The standard coatings on domestic vacuum tube solar thermal water heaters using multilayers of materials like nickel sulphide and oxide and aluminium nitride already achieve absorptances in the visible of 94% while having emittances in the long infra-red (to stop re-radiation) of 8%. The gain from going from 94% to 99.95% is very little and the cost of going from a relatively low technology of such multilayer films to carbon nanotubes is at present huge. It will have to fall a long way to make it worth such a small gain

The article does not quote the infra-red emittance of the carbon nanotubes but this is a vital parameter. Since by Stefan's law, radiated energy rises as the fourth power of the absolute temperature and these utility sized solar thermal units operate with their absorbers at a much higher temperature than domestic units the re-radiation losses will be much higher. A small increase in infra-red emittance will easily wipe out the gain from a slightly higher visible absorptance.

I think that reducing reflectivity helps quite a bit, that was one of the main effects in improving efficiency in BigGav's second PV link on multicrystalline silicon. But, the main limit for this type of PV is the electronic bandgap which ends up only being responsive to a portion of the solar spectrum. Usually, PV is used with concentrators when it is multijunction so that there is more than one bandgap. That material is expensive because it is complex to manufacture and it is optimized to run hot, though not as hot as concentrated solar thermal. It is sometimes worth it though because you need less of it. What I have in mind is to use regular silicon and run it cold, but then use the low grade heat from the cooling water to save some energy in home water heating thoughout a small town. Because there is no need for really high tempertures and because existing infrastructure is reused, the cost comes in pretty low. Lower reflectivity in the panels means that much more heat for the water.

Chris