Was specimen polished or fractured?

Topographical features of specimens can shield EDS detector.

Dear Vladimir;

Many thanks for your answer. 

The specimens were fractured, all included SiAlON as sintering aid. Several other specimens (with similar topography) investigated and results were excellent. The problem is just about SiAlON containing samples!

which element you can't see and which model of SEM-EDX you are using? KeV you are using? Sample treatment?

coating error?

Si, Al, C, N, B, O and Zr; all elements can be detected except B due to its low scattering

factor. The problem may arise from the sample preparation.

Please post a spectrum so that we ca have a look wher the problem could be. Additional line are commonlyfluorescence peak, escape peaks and sum peaks. The first are related to fluorescence of the detector material Si (caused by incoming x-rays) which is not problematic in your case since you have Si. On the other hand they are minimized nowadays. Escape peaks are produced when instead of all of the energy of the incoming X-ray is converted to electron-hole pairs. The energy measured for the incoming X-ray is therefore reduced by the magnitude of the Si Kα X-ray ( 1.74 keV), and an escape peak is detected in the EDX spectrum at an energy 1.74 keV less than the energy of related major peaks in the spectrum.

The last one, the sum peak is produced when the pulse processor cannot distinguish between two X-rays that arrive almost simultaneously. Instead of recording two X-rays with either the same or different energies, one X-ray with energy equal to the sum of the energies of the two incoming X-rays is recorded and plotted in the spectrum. 

But as I said, the best is to post a spectrum so that we can help a bit better.

Dear Behzad, as I understand (may be incorrectly) from your description of spectrum it consists just of few impulses (“sharp vertical lines”), which is typical for the case of shielded detector. It means that straight line connecting point of interest on your specimen and detector was intersected by some feature (“wall”) of specimen topography; therefore no radiation from the point of interest was reaching detector.

Dear Yue;

No element is detected due to undetectable peaks (as mentioned above, the outcome is just several unstable vertical lines). SEM Model: MIRA3 TESCAN, HV: 10KV

Due to their electrical conductivity, samples were not Au/Pd coated and simply fractured.

@Ferenc: Samples were not coated.

@S. Alleg: investigating the problem from preparation process aspect, I tested some samples with similar preparation process but different compositions. the results were clear.

@Gert Nozle: Thank you for your detailed guidelines, unfortunately as I mentioned in Question, no spectrum obtained from the analysis. I think Vladimir surprisingly illustrated the problem more clear than which I explained! now, I am seeking for a solution!

@Vladimir: Thank you for your help and illustration (showed that you are undoubtedly fully experienced in the area of Electron Microscopy). Attached, please see one of the SEM Micrographs. is it possible in such a sample that the connection line being intersected by the sample's feature? if so, do you have any suggestions to solve the problem?

If it was a spot analysis with the spot located in the middle of the picture, then definitely X-rays were blocked. Central spot lies pretty dip and is surrounded with relatively high crystals which would prevent X-rays going to detector.

However often it is difficult to determine from a picture which feature can block X-rays, but when you are getting an empty spectrum (no lines, no usual background, just a few “spikes”) it is reasonable to assume that X-rays were blocked.

You can try to obtain spectrum from other spots, those which look higher than surrounding features. You can rotate specimen. You can tilt specimen toward detector. You can look on your specimen without microscope and check whether one side is much higher than the other one.

Nevertheless it would be nice to see such kind of spectra. We are also looking on fractured surfaces with all consequences since the shielding can occor also from parts far away from the imaged region but I must admit that I've never seen only some "vertical lines"!

@Gert Nolze

As the spectra were not clear, unfortunately I did not save them. Anyway, I will send you some of them, as soon as I proceed another EDS analysis on my samples!

@Vladimir Dusevich

The strange phenomena is that I got no meaningful spectrum from 6 samples containing SiAlON, even by testing several spots, tilting the samples etc. :( 

10KeV is used for light elements Z

Although we cannot say anything relevant at the moment we should wait for a collected spectrum I am also surprized since even in case of shielding there are usually many backscattered electrons leaving the surface and generated somewhere else radiation (e.g. at the edges of the shielding objects if they are still covered by the collimator). It is certainly an interesting topic for discussion :-). 

@Gert

Yes, I am surprised too. I will carry out another analysis on Wednesday, and share the results with you. I want to be sure that if the EDS detector works correctly, so I will use another SEM.

Thank you for your interests and helps.

elactrons are electrons, anyway which SEM you are using :-). I still hope Vladimir is right...but first it is good to see a single spectrum. Good luck!

I think your detecting wave is very week, you should change applied current to have enough energy for detecting of the materials with different Z.

if using uncoated samples, perhaps you should consider some surface charging, even if you "don't have to"... for some types of materials it's just not working (according to my experience) even if you have a clear SE image...

