Does Size Matter?

[Steve]

Is there any difference between a 4.5um column and a 5.0um column? Would it be helpful if columns manufactures gave the range of particles sizes for their column so a 5.0um column may become a 3.5 - 6.5um column or 4.5um - 5.5um column? Or should we leave things well enough alone?

[Consumer Products Guy]

4.5 and 5.0 seem indistinguishable to me; which category would a column of 4.7 fall in, like would be within spec for each....

[Chris Pohl]

Steve,

Of course size matters, ask any woman.

Seriously, the particle size described by most manufacturers is the average particle size. So, on average, the 4.5 µ resin will exhibit 11% higher efficiency and 23% higher pressure. Of course, since these values are averages, it's entirely possible that you might get a column from each supplier with identical pressure and efficiency. But on average, the 4.5 µ resin will give better efficiency and higher pressure all other things being equal.

[HW Mueller]

Steve, how about focusing on the primary things, like a cleaner silica?

[Steve]

Chris:

I understand the theory, but with using average size for columns I can take 3um silica and an equal number of 7um silica and make a 5um column without ever having 5um particles? Is there an easier way to let people know that the column they are using has a 1 or 2 sigma deviation?

HW:

Our new silica "Nucleodur" has some of the lowest metal contamination in the industry. " Try it you'll like it"

[Uwe Neue]

The standard way of dealing with a description of the particle size distribution is to give the "90/10 ratio" i.e. the ratio of the size of the particle at the 90% point of the distribution to the size at the 10% point of the distribution. Commonly, the volume average is used for creating both the value of the average and the 90/10 ratio.

The separation performance ( C term) of a column packed by mixing equal volumes of 7 and 3 micron particles is indistinguishable from the performance of a column packed with 5 micron particles with a narrow distribution. However, the backpressure of a column with this mixture will increase, since it depends on the population average of the particles. Since it takes much more 3 micron particles than 5 micron particles to make an equal volume, the pressure increases quite a bit.

This is why manufacturers strife for narrow particle size distributions.

Steve, you will find the value of the 90/10 ratio in the Certificates of Analysis that come with the Waters columns. If you want to know more about particle size distributions, you can get hold of my book. The material just mentioned is covered in more detail there.

[Steve]

Uwe:

Are you basing your 90/10 ration on number of particles or volume of particles?

Assuming the 90/10 ratio is done on lots of silica and not individual columns and that only a small portion of that lot is analyzed. Where is the confidence level listed on the CofA?

If your testing 90/10 on total number of particles then you must throw away a lot of batches.

If your testing 90/10 on volume then throw away your CofA as useless.

[JI2002]

If a column is packed with equal number of 3u and 7u silica particles, its performance is close to a normal 7u column in terms of back pressure and plate number, which means its efficiency and pressure are much lower than a normal 5u column.

If a column is packed with equal volume of 3u and 7u silica particles (average particle size is about 3.29u), the plate number and back pressure of the colume should be higher than a normal 5u column.

[Uwe Neue]

JI2002:

Your analysis does not agree with the measured data. When you determine the "average" particle size, you need to specify, how you do the averaging. You can do the averaging by volume, cross-section, linear diameter or via the population. The common measurement techniques give you a distribution of the volume or a distribution of the population.

How each of these parameters affects the chromatography, can be predicted to some degree in a straightforward way. The mass-transfer depends on the particle cross-section. Therefore one expects a close coincidence of this term with the volume-averaged particle size. The measurement confirmed this to be true.

When we did our study, I was not sure what exactly the outcome would be for the backpressure. We knew beforehand, that a wider particle size distribution gives higher backpressures. It turned out very clearly and without any doubt that the backpressure correlated with the population average, and not with anything else.

The initial study consisted out of 7 data points. Recently, the general findings of this study have been reconfirmed.

[JI2002]

In my comment above, I meant to say that a column packed with equal numbers of 3 u and 7u silica particles is similar to a normal 6 u column in terms of plate number and backpressure. To me, average particle size of equal numbers of 3 u and 7 u particles is 5 u.

When a mixture of 3u and 7 u particles are packed into a column, it's very complicated to calculate the pressure drop and plate number. So I use a simplified model which assumes that part of the column is packed with 7 u particles and part of the column is packed with 3 u particles. Because all the calculations are based on empirical formulas and the simplified model, they could be off a bit from the ture value.