I'm a student and have recently done some XRD on some iron ore mine tailings which came from an Australian mine tailings storage dam. The redness in colour and industry source led me to reason that a significant amount of iron may still be present in the tailings. I currently have access to 2 X-Ray Diffractometers each with a different x-ray source (Copper vs Cobalt) and I've heard that there can be significant differences in diffractograms due to the scattering of copper vs cobalt x-rays by iron. I wanted to see this effect in action so I ran the sample on both XRDs.


Below are the XRD instrument details, I am using Match! for PhaseID and TOPAS5 for QPA.

Parameter XRD1 XRD2
Instrument Bruker D8 Advance Bruker D8 D
Emission Profile Cu Ka 5 Berger Co Ka 7

Results: Assuming I've read and analysed the bulk XRF analytics right, the sample seems to contain mostly phosphorus at 45.2 wt% and manganese at 23 wt% (which I presume comes from the clay and soil covering used to contain tailings etc.) with only 1.3 wt% coming from the iron itself.

Looking at the following diffractograms where the blue trend is from XRD1 (D8A) and the green trend is from XRD2 (D8D) I am surprised by the stark differences in pattern, namely the transpositions of the background (overall elevated intensities) and 2$\theta$ positions (there seems to be a lag between the patterns).

XRD Patterns

The Questions:

  1. How do the different emission profiles affect the diffraction patterns for this sample and what is going on physically?

  2. Does the difference in emission profiles explain the higher background, translated 2$\theta$ positions, both, neither, or only one of the differences seen between the patterns?

  3. How could I know quantitatively if an XRD emission source is good/suitable or bad/unsuitable for analysing samples containing iron (or any other element) components?

  4. Is it reasonable to presume that the instruments have been mixed up in the naming files because the blue trend (Co emission XRD2) is lower than the green trend (Cu emission XRD1)?

  • 2
    $\begingroup$ That is a really broad question. My wife just got back from a one-week intensive short course on XRF and there are a lot of details to pay attention to in order to get accurate composition analysis. You might want to ask if there is a standard (maybe silicon?) to compare the two instruments. A lag in 2-theta is expected because the wavelengths (energies) are different for the two systems. A table of K-alpha energies are at hyperphysics.phy-astr.gsu.edu/hbase/Tables/kxray.html which may help as well. $\endgroup$
    – Jon Custer
    Oct 26, 2022 at 13:36
  • $\begingroup$ A 10 wt% $CaF2$ standard was used in the sample which was loaded and analysed with both XRD instruments. $\endgroup$
    – Hendrix13
    Oct 26, 2022 at 13:45
  • $\begingroup$ Fair enough. Now, take the Cu and Co x-ray energies (wavelengths) and calculate the 2-theta shift of a given lattice spacing. $\endgroup$
    – Jon Custer
    Oct 26, 2022 at 13:51
  • $\begingroup$ You mean by just rearranging the $$2d(sin (\theta ) = n \lambda$$ Bragg equation for $$2\theta$$ and assuming n =1 right? $\endgroup$
    – Hendrix13
    Oct 26, 2022 at 14:04
  • $\begingroup$ That is pretty much it (you may need to watch for n=2 peaks as well - it has been a long time since I though deeply about diffraction). $\endgroup$
    – Jon Custer
    Oct 26, 2022 at 14:14

1 Answer 1


1.) Assuming you use a cathode tube as source for the X-ray radiation, the target material emits bremsstrahlung what (on first approximation)* looks like a continuous emission (:

enter image description here

(image credit above Wikipedia article)

On top of this you have a number of sharp emissions which are characteristic of the element's relaxation of inner (non-valence) electrons. This is why you read experiments described as using e.g., Cu $K\alpha$, or Mo $K \alpha$ radiation. You don't need to use solid metals, indeed liquid metal target like molten Ga or In offer X-ray sources of high brilliance (though smaller in diameter than conventional tubes).

The underlying principle for the diffraction experiment is the Bragg equation, $n\lambda = 2d \sin(\theta)$ where $n$ is an integer, $\lambda$ the wavelength of the X-ray radiation, $\theta$ the glancing angle of the diffraction experiment, and $d$ the distance of the lattice planes characterized. (It just so happens that these interplane distances are about of the wavelength of X-ray radiation; given a well suited repetitive pattern, diffraction equally may be observed with visible light (say scale of about $\ce{500 nm}$), or even sound.)

Both single crystal (XRD) as well as powder (PXRD) diffractometers go great lengths to pick and select only the characteristic wavelength for the experiment. Again, depending on the target material of the X-ray tube and thus characteristic emission spectrum, both design and material for the monochromator varies (e.g., for Mo radiation, graphite) widely.

