(i) EDX is a local probe. The probed volume can be as low as ~1 cube µm for SEM EDX, and as low as ~1 cube nm for TEM EDX. Laboratory XRD can have a probe spot larger than ~1 square mm. You might have local concentration gradients, which you will not detect with XRD but sure are detectable with EDX.

(ii) EDX precision. If you do not know what you are doing, EDX can give you bad concentration values. Precise determination of concentration via EDX involves knowing sample thickness, scattering power of atoms, etc. which might not be accounted for correctly in an automated evaluation routine provided by the manufacturer. A good EDX evaluation is performed with standards. If you didn't do so, you might have to put an errorbar as large as 10 percentage points (+ and -) on your data point.

(iii) Crystallite size. Depending on the crystallite size, your XRD signal can shift and broaden. For small crystallites you might have to do a refinement to extract the true value (e.g. Rietveld).

(iv) Contamination. The excess oxygen signal in EDX could originate from organic contamination, which you won't detect in XRD. But you might expect a strong carbon signal as well, if that were the case.

XRD is essentially sensitive only to crystalline structure i.e. the origin of crystal peaks in your diffraction pattern, while the amorphous part of your sample cannot be detected. EDX in contrast doesn't have such limitations. It shows information that comes from crystalline and amorphous phase simultaneously and it cannot be differed.

The discrepancy can be assigned to the local inhomogenity of your sample, as the small area of your sample that you analyze cannot be an objective representation of your whole sample and likely can contain an amorphous phase that contains more oxygen. This is one possibility (and contributes with certain amount of discrepancy to your result), and the other is more related to the EDX analysis and quantization method of your X-ray spectra.

Depending on the conditions you have been using to obtain EDX spectra and quantization the results can be differently interpreted. For the oxygen content, EDX spectra is quantized using the K_alpha line at 0.525 keV, while the Zn can be quantized using the K_alpha line at 8.63 keV, or L_alpha line at 1.012keV. It is general recommendation for obtaining a good EDX spectrum to work with SEM voltage at least 50% above the highest energy line in question based on which you want to perform the analysis of your sample. Using for example only 10kV for analysis in this case and doing quantization using K_alpha line for Zn can be accounted for the discrepancy as to little signal from Zn at that high energy was able to reach the detector. Therefore it is recommendable to do the quantization using spectral lines of similar energy; in this case K_alpha line for oxygen and L_alpha line for Zn. So, if the analysis has been done not carefully according to these remarks it is most likely the reason for discrepancy in your results.

I hope this will help you out.

(i) EDX is a local probe. The probed volume can be as low as ~1 cube µm for SEM EDX, and as low as ~1 cube nm for TEM EDX. Laboratory XRD can have a probe spot larger than ~1 square mm. You might have local concentration gradients, which you will not detect with XRD but sure are detectable with EDX.

(ii) EDX precision. If you do not know what you are doing, EDX can give you bad concentration values. Precise determination of concentration via EDX involves knowing sample thickness, scattering power of atoms, etc. which might not be accounted for correctly in an automated evaluation routine provided by the manufacturer. A good EDX evaluation is performed with standards. If you didn't do so, you might have to put an errorbar as large as 10 percentage points (+ and -) on your data point.

(iii) Crystallite size. Depending on the crystallite size, your XRD signal can shift and broaden. For small crystallites you might have to do a refinement to extract the true value (e.g. Rietveld).

(iv) Contamination. The excess oxygen signal in EDX could originate from organic contamination, which you won't detect in XRD. But you might expect a strong carbon signal as well, if that were the case.

Hi Lakshmi,

I am not sure about the technical reason behind the 1:2 ratio of atomic % of Zn to O. However, quantitative analysis of atoms below the atomic number of 10 might be very difficult with EDX (unless the instrument is specifically equipped to do so). Did you try the elemental analysis using any other technique such as XPS, XRF, ICP, etc. If you still get the same result for atomic percentages using any of these techniques, then you can look forward for a specific reason behind it. I would rather suggest to analyze your sample with one of these techniques.

Regards,

Sreekanth

The EDS results of O are normally inaccurate, because there always exists oxygen even in the high vacuum. This is from my experience.

I agree with Luo. At the surface, there are additional Zn-O-H species (as in almost any solid surface) and other contaminants as well.

Another question to Lakshmi is: Did you try to carefully etch the ZnO surface carefully enough under totally inert atmosphere before measuring EDX? That may clarify things.

Another question is: How deep did beneath the surface do you go to measure atom ratios?

These may help make solid conclusions. I wish I can here from you.

Your result is really suspicious - too neat Zn+O2 numbers. It looks like presets for your software did not really used calibrated intensity of O, but it was set for something like “stoichiometry” plus “O by difference”.

For good EDS results for O you need to use standards; never use standardless analysis. Good standard for you is, of course, certified ZnO (you can buy it from electron microscopy suppliers). Then you should use full quantitative routine without any preset values. But first check if you have any other elements in your specimens beside Zn and O. If only O and Zn are present, it is possible that best voltage for you will be about 2-4 kV (analyze C K line and Zn L line). Lower voltages give better signal/background ratio for O. I do not think you have any detectable contamination with O from vacuum. You should have thick layer of your ZnO specimen, at least 3 microns.

