The Reynolds Cup is a very popular event on the CMS calendar. In 2014, the cup was supported by The Clay Minerals Society, the German Clay Group (DTTG) and the Clay Minerals Group of the Mineralogical Society of Great Britain & Ireland.
The Reynolds Cup has developed into a story of success (sometimes even mentioned as “world championship” in quantitative mineralogy). The participation has increased year on year. Whereas in 2002 from 40 registrants 15 submitted their results (rate 38%) the number increased in 2014 again to 81 registrants from 21 countries with 67 sent in results (rate 83%).
The specific aim of the 7th contest was to evaluate proficiency of analysts working on clay-rich materials representing a sulphate rock (gypsiferous Keuper), an activated bentonite and kaolin clay. The analytical challenges were again numerous. The high amount of phases in each sample (about 20), many of them as trace phases, the presence of different smectite, well and poorly ordered kaolinite beside halloysite as well as of amorphous materials has placed high demands particularly on the qualitative analysis. Furthermore, the sample preparation was a special exercise with the occurrence of different sulphates. Participants were allowed approximately ten weeks to submit their results. As in the last Reynolds Cup, clay minerals/phyllosilicates were judged more strictly by requiring participants to quantify to a greater level of detail in terms of clay mineral/phyllosilicate classification. The entries ranged from 60 to 400% were ranked according to the sum of absolute errors for the three samples which ranges from 60 to 400%. Due to the fact that several entries are within established uncertainties in the compositions of the pure mineral mixtures, the total clay bias, the sum of relative errors, and a penalty for non- and misidentified phases were applied as additional judging criteria to select the one winner.
The three top contestants were announced at the 51st Annual CMS meeting in College Station, TX in May 2014 and presented with plaques. Only one of them has not been placed at least once in the TOP-3 previously. Interestingly, the 21 TOP-3-places in the RC-history are shared between 11 participants/groups only, which mean some participants consistently achieve excellent results.
Reynolds Cup presentation – from left, Jan Dietel (3rd), Helen Pendlowski and Steve Hillier (2nd), and Michael Plötze (contest organizer)
First Place goes to Reinhard Kleeberg and his coworkers Ulf Kempe and Robert Möckel (TU Bergakademie and Helmholtz Institute Freiberg, Germany). Reinhard is not only placing in the TOP-3 in all but the 2004 contest in which he prepared the samples. He is thus also the first who could repeat his victory in the Reynolds Cup.
The winning team of the 7th Reynolds Cup – from left, Dr. Ulf Kempe, Dr. Robert Möckel, Dr. Reinhard Kleeberg, Sabine Karbautzki, Christine Anders, Kristin Unger (TU Bergakademie and Helmholtz Institute Freiberg, Germany).
Relegated to Second Place only because of the larger clay bias was the team of Steve Hillier with his teammates Helen Pendlowski, Nia Gray, and Ian Phillips. Steve and his team from The James Hutton Institute, Aberdeen Scotland have placed in the TOP-3 in all but the 2010 contest in which they prepared the samples.
The 2nd Place team of the 7th Reynolds Cup – from left, Steve Hillier, Helen Pendlowski, Nia Gray, and Ian Phillips (The James Hutton Institute, Aberdeen, Scotland).
Two entries tied for third place: The group from the BGR in Hannover (Germany) which ended up in the two previous Reynolds Cups on the 5th place could make it on the third place this time. This team including Kristian Ufer (who was formerly working together with Reinhard Kleeberg), Stephan Kaufhold and Reiner Dohrmann had a big advantage with their very recent publication about the mineralogical analysis of activated bentonites and therefore not surprisingly the lowest clay bias on the bentonite sample.
The 3rd Place team of the 7th Reynolds Cup – from left, Kristian Ufer, Stephan Kaufhold, and Reiner Dohrmann (Technical Mineralogy and Clay Mineralogy, Federal Institute for Geosciences and Natural Resources and State Authority of Mining, Energy and Geology, Hannover, Germany).
