In SEM the image magnification directly proportional to the scan length and this is usually calibrated by the engineers during the service. Usually however, at least on the less expensive and sophisticated instruments, the scan length and image magnification is only accurate at 1 working distance (usually the instrument default) e.g. 10 mm on a JEOL-7000F (setting e.g. WD = 4 mm changes the image magnification quite a bit). In my experience, under optimal conditions on a well calibrated instrument using the image magnification for measurement purposes is fine and accurate down to the resolution limit of the microscope. This you can of course check with your engineer or by getting a calibration grid.

TEM and HRTEM have of course a higher resolution than a SEM. The image magnification does however work in a completely different way. The microscope reads the image magnification for each microscope magnification from a calibration table, which has to be set up by the instrument responsible/engineer. The image magnification also varies with the objective lens current, which often should be set to a specific reccomended value. In my experience, the image magnification on most TEMs is off by ~2-5% on most TEMs in case somebody did not pay special attention to this. Also note that since the magnification comes from a pre-defined table, this means that the image magnification may be perfect on say 400.000 kx magnification, but completely off at 200.000 kx. Best way to be sure is to calibrate the image yourself. If you work with crystalline materials you can use the lattice fringe spacing (e.g. by FFT) to calibrate high resolution images and then correlate those to images of lower magnification, where lattice fringes may not be visible. If you work with non-crystalline materials, you can add a standard to your sample e.g. graphite and measure the latice fringe spacing of the standard. The big problem is that many users trust the calibration of a TEM blindly, when it is often completely incorrect (can be 15% off).

In my opinion, both SEM and TEM are both suited well for measuring particle sizes. A SEM will of course not work well for small particles (10 nm) but works well for + 100 nm+ particles. Also, naturally in TEM you measure the 2D projection of a 3D object, whereas in SEM you will in most cases measure on an image with a certain depth of field (depends on the position of the detector). Try to measure each sample using a standardized protocol and check the calibration of your images and you should be fine.

In SEM the image magnification directly proportional to the scan length and this is usually calibrated by the engineers during the service. Usually however, at least on the less expensive and sophisticated instruments, the scan length and image magnification is only accurate at 1 working distance (usually the instrument default) e.g. 10 mm on a JEOL-7000F (setting e.g. WD = 4 mm changes the image magnification quite a bit). In my experience, under optimal conditions on a well calibrated instrument using the image magnification for measurement purposes is fine and accurate down to the resolution limit of the microscope. This you can of course check with your engineer or by getting a calibration grid.

TEM and HRTEM have of course a higher resolution than a SEM. The image magnification does however work in a completely different way. The microscope reads the image magnification for each microscope magnification from a calibration table, which has to be set up by the instrument responsible/engineer. The image magnification also varies with the objective lens current, which often should be set to a specific reccomended value. In my experience, the image magnification on most TEMs is off by ~2-5% on most TEMs in case somebody did not pay special attention to this. Also note that since the magnification comes from a pre-defined table, this means that the image magnification may be perfect on say 400.000 kx magnification, but completely off at 200.000 kx. Best way to be sure is to calibrate the image yourself. If you work with crystalline materials you can use the lattice fringe spacing (e.g. by FFT) to calibrate high resolution images and then correlate those to images of lower magnification, where lattice fringes may not be visible. If you work with non-crystalline materials, you can add a standard to your sample e.g. graphite and measure the latice fringe spacing of the standard. The big problem is that many users trust the calibration of a TEM blindly, when it is often completely incorrect (can be 15% off).

In my opinion, both SEM and TEM are both suited well for measuring particle sizes. A SEM will of course not work well for small particles (10 nm) but works well for + 100 nm+ particles. Also, naturally in TEM you measure the 2D projection of a 3D object, whereas in SEM you will in most cases measure on an image with a certain depth of field (depends on the position of the detector). Try to measure each sample using a standardized protocol and check the calibration of your images and you should be fine.

We just estime the size of nanoparticles using SEM/TEM. As i see it, SEM/TEM can not be used to replace particle size analyser in order to obtain accuracy sizes of nanoparticles.