If the substrate is Si wafer or glass and your film is extremely thin then Oxygen from the likely SiO2 on the Si wafer or from the glass can be a source of this large O signal.

hi,

at what voltages u taken the EDX results?

I would recall the answer of Vladimir. That needs to be rechecked.

I also believe that going deeper should give better results.

There is a general reason other than the instrument chemistry, As in EDX the analyst will be targeting roughly any place in which the O % vary with place to place. For best EDX i will recommend TEM with EDX,

Ms Lakshmi

1) as Martin Süess wrote, local versus global information

2) acceleration voltage and line used for quantification: I would suggest to use the Zn L-line and the O K-line if you can, to have comparable excitation volumes (I had once the same problem with ZrO2 - you need to use the closest metal line to O unless you have a correction for sample thickness etc.). With Zn-L and O-K you can use then a low acceleration voltage (e.5 kV). If you want to use Zn-K you have to use a higher acceleration voltage and bear in mid, that you are probing for Zn-K a much larger volume than for O-K (escape depth is quite different) and that this might give you a wrong ratio, but rather to the low side while as other people already hinted surface contamination with O containing species will result in higher oxygen signals than representative for stoichiometric ZnO and therefore a 1:2 ratio Zn:O, but in this case you should also have a strong C contribution as Martin Süess hinted.

The answer lies in the physical properties. ZnO is white and is hexagonal (wurtzite) while ZnO2 is yellow and is cubic. What structure do you have?. Run TGA and DTA on your sample. You may have a mixture. ZnO2 decomposes at 212°C while ZnO is stable up to beyond 1000 °C. - RC Ropp

I think, you have a personal error in the writhing your results.

You should re-check your assigning data.

In fact you have synthesized ZnO correctly, but you dedicated the ZnO atomic ratio to oxygen atom and counter.

You can confirm your data by the using of ICP analysis.

For another comparison you can perform an EDX analysis from a ZnO purchased from Merk Co.

It should corrected as: Zn atomic % (66.63) and O (33.37)

Hi, it's very difficult to perform a quantitative analysis of light elements such as O in any compound by EDXS, even if a standard is available, because of

a) unreliable k-factors for low-energetic X-ray lines,

b) low detector efficiency of X-ray detectors for low-energetic X-ray lines, implying low count rates and hence unfavourable statistics, unless they are windowless,

c) effects from self-absorption that depend on sample geometry, tilt and thickness &

d) possible effects from other lines via fluorescence etc,

which means you will have to take all these effects into account if you want to trust your result.

Thomas

@Majid Ghashang,

Dear Majid, I couldn't understand the last line of your reply..."It should corrected as: Zn atomic % (66.63) and O (33.37)". Could you elaboratea little bit on that.

Regards,

Rajeev

Dear professor Rajeev Kumar

Thank you for your comments

I was synthesized ZnO via different methods such as homogeneous precipitation method.

I performed EDX analysis for the most of the as synthesized samples.

For example the atomic ratio of a sample was: Zn (63.11%) and O (33.62%).

El AN Series unn. C norm. C Atom. C Error

[wt.%] [wt.%] [at.%] [wt.%]

---------------------------------------------

N 7 K-series 0.83 0.97 3.27 0.3

O 8 K-series 9.77 11.42 33.62 1.5

Zn 30 K-series 74.94 87.61 63.11 1.9

---------------------------------------------

Total: 85.54 100.00 100.00

Although EDX analysis cannot performed 100% accuracy but this method can be used for determination of atomic ratio when we connivance from their errors.

I think, you have a personal error in the writhing your results.

You should re-check your assigning data.

In fact you have synthesized ZnO correctly, but you dedicated the Zn atomic ratio to oxygen atom and counter.

You can confirm your data by the using of ICP analysis.

For another comparison you can perform an EDX analysis from a ZnO purchased from Merk Co.

I think the atomic ratio in your synthesized sample is: Zn (66.63%) and O (33.37%)

I think you displace the atomic ratio of Zn with the atomic ratio of Oxygen. In fact you have unwanted error.

With regards

Majid ghashang ([email protected])

Quantitative data of lower atomic number elements through EDS is not reliable. So, instead of doing EDS, try XRF or XPS. It would give you a clear picture for sure.

I suppose you've already taken it into account but, just in case, I'll tell you that in the case of my ZnO thin films EDS results indicated excess of O signal because of the SiO2 substrate. Such oxygen excess matched the amount of Si from the substrate, so no doubt it was caused by the SiO2.

Dear Lakshmi

Analyze a commercial ZnO sample (that you know has 1:1 Zn:O relationship) in the same way you analyzed your sample.

Then, correct your sample results, using the data from the known ZnO.

Regards

You need to make an EDX measurement with a standard.

XPS is a good idea also.