Hi Muhammad,

althouh you see the iron oxide megascopically the iron oxide crystallites themselves in your sample may be too small. Peak width increases (while peak height decreases) with decreasing cystallite size. Please increase your XRD measurement time significantly  to improve the signal to noise ratio of your XRD spectrum.

You will then propably see the iron oxide peak(s); do not expect really sharp ones.

Hi Muhammad,

althouh you see the iron oxide megascopically the iron oxide crystallites themselves in your sample may be too small. Peak width increases (while peak height decreases) with decreasing cystallite size. Please increase your XRD measurement time significantly  to improve the signal to noise ratio of your XRD spectrum.

You will then propably see the iron oxide peak(s); do not expect really sharp ones.

Dear Gerhard Martens,

Thanks alot for your kind reply and information. Actually, i myself run the XRD since last ten years and whatever, you are suggesting i did it so many times but with no success. Even, we analyzed the samples for the known peaks of common iron-bearing minerals such as the magnetite, hematite, siderite, limonite, pyrite etc with enough scanning time e.g., 2sec/step having step size of 0.02 but inspite of that there is no any peak for any iron bearing minerals, which are frequently found within sandstones. That is my concern that why it is not detected in XRD patterns though, it is observed by other techniques, which i mentioned in my question.

Hi Muhammad,

thanks for your reply. The result is very astonishing.

May Ian J Slipper comment on this. She has also got a very lot of experience in this field.

Hi Ian, may you comment?

XRD for clay minerals give excellent results and for sandstone, XRD limits itself. 

Following links may solve your issue.

http://www.iaeng.org/publication/WCE2013/WCE2013_pp2171-2176.pdf

http://www.icdd.com/resources/axa/vol46/v46_30.pdf

http://www.claysandminerals.com/sites/www.claysandminerals.com/files/documents/2000-Accurate-quant.pdf

Regards

http://jsedres.sepmonline.org/content/69/5/1050.abstract

Hi Ijaz;

what is the extract of your papers presented above?

Do you think that the iron of Muhammads sample(s) may be bound in an amorphous/glassy state? So 'real' diffraction cannot be really seen due to the very, very broad peak (The excessive peak width in such a case is not attributed to small crystallite size but to lack of crystallinity). This can be indeed a solution to Muhammads question.

Thanks for this idea. 

@Gerhard, I did not reply on this because I thought that you had covered the essentials of this problem with your answer.

Quartz being super dominant would wipeout the pattern leaving anything else struggling in the background, add to this the low % by weight of the Fe oxides, plus the possibility  of disordered structures and I am not surprised that they are not seen well.

Commenting on the papers suggested by Ijaz: Mubiyai does not identify FeOxides, the highest Fe content was implied to be glauconite. the ICDD paper is concerned with a heavy separation only and is largely irrelevant here. The only mention of interest is goethite (which is worth considering). The Hillier paper is an excellent treatment of the clays in sandstones, but again is irrelevant to this discussion. I don't have current access to the JoSR paper, but from the abstract appears to be a general techniques paper rather than the problem of FeOx in sandstone.

I would like to see the original raw data, so that I could have a look for myself (Muhammad: can you post the data?). I would also be interested to see the SEM images and EDX results of the surfaces of the grains. 

@ Ian

Hi Ian,

many thanks for your remarks on Muhammads questions and many thanks for your kind reply and your valuable time looking into Ijaz's proposed  papers.

Best regards

Hi Kenneth,

does then the" 'lack of crystallinity' regardless of size"  in your sense mean a lost of long range order;  thus beeing almost amorphous having only a short range order over 1 to about 2 (or max 3 or 4) atomic distances combined with a broadening of  pair distribution function with increasing distances?

This is what I meant above with 'glassy/amorphous' state (like liquid water). Such state indeed produces very broad peaks, while the total volume of such a material can extent to mm range or even larger and does not obey the Scherrer equation (the one which is often discussed here in RG with respect to crystallite size) .   

Also much thanks from my side giving here brain opening ideas.

Regards

IJAZ 

Hi Ian,

Thanks alot for for your kind reply. Please find here the XRD image, EDX and a few SEM Images for a sandstone sample.

Hello Muhammad, thank you for the files. These are puzzling, and some of what is to follow may sound harsh, but I am trying to help.

Firstly (and I seem to say this a lot in my replies) the data are of poor quality. The Y-axis indicates that you only have barely 70 counts in the 100% peak. For a sandstone I would expect this to be in the thousands or 10s of thousands in order to gain enough information from minerals in the background. This is the principal reason why you cannot determine your low level iron oxides. In your reply to Gerhard you mention that you used 2 seconds per step, but the information bar suggests the count time is only 1 sec - Is this a scintillation/point counter? I cannot see clearly on this image but I think you just do not have enough data for an analysis. So the first thing to do is increase that count time significantly. How much material did you have? Sometimes with very little material (10mg) traces appear like this; I hope you are using an 'infinite' thickness of sample (0.5-1g).

The data are background corrected, I would like to see the raw data. If iron is present in any significant amount then it will affect how the background looks (with Cu radiation), so don't throw away the background.

I see you have gone through 6 procedures of X-offset. This should not be necessary. If quartz is present, then a comparison with PDF 00-46-1045 should allow you adjust the sample displacement to bring the peaks to the correct position.

