9th RC Competition

The 2018 Reynolds Cup was supported by The Clay Minerals Society, the Deutsche Ton- und Tonmineralgruppe e.V. (DTTG), Qmineral Analysis & Consulting and the University of Leuven (Belgium).

Already the 9th edition of this round robin caught the interest of 88 registrants from 28 countries of which 73 finally submitted their best and final quantitative result. The 73 final participants were affiliated with either academia and research institutions (78%) or commercial laboratories (22%).

As usual the three samples of RC2018 were comprised of mixtures of purified, natural and synthetic minerals commonly found in clay bearing rocks and soils that represent realistic mineral assemblages. The prepared mixtures represent the fine fraction of an evaporitic deposit, a weathered marine deposit and a carbonatite deposit. This information was relayed to the contestants with their samples. The non-clay minerals and clay minerals/phyllosilicates were judged in groups as in previous competitions.

The focus of the 2018 RC contest was on the quantitative differentiation of clay minerals which resulted in complex mineral mixtures. Similar to previous editions, 96% of the participants relied on X-ray powder diffraction as a primary investigation technique combined with other complementary techniques such as various chemical analyses, thermal analyses and spectroscopic techniques. Approximately 2/3th of the participants also investigated oriented slides with X-ray diffraction for clay mineral identification. The graph below very clearly illustrates that the use of oriented clay slides very strongly affects the total bias, and thus the final position in the ranking.

The three top contestants were announced at the 55th Annual CMS meeting in Urbana-Champaign, Illinois, in June 2018 and presented with plaques. The top finishers are presented below:

First Place Winners

The team of The James Hutton Institute, Aberdeen, Scotland Stephen Hillier (middle), Ian Phillips (left), Helen Pendlowski (right).

Stephen Hillier presenting the cup.

First place, bias 70.5 %

After winning in 2008, the James Hutton team with their captain Steve Hillier win their second Reynolds Cup with a significant margin. Steve’s overall RC record is nothing but impressive, being a top 3 finisher in all (!) previous RC contests except for 2010 where he was the organizer. Congratulations to the team of the James Hutton Institute

Second Place Winners

Bruno Lanson (middle), Nathaniel Findling (right) and Doriana Vinci (left), ISTerre – Univ. Grenoble Alpes/CNRS, France and Univ. degli studi, Bari, Italy

The team of Bruno Lanson delivered a fantastic performance. Especially the very low bias for the second RC18 sample (weathered marine deposit) is impressive. Unfortunately, the team made a few mistakes in sample 1 adding a significant amount of bias which prevented them from getting close to the first position. Congratulations to the team of ISTerre.

Third Place Winners

Michael Ploetze, ETH Claylab, Zürich, Switzerland

Third Place, bias 95.6%

Swiss quality ! Michael is one of the usual suspects in the top ranking of the Reynolds Cup after winning the 6th edition in 2012. This years’ edition he provided the most accurate result for the fine fraction of evaporitic deposit sample. Well done !

These guys came close:

Fourth Place Winners

Team of Polish Academy of Sciences
with proud team leaders Arkadiusz Derkowski and Marek Szczerba.

Fourth Place, bias 102.2%

This creative team of the Polish Academy of Sciences benefits from the knowledge and experience of
Arkadiusz Derkowski, Marek Szczerba and Jan Srodon combined with several enthusiastic young researchers.

Fifth Place Winners

Mark Raven (right) and Peter Self (left)
CSIRO Adelaide, Australia

Fifth Place, bias 104.0%

Also Mark and Peter are usual suspects in top 5 – they get things done ! They won the Reynolds Cup in 2010 !

How the Top 3 teams did the job:

Steve, Ian and Helen, the James Hutton team: Our method was essentially identical to the method we used in the previous RC8, RC7 and RC6 competitions. That is – we used full pattern fitting of prior determined, in our case measured, X-ray diffraction patterns to model the XRD pattern of the bulk sample and to obtain weight fractions of the minerals identified using a reference intensity ratio method. An example of the measured pattern (black) and the weighted sum of component patterns (red) that was fitted to it for sample RC9-1, showing all the component patterns used to make the fit, is shown below (note the log intensity scale).

To ensure precision in this process we micronize and then spray dry the samples to obtain random powders and our standard patterns are also obtained on similarly micronized and spray dried minerals. Quantitative XRD of the bulk sample was supported by the analysis of XRD patterns collected from oriented specimens of clay size fractions separated from split portions of the samples. This is an essential step for precise identification of some of the clay minerals in the three mixtures; as emphasised in the Reynolds Cup ‘Key Recommendations’ identification is a key a priori step to a successful quantitative analysis. In addition to the bulk samples random powder XRD, and clay size fraction XRD, we examined, using optical microscopy, the silt/sand size fraction which was left over after removing the clay fraction. This was done simply to confirm some mineral identifications and to cross check identifications with XRD data. Finally, we measured the bulk chemical composition of the samples and compared it with a chemical composition generated by assuming compositions for all the minerals quantified by XRD. This later ‘validation’ procedure is about the only way to independently check if a quantitative mineralogical analysis on an unknown sample is reasonable.