Volker is right. Though it depends on your sample, since generally more polydisperse particles require that you measure more particles before you can get an idea about the size distribution. In the case of monodisperse particles (say 5% standard deviation) the mean diameter usually converges fast (50 measurements) but I generally try and measure at least 300 particles. You should of course measure even more if you are interested in the shape of the distribution or if your sample is more polydisperse. With automated image analysis techniques, e.g. particle analysis in Image J, you can easily measure thousands of particles, but these methods put a high demands on the image quality and often requires non-overlapping particles with a uniform orientation.

Thanks to Erik Wetterskog,Ye Kuang and Volker Klemm..

Mr.Erik answer is very informative.,But basic problem with SEM/FESEM techniques is that particle dispersion is a key issue. The image are most of the times are overlapped and egdes if any are blurred. Can any one suggest that how we can quote the size of nanoparticle based on SEM/FESEM images? For example, can it would be correct to say "under 500nm magnification the size of particles are 20-50 nm? In TEM/HRTEM and particle size analyser(DLS) there is option to study particle size distribution but in SEM/FESEM complete sample need to be analysed. And surely Image J software needs high quality image to differentiate between particles ( without any fluroscence) and back ground.

I have seen a few cases of particles analyzed with the FESEM, but they were always relatively large ones. The huge advantage of the SEM is that you can easily survey different parts of the specimen to check for uniformity.

I still believe the TEM is a great tool to get information on the features of single particles, but if you need a good statistical basis you have to rely also on other techniques.

I tend not to use DLS, as you need to disperse your particles.

For nano monocrystalline particles (the majority of the cases) the suggestion I give, whenever possible, is to complement TEM with modern XRD line profile analysis techniques (forget Scherrer formula): you can get very accurate information on the distribution on a large number of domains (that would correspond to the particles in this case), you are phase-sensitive (good luck in separating e.g. spherical hematite and ferrite particles using the TEM) and you can also have some information on the shape and on the defects (hard to quantify with the TEM)!

Moreover calibration is usually not an issue with XRD if you know your machine and the man time is invested only in data analysis (from minutes to hours). I know a FESEM or a TEM micrograph are more likely to hit the cover of some journal, but well, it's a huge investment of time!

I would like to include that it is also dependent on the material used in the nanoparticles. For example if the particle are organic, the electron beam might degrade the particle and the size will be dependent on how long time you have had the beam focused on the particles. Another technique like AFM might be useful for this type of particles (for example cellulose nano crystals).

However, AFM also have drawbacks and need skill full person to operate and interpret the imaging. For example tip broadening, careful calibration and optimal feedback control is crucial when looking for really small sizes. The advantage is that one can do analyse also in liquid.

i agree with who that interested measure the size of particle using TEM with also compare with XRD. .Also if the the particle like single crystal will shows high accurate result between XRD and TEM.

regards for all

Unfortunately, in FESEM/SEM measurements the results may strongly depend on sample preparation. Routine procedures, like coating, can noticeably change size/morphology. For instance, see the ref. below for coating effects.

Drying of solution also changes the sample a lot since agglomeration readily occurs.

Does DLS measurements also depend on sample preparation?

---

"A SEM study of nanosized metal films and metal nanoparticles obtained by magnetron sputtering"

http://dx.doi.org/10.1007/s11172-011-0399-x

Thanks to all for their valuable discussion and suggestions. Recently I have taken organic polymer and metal sulphide nanoparticles for FESEM analysis. For the organic samples gold coating was perfomed and samples were analyzed at 500nm magnification at 15Kv. Sample was prepared by drop cast method and water was used as solvent. As per technician suggestion " higher voltage will destroy the sample and lower voltage will give lesser resolution." As it is evident in both SEM and FESEM voltage play a key role and there is chance of destruction of samples for better images.

It would be helpful if it is known that how long coating and how much voltage can be applied over the samples to get better quality results. Where as in case of metal sulphide nanoparticles a higher voltage may give reflection , images will be distorted. Is it good practice to write about coating in final results? Because it is certainly changing the size and diameter of the samples.

If you coat the samples, you should certainly report it. Your coating, if its sputtered gold consists of small gold (nano)particles. The longer the coating, the larger the particles (but I have little experience, since I never coat my samples). If the coating particles are in the same size range as your metal sulphide particles, its rather fruitless. You will be imaging the coating.

