Article pubs.acs.org/EF
Quantitative X‑ray Fluorescence Analysis of Biomass: Objective Evaluation of a Typical Commercial Multi-Element Method on a WD-XRF Spectrometer Lars Klembt Andersen,*,† Trevor J. Morgan,† Aikaterini K. Boulamanti,† Patricia Á lvarez,‡ Stanislav V. Vassilev,†,§ and David Baxter† †
Institute for Energy and Transport, Joint Research Centre, European Commission, Westerduinweg 3, 1755 LE Petten, The Netherlands ‡ Instituto Nacional del Carbon, C/Francisco Pintado Fe 26, La Corredoria, 33011 Oviedo, Spain § Institute of Mineralogy and Crystallography, Bulgarian Academy of Sciences, Acad. G. Bonchev Street, Block 107, 1113 Sofia, Bulgaria S Supporting Information *
ABSTRACT: With the increasing utilization and trade of biomass, there is a growing need for quick and reliable quantitative chemical analysis of the inorganic elemental composition of biomass. X-ray fluorescence (XRF) spectrometry performed directly on the raw biomass with limited prior sample preparation is an attractive method to this end. However, reliable calibration of XRF spectrometers for universal multi-element analysis can be very hard to implement mainly due to problems with matrix corrections. XRF users thus often rely on commercial precalibrated or standardless methods delivered with their XRF spectrometer. These methods are often sold without any guarantee on performance. Given the actual frequent use of these methods, along with their potential as ready-to-use methods for multi-element analysis of biomass, we investigate here the performance of a typical commercial precalibrated/standardless method recently purchased with a 4 kW wavelength dispersive (WD) XRF spectrometer. The accuracy (trueness and precision) is determined by analyzing the certified inorganic elements in 13 certified reference materials (CRMs) of diverse vegetal/plant origin. The certified elements detected by the XRF are Na, Mg, Al, P, S, Cl, K, Ca, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Rb, Sr, Mo, Ba, and Pb. The relative systematic error (bias, trueness) is typically better than ±20% for elements in the range 25 to 100 ppm, better than ±15% for the range 100 to 1000 ppm, and better than ±10% for concentrations above 1000 ppm. The relative precision (measured as the relative standard error) is better than ±5% (typically ±1−2%) for concentrations >25 ppm. Quantifying elements below 25 ppm is possible in some cases, but it requires a more detailed study for each specific element. Occasionally, larger relative biases of up to ±40% can occur for certain elements in certain samples, so care has to be taken to carefully test the applied method for the particular samples and elements of interest. The XRF method can further be used to estimate the ash yield from biomass combustion with a relative bias (trueness) typically better than ±15%. This involves the determination of silicon (Si) and titanium (Ti) by XRF. The choice of matrix composition in the matrix correction model and the influence of sample moisture and sample grain-size are also addressed. The instrument and method are described in detail allowing for comparisons with other similar XRF spectrometers often already available in industrial analytical laboratories. ‘sample matrix effects’ including inter-element interferences, there is often not a simple proportionality between element concentration and intensity of the X-ray fluorescence emitted from this element. One approach to solving this problem involves setting up a calibration based on standard reference materials of similar biomass origin to match the sample matrix. Such a fully calibrated method is generally preferred if only a single or a few elements are to be determined. The other approach is the so-called ‘standardless’ method based on fundamental physical parameters. Combinations of the two methods also seem possible. Modern instruments are often sold with commercial precalibrated semi-quantitative analysis programs consisting of fundamental parameter methods, maybe combined with
1. INTRODUCTION With the increasing utilization and trade of biomass, there is a growing need for quick and reliable quantitative elemental chemical analysis of biomass, e.g., with producers, traders, and end-users for quality assurance and potentially also as a tool in ascertaining the type and origin of the biomass. X-ray fluorescence (XRF) spectrometry performed directly on the raw biomass with limited prior sample preparation is potentially attractive for simultaneous quantitative elemental analysis of many elements including ash-forming and nutrient elements (e.g., Na, Mg, Al, Si, P, S, K, Ca, Mn, and Fe) and environmentally important elements (e.g., S and Cl) including heavy metals (e.g., Tl, Pb, Cd, As, and Hg). When performing quantitative chemical analysis with XRF spectrometry, one of the fundamental issues is the calibration method used to convert measured fluorescence intensity to calculated concentrations of each element.1−4 Because of so-called © 2013 American Chemical Society
Received: August 2, 2013 Revised: October 31, 2013 Published: November 7, 2013 7439
dx.doi.org/10.1021/ef4015394 | Energy Fuels 2013, 27, 7439−7454
Energy & Fuels
Article