By success in the 9th Reynolds Cup it is very satisfying to demonstrate that a simple method like full pattern fitting of prior determined XRD patterns can easily compete with much more sophisticated methods like Rietveld approaches. Furthermore, there are fewer potential pitfalls with a full pattern fitting approach and despite its simplicity of application it is demonstrably a very powerful method of quantitative analysis, capable of the accurate direct analysis of crystalline, disordered and amorphous phases in complex mixtures.

Bruno, Nathaniel and Doriana, the ISTerre – Univ. Grenoble Alpes/CNRS team: A ~1.2g aliquot was milled in alcohol (McCrone mill) to prepare a randomly oriented powder (front loading) for XRD analysis. After data collection, the same powder was mixed with 30 wt% Al2O3 to estimate the content of amorphous material. Phase identification was performed on the 1st data set using Bruker’s regular software (a rather ancient version of it). Next both data sets were processed with the Rietveld code BGMN and its Profex interface. Additional identification was performed for “obvious” intensity remnants.

A 2nd aliquot (~2.5g) was then used to extract the <2µm clay fraction. Potential carbonates and OM were removed first by using a Na acetate/acetic acid buffer and hydrogen peroxide, respectively. The extracted clay fraction was then saturated (Na) and oriented slides were prepared by pipetting a clay slurry onto a glass slide. XRD data was collected in both AD and EG states. Identification was performed first and then “refined” using the Sybilla software to fit the data. The suspected presence of halloysite in sample RC9-2 was confirmed with a formamide test.

Part of the extracted <2µm fraction was used to prepare a randomly oriented powder to differentiate di- and tri-octahedral clay minerals (but again we failed to identify both di- and tri-oct smectite in sample RC9-1). Part of this fraction was also used for SEM/EDX analyses for the same purpose (and the same lack of success) and to assess the actual composition of the clay minerals identified.

The >2µm fraction “residue” of this extraction was used for SEM/EDX observations i) to confirm the chemical composition of phases identified with XRD and ii) to possibly identify minor phases that were overlooked in the 1st identification/refinement round. This fraction was also X-rayed.

Michael Ploetze – ETH Zürich First step was the homogenization of the sample and splitting (rotational sample splitter) in portions. The main tool for qualitative and quantitative phase analysis was the XRD.

A) 1 g of the bulk sample was micronized in ethanol with a McCrone mill and oven drying at 60 °C. Before drying, the slurry was qualitatively checked for magnetic minerals with a permanent magnet.

The X-ray pattern of randomly oriented specimen (front loaded with a blade to minimize preferred orientation) were collected with a Bruker AXS D8 Advance II (CoKalpha, Lynxeye XE-T detector, automatic divergence slit and air scatter slit, 2.5° Soller slits primary and secondary, 4-90°2Theta range, step size 0.02°2Theta, 3 s per step). For a “special look” on the 060 region for di-/trioctahedral clay minerals the measurements were repeated in the 58-80°2Theta with longer counting time.

After the measurement, 1 mg was taken for FT-IR analysis (KBr technic). The rest was mixed with corundum (final content 20 wt.%) by short milling in the McCrone mill and the XRD measurement repeated for quantification of amorphous content.

B) From another 1-g-aliquot of the bulk the 20 µm fraction was separated by sieving with ethanol. Textured specimens (smear slides) were prepared from the fine material and X-rayed air-dry and after different treatments (ethylene glycol, formamide time dependent for halloysite and kaolinite, guanidine for vermiculite, heated at 550 °C).

Evaluation: The qualitative analysis was carried out with Diffrac.Eva (Bruker AXS) comparing the pattern with the PDF-2. Results of the diagnostic measurements of the textured specimens were taken into account. The results were checked for plausibility according to literature and own experience. Unfortunately, the formamide test for halloysite in sample 2 failed although there were indications in the FT-IR spectrum.

Quantification was carried out with Rietveld analysis. The Rietveld program used was Profex/BGMN, with careful selection of models and refined parameters (refinement range, background, real structure parameters).

The total inorganic (TIC) and organic carbon (TOC) (0.1 g sample) and the cation exchange capacity (0.5 g sample) were determined on a third portion for cross checks of quantification.

Thanks to the winners for sharing their knowledge, and thanks to all participants for their efforts in raising the level of mineral phase analysis.

Rieko Adriaens, RC2018 organizer
Qmineral Analysis & Consulting
Heverlee, Belgium