (not included in the paper for simplicity reasons) of the models for each extraction reagent were used to interpret the relationship between analytical response and the evaluated parameters (microwave temperature and coal amount). The three dimensional response surface plots for water extracts are illustrated in Fig. 2, the plots for 5 M HCl and 2 M HNO 3 reagents are shown in Figs. S3,S4 (see Appendix), respectively. These three dimensional response surface plots show that, maximum analytical responses for most of the investigated metal ions (for all the examined extracting reagents) were achieved at 200 °C and 0.1 g for temperature and time, respectively. Hence, the overall interpretation of the multivariate optimization revealed that, the optimum conditions of the proposed MW-ASE were: 0.1 g sample amount, 200 °C microwave temperature and 5 min extraction time for H 2 O, HCl and HNO 3 . Additionally, the analytical responses achieved were compared with the predicted values of the RSM model and there were no significant difference at a 95% confidence levels. Therefore, the obtained optimum factors were used during validation and application stages.
Validation of the proposed MW-ASe method. The optimum conditions of the proposed MW-ASE method were then applied in three coal CRMs (SARM 18,19 and 20) and the results are shown in Fig. 3. The latter demonstrates that, Ga, Sr and Ba were easily extracted with pure water, therefore, these three metals can be regarded as highly mobile metal ions and can be easily introduced into the water bodies during rainy seasons. Recently, our research group reported MW-ASE method for determination of sulphur forms in coal samples www.nature.com/scientificreports/ and it was also observed that pure water can effectively extract ± 20% of sulphate ions in SARM 20 28 . Therefore, it can be concluded that Ba, Ga and Sr can be associated with the sulphate, carbonate and phosphate ions of the coal. Additionally, several research groups have also identified these three metal ions as easily mobile and therefore, are a threat to the environment 5,8,28 . From Fig. 3, it can also be observed that most metal ions showed strong affinity towards the HCl acidic environment, except for Ga. These observations suggest that most of metal ions are associated with the sulphate minerals of the coal and can only be mobile in acid conditions. The diluted hydrogen peroxide extracts showed minimum amount of metal ions, these results show that there are lighter interactions between elemental impurities and the organic content of the coal. The proposed sequential extraction was also compared with the sequential extraction method reported by Laban et al. 5 and comparison results are illustrated in Fig. 4. The analytical results shown in Fig. 4 illustrate that the proposed the proposed MW-ASE procedure is quite comparable with the literature reported results described by Laban and Atkins for all investigated metal ions, expect for Pb. The latter showed results that were below detection limits in the literature report. This is because, the literature work used less sensitive spectrometric technique (ICP-AES) as compared to the ICP-MS used in the current study. Therefore, it can be concluded that, the proposed method is more sensitive to Pd detection as compared to Laban    . It has to be noted that, even though the pattern of extraction was similar, but the proposed method showed moderate extraction recoveries (79-89%), while the literature report showed excellent extraction efficiencies (≥ 100%). The strong affinity of Cr towards organic components was also reported by several researchers 5,8,29,30 . Cobalt also reveals strong interaction with organic components of SARM 18, but was more bound with sulphate minerals in SARM 19 and 20. The rest of the metal ions were also sulphate bonded, except for strontium in SARM 18. Trace elements such as Mn and Pb are known to dominate in coal as carbonate or sulphide minerals, hence are easily solubilized in acidic medium 5,31 . The rest of the metal ions were mostly sulphate bonded, except for Sr in SARM 18. In overall, Laban and Atkin's extraction recoveries were higher as compared to those of the current study. This is due to the kaolinite and quartz bonded minerals that could not be distracted with the chemical reagents used in the proposed MW-SAE method. It has to be noted that, HF reagent used for the destruction of the organic minerals by Laban and Atkin, can also decomposed clay minerals. The current study replaced notorious HF with environmentally friendly dilute H 2 O 2 for the decomposition of organic matter 26,28 . Therefore, Cr partially remained in the clay part of the coal; hence cannot be quantitatively extracted without the use of hydrofluoric acid (HF) 5,8,29,32 . Table 5 shows the ICP-OES/ MS results that were obtained from the sequential extraction of three coal samples (CSA, CSB and CSC) received from one of the South African coal mines. From Table 5 results it can be observed that, Ga, Ba and Sr were the only metals that showed solubility toward water, this means that during wet weather conditions, these metals can be easily leached to the environment. Additionally, Ba and Sr concentration results show that, these two metals were most dominating across the whole spectrum of the three coal samples. It is also worthy to mention that, the sum concentration of most of the investigated metal ions correlated well with the concentration levels that were previous reported using total digestion methods 26,27 . This correlation confirms that, the accuracy and reliability of the proposed MW-ASE method was excellent. It worthy to indicate that, Be, Sc, Pb and Th in CSA showed slightly lower extraction efficiencies as compared to one literature report 27 . The similar trend was observed for Be and Th in CSB and for Be and V in CSC. The lower recoveries of the few metal ions (Be, Sc, V, Pb