2.) To record the diffraction signal, you either a point detector (similar to a Geiger counter which moves in respect to X-ray source and sample), or a fixed CCD array (similar to a blown-up camera, horseshoe like bent around your sample). They differ in their sensitivity for X-ray radiation (which equally depends on the wavelength used in the experiment). An other plausible reason that green and blue trace in your superimposed diffraction pattern appear moved against each other are different strategies to counter (as in exclude a priori by experimental design) e.g., by variation of the time of exposure to collect intensities, averaging multiple single recordings to cancel out noise in intensity while varying along the $2\theta$ scale. There equally are many mathematical methods to discern the pattern of noise from pattern of signals and dedicated programs (e.g., fityk) before one even starts to assign the diffraction peaks. Though the blue trace looks like to be better in terms of signal/noise ratio than the green one, the two traces include differences (presence/absence of signals) which can not fully attributed to variation of Cu/Co only. A potential pitfall is preferential orientation by sample inhomogenity (e.g., not well enough grinded crystals) when filling a capillary sample holder/transmission geometry (sample is between X-ray source and detector) vs. placing the powder sample on a rotating dish/reflection geometry.

3.) Of course, you select a brilliant X-ray source for your sample to obtain a good signal/noise ratio, however this equally should account for the sample's composition. While inorganic samples are much less prone to decompose under X-ray radiation than organic/biologic samples, your incident X-ray radiation actually may initiate X-ray fluorescence in your sample. Then, the metal atoms' of your sample equally may contribute to the signal recorded. If I remember correctly from the colleagues of inorganic chemistry, samples with Fe are susceptible to this type of interaction with Cu $K \alpha$ radiation. Check this.

4.) You did not show an emission spectrum (or two emission spectra), but diffractograms. If peaks are narrow (or broad) has to be taken into account to characterize your sample as well as the peaks' intensity pattern. At the end of performing a Rietveld refinement, there should be a good match of the diffraction pattern recorded with a hypothetical diffraction pattern based on the sample's composition -- both along $2\theta$ as well as the intensities. This is why publications add a difference trace below the diffractogram recorded during the experiment:

enter image description here

(image credit to Wang et al, CrystEngComm, 2018, 20, 699-702)

Engaging a PXRD experiment and Rietveld refinement is suitable both for pure compounds, as well for the analysis of mixtures like alloys, ashes, clays, etc. So you may reach your goal to characterize your sample, ideally, you validate the results with a technique independent of PXRD (since it is a mineral e.g., by X-ray fluorescence).

*) Of course this is not true, electromagnetic radiation is discrete along the energy scale (Planck).

  • $\begingroup$ Thanks for the answer! So following on, (2) how do I minimise preferred orientation?, (3) what metric quantitatively can I use to measure whether a certain x-ray source is good or bad, the metal's x-ray fluoresence number?, (4) I've since done a rietveld refinement and can provide the picture if need be. $\endgroup$
    – Hendrix13
    Nov 2, 2022 at 1:00
  • 1
    $\begingroup$ (2) E.g., don't let needle-shaped crystals stack like matches in the sample capillary (transmission geometry); don't let plates (e.g., mica) stack like a pile of newspapers on the rotating disc of the sample holder (reflection geometry); grind (not squeeze) the sample finely, uniformly in size (diameter-like) and dimensions (e.g., materials-talks.com/…). 3) I'm unaware of a fluorescence number, yet one can compare the present setup with setups used for characterizations of similar samples (literature). $\endgroup$
    – Buttonwood
    Nov 2, 2022 at 16:53
  • 1
    $\begingroup$ An example would be to consider the same X-ray source, but instead of K$\alpha$ to pick $K\beta$ instead (example X-Ray Powder Diffraction: Why Not Use CuKβ Radiation? by Otto, doi.org/10.4236/jasmi.2018.83004). Equally, some of the $R$ values of the model solved and refined are not only about the plausibility of the model (interatomar distances/angles), but account about the separation of (wanted) diffraction signal and (still recorded) background ($R_B$, or Bragg residual). Eventually, not a single criterion, but multiple criteria describe the quality of the experiment. $\endgroup$
    – Buttonwood
    Nov 2, 2022 at 16:58
  • $\begingroup$ Thanks for the great insights especially for (2) as I've only been using rotating discs for everything I do. Now I know to preferentially use that for needle-like samples and sample capillary holders for plate like materials. For (3), good points, another good way I guess is just to compare the background from the two XRDs I have access to. $\endgroup$
    – Hendrix13
    Nov 3, 2022 at 4:32
  • $\begingroup$ The comparison of backgrounds were sensible with fixed incident radiation, if sample crystallinity similar enough (if not identical sample), the same detector and same detector parameters were used. E.g., the croissant-shaped image plates for PXRD allow a quick recording along $2\theta$, but with fixed resolution (in $2\theta$). PXRD's point detectors may move in different angular speed ($2\theta/t$), often in much greater angular resolution than an image plate (fixed by CCD pixels dimension & design) can offer. Some Empyrean PXRD (e.g,, cmca.uwa.edu.au/facilities/xrd/empyrean) $\endgroup$
    – Buttonwood
    Nov 3, 2022 at 21:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.