Relative errors were in the range from not detected by XRF to +47% depending on the concentration and the analyzed element. However, for many values this standardless XRF method gave very good results with errors being less than ±10%. This approach, i.e., testing the standardless XRF method against at least one reference material of similar type as the investigated samples, seems to be what should minimally be done to ensure that the XRF data of the analyzed samples are at least somewhat reliable. However, we have also noticed several publications where XRF analysis data are simply stated as ‘analyzed by XRF’ with no information on the method used or tests performed on the method’s performance. One could suspect that in some of these cases, the precalibrated/ standardless method delivered with the XRF instrument was simply used with little checking of its actual performance. In these cases, it is important to ask the following question: For which samples, elements, and concentration ranges does the method give reliable results? Also, it is not completely clear if testing with a single reference material is actually sufficient to reliably establish the performance of a method. The present study demonstrates one possible approach to comprehensively assessing the performance of an XRF method for raw biomass. This is done by analyzing 13 CRMs of very diverse biomass origin including lichen and marine biomass (Table 1). Given the broad range and relatively large number of
calibrations by reference materials, for analysis of nearly the whole periodic table (fluorine to uranium) in several different types of samples, e.g., biomass, coal, minerals, soils, metals, plastics, etc. The term ‘semi-quantitative’ is used here by the manufacturers for methods (believed to) giving only estimated (uncertain) values or for those without guaranties on their accuracy. The present study investigates the actual performance of such a semi-quantitative standardless method when used to analyze raw biomass (i.e., not biomass ash). The method is based on a combination of calibration with reference materials and fundamental parameters. Note: we use here the term standardless for such combined methods as well to indicate that they are different from the fully calibrated methods based only on calibration with matrix matched reference materials without using fundamental parameters. The application of both fully calibrated methods (using reference standards) and standardless methods in multi-element XRF analysis of vegetal and plant materials has been discussed in two illustrative publications.5,6 Biomass consists mainly of organic constituents and moisture (water), and along with these, various inorganic constituents, e.g., as mineral phases and ions.7−9 Some typical elemental concentrations are given in the literature.7−9 Here, it is noted that the composition of biomass varies significantly both between biomass types and within each biomass type. This large variation could potentially complicate the calibration of the XRF spectrometer by means of certified reference materials, and this, in turn, could favor the use of standardless methods. Likewise, the general lack of commercially available biomass certified reference materials (CRMs) probably makes it complicated for some XRF laboratories to develop a calibration for biomass based on reference standards, unless they can engage in less producing their own reference standards by other analytical techniques. On the positive side, biomass samples could well show less (variation in) matrix effects and inter-element interferences compared to those of other types of samples due to (1) the low concentration of inorganic elements and (2) these elements being imbedded in a fairly constant matrix of mainly organic constituents, and (3) these containing mainly the light elements (C, H, and O) which mainly influence the X-rays through mass energy-absorption and associated scattering. These observations indicate that biomass could be relatively easily analyzed by XRF by either calibrated methods using reference standards or by standardless methods. However, XRF investigations on raw biomass (i.e., not biomass ash) and specifically investigation of method performance by analysis of reference materials of raw biomass, plant, or vegetal samples are fairly rare. One publication gives a good example of elemental analysis of vegetal and plant materials by standardless XRF utilizing the scattering in the sample of the incoming excitation X-rays in performing the matrix correction.10 The method was tested by analyzing for 10 elements in 5 CRMs of vegetal/plant origin. Errors (bias and trueness, i.e., the relative difference between XRF results to certified mean values) were found in the range −31% to +257% depending on the concentration and the analyzed element. Notably, only 5 out of 38 reported measured values showed a relative bias larger than ±15%. Some more recent studies report on standardless XRF analysis of CRMs (tea, marine sediment, and iron ore) using a commercial standardless analysis method on wavelength-dispersive XRF (WD-XRF) instruments.11−13 In each study, the standardless method was tested by means of a single CRM of similar type as the samples to be analyzed.
Table 1. Certified Reference Materials Used in This Study (As-Received) CRM name and origina Apple leaves SRM 1515 Peach leaves SRM 1547 Spinach leaves SRM 1570a Tomato leaves SRM 1573a Pine needles SRM 1575a Aquatic plant BCR-60 (Lagarosiphon major) Beech leaves BCR-100 Hay powder BCR-129 Sea lettuce BCR-279 White clover BCR-402 Plankton BCR-414 Lichen BCR-482 Aquatic plant BCR-596 (Trapa natans)
appearance and grain sizesb fine powder 10−20 μm; flour feel fine powder 20−30 μm; flour feel fine powder 10−20 μm; flour feel fine powder 20−30 μm; flour feel fine powder 20−30 μm; flour feel fine powder 10−20 μm; flour feel with grains of 50 × 200 μm (∼10%) powder 10−50 μm; flour feel with grains of 100− 150 μm (
