Xinji Zhu Have you got any dimer or other oligomer in your system that will distort the PDI? If you're getting a z-average around 20 nm then it seems you've got a clean system. Does the pH change the configuration of your material? In theory your system should give a peak as 'sharp' as a 20 nm latex (and thus a PDI < 0.05). There's is no answer to any acceptable number for the PDI. Simply it has to be fit for purpose. If you have access to SEC/GPC you may wish to look for larger material in your system.

I assume you know the structure and thus the molecular mass of your protein? If so, at a hydrodynamic radius of 10 nm you'd expect a molecular mass of 100 - 1500 kDa dependent on structure. How does that compare to your expectation?

Dear Dr. @Alan F Rawle

by intensity or percentage graph, there is only one peak in my results. my protein is less than 100 kDa, so z-average(around 20nm) is overestimated or formation of dimer?

PDI should be less than 0.08 for monodisperse samples.

Mohaddeseh Shokrimoghaddam Hmmm, the strict term 'monodisperse' would mean a PDI of 0....

Xinji Zhu At a molecular mass of 100 kDa the expected size (hydrodynamic radius) is as attached. Have you some information as to the form/shape of your material? Is there contamination or oligomer formation suspected or expected?

Dear Dr.@Alan F Rawle,

Here are some results for Intensity and Number graphs, the results give me three size parameters (Z-average(~9nm), peak by intensity(~10nm), peak by number(~5.5nm)). In the intensity graph, there is a small second peak at above 1000nm(maybe contaminate in my sample). Based on the molecular weight of my protein (~70kDa), the size should be less than 5nm as you mentioned. I used 10mM Tris buffer at pH7.0 with 10mM NaCL. I wonder why there is a big difference between Z-agerage with peak by number?

Theoretically the PDI value greater than 0.7 indicate broad size distribution. Therefore, as I feel 0.1 is fine. I have never got the value of 0.05 PDI for any sample in last 20 years. I may be wrong but practically less than 0.5 PDI gives monodispersed distribution that I have observed several times.

Agnishwar Girigoswami If you measure a standard latex (that you should be doing for performance verification/OQ) then you'll get PDI's < 0.05. There is no PDI that is acceptable or unacceptable except for a standard latex. PDI's for 'your' product simply must reflect 'just good enough' and 'fit for purpose'. In certain circumstances (~ > 50 nm) then laser diffraction can be used as an orthogonal technique.

Xinji Zhu Thank you for posting (part of) your report. This is the first time in my years on RG that I've seen a posted report with the result quality marked 'Good'. Well done! You are to be congratulated.

First thing, do not generate or use number distributions from DLS. This practice is 'deprecated' in the international standards (ISO22412:2017 & ASTM E2490-09(15)). A volume distribution transformation may be reasonable given your data.

OK, now you're talking of a molecular mass ~ 70 kDa. Again I can use the calculator in the Zetasizer software to estimate a size. See attached. This gives an estimated hydrodynamic radius of ~ 3.7 nm for a globular protein. Again you don't tell me if this shape is reasonable or not. This would give you an estimated diameter of 7.4 nm. Other shapes give large estimates for the diameter up to about 14 nm for a linear polysaccharide. So your z-average is in the right ball-park. What else can I see? I'll only comment on the intensity plot. The (automatic) attenuator is at 10 so this tells me that your material is either small, in low concentration or poor optical contrast (relative refractive index). I'm sure you'll agree! If you could increase your concentration by 100 or 1000 times then this may be helpful in giving more signal. The intercept is pretty high at ~ 0.8 and this is another sign of good data. You're using the disposable 40 microliter cuvette. Do you really have a sample size limitation here? This is almost certainly the cause of the apparent large (and false) peak at around 3.5 microns (flare from scattering from the wall of the cell). This distorts the z-average upwards. For such small material as yours I'd be considering a quartz or glass cuvette for better optical quality. However, transforming your data to volume is reasonable in this circumstance and this will get rid of this artifact peak.

The PDI is high for a single pure protein. However, we'd need to know whether this is caused by a small amount of debris in the system. It's tough to clean a 40 microliter cell before measurement. Or caused by possible dimer formation (which will only broaden a DLS peak). There may be clues in your correlogram if you post this.

All-in-all I think your results are believable, More information as to the nature of your protein would allow more options to be explored.

Dear Dr. Alan F Rawle

Thank you for all the information. That's really helpful. My protein is a global protein and the concentration I use is 2.5mg/ml. I don't have a lot of protein and therefore use 40ul for my test.

To add to the excellent answers above, while interpreting DLS data for polydispersity index (PDI) it important to keep in mind the inherent errors of DLS measurements. Direct conversion of the DLS intensity distribution to number or volume distribution has strong implications. For instance, a long chain polymer may have high scattering and based on this scattering intensity if some one approximates a sphere then the number density can be quite off from the reality.

It is known that the hydrodynamic radius that DLS measures depends strongly on the nanoparticle morphology. This is indeed one of the reasons why DLS results are a bit off from the real size.

But even if this could be corrected, DLS intensity distribution is still inherently off towards larger size due to polydispersity. It can be shown quite clearly that as the nanoparticle (any nanoparticle that scatters) polydispersity increases, the DLS intensity distribution shift to larger values and also the distribution broadens. In this article we did a reverse theoretical study:

Article Effect of Size Distribution, Skewness and Roughness on the O...

We generated normal (also log-normal skewed distribution) nanopartice size distributions with a constant mean and varying standard deviation (i.e. polydispersity) for a fixed number of total atoms (or mass). Then we calculated how the DLS intensity distribution would vary for these different size distributions. The trend is quite clear. For further verification, we have shown how the DLS distribution is independent of laser wavelength (different manufacturers have different laser wavelengths). Also, it is shown that the choice of optical constant has no effect on the intensity distribution. This means that these results are applicable to any nanoparticle (lipids, polymers, etc). Importantly, for highly irregular nanoparticles, the effect of polydispersity on the optical intensity distribution can be even stronger.

In the Figure added:. (a) Effect of polydispersity i.e. standard deviation, σ on the theoretical scattering intensity distribution (theoretical raw DLS data) at 660 nm incident laser for a fixed mean of 36 nm. (b) Intensity distribution peak vs. σ fitted to Eq. (11). (c) The dependence of the theoretical scattering intensity distribution from optical data source. (d) The dependence of the theoretical scattering intensity distribution from incident laser wavelength.

Even with these issues, the power of DLS as a fast technique can never be undermined!