During the last 50 years, the cement industry has been relying on XRF analysis for the quantification of the elemental composition of cement industry-related materials. Sample preparation by pressed powders was commonly used during the first decades, but for the last 20 years borate fusion saw an important increase in popularity and both techniques are now accepted for those analyses.
The 21st century saw a significant change in the management of the production of cement through the increase of production of cements with alternative raw materials and additives involving secondary fuels. This complication of the cement matrix and the use of calibration reference materials from various sources in the world make the use of pressed powders more complicated due to the difficulty to matrix match the calibration standards and the production samples from the plant. The use of the borate fusion preparation allows for more accurate analysis and requires less calibration curves because it removes particle size and mineralogy effects [1, 2]. However, to facilitate the lab work, a single fusion method for the preparation of all cements, all process materials and all raw materials is desirable to allow compliance with the ASTM C 114 and ISO/DIS 29581-2 specifications of precision and accuracy.
To reach these objectives, a robust analytical method using a fully automated fusion fluxer has been developed for the quantification of all elements of interest for the cement industry. This single method was used to prepare all cements, all process materials and a very large range of raw materials. Two sets of certified reference materials (CRMs), one from the National Institute of Standards and Technology (NIST) and the other from the Japan Cement Association (JCA) were used to verify that this fusion method allows a matrix match for cement from different origins. The evaluation of precision and accuracy was performed according to the instructions provided by ASTM C 114  and ISO/DIS 29581-2 .
Apparatus and instrumental conditions
1. A Claisse M4 Fluxer was used to generate all fusion disks. Its auto-regulating gas system and pre-set fusion programs allow for the most repeatable and reproducible fusion conditions.
2. A Fisher Scientific Isotemp muffle furnace was used for the LOI determinations and preparation of ignited samples. The LOI method used for all cement types and clinker included ignition at 950 degrees Celsius in a clean platinum crucible for 60 minutes.
3. A Bruker-AXS S4 Explorer sequential WDXRF spectrometer was used for data generation.
Global sample preparation method
First 0.6000 g of ignited sample is weighed with plus or minus 0.0001 g tolerance in a Claisse Optimix platinum crucible. Then, 6.0000 g of Claisse LiT/LiM/LiBr 49.75/49.75/0.50, borate flux is weighed with plus or minus 0.0003 g tolerance on top of the sample. A mini-vortex mixer is used to mix the sample with the flux. The mini-vortex mixer speed was controlled so as not to lose material because variance from the ratio of flux to sample weight causes error in the results .
The maximum fusion temperature used for the fusion on the Claisse M4 Fluxer is 1025 degrees Celsius because it is known that over the critical temperature of 1050 degrees Celsius flux begins to volatilize without consistency which changes the sample to flux ratio . Other compounds like silica begin to volatilize without consistency as well . Molten flux was poured in a 32 mm diameter platinum mold.
Preparation for calibration, selection of control samples and preparation for validation
As discussed previously, one objective of this project was to calibrate the WDXRF with two sets of CRMs from different origins: NIST Standard Reference Material Series 1880a, 1881a and 1884a to 1889a, and JCA Reference Materials for XRF analysis 601A Series XRF-01 to XRF-15. The second objective was to comply with the requirements of ASTM C 114 and ISO/DIS 29581-2. Those standard methods have two different philosophies. ASTM C 114 uses CRMs to verify precision and accuracy on two different days . ISO/DIS 29581-2 validates repeatability of the method using one or more CRMs, as control samples that are not included in the calibration over at least two weeks . Also, for verification of ASTM requirements, results should include LOI, and for ISO, LOI free results are needed. Finally, since two CRMs were used for the ISO validation, the selected samples had to cover both the high and low concentrations of all elements as prescribed in the standard method .
For the calibration of the WDXRF instrument and for qualification of the Global Fusion/XRF method with ASTM C 114, two sets of glass disks were prepared for every CRM, one on the first day and the second on the next day, not less than 24 hours apart. For the validation of the analytical method with ISO, 10 glass disks of each control samples (JCA XRF-03 and JCA XRF-14) were prepared over 15 days. The control sample glass disks were analyzed on the same day they were prepared.
RESULTS AND DISCUSSION
Robustness of the fusion method
It was proven that ignition of the sample is absolutely necessary in the analytical process for a global fusion method. A preparation with a sample to flux ratio of 1:10 with 6.6000 g of total mass takes a fusion program of 13 minutes heating at 1025 degrees Celsius on the M4 Fluxer to prepare stable glass disks with high alumina and/or high silica samples. The cooling process is done with forced air for 5 minutes.
More than 200 different samples from 20 different material types were fused with the global fusion method which showed good efficiency to prepare homogenous and stable glass disks with all of the materials.
ASTM precision and accuracy
The ASTM C114 precision test was applied as it is described in the method . The duplicates are the two disks prepared on two different days for every CRM. The maximum difference for all elements is compared to the ASTM precision limit. The maximum values obtained for all elements meet the specifications and well within the limits.
The ASTM C114 accuracy test was applied as it is described in the method . The absolute maximum error for all elements is compared to the ASTM accuracy limit. The maximum values obtained for all elements meet the specifications and well within the limits.
ISO precision and accuracy
The ISO/DIS 29581-2 precision test was applied as described in the method . The absolute differences were calculated from successive results of the control samples. The maximum absolute difference for all elements is compared to the ISO expert precision limit. The maximum values obtained for all elements meet the specified limits for both control samples.
The ISO/DIS 29581-2 accuracy test was applied as described in the method , but without averaging the results of the different preparations. The accuracy values were calculated as difference of the results from the 10 preparations over 15 days against the certified values. The absolute maximum error for all elements is compared to the ISO expert accuracy limit. The maximum values obtained for all elements are in the requirement limits for both control samples.
A universal sample preparation by fusion method for XRF analysis has been developed by Claisse to process various cement types and all the raw materials found in a cement plant. Despite its versatility, the universal borate fusion method complies with the precision and accuracy requirements of ASTM C 114 and ISO/DIS 29581-2