Hello, I prepared bimetallic metal/metal oxide nanoparticles embedded in N-doped carbon materials and annealed at 450 and 750 degree C and grinded to fine particles. Then, I measured the XRD of these materials with slow scan rate (1 degree per min) and the patterns are attached here. It shows high noise in the patterns. May I know if this kind of noise is usual regarding my materials? If not, what technique(s) should I follow to reduce noise ratio? Any kind suggestion will be gratefully accepted. Thank you.
Regards.
Dear Md Mahbubul Alam ,
I think, the concentration of your NPs is quite low, so you will have a low intensity diffraction signal from them.
From the experimental point of view, you have to significantly extend the measurement time to lower the relative noise.
Alternatively you may add up many short time measured pattern.
From the data evaluation point of view you may smoothen your pattern by software ( e.g. via ORIGIN program). But this alternative is not the primary choice.
In addition, I see, that you pattern suffer from high background, which also contributes to the noise.
Such a background, or at least part of the background, may arise from fluoresence of your metal atoms.
If you have got Co, Fe, Mn , Cr, V etc. in your sample the K-fluorescence will be excited by the Cu radiation of your x-ray tube. If so, you should use a Co based x-ray tube to avoid at least the fluorescence from Fe.
Dr. Gerhard Martens, I'm highly grateful for your elaborated answer as well as your very kind suggestions. Actually I was planning to form metal single atoms nanomaterials, therefore my metal loading is really less (as you guessed). Also the X-ray source was Cu radiation. So, as per your answer, it is very obvious the metal fluorescence contribution in high background.
I learned today how to smoothen my XRD pattern by using ORIOGIN software. but before utilizing this, I want to deal with noise more scientifically.
One more thing, may I have any reference regarding the metal fluorescence interruption with metals in sample during XRD to increase noise? It will be highly appreciated.
Thank you again for your very kind and cordial suggestions and discussion.
This is quite common, although it could be improved. I would recommend increasing the measurement time and therefore the total recording time of the diffractogram. On the other hand, sample preparation is also very important: there must be enough sample in the sample holder and it must be as flat and homogeneous as possible. Increasing the proportion of nanoparticles would also improve the result but, surely, with the aforementioned, it improves a little more.
Thanks Md Mahbubul Alam , for feed back.
With respect to enhancement of noise due to fluorescence background, I have no reference at hand. But I think that is not necessary.
Any background B will contribute an increase of the noise to that of the net signal S.
Noise of net signal S : sqrt(S)
Noise of superposition of signal S and background B: sqrt(S+B)
Here it is of no relevance, whether the background arises from fluorescence or shows up as 'diffraction' signal from any amorphous part of the sample (e.g. the matrix).
Due to the fact, that you have 'doped' your carbon composite by NPs the relative amont of the NPs in the sample is quite small. Thus your background is mainly due to the amorphous part of your sample matrix material.
To me the step size may be too small. You can bin down the existing measurement by factor of 2 and see how the spectrum improves.
Dr. Francisco Manuel Marquez , Thank you so much for your valuable and very kind suggestions. So, next time I'll be more careful to follow your guidelines.
Dr. Gerhard Martens , Thanks again for your nice explanation. Actually I wanted to make a catalyst of high carbon and low precious metal content. So may be with metal fluorescence, amorphous carbon content is contributing to make high noise. However, I will be more careful as per your suggestion to some good results of least noise. Thank you.
Dr. Grzegorz Cios , I regret that I couldn't catch the point you expressed, and therefore your suggestion. A little bit elaborated explanation will be very helpful for me. Because I 'm desperate to obtain some good xrd result. Thank you so much.
Dear Md Mahbubul Alam ,
by binning, which is suggested by Grzegorz Cios , is meant, that
a) you may add up the counts of an adjacent measurement point (to the left or the right) and
b) assign a 2theta value to the result, which is the mean 2theta of both x-axis values., and
c) do this for all measurement points of your pattern.
The binning has the effect of doubling the measurement time und thus reducing the noise by sqrt(2). It is some kind of software smoothing.
However the pattern resolution in 2theta will degrade a bit. On the other hand, you deal with NPs, which by nature will have a bit broader peaks compared to non-NP material.
You may also add up the counts of the left and the right neighbor of each data point to the data point at the center.
When doing this for all data points you will have a smoothed xrd pattern of effective triple measurement time, but at the cost of further degraded angular resolution. The noise will go down by sqrt(3).
These two kinds of smoothing are only simple ones. Modern smoothing softwares apply more advanced kernels, which make use of a sum of specially weighted neighbor intensities or more complicated functions of them. The goal of all these advanced kernals is to minimize the accompanied degradation of the angular resolution of the XRD pattern, while reducing the noise.
Yes, this is what I meant Gerhard Martens and as long as measured peak consists of >10 (some papers say 6-9) 2theta measurement points (or binned measurement points) fitted peak should have good enough resolution. Md Mahbubul Alam if you will post here the measurement files I can show you the effect of binning.
- Badges
- Science method
- Similar topics
- Methodology
- Crystallography
- X-ray Diffraction