The problem with CNTs and DLS is that they have a tendency to "clump" together and dispersing them into single particles is very difficult. However, we have been able to measure the length, diameter and chirality of CNTs in a single experment using the new Zetasizer Helix instrument (which combines DLS with Raman spectroscopy). The white paper below describes this in more detail, and includes information about the sample preparation steps required to produce a stable dispersion of individual nanotubes.

Hope this helps.

http://www.malvern.com/en/support/resource-center/Whitepapers/WP140627SingleWallCNTHelixDlsRaman.aspx

Easiest way is using carboxylated CNTs which are dispersible in water and are commercially available. Other way is by controlled oxidation of CNT with H2SO4/KMnO4 which yield water dispersible CNT,  at low concentrations.. Secondary modifications like poly-electrolyte modified CNT or PEGylated CNT or DNA modified CNT can improve dispersibility. 

The problem with CNTs and DLS is that they have a tendency to "clump" together and dispersing them into single particles is very difficult. However, we have been able to measure the length, diameter and chirality of CNTs in a single experment using the new Zetasizer Helix instrument (which combines DLS with Raman spectroscopy). The white paper below describes this in more detail, and includes information about the sample preparation steps required to produce a stable dispersion of individual nanotubes.

Hope this helps.

http://www.malvern.com/en/support/resource-center/Whitepapers/WP140627SingleWallCNTHelixDlsRaman.aspx

You should keep in mind, what is the purpose of your measurement. Likely, to characterize the particles or their dispersion, not to have nice DLS results. When you have a powder, which you want to use as a powder, you should characterize the powder in dry state. If you want to apply the CNT in dispersion, you have to characterize this dispersion. Then it would be very helpful to know particle size distribution including the coarser agglomerates.

As suggested by Dr. Sobisch, I would also recommend measuring sizes rather via techniques that are suitable for powders, e.g. TEM. However, if you are determined to use a solution and analyze by the Zetasizer, you can try dispersing the CNTs in the DMF - for most types of the tubes it gives much better dispersion than water. Still, the tubes are not likely to disperse spontaneously, you probably will need to apply sonication, preferably by tip sonicator, using low frequencies (too high frequencies may damage the tubes). Also, keep in mind that not all types of cells are suitable for measuring organics. For more information, you can try looking into this technical note on the cells selection.

http://www.atascientific.com.au/pdf/Cells-for-Zetasizer-Nano(MRK1751-01).pdf

I can suggest you this good article about determination of CNT dispersion using zetasizer. Dispersion parameters for 4 different CNT types in water are described.

http://www.sciencedirect.com/science/article/pii/S0008622310002538

We have developed a characterization protocol including TGA, UV-vis, Raman, and FT-IR spectroscopies and dynamic light scattering for the characterization of CNTs and graphite derivatives in solution, and hence we used routinely the Zetasizer nano to obtain a qualitative evaluation of the aggregation of carbon nanostructures in solutions.

Briefly,

1) sample of CNTs were extracted with 5.0 mL on an organic solvent (e.g., DMF, chlorobenzene, chloroform, N-methylpyrrolidone, methanol) by means of a sonication (Misonix 3000; power: 1 W/mL, pulse on: 3 s, pulse off: 3 s, time: 1 min) and a centrifugation (MR23i Jouan ultracentrifuge equipped with a SWM 180.5 swinging bucket rotor - Thermo electron corporation, 5 min @ 3000 rpm) procedure.

2) 10 to 50 microliters of the supernatant (free of larger aggregates) are diluted to 2 mL in a quartz cuvette and placed in the Zetasizer nano cuvette holder immediately after centrifugation. Each DLS measurement was averaged over 20 runs (10 s per run). It is worth noting that the less soluble CNTs gave DLS data poorly reproducible for the presence of large aggregates. The distribution of particle sizes was obtained by plotting the number size distribution provided by correlograms taken with the software program by Malvern (full text @ https://www.researchgate.net/publication/51466112_The_continuous-flow_cycloaddition_of_azomethine_ylides_to_carbon_nanotubes).

Although the spherical model assumed by DLS is not fully pertinent to the real geometry, nevertheless (i) it can give a reasonable estimate of the order of magnitude, and (ii) it allows comparison of samples with similar geometry, as in our case. For CNT derivatives we often found aggregates with sizes between 70 nm and 400 nm (see: https://www.researchgate.net/publication/254261135_Synthesis_and_characterisation_of_a_trithiocarbonate_for_the_decoration_of_carbon_nanostructures and https://www.researchgate.net/publication/261923267_Chemistry_of_Carbon_Nanotubes_in_Flow).

Article The continuous-flow cycloaddition of azomethine ylides to ca...

Article Synthesis and characterisation of a trithiocarbonate for the...

Article Chemistry of Carbon Nanotubes in Flow

Thanks for the valuable answers.