Imaging pristine metal sulphide particles, can be done without a coating as they are semiconducting. However, if the particles are embedded in a polymer matrix it will be difficult. A high acc. voltage will burn trough the polymer (which in fact can help you see the particles, however perhaps not reccomended). A low accelarating voltage will allow you to image the surface of your composite. Many modern SEMs can attain a high resolution even at low accelarating voltages. The biggest difference, when imaging at different accelarating voltages, besides the beam damage, is the interaction volume. High voltages causes the electrons to penetrate further into the sample, whereas low acc. voltage will allow you to image primarily the surface of your specimen. Really low acc. voltages (< 1 keV) allows you to image surfaces of non-conducting samples (e.g. silica, cellulose) without destryoing them. There are often methods in the instruments (e.g. "gentle beam modes") where a bias is applied over the sample, to reduce charging.

Perhaps if you described your material and your expectations in greater detail, it would be easier to come with better input. It could be that TEM is the way to go. What is the size of your primary particles? Do you wish to image the agglomeration state or are you interested in the mesoscopic structure? You mentioned a polymer, is it a composite or is it a ligand coating?

The quality of the images, will as people have mentioned, strongly depend on the sample preparation and the material you image. Contamination of organic matter, is generally problematic and will drastically lower the resolution due to charging as it is carbonized by the e-beam.

@Erik Wetterskog,

My samples are Cadmium sulphide nanoparticles of range 80-100nm as it was shown in SEM. I have found different morphology of particles which were not so clear in SEM. Usually we take samples over glass slides or aluminium foil through drop cast method ( sample was sonicated in water or enthanol solvent then a single drop is taken over a glass slide) and dry it by evaopration . The gold coating is done by sputtering for 2-3 min. I have various range of particles ( size range 20-100nm) . I need to know their size -distribution and morphology too. About organic nanoparticles, they are PLGA particles of size range 150-220nm. Same process of sample preparation is done ( drop cast method) for FESEM. Coating is done in same way ( gold coating). Organic samples were unable to withstand acc. voltage of 20kV. I need to know the size and morphology. Single particle is always preferred. So i do sonication to avoid agglomeration.

Ritu Raj,

I already saw very good answers on your question. I want to add that the statistics of size distribution is very importation. You can focus only at tiny region of your sample using electron microscopy ( any SEM, TEM, STEM). In my opinion, its better to get several low mag images in different areas and measure the sizes and get the size distribution statistics.

In order to see the morphology, you should magnify each particle and get the images. Again, try to get more images in different areas and make a histogram of possible shapes.

Other important aspect is that you should try other techniques also for size determination such as powder XRD, DLS. The result from electron microscopy should match with the results from XRD and DLS.

As per my knowledge u can use Image j software for calculating Nano particle size. this software give particle size of each nano particle in the sem image then u can use Origin or other software for calculate the accurate size of the particle

In my opinion almost no method provides a complete answer for accurate nano particle 3D size determination. In many instances nano particles are not spherical making the process more complicated. It will be a long reply to include advantages and disadvantages of each method, which some papers have already discussed (do not have the reference now, unfortunately). For example SEM provides top-down image but height information is hard to get. On the other hand AFM provides accurate height information but lateral dimension is usually not so accurate. DLS has its won issues. Even TEM type of instruments usually provide one cross-sections of the nanoparticles.  As Volker Klemm replied before, statistics is very important. Our group has developed another method for size determination using ubiquitous optical microscopes. Combination of  several suitable methods  and use of statistics  would be a better combination for accurate results, in my opinion. 

“Nanoparticle size determination using optical microscopes,” Appl. Phys. Lett. 105, 163105 (2014); http://dx.doi.org/10.1063/1.4900484 

dx.doi.org/10.1063/1.4900484

A recent paper shows an inherent bias in measurement using SEM/AFM/TEM

Combining SEM with AFM provides a more accurate size determination

http://link.springer.com/article/10.1007/s00216-016-9909-x

You can use ImageJ software to determine the average size of nanoparticles in SEM, TEM image. See it here: http://bit.ly/2sbb5B5

As said, by image processing software(s) you can determine but you must do this method for many figures from different zones and report your result statistically like this: mean±standard deviation.