and Th) might be due to the strong affinity of these metal ions with the quartz minerals of the coal [33][34][35] . However, most of the challenging metal ions were not detected on the previously reported workor they were reported at lower extraction recoveries 26 . It has to be noted that, 3 mL of 7 mol −1 HNO 3 was only added in order to enhance the extraction efficiencies of the investigated elements. Therefore, the proposed MW-ASE method reported 50% less consumption of HNO 3 when compared with literature reported studies [36][37][38] . Therefore, the proposed MW-ASE procedure can be used as an alternative environmentally friendly method for sequential extraction of metal ions from various coal samples and related matrices. www.nature.com/scientificreports/ comparison of current MW-ASe with literature reports. The proposed MW-ASE method was compared with other sequential extraction methods that were performed in coal related matrices for determination of various metal distribution. All the ten compared methods and the current method are illustrated in Table 6. The latter shows that, seven publications described the metal mobility in coal samples 3,5,[11][12][13]39,41 , and the other three studies reported metal distribution in coal fly ash matrices 7,9,40. Furthermore, this table shows that, all the published literature methods reported the use of either large reagent volumes (≥ 20 mL), notorious concentrated inorganic acids or both 3,5,7,9,[11][12][13]39,40 . However, with the proposed MW-ASE method, it was possible to use reduced diluted reagent volumes of 12 mL and environmentally friendly reagents (H 2 O and diluted H 2 O 2 ). This was due to the rapid interaction of reagents and the coal, which was enhanced by the use of microwaves, as reported by Laban and Atkins 5 . Another impressive feature about the proposed methods was its short extraction time (0.34 h) as compared to the published literature reports, with extraction time ranging from 2.5 to 86 h. It has to be noted that, 2.5 h sequential extraction was also facilitated by the use of microwave system 5 . However, the only limitation about the proposed MW-ASE method was the use of high temperatures (200 °C). The latter were also reported for all the sequential extraction methods that involved the use of microwave systems 3,5,12 . The mostly investigated metal ions on various sequential extraction reports were transition metals and good recoveries (≤ 100%) were obtained. However, all the literature reported sequential extraction methods showed poor precision (≥ 10%), except for the work reported by Yang et al. 3 , Dahl et al. 9 and the proposed MW-ASE method. www.nature.com/scientificreports/ conclusion An improved microwave assisted sequential extraction (MW-ASE) method followed by ICP-OES/MS analysis was successfully developed for rapid and effective determination of metal distribution in South African coal samples. The most influential extraction parameters (coal amount, extraction time and temperature) affecting metal sequential extraction efficiencies were successfully optimised by using 2 3 full factorial design and response surface methodology. The optimum conditions of the proposed MW-ASE were 0.1 g, 200 °C and 5 min for coal amount, microwave temperature and extraction time, respectively. Then, the optimised MW-ASE method was applied in three coal CRMs (SARM 18,19 and 20) and three metal ions (Ga, Sr and Ba) showed solubility towards water, irrespective of the CRM. Hence, these three metals were regarded as highly mobile metal ions and are expected to leach out into the water bodies during rainy seasons. The rest of the metals were only mobile in acidic conditions, therefore, these metals can be regarded as immobile.

Application of the proposed MW-SAe method in real coal samples.
For validation purposes, the proposed MW-ASE method was compared with published literature work reported by Laban and Atkin using the same coal CRMs investigated in the current study. It has to be noted that, the patterns of extraction recoveries were similar for all metal ions in three different CRMs, but the proposed MW-ASE method showed moderate extraction recoveries (79-98%), while Laban's' work showed higher extraction recoveries (≥ 100%). The lower extraction efficiencies reported in MW-ASE method were due to the kaolinite and quartz bonded metal ions, which could not be distracted with the environmentally friendly diluted H 2 O 2 used in the proposed procedure. However, Laban and Atkins conducted their last sequential extraction step by using notorious HF reagent, in order to also decomposed refractory clay minerals. It has to be noted that, the extraction efficiencies of the proposed MW-ASE method were also compared with extraction efficiencies that were previously reported for digestion methods and the results were quite comparable, except for few metal ions such as Be, Sc, V, Pb and Th. These metal ions were suspected to have strong affinity with the quartz minerals of the coal (see Table 5).
Furthermore, the current study was compared with other published sequential extraction methods as shown in Table 6. The latter shows some improvements when looking at the toxicity and cost. This is because; diluted environmentally friendly H 2 O 2 was used for extraction of organic bound metals instead of concentrated notorious inorganic acids. It has to be noted that, the proposed MW-ASE method reported the use of 3 mL of 7 mol −1 HNO 3 , in which was only added to enhance the extraction efficiencies of the studied elements. However, HNO 3 consumed by the proposed MW-ASE was 50% less when compared with the previously reported studies. Additionally, the proposed sequential extraction method was more accurate for direct determination of water soluble and organic bound metal ions in coal samples, in a short period of time (0.34 h) as compared to other published work (2.5 to 86 h).