It is possible to analyze B2O3 in Colemanite by WDXRF. You need to make sure you have the right multilyaer analyzing crystal in the spectrometer. Each vendor has one capable of the analysis and they go by different commercial names but they all have extremely large D spacings. You should have more than enough signal, but the trick will be the sample preparation. The escape depth for BKa is extremely small, so you will need to be sure your sample is very homogeneous, ground to a very small particular size ( I would advise < 50um) to minimize grain size effects and pressed with a very flat surface. You should also be careful to not use a binder containing Boron. Cellulose should work fine. Precision in the sample dosing and pressing is also key as the surface needs to be very flat and the same from sample to sample. If the application is a critical one for you I would advise considering using some type of automated sample preparation to get the most consistent results. In short the application is quite possible, but you need to take great care in the sample preparation to insure good results.

I am application scientist at Rigaku in Germany. You seem to use our ZSX Primus II. David already gave the right hints concerning measuring conditions and sample preparation. So, I will not repeat this. In order to measure Boron you should use the S8 slit and the RX75 crystal.

I have seen application notes for analyzing boron by XRF, but did not think the results could compare with the wet-chemical analysis. Our lab analyzes colemanite as is done in Senem's lab, using fused beads for the WDXRF. We also analyze ulexite in the same manner. These materials are a challenge as ~1/3 of the sample is determined by XRF, ~1/3 is %LOI and ~1/3 is B2O3. Have either of you actually compared the XRF and wet-chemical B2O3 analyses? How low of a concentration can be quantified?

I would really be interested in seeing data obtained by WDXRF compared directly to the wet-chemical analysis. This would be a great time-saver for our lab, if the precision and accuracy can be demonstrated.

Excellent points! I am interested in your approach using fused beads. I am assuming you are able to swamp the B blank because the concentration of B2O3 in the Colemanite and Ulexite is so high. I'd be interested in hearing more about your approach and recipe. What level of B2O3 precision is required for your B2O3 analyses in the Colemanite and Ulexite ?

We don't have a comparison between wet analytical and XRF. I guess with the pressed pellet method and the ZSX Primus II you have, in case your spectrometer has the right Hardware, a precision of +/- 0,01% could be achievable, taking into account a long enough measurement time. 

Are you using a borate flux for analysing the B2O3 as fused bead or is there another possibility? I could imagine sing Li-borate fluxes for analysing Boron is not that nice.

To clarify, our lab determines B2O3 with wet-chemical analysis only and it is quite time intensive and requires significant training of a technician to get them to the point they can obtain high quality results. Our current reported uncertainty is +/- 0.08 wt. % (absolute). The fused beads are for WDXRF only - the typical 1:10 ratio with Li-borate/LiBr flux. Our XRFs are PANalytical - sorry Pol. I do think the Rigaku XRFs are high quality too though. : )

Our XRFs do not have the crystal for boron, but one of our other sites did order one. Just a thought - would there be a non-B flux that might be used with these borate minerals? Could Na2CO3/K2CO3 be used to make a bead? It is used to get these materials into solution.

Thank you for the suggestions and explanations.  I will try to make PPPs (pressed powder pellet) and make a calibration program out of it, maybe just for Boron, because I dont think I can get the precision and accuracy I need for light elements with PPP. I may continue to analyze other elements with my colemanite program using fused beads.
But a non-B flux sounds very nice, if there exists one.  I think I will start with checking of repeatibility of suggested sample preparation techniques. If I can achieve high repeatibility, I can go on pressing the standard powders for colemanite.  I will let you know about the results.

Somebody mentioned the need for very fine grinding. For what it´s worth, this method has given me good results:

• Use a Herzog vibratory mill with 100 ml capacity
• Grind 15 gram material adding 3 drops of Tri Ethanol Amine
• Grind for 3 minutes
• Press pellets without binder as mentioned by Lynn

For cement this gives a 50% particle size of 15-20 microns with a maximum of around 50 microns. I have not analyzed B, but for my lightest element (F) it is very repeatable.

I would suggest that colemanite with its high B2O3 content should be almost self fluxing. The colemanite should decompose releasing H2O to make a glass of sorts CaO.xB2O3 and these will melt in the temperature range from around 1000-1300°C. So all that is required is a suitable alkali flux to reduce the melting point further and to improve fluidity of the glass.

I would use Li2CO3, but the other carbonates that Lynn suggested Na2CO3, K2CO3 will all work. Others like NaNO3, KNO3 and even LiNO3 are also suitable. The disadvantage of the Li2CO3 decomposes/reacts at a higher temperature – only at around 850-900°C while the Na2CO3 and K2CO3 decompose at 550-800°C. I would use Li2CO3 because the resulting Li2O would absorb the X-rays less and give a greater penetration and escape depth which will be better for the B2O3 analysis. Both Claisse and Scancia sell Li2CO3 as a standard high purity product.

The trick is to control the fusion because of the gas evolved. The colemanite would probably start to lose the water of crystallisation at 300-500°C, then the carbonate has to be decomposed/reacted along with the further decomposition of the colemanite. Finally the glass may need to be annealed to remove any dissolved gas.

This is not too dissimilar to one of our fusion methods, from our experience we fuse at 3 different temperatures 650°C for around 6 minutes, ramping to 850°C with another hold time of 6 minutes and then the final fusion at 1050°C for a hold time of 6 minutes – then casting. If the glass contains bubbles then the next strategy would to allow the glass to cool after the initial fusion – down to around 900°C so that it begins to solidify. Then re-heat it up to 1050°C before re-casting it. It may also require a wetting agent to be added manually before casting – using ammonium iodide pellets. This is possible using a Katanax fusion machine. You could also consider adding a little LiBr (50mg or so) to the initial fusion.

I would estimate that about 12g of sample (I’m estimating a 30% LOI) and 2g of Li2CO3 would get you around a 10g bead.

A mass balance would be required to correct for LOI, so I’d determine the bead mass after fusion and correct for dilution/preconcentration. It sounds like a really interesting application. The big trick with this application will be controlling the evolution of gas and preventing the sample from spitting. I have never fused such a large sample with such a lot of potential gas coming off, it may take a lot more time than I have suggested. If you heat too quickly the sample may foam over the top of the crucible and damage the fusion equipment.

Good luck I hope this helps.