@ Ferenc: The surface charging is certainly a general challange which need to keep in mind, but how this should cause the described "lines"? As far as I understand this effect, charging mainly reduce the kinetic energy of the electrons and therefore falsify the assumed acceleration energy so that a quantification becomes errornous.

@ Ali: if you change the current you only increase the count rate (energy density) but not the characteristic of the spectrum since the kinetic energy of the electrons producing X-rays keeps unchanged. Or do you recommend the change of the acceleration voltage? But from my point of view for the detection of Si, Al, N and O you don't need a higher electron energy.

@ Gert Nolze: Dear Gert, thank you so much for your comment on my answer. As I knew, an increase on energy density can increase the count rate (the intensity of detected wave), so I thought that may be good suggestion. I should revise my knowledge about and thank you once again in advance for your useful comment.

I think, considering your description, that the problem may be in the fact that you're watching a surface of a fractured ceramic after sintering and want to do the EDS analysis on that area. In this type of samples the fracture surface irregularity may prevent the secondary electrons scattered by the sample (which are used for local chemical analysis) that does not reach the EDS detector and, therefore, any element was detected. I suggest, if it is possible, that you first rectify the sample surface in the area you want to observe, in order to eliminate the surface roughness and to use the largest possible area for EDS analysis.

@Manuel

Thank you for your recommendations, it seems that the problem was due to Detector Calibration!!!!

Although I know that it somehow "embarassing" to talk about it: but could you tell us what exactly happened since this - from my point of view - could accidentally happen to everyone. in so far it would be nice what we should prevent. It would be still interesting to see a non-calibrated spectrum in order to see when we should be careful.

Many thanks!

Hi, Behzard,

Ok, it is good to know that the issue is apparently solved .... Note however that the problem of a poor detector calibration gives rise to a similar fault that I described before, ie, a "empty" spectra peaks in samples crystal EDS analysis are always related with problems of signal detection.

@Manuel and Gert

Thank you for following this funny problem. Unfortunately I am not authorized to work with system directly and I always ask SEM Operator to perform needed analysis!

Today morning, using another SEM (FE-SEM TESCAN MIRA3), samples were characterized without even minor problem! and operator said: "The previous problem was probably due to detector calibration"! Do you know what dose it mean??? maybe it was related with poor signal detection or any other causes. anyway, he did not try to clear the problem and solution!!!

Behzad,

When we speak on "detector calibration" this usually means that the EDS detector is not aligned properly with the electron beam which irradiate from the sample and / or "something" is obstructing the passage of these electrons between the sample and the detector such as a very high surface roughness that prevents these electrons from reaching the detector (EDS detector it is normally placed laterally at an angle of incidence close to 45 ° relative to the sample position. Note, this angle depends largely on the type of equipment, etc. ).

Hoping to have helped

Manuel Ribeiro

Dear manuel;

Thank you in advance for your helpful information. I hope i could pay your helps back.

When I talk about detector calibration than it is not related to the sample-detector alignment since this is usually defined by the so-called analytical point of the SEM and for a fix detector installation also given by a specified working distance +-5mm. The detector calibration actually is responsible for a correction of an electronic drift or a changing efficiency of the chip which need to be checked from time to time so that the EDS software is still able to identify the correct wave length and quantify the signal in an acceptable quality. These calibration procedures also enable to recognize whether a system has ice or oil on the window (Be or polymer) and needs to be cleaned in order to interprete incoming x-ray quants correctly. The calibration also allows to monitor the energy resolution of the detector which decreases over the time as well.

Detector calibration is needed EDS detector it is normally placed laterally at an angle of incidence close to 45 ° relative to the sample position.

@ Karunesh: This is only forced by the available ports in SEMs. You can practically use any position as long as you still see your sample surface, however the quantification errors become more essential for small angles (gracing incident). In fact, the best position of a detector would be vertical to the surface, i.e. parallel to the surface normal. Practically you could even use a half sphere detector in order to collect all x-ray quants. First adaptations of such detectors are multiple chip detectors which cover a bigger solid angle and do not loose energy resolution. In SEM they are known as Quad detectors (Bruker) and are in fact horicontally aligned to the sample. The beam is going through a hole between all four chips. This is also a common configuration in modern TEMs where each count "counts" :-).

Coming back to your statement: the setup of an EDS detector requires the geometrically alignment of the detector with respect of the sample, what automatically means that for tilted samples the applied corrections are wrong! However, with the original calibration of an EDS detector this has nothing to do! There only the signal processing of the electronic needs to be calibrated that the counts delivered by the chip are accurately processed and displayed! It is comparable to XRD where you also need to know the point of Zero (usually they are using as reference point 4 degrees), and the determined energy resolution in EDS is comparable to the the exact alignment of slits and sample/detector focusing.  

Similar issue faced by me. I spin coated nickel nanoparticles without removing impurities like Na, Cl etc. And , unexpectedly I'm getting no presence of Nickel element in EDX... I were wondering what could be the reason? (NPs size is around 1nm. And impurities looks bigger ~ micron size.)