Get a lab calibrated sample set in your particle size range and run it to see which machine is out of calibration. ■

Because there is a sahpe effect on size measurements. It will be difficult to generate the same shape from batch to batch. You will get good correlation if you consider the shape information aswell.

get back if you needmore information. ■

The insitec and coulter counter measure different propeties of the particles. The coulter counter measures the volume of the particle and is independent of the particle shape, colour , refractive index etc, the insitec measures the light scatterd(diffracted) by the particle and is affected by colour, shape and refractive index, it's also affected by multiple scatter, where the light scattered by one particle hits another and is scattered again. If your correlation is changing it's probably down to a change in colour, shape or ri in the particles - alternatively since the insitec measures at much higher concentration their may be some surface charge effect which changes when the sample is (a) put into a conducting liquid, and/or (b) when it's diluted down for analysis on the CC. ■

Overall, the response Brian Miller is the most correct, in my opinion. Chances are that your Insitec (Laser Diff) is sizing higher than that of the Coulter Counter. The reasons Brian Miller stated are all 100% correct. In addition the fact that the Laser Diff system you are using has much less detectors will also give you less resolution.

Now, please realize that the Insitec is meant to be an on-line system where the goal is to give you notification of process changes. That is just the nature of the system. You will get lower resolution results but it is in-process.

On the other hand, the Coulter Counter is VERY HIGH resolution but is a lab benchtop system.

Are you getting larger sizes from the Insitec? ■

I would address you by name, but I dont know what it is.

Never the less, well, the fact that you have a particle that has a circularity close to 1 is an ideal sample for making a correlation study. In fact, we are in the midst of doing a correlation study of perfect spheres (circularity close to 1) and irregular shape particles with different technologies.

Yes, with a perfect sphere, you should get the same results from an ESZ (Coulter Counter) and a Laser Diffraction system. The ESZ makes the equivalent spherical diameter by volume displacement and the LD systems report equivalent shperical diameter by light scatter. in theory both should be similar in results (i.e. technique to technique accuracy) if you have the correct refractive index for the modeling in the LD system AND if the LD system has enough detection points to ensure you can see the process shift.

Note that your particle is black and that has a high degree of absorbance of light. Please be sure you have the proper refractive index and you are not over saturating the system causing multiple scattering. However, I guessed that the size seems larger on the Insitec due to the lack of detectors it has. Note, remember that the nature of the instrument is to give you a rough indication of process change not necessarily to give you accurate results.

Overall, I think you are doing a good thing in getting a good idea of the correlation between the in-process systems you have and the bench top systems you are accostomed to using. This will be a good point of reference.

In terms of process monitoring, the benchtop system will give you more accuracy and will be able to detect slighter shifts in your process but you give up the fact that you do not have it in-process. That is the price you pay.

I hope this helps. ■

It appears that Mr. Miller and Mr. Buza are both on the right track with there comments. As a real simple possbility to verify.... you are comparing Volume to Volume results, and not Number form the CC to Volume on the laser diffraction correct?

This might be too obvious of a question, put I like to start with the simplest possibilities first. ■

Some good comments and questions. However, comparing online data with offline data is not a trivial task at all and I am surprised that the most obvious questions have not been asked so far just to avoid you are not comparing apples with pears. They have nothing to do with the method you use but with the setup of the experiment:

1. Are you 100% sure about the representativity of your samples both online and offline?

2. Have you done the Insitec installation on your own or did a Malvern Engineer the installation and commissioning?

3. Where and how do you take your sample for the offline analysis? Do you use the outlet of the Insitec and collect the sample? If yes, by what means do you collect the sample and are you sure you are not loosing a portion of the material? In case you take a downstream sample after the Insitec when the material has been separated from the gas: How can you be sure that you measure exactly the same material offline that has been measured offline? Think of backmixing, material built-up in cyclones, hoppers or filters etc.

....

There are more questions to add. But I guess you get the idea.

To evercome the uncertainties above I would propose that you come up with a series of offline samples that cover your full product range, you devide each sample into two equal samples (that alone can be difficult enough and a subject for a 2 days workshop) and then measure each sample offline on both instruments using your standard SOPs to compare the results. I am pretty sure you will get a good correlation. Usually the R2 is 0.95 or better if you stay within reasonable areas of the measurement range. Once you have worked out the offline correlation you can then think of where the differences occur comparing the online result with the offline result. The answer to that is very likely hidden in the above questions.

We are quite often involved in the same type of scenario so I would heavily recommend to contact your Malvern representative in the US. They will certainly be able to help you with some further hints.

Oh, and forget about the statement regarding the sensitivity. It is an arbitrary assumption that online measurement equipment has to have lower sensitivity compared to benchtop devices...

Best Regards,

Oliver Schmitt ■