The other third-placed contestants are Jan Dietel and Jasmaria Wojatschke. Neither has been placed in the TOP-3 previously and both were still working on their PhD at the economic geology and mineralogy research group at the University Greifswald (Germany) at the time of the award.
The 3rd Place team of the 7th Reynolds Cup – from left, Jasmaria Wojatschke and Jan Dietel (University Greifswald, Institute for Geography and Geology, Germany).
Falling just short of the TOP-3 finishers and thus deserving honorable mentions for their great efforts are Katja Emmerich and Annett Steudel (KIT Karlsruhe, Germany) and again as in the previous 6th RC Rieko Adriaens and Gilles Mertens from Belgium (KULeuven University Clay lab and Qmineral Consulting, Heverlee).
Congratulations to all our participants and particularly to our winners!
What were the methods applied
XRD was applied as primary tool for identifying and quantifying minerals by more than 90% of the participants. For evaluation and quantification more than half of the participants used the Rietveld-method. As key complementary techniques grain size separation and their separate analysis particularly of oriented samples was applied by almost 70% of the participants. Other additional methods were i.e. chemical analysis and electron microscopy as well as IR spectroscopy and thermal analysis. Only 6% of the participants did not use XRD at all.
And how did the Top 3 do it?
Reinhard’s team described their procedure:
Our qualitative mineralogical analysis is based on information from conventional laboratory XRPD and a combination of optical microscopy and SEM/EDX. Especially, the investigation of optically selected particles of the >20 µm fraction by SEM/EDX helped to select or improve the crystal structure models used in the Rietveld quantification. Some minor phases what have been overlooked in the XRPD patterns could be identified in the >20 µm fractions. Moreover, some grains of questionable phases separated under the optical microscope which could not be verified by SEM/EDX were investigated by XRPD and IR spectroscopy to confirm special hypothesis, e.g. halloysite and ferrihydrite in sample 3.
One key method for the identification of the minor phases in the very complex sample 1 was the removal of sulphates and carbonates by EDTA treatment and the analysis of the residual. The identification of clay minerals was done conventionally using oriented samples on glass slides. Several treatments of clays (air drying, Mg and K exchange, ethylene glycol saturation, formamide intercalation, heating) have been applied for identification purposes, specifically to the problems of the individual samples. The identification of the R1 I/S mineral in sample 3 was supported by a manual modelling of the basal series in the EG saturated stage using the SYBILLA software.
The quantitative mineralogy was solely derived from Rietveld analysis (BGMN/AUTOQUAN software), performed on diffraction patterns of McCrone milled, side-loaded powders. These patterns were measured with Co Kα radiation, automatic divergence slit, on a diffractometer system URD-6 built in 1989 but equipped with a Peltier cooled solid state point detector Meteor0D. We measured patterns of the bulk material and of the sieved fractions <20 and >20 µm, each added by 20% of corundum as internal standard for estimating the amorphous content. Approaching consistence between the quantitative results obtained from the bulk sample and from the weighted sum of the fraction was used as criterion for plausibility. We believe that this strategy helps to avoid critical errors of the Rietveld analysis.
As there was not enough material for chemical analysis left and as the benefit of chemical data is limited when complex mineralogical compositions have to be quantified, we did not run quantitative XRF or any other quantitative chemical method. Due to this fact we were not able to set chemical restraints, e.g. for TiO2, PO4 and Sr bearing minerals. Some hints could be derived from semi quantitative SEM/EDX of the powders only.
We are grateful to Michael for putting together such complex samples. Errors like the incomplete identification of clay minerals as well as serious bias in the quantification of the amorphous content are taken for wake-up calls to check our methods critically. Other open questions like a serious differentiation and quantification of 2:1 minerals in complex mixtures also deserve efforts in our methodical work.
