This depends on the type of characterization method you have in mind. Do you plan to perform mechanical or electrical characterization, or anything else?

What kind of CNTs do you have, multiwalled or single wall, diluted in water?

I made experiments of MWCNTs with standard tapping mode cantilevers with resonance frequencies in the range of 100-300kHz. I tried different substrates, mica and HOPG substrates (see attached fig) gave good results. But depending on the ideas you have in mind, this could be different in your case. Maybe you should give more details here.

I plan to perform a mechanical characterization (semi-contact model). My carbon nanotubes are single-wall CNTs. I disperse them in SDS solution.

It depends on the nature of the bonding between nanotubes and surfaces, you can see in my paper a dull description how gently measuring height

Article Nanotubes Oxidation Temperature Controls the Height of Singl...

i think SEM is the better way to characterized CNT ,the AFM in the two modes ;tapping or contact are great in mapping well alinged structures like TNT or AAO or any powder , see a live test of ATO at Ti substrate using contact mode AFM.

http://www.youtube.com/watch?v=ZA1jJi4FeaQ

As many people have said, i believe that you can use HOPG as substrate and AFM in both tapping and as well non-contact modes.

Or you could use single crystals as substrates and perform STM measures.

You should have fine substrate with less tilt, i suggest u that first try to make a film with them (u can dispersed them into solvent and after making film evaporate the solvent)

I have using this method for PPy/CNT nanoconposites:

https://www.researchgate.net/publication/258358234_Facile_synthesis_and_characterization_of_polypyrrole-multiwalled_carbon_nanotubes_by_in_situ_oxidative_polymerization

Article Facile synthesis and characterization of polypyrrole-multiwa...

Since you disperse SWCNTs with SDS, gold maybe better for you. you should wash it with water and further immerse it into HNO3 for some time to remove SDS, you may have clean SWCNT for AFM.

https://www.researchgate.net/publication/229004802_Absorption_spectroscopy_of_surfactant-dispersed_carbon_nanotube_film_Modulation_of_electronic_structures?ev=prf_pub

Article Absorption spectroscopy of surfactant-dispersed carbon nanot...

For AFM, you can use any substrate. As i know, tapping mode is good for characterization and its force constant in the range of 2.5-3 N/m for etched silicon tapping mode tips and depends on tip material.

I suggest starting with mica as the substrate for the least artifacts.

For the study of CNTs should be fine cantilevers with a high spring constant.

SDS dispersed CNT can show false size of CNT. first of all check dispersion by UV or SEM. if CNT's are dispersed well then go for contact mode on mica substrate. SDS dispersed CNT will have good interaction with mica hence a better image can be taken in contact mode. good luck

I think mica substrate and tapping mode will go well with you... you may observe the self-organization of the SDS around the carbon nanotubes so you may work on concentration below the cmc of the SDS... good luck

The strongest adhesion of a CNT on a surface will be provided by graphite. We have measured adhesion of CVD grown single wall CNTs on various substrate, and we get the following numbers :

Graphite 6.6 nJ/m

Gold 5.1 nJ/m

Mica 4.2 nJ/m

Al 3.8 nJ/m

Pt 3.2 nJ/m

Si 3.1 nJ/m

(T. Li, L. Bellon, article in preparation and T. Li, PhD thesis, 2013). Obviously the adhesion energy will depend on the nanotube diameter, but the relative order should be conserved. You can read about the measurement methods and the results for Graphite and Mica in the following article :

https://www.researchgate.net/publication/48169928_Carbon_nanotubes_adhesion_and_nanomechanical_behavior_from_peeling_forcespectroscopy?ev=prf_pub

So using Graphite instead of Si will give you twice larger adhesion of the nanotube on the substrate. If Graphite is no option, gold is your best choice.

As for the cantilever stiffness, it depends on the imaging mode you intend to use. I imagine tapping mode is best suited if you want to avoid dragging the nanotube on the surface, hence a large stiffness...

Article Carbon nanotubes adhesion and nanomechanical behavior from p...

This is the AFM image of my SWCNTs on mica. I use tapping model with a cantilever of 125 kHz. Are those white dots SDS salts? They always appear in my AFM image.

What is exact question? What do you want from AFM characterization of CNTs? Length/diameter distribution? Degree of bundlization? Density of defects? Depending on problem you will have different answers. Bare Si, mica or HOPG may be good choice if you use thick dispersion and/or need specific (force/electric/magnetic what ever) measurements. If you like to deposit from very diluted media (for example if you prefer individual CNTs rather bundles) you might need functionalized surface - both mica and Si may be functionalized with amino-groups such as APTES which increase roughness of substrate but create reasonable surface density of deposited CNTs.

If you aim to study topography related aspects of CNTs, provided suggestions are fine (especially those of Ludovic). However, if you are trying to study electrical properties such as traped charges in CNT, defect states, conductivity of SWCNTs, etc you must use conductive probes in both contact and non contact modes (mainly non contact). for this, people mainly use Pt or TiN coated cantilevers. Regarding the substrate; for MWCNT which are metallic you should use insulator substrates like SiO2 or Al2O3. This also the case for the SWCNTs especially those of which that are metallic.

Dear Yin.

I would suggest to use gold. We used our own type of substrate preparation. The full articles is available at researchgate on the following link:

https://www.researchgate.net/publication/256984331_Preparation_of_Atomically_Flat_Gold_Substrates_for_AFM_Measurements?ev=prf_pub

With kind regards,

Uros.

Article Preparation of Atomically Flat Gold Substrates for AFM Measurements