A comparative assessment of biomass ash preparation methods using X-ray ﬂuorescence and wet chemical analysis

(cid:1) Signiﬁcant unburned carbon in biomass ash produced at 550 (cid:1) C is demonstrated by STA. (cid:1) Analyses using XRF and wet chemical methods presented for various biomass ashes heated to 850 (cid:1) C. (cid:1) Biomass ash heated to 815 (cid:1) C gives improved consistency and reliability in XRF ash analysis. X-ray ﬂuorescence (XRF) spectroscopy is a rapid method used to determine the composition of biomass ash, but the accuracy of the method is sensitive to various factors including ash preparation methods. In this study different types of biomass ash were examined by using wet chemical analysis (WCA) and com- pared with the respective XRF results. The biomass ash was initially prepared in accordance with the European Standard method at 550 (cid:1) C. At this low combustion temperature the amount of residual unburned carbon is signiﬁcant. To eliminate this, the ashes were heated at higher temperatures: a batch of twenty biomass ashes were heated at 850 (cid:1) C and a batch of ﬁve heated to 815 (cid:1) C. At these higher tem- peratures there may be loss of inorganic components by vaporisation. Variation in these effects may lead to unreliable results. The relationship between XRF and WCA results are given by regression equations. The ashes processed at 815 (cid:1) C show better agreement between the two analysis methods.


Introduction
The inorganic content and composition of biomass is important in relation to combustion behaviour in furnaces since it influences corrosion and ash deposition on heat exchange and other surfaces in the combustor. During combustion, the organic matter burnsout leaving metal oxides and other metal adducts. Knowledge of the melting point of these compounds is important in furnace operations [1][2][3][4]. Certain metals can also affect the rates of pyrolysis [5] and bio-refinery processes [6]. Inorganic content varies in different types of biomass but may also vary significantly in similar types of biomass from different regions. Some wood-pellet biomass fuels have low silica and low potassium and usually high calcium content whilst agricultural residues such as straw can show high silica, high potassium and low calcium content.
The European Standards for the determination of major elements in solid biofuels [7,8] set out procedures for forming the biomass ash and for determination of the major metals content by acid-digestion and subsequent spectroscopic analysis. Other analytical techniques for ash analysis commonly reported in the published work include X-ray Fluorescence (XRF) spectroscopy, for example [6], since it is a relatively rapid and convenient method. However, XRF spectroscopy is sensitive to interactions between elements in the matrix whereby the respective fluorescent emissions interfere with or obscure each other. This can be corrected for, although the analysis depends on use of known reference samples. Other sources of error in the XRF method include flaws in the sample matrix and inhomogeneity in the ash solid [5]. The way in which the ash sample is prepared from the biomass is therefore crucial for obtaining reliable results. This also applies to other diagnostic methods such as ash melting test methods [9] which rely on representative ashing techniques. This study compares two different ash preparation methods using the results obtained from ''wet" chemical analytical techniques with those from XRF analysis.  According to Niu [2], different ashing rates and different ashing temperatures lead to different ash compositions, which affect the ash fusion characteristics (AFC). Therefore, it is essential to establish an appropriate standard for ash preparation for determining the biomass AFC. The fusion characteristic of the biomass ash is affected by the high-temperature behaviour of the mixtures, which are often eutectics, and is not a simple function of the major element content in the ash. Changes in the proportions of certain metals can therefore have significant effects on the fusion characteristics. At temperatures above 815°C, potassium and chlorine begin to be released to the gas phase, reducing the relative content of K 2 O in the resulting ash.

Experimental methods
Twenty solid biomass fuels were selected for this study. These are typical materials used in large scale power plant, many of them being sourced from power plant operators. The selection includes forestry biomass, energy crops, agricultural residues and a torrefied biomass thus providing a range of differing ash compositions. The reference names adopted for these sample fuels are given in Table 1.
The samples of ash for analysis were prepared from portions of the raw materials following the British/European Standards [7,8]. Samples with <2.0 mm particle top size were produced in crucibles in an electric furnace at a temperature of 550°C for 14 h duration. In order to ensure any unburned carbon was removed, the samples were further heated. Two methods were used: (i) in the first batch (Case 1) the temperature was raised to 850°C for 2 h; (ii) a second batch of ash from five biomass samples (Case 2) was prepared by heating to a temperature of 815°C for 2 h. In each case, the ashes were agitated every 30 min. The resulting ashes were manually ground and sieved to <106 lm.