You have smoothed the data. This is always a bad thing. There is no reason ever to smooth data. It can only throw away information. Every peak tells a story!

The identifications are not that convincing.The quartz pattern (I do not know which you have used as the PDF number is switched off) does not match your data very well, and qtz is one mineral that should fit perfectly. The first two peaks should be at 20.8 and 26.6 degrees, yours are too low (even after 6 X-offsets), additionally the peaks at higher angles are largely missing. This is puzzling. You may need to check your calibration with a recognised standard before proceeding. The identification of calcite does not match the data. There are very many peaks in the background unaccounted for.

The SQ values are largely meaningless; for this to have any validity all phases must be identified and the intensities of all 100% peaks matched to the data, provided that preferred orientation is absent.

Thank you for the EDX data. Is this from a bulk sample or specifically at a point where you saw a grain coating? Sorry, I could not see any SEM images in doc file, just one EDX spectrum. These data show the presence of ?C, O, Na, Mg, Al, Si, ?S, K, Ca and Fe (Switch off the L-line labels from S, K, Ca to make it clearer). Again you only have a maximum of 16cps in the highest channel, this does not seem high enough for good counting statistics. However, it clearly shows the elements that are missing from your XRD qualitative analysis (Na, Mg). When you have better XRD data, use the EDX results to filter the elements that may be found in the search-match procedure. I am guessing that you are using EVA, it is possible to do this.

All of the arguments above concerning degree of crystallinity are moot until you can provide a good diffractogram.

Dear Ian,

·         Thanks a lot for your detailed reply, information and help to improve our XRD results. Yes, you are right; in this sample the intensity of counts on major peak (100%) is up to 70 but in other sandstone samples, it is up to 1000 counts or even greater but these samples are white or grey in color and there is no any signature of iron bearing minerals. Kindly, tell me that it is always the case that in all sandstone samples, the counts should go up to thousands and then we may identify the minerals of low % such as the iron-bearing minerals and other etc.

·         In case of analytical data, you are right the samples were analyzed at 1 sec/step rather than 2 and it is typo-mistake.

·         The detector is scintillation counter

·         The sample is enough and it is up to 1g.

·         Yes, you are right. The background is removed and here we normally remove the background through software. What do you think? Is it not a correct method?

·         A small shift is done with the X-offset to overlap the peaks of quartz, calcite etc during the evaluation procedure.

·         Yes, the data is smoothed to make the peaks sharp but I will not do it again as you suggested if required information is deleted.

·         The identification is done through the EVA software. The identification of quartz is correctly matching to its two major peaks as you mentioned (i.e., 20.8 and 26.6). But as you know that there are different patterns of every mineral and so far the quartz. There are different structural varieties of quartz and so far there is always a little shift of two major peaks of quartz. It is because of this that there is a little shift in two major peaks of quartz in our sample, while all the other peaks of quartz at higher angles are there with the red colour.

·         Though like the quartz, there is also a little shift in the major peak of calcite near 29 angle but inspite of this, it is calcite because we have the control of petrography. Yes, there are some un-identified minor peaks. Though, we tried to identify them but with no success.

·         Yes, in this case the SQ is meaningless but it supports a little to petrographic findings and therefore we are using it.

·         In case of EDX, it is the bulk analysis not a point analysis. All the elements are matching with the SQ analysis of XRD except the Fe, Na and Mg. Iron is the problem that is the focused of my original question while the concentration of both Na and Mg is < 1% and so far there will be minimum chances of any typical mineral having both of these element or any one. Yes, I am using the EVA while the diffractometer is of Bruker-AXS and it is D8-ADVANCE.

·         Lastly, I am very much thankful to your valuable suggestions and hope that through these discussions, we are solving the issues which we face.

Hello Muhammad,

thanks for the clarifications. Firstly recollect your data at 5 or 10 seconds per step for this sample. It will take time because you have a point counter but it will be worth it. I also use D8 Advance and EVA, but I use a Lynx-Eye PSD which makes the job so much faster - if you can afford it, it really is worth it. Being 10x faster it also has the capability to remove fluorescence due to Fe so the backgrounds give a lot more detail.

If you have clays or other poorly crystalline material - yes the total counts will always be lower. I am currently working on some sandstones and my total counts in the qtz 100% peak are 250,000. This is allowing me to identify and quantify components in the background at levels below 1% using TOPAS.

Use background subtraction before using Search-match, this is essential with EVA, but for display purposes it is best to show the real background. When doing the subtraction use Append instead of Replace, then you always have the real data to hand. In this way you can see whether the sample contains significant amorphous content.

Do you rotate the sample?

Do not use the offset to match calcite. Only use quartz pattern. I find 00-46-1045 to be the best. I tend to use Sample Displacement rather than X-offset. X-offset should not vary if the D8 has been calibrated correctly, but sample height error will vary nearly every time you prepare a sample, maybe only 0.1-0.6 of a millimetre, but it is common to adjust it this way. Low background holders can also give sample displacement errors. The reason I say don't use calcite is that the structure is variable, depending on the Mg content. It is not reliable as a calibration material, whereas quartz is. I have never found a quartz sample in a natural material that deviates from this pattern.

Good luck,

please post your new data as a raw file.

Dear Ian,

Kindly find here the newly generated data according to your suggestions. Possibly, it can help.