Steve Hillier said:
How did we do it? Pretty much the same as in 2012, firstly we sampled about 2.5 grams from the approx. 4.8 supplied and McCrone milled then spray dried these portions at 60°C from ethanol. The powder patterns from these were recorded on a new instrument in our lab that we are keen to calibrate for quantitative work, it’s quite a bit of work to run all the standards required, but once it’s done for the common minerals, you can slowly add to them over time, as and when you need to. We then set about the identification of the minerals in the powder patterns – spray drying and the resulting randomness of the power patterns allowing us to rely on relative intensities as much d-spacing in the identification steps. At the same time we separated <2 µm clay size fractions from another small portion (about 0.4 g) of the remaining sample, to make a more detailed clay mineral identification, in order to inform the choice of standards to be used for clay minerals in the bulk sample analysis. We also saved the coarser fractions from these clay mineral separates and examined them under a binocular microscope to cross check with the list of minerals identified by XRD. Once we were happy with the mineral identification, we started the quantification process consisting of fitting measured diffraction patterns for all and every mineral identified to the XRD pattern of unknown mixture using the Solver add-in in an Excel program. We iterated through this process trying to improve the fit between measured and calculated (weighted sum) patterns, a process that feeds back to the identification as it becomes quickly obvious if any major minerals have been missed. Finally we made two cross checks, 1) that a chemical composition calculated from our mineralogical analysis would generate a chemical analysis similar to that which we had measured on another small portion (0.9 g) of the sample by XRF, and 2) that CEC values calculated from the mineralogy were similar to those measured on another portion (1.0 g) of the sample by a cobalt hexamine method. Sample RC7-2 had our highest bias, probably due to the calibration of the amorphous phase used in this sample which can easily ‘trade’ against the di-octahedral smectite pattern, so we underestimated amorphous and overestimated smectite, although we also failed to recognize the minor amount of illite-smectite in this smectite dominated sample.
Reiner Dohrmann summarized the exhaustive procedure with application of many complementary methods as follows:
XRD of untreated material as shipped; inspection with optical microscope; 20 µm or 6.3 µm wet sieving (partly with ethanol); SEM; XRD of bulk material (pure and with 10% ZnO as internal standard); XRD of <20 µm and >20 µm fractions (partly <6.3 µm); XRD (AD and EG intercalated) of oriented specimen, partly at controlled relative humidity (0% – 90%); DTA-MS of bulk materials; Rietveld refinement of bulk and all fractions; LECO C+S determination; IR of bulk and all fractions; CEC determination by Cu-trien intercalation; XRD of bulk and some fractions; calculating normative chemical composition from Rietveld refinement and adjusting with XRF data; specific surface area (only sample #RC7-3); chemical analysis of solutions collected from washed samples during fractionation (anions and cations).
Here is the procedure by Jan and Jasmaria :
First of all we examined the samples by optical microscopy and thereafter using random powder XRD for identifying critical minerals with a tight stability range or well soluble mineral phases. Secondly we took 0.900 g of each sample mixing it with 0.100 g zincite (ZnO) as an internal standard and grinding it with a McCrone micronizing mill with ethanol (agate grinding elements, 8 minutes grinding time).
About 2 g of each sample was used to separate particles <2 µm and >2 µm. Both subsamples were prepared as random powders as well as preferred oriented samples using sedimentation on glass slides. These oriented samples were measured by XRD in an air dried condition and then treated with ethylene glycol at 60 °C fo r 20 hours and thereafter heated to 550 °C. Sample 3 additionally was treated with formamide to distinguish kaolinite/halloysite.
About 2 g of sample 1 was dissolved in distilled water for removal of soluble mineral phases and some HCl solution was added to remove remnant carbonate phases. The remaining sample was again prepared as random powder and preferred oriented sample.
Random powders were prepared by front-loading into aluminum cuvettes and measured on a Bruker D8 Advance diffractometer with Fe-filtered Co radiation, generated at 40 kV and 30 mA, equipped with fixed divergence (0.499°), antiscatter slit (8 mm) and a LynxEye 1D detector.
All random powder samples were first analyzed by using EVA (from Bruker-AXS) with PDF-2 database, release 2009. The following detailed qualitative and quantitative analyses were done with the help of the Rietveld method using the program BGMNwin. Structure models were selected carefully using real structure parameters and model modifications were done to better describe the observed diffraction pattern.
2014 Principal Organizer
Michael Plötze
ETH Zurich, Switzerland, August 2014.