A possible reason for this could be charging of the polymere during EDX. Charging on polymers could be so strong that the e-beam is deflected from the samples and hit the inner of the electron microscope at some arbitrary position. You should try to work with very low (or in some case very high ) Primary electron energy (e.g. 2kv or 30kv). Perhaps this helps. Otherwise coating of the polymer with a conduction element can be tried.

Some older EDX Systems are not very sensitv to light elements (e.g. Boron which is just below carbon is generally difficult to detect).

If you don't observe carbon in EDX even you think that your sample includes carbon, there are several reasons for this.

Are you sure that the electron beam hits to your sample ?

The depth of observations in EDX is around 1 µm (depends on the energy). You don't give any information about the sample. If your sample doesn't include carbon in this depth, may be it is normal.  

Kirsten told about the charging up problem which is very painful for polymer observations and can be avoided by optimizing the primary energy beam or by metal coating the polymer sample. 

What type of detector / window are you using? If you have a Be window then you will not see carbon. You need a type of ultra-thin window to be able to detect elements below Na. If your detector is old it may also have contamination on the window. If you use LN you may have ice on the window. 

As mentioned above polymers will charge up, so if you have the ability to use a low vacuum or variable pressure SEM then that will also solve the charging issue.

For polymers you can work at 1 or 2kV which will be enough to excite carbon. Difficult with a W-gun so try to use a FEG-SEM. What instrument do you have?

Put a carbon/graphite standard sample in and run that using some the suggestions to make sure you can detect it before looking at your material.

You may try coating the sample with platinum or gold-palladium. Coat it for more than 150 sec at 10 mA. For some SEM instruments it is required to keep the electron energy more than 15 kV. Use thick sample for analysis. It will be better if you give more details about your sample and its preparation for SEM analysis.

Sorry, too late for response. I lost the connection

I haven'nt check the detector, because i used SEM-EDX not in my university. Okay, i will check it. This is my SEM-EDX data.

Khusna, You have a very strong signal from carbon. This is the first peak at 0.277keV. The problem is with who ever did the analysis for you has not recognised it as such. 

This is my 1st time SEM-EDX. We just see on the bellow oxide data, not the signal.

What is the different between ren signal and green? We call some oxide and element manually. Thank you for the next answer......

Khusna, the numbers don't mean very much because the data analysis has not been done correctly because elements have been omitted from the quant. I do not know your particular software but it looks to me like the red is your primary data and the green regions look like energy windows that have been added for the identified elements centred on their peak positions and these have been overlaid on your raw data. It is not very clear, however. It looks to me like the first peak at 0.277keV  is C Ka, the second at 0.525keV is the O Ka, this has been picked out in green with an identification marker. The third, red, appears to be at around 2.12keV, but is not identified. The nearest candidates are P Ka but this is at a lower energy (2.013keV), Nb La (2.166keV), but you would expect to see the other L series lines, Au Ma (2.122keV) but you would expect to see the L series starting at 9.17keV, so this line is a mystery (was it gold coated?). The fourth in green is at 2.3keV and is the S Ka line.

If this is the first time you have used the technique, you really should do some reading up beforehand so that you can understand the data. It is very dangerous to use methods you do not understand, especially when quoting the quantification results. There are very many ways that this can go wrong and only a few in which it is done correctly. 

There are many good books to read, but I suggest you start by following the EDS pages in the MyScope website that I have put in the link below - good luck.

http://www.ammrf.org.au/myscope/analysis/eds/

Thank you, Ian. You are right, that i have to know about EDS. During this time, we just know that we can make conclusion from numeral data. 

Ok, i get your link, and start to read it. DOnt be feed up for answering my question. 

Khusna, no problem, glad to help.

I thought about your spectrum overnight and looked at it again this morning. I see the Duane-Hunt limit is around 9keV, so this indicates that your incident beam energy was probably around 9kV, so that means the unidentified peak is most likely to be Au M, as the incident energy is not sufficient to excite the Au L lines. I suspect therefore that your sample was gold coated. If your heaviest element of interest is S then you can afford to reduce the incident beam energy to 6kV which is enough overvoltage for S (2.307 * 2.5 = 5.77). This will help to mitigate beam damage to your polymer.

[EDIT- I see the V acc was 10kV on the SEM image]

Another thing to consider is the length of acquisition time. I notice two things, the maximum number of counts in the spectrum is only around 1200, this is not enough for good peak statistics. It is also reflected in the error % figure that has been generated by the quant software, 5% is very high for S, and almost worthless, you need to aim for error figures of around 0.1- 0.2%

Another thing to note is that the quant values are normalised, this means that it has been done without standardisation. The results can only be used as relative figures and not absolute figures.

Whoever gathered these data for you need to do it again with a longer acquisition time, to improve peak statistics and lower the error. If the analyst is not very familiar with this make sure the specimen is at the correct WD for the detector, this can make a big difference to the counting efficiency. Also all elements need to be identified in the quant software that includes both C and Au. The Au needs to be considered for peak deconvolution but then removed from the final quantification. 