Wet chemical analysis
A sample of each ash (in both cases) was prepared as above for wet chemical analysis by acid digestion. Two analyte solutions were produced referred to as ''Solution A" and ''Solution B". Solution A is from the reaction of ash with sodium hydroxide at 650°C to produce sodium silicate which was then dissolved in HCl to give silicic acid in solution. The analyte is used only for the determination of silicon content. Solution B was prepared from acid digestion by hydrofluoric, hydrochloric, sulphuric and nitric acid sequentially. The content of ten major inorganic elements from this analyte were quantified using the following analytical methods: (i) Atomic Absorption Spectroscopy for Na, Mg, K, Mn, Fe, spectrophotometry for Si, P, Ti and EDTA titration for Ca.

XRF analysis
The ash samples were prepared for XRF analysis by forming into fused glass discs. For Case 1, these consisted of 0.5 g of sample (ground to <106 lm), 5.0 g of flux, 0.05 g lithium bromide and a fusion temperature of 1200°C. For Case 2, flux content reduced to 4.5 g and fusion temperature to 1100°C. The samples were analysed using an X-Thermo ARL Advant XP sequential X-ray fluorescence spectrometer.

Simultaneous Thermal Analyser (STA)
Studies were undertaken using a Netzsch 449C Jupiter Simultaneous Thermal Analyser (STA), coupled to a Netzsch QMS 403C Aeolos Quadrupole Mass Spectrometer. This allows the simultaneous application of thermogravimetry (TG) and differential thermal analysis (DTA), which monitors the temperature difference between the sample and an inert reference material. In this study, 10 mg of each ash was heated from room temperature to 1400°C at 10°C/min under a gas flow rate of 80 ml/min of 12.5% O 2 in He. The gases evolved were transferred to a heated fused silica capillary to a mass spectrometer; the monitored species were H 2 O, CO, CO 2 , K and KCl.

Results and discussion
The XRF results were compared against the wet chemical analysis results for the respective ash samples for Cases 1 and 2.

Case 1
The results obtained from the wet chemical and XRF analyses for Case 1 are presented in Table 2. The results are plotted against each other in Fig. 1. for comparison. A linear regression function can be obtained from the data as: With the coefficient of determination, R 2 = 0.937.

Case 2
The results obtained from the wet chemical and XRF analyses for Case 2 are presented in Table 3.
The results are plotted against each other in Fig. 2. for comparison. A linear regression function can be obtained from the data as: With R 2 = 0.997. The coefficient of determination, R 2 , for this function is higher than case 1 -a clear improvement in the correlation.

Simultaneous thermal analysis
STA-MS analysis plots showing the mass loss upon heating four example ash samples alongside the corresponding gas evolution profiles are presented in Fig. 3(a)-(d). The mass loss curves are complex with gradual mass losses up to the final temperature of 1400°C. There is a notable mass loss from CO and CO 2 in all cases indicating unburned carbon in the ash. Furthermore, much of the mass loss from oxidation of the carbon occurs above 550°C. Whilst the decomposition of calcium and potassium carbonates could contribute to the signal it has been shown [10] that in the case of the former the main decomposition occurs at 700-830°C, and in the case of the latter 1200-1400°C. Thus this demonstrates that the ashing temperature of biomass at 550°C (in accordance with European Standard method) in not adequate to ensure that there is minimal residual carbon in the ash. The measured K and KCl losses are very low and none of the samples show significant change between 550°C and 850°C.

Discussion
There have been concerns about the accuracy of the XRF test method where low temperature (550°C) biomass ashing is used and results for different metal oxides compared to wet chemical analysis show different degrees of reliability. This arises because of unburned carbon that may be present which has been demonstrated in STA-MS analysis of four ash samples. Where higher ashing temperatures are used, there is a risk of some vaporisation of volatile inorganic components. However, the STA-MS analyses show that heating ash up to 850°C, removal of unburned carbon can be achieved with little loss of other alkali metals.
Two different methods for preparing biomass ash samples for XRF analysis have been evaluated using wet chemical analysis to determine the accuracy of the XRF results. The results of ''Case 1" show that there is correlation between the two sets of results but with a notable level of discrepancy indicated by a coefficient of determination less than 0.94. The results for Case 2, (with a reduced set of samples) show a much better agreement in the data implying that the sample preparation method is improved. In the sample preparation for XRF analysis, it is then important that ashing temperature and sample fusion temperature are carefully and consistently chosen.

Conclusions
The main conclusions from this investigation are as follows: (1) Biomass ash prepared at 550°C ash can give inaccurate results from XRF analysis owing to significant unburned carbon retained in the ash. (2) Processing the ash at a temperature above 800°C produces an ash composition more suitable for XRF testing. (3) Whilst there is correlation between XRF analyses and wet chemical analyses for twenty different biomass ash samples prepared at 850°C, the method using biomass ash prepared at 815°C shows better agreement between the two analytical methods.