I am working on developing a model that is based on the Ostwald ripening of carbides in an Austenitic Stainless steel. I have a ton of images from the SEM. I have to measure the "carbide size distribution" to figure out the rate of growth.
I have Imagepro (which I am not very fond of) and also ImageJ (which is apparently good).
Can any researcher let me know the best way to measure the carbide size distrbution from SEM pictures?
I used ImageJ to measure size distribution and density of carbides this summer and this method gave me good results. Although you can try to use matlab and write a code that does it automatically. But the manual method allows you to be more accurate and to select the particles you want to measure after an image treatment.
As you mentioned about ImageJ. In this connection I want to share my experience. I took separate measurement of w & h from 100 particles. And after filtering 200 data set to w & h I took its average. Then defining binning range I calculated the distribution histogram. (Please take care about the aspect ratio).
In Image J when you the scale according to microstructure scale and measure one particle in your microstructure there is one option to measure w & h. This measurement also give you idea about circularity of the particle as well as aspect ratio. And then I took its average to measure 1 particle. In this way I took 200 data to measure 100 grains.
I used ImageJ to measure size distribution and density of carbides this summer and this method gave me good results. Although you can try to use matlab and write a code that does it automatically. But the manual method allows you to be more accurate and to select the particles you want to measure after an image treatment.
Leica QWIN... it is not for free, but I made good experiences with this software.
@Tatina you are r8, but when calculating grain and/or particle w&h usually refer length and width of the particular. Tht's why I'v mention the issue of aspect ratio :-)
Your SEM results may suffer from uneven distribution or an unrepresentative sample. The fact that you've needed to take a ton of images suggests that you're seeing variability, and we might assume the reason for the assessment is either quality assurance or process development, neither of which should involve guesstimates.
Quantitative measurements (how many) of qualitative results (type) benefit from fractionating several separate samples, perhaps by chromatography, density gradient or filtration. SEM can then be done to confirm the WxH of the particles in the fractions; all images within each sample should all be alike, proving the success of the fractionation. Then your assessment should fall into place nicely: the same mass of a particle with a different ratio (e.g. 1x4 vs. 2x2) will fractionate differently, and you need only measure the typical particle found in each fraction.
there is a software named WINROOF. One of my collegues use this software. It seems quite good. however, it is not free!
imageJ is good option, but is important in all the studies of distribution and size of particles or phases perform measurments in a good number of images, all of them in the same conditions, (magnfication, etcetc.)
I guess the solution or probable solution of your problem lies in some aspects of the precipitates under consideration.
1. Shape of the carbides.
2. Minimum diameter or length of the carbides.
3. Growth mechanism influencing parameters.
4. Quality of SEM images.
Area fraction measurement of the carbides can give a reasonable approximation but can be misleading in terms of growth kinetics. This kind of measurement is only acceptable when carbides have faceted interface or have the spherical one.
On the other hand, a needle shaped carbides need the aspect ration measurement where the area fraction/diameter can be misleading.
Finally the conclusion arrives that you need to identify the growth mechanism and type of data required before defining the measurement technique or data type selection.
Image J is the appropriate software for the analysis. But with SEM images the accuracy of the analysis will decrease. It will be better if you use optical images for the analysis.
The accuracy of image J is based on the number of pixels it can measures on your image. So you must have good quality image (resolution) to have a good accuracy for your resuts.
Higher is the magnification, smaller would be the number of carbide particles under consideration. This may give error in the size distribution. In case of optical micrographs due to low magnification large number of carbide particles are considered and this may lead to a good accuracy.
But, if size of the carbide particles is very small and is out of the range of optical microscope then SEM is naturally a better option.
Magnification is not a problem for SEM, you can choose any suitable magnification. Etched metallurgical cross sections are often better visible under optical microscope because etching creates less contrast for SEM, so SEM needs overetching – not good for quantification of small features. SEM is superior to optical scope if you can use backscattered electrons (BSE). If BSE images have good contrast, you can use your sections without any etching at all and these images are very good for quantification. You may wish to try BSE for your carbides. Good luck.
Most of the soft-techniques of size measurements suffers from the fact that the areas of every precipitates in the image field is not highlighed to account for the exact area of the precipitate. This is because the contrast level in each precipitate is varying and therefore the area of a precipitate is either under estimated or overestimated. The reasonable measurements requires adjustment/manipulation of number of parameters. If you are not cofident with the soft-techinues, try the manual method which is time consuming but measures what you see in the image.
You should read more references about image handling/crystal size distribution and the ImageJ instructions. I advise you to get good images with favorable magnification at first. Then, you can handle with your images by using ImageJ and other softwares. I used ImageJ software to determine the size distribution of precipitates/crystals in the past years. Finally, I wish you happy.
Using un-etching sample SEM backscattered electrons (BSE which have good contrast, for your carbides. To eliminate the fact that the area of a precipitate is either under estimated or overestimated. You need to calibrate your BSE measurement parameters with the reference of the measurement from optical microscopy
- Badges
- Science method