Remember that trying to analyse light elements quantitatively by using X-ray analysis is fraught with problems, at best it can only be an approximate semi-quant result. No problem with qualitative work however. 

Good luck.

I will try to measure SEM EDS again, based on your suggest.

Thank you

Hi Khusna,

Ask your FE-SEM operator to close the chamber view and Increase the voltage and beam intensity to have a better count as suggested by Dr. Ian J Slipper. Hope this will resolve the problem.

Good luck

Raeesh, there is no need to increase the voltage. It is plenty high enough for this material as I detailed in my reply above. A higher beam current will give better count rate, but is likely to damage the specimen more. Longer counting times are what is required, increased aperture size will also help.

Thank you Dr. Slipper,

I tried to do the EDAX analyis using voltage of 10 kV but i was not able to have count no more than 200 at beam intensity of 16 with chamber view in off mode. Please suggest, how to overcome the problem? The material contains P, C, N and O. Will this much count will give the reliable data? And one more thing does the larger WD play any role in having better count.

regards

raeesh

Raeesh, If your material is stable then you can use higher accelerating voltages (Vacc) to obtain high count rates. One SEM I work with to look at geological materials uses 20kV as a 'standard', Another SEM I vary the Vacc down to 5kV depending on the material. As long as the Vacc is x2.5 the excitation energy of the line of interest of the heaviest element then you will get the characteristic X-ray line from that element. So in your example of P, C, N, O the heaviest is P at 2.013keV. That would mean you need at least 5kV to generate X-rays from the P. So 10kV is plenty and will increase the count rate significantly. But, your material is elementally light and probably beam sensitive, so I would be using the lower value. If you are not getting counts then the problem lies elsewhere.

You mention beam intensity. Different manufacturers use different terminology for this, intensity, beam current, 'spot size', 'resolution', probe current, condenser lens, all controlling the same parameter. Also different manufacturers display this differently, sometimes with a nano or pico ammeter, sometimes with a decade dial, or an arbitrary set of numbers of spot size. I do not know what instrument you have, but generally yes the intensity should or spot size should be increased. You need to put a standard sample of a metal such as cobalt under the beam and ensure that you are getting the correct count rate as when the instrument was installed, this is a routine check on any EDX system.

As far as chamber view is concerned, I do not have these on my scopes, but it would only cause a problem to you if you were trying use the backscattered electron detector at the same time. So off mode is ok.

Things you need to consider regarding WD: Is the sample at at the correct height for the detector. I cannot give you a value as all SEM/dectector combinations exist and you need to check with the manufacturer what the recommended WD is for your instrumentation. If the sample is above or below this correct height then the count rate will drop off. It is important in EDX to set the WD at the correct value by first adjusting the focus, only then move the Z-height until the specimen comes into focus. That way you know you have correct focus at the correct WD for the detector.

Other points about the geometry of the system, the sample needs to be flat and horizontal. If it is rough then you will different results from different parts of the sample.

Other considerations: What aperture are you using. Some SEMs are set up for higher resolution imaging and this configuration will give poor EDX. So using a larger objective aperture will drastically increase your count rate.

Process time: There is an inverse relationship between count rate and peak resolution that is controlled by the process time, a long process time will give you high peak resolution but poor count rate, and a short process time will increase the throughput of counts but will give poorer peak resolution. You need to adjust the process time to give a dead time of 30-50% for the best count rates. (This will depend on the type and generation of your detector and pulse processor, so again check the manufacturers instructions.)

Counting time. In you message you say 'have count no more than 200' is that total counts or count rate? I could reproduce this 200 counts using a very short counting time for a high count rate. You might just need to increase your counting time. Again put a Co metal standard under the beam at make sure you are getting the correct count rates (This will depend on detector/pulse processor) but an old system I have used 2,500cps, a newer one used 4,500cps, but modern systems can give you 10s thousands cps.  Older systems used to use 60seconds as a 'standard' count time, but with modern equipment this can come down to a few seconds, again it depends on your count rate. You need enough counts to generate good peak shapes and reduce statistical errors.

I mentioned this above but I will repeat, trying to quantify light elements by EDX is not the best use of the technique. It is not easy to get reliable results. You need to use standards to calibrate and ensure that what you get is real. It is very easy to generate numbers, with very high precision, from an EDX which are meaningless.

So my advice to you is to learn the  technique thoroughly before trying to use it or else you can go away with false results. Lots to think about.

Sorry Khusna for hijacking your thread!

Thank you Dr. Slipper for explaining thing in detail.

Sorry khusna for coming in between but i thought it is always good to get some idea from experienced person like Dr. Slipper.

best

raeesh

No matter, Raeesh. Glad to share it. 

Dr. Slipper  : I still stand in line for SEM-EDX instrument. I will try what your suggest to me, at the same time studying about SEM-EDX from your link shared