Pigouvian CO2 Taxes or Cap-and-Trade?

Now that the supreme court has ruled that the EPA's mandate includes regulating CO2 emissions (link), I have been asking people what sort of regulation they expect; traditional regulations, CO2 taxes, cap-and-trade or something else? From the limited responses I have gotten people expect a cap-and-trade type of regulation to emerge.

Asking around has started me re-thinking about whether Pigouvian CO2 taxes or cap-and-trade regulation. I think the issue of estimating marginal costs for CO2 emissions is the most important factor for favoring one scheme or another. Pigouvian CO2 tax schemes require good estimates, while cap-and-trade schemes don't require estimates at all, but Pigouvian schemes provide flexibility and decentralized decision making in finding the optimal total CO2 emission level, while cap-and-trade schemes either don't allow changes or require centralized decision making for total CO2 emission levels. I initially favor Pigouvian taxes because they have more theoretical appeal and because of the flexibility they provide, and if good marginal cost estimates are possible then Pigouvian taxes will probably provide the most efficient outcome. If good marginal cost estimates for CO2 emissions can't be found, but, the science works out that there is an emissions limit below which emissions are relatively harmless and above which emissions are quite harmful then marginal costs spike around this limit and cap-and-trade solutions will be relatively efficient because we know enough about the marginal cost curve to set a cap level that will be quite close to the level where marginal costs are equal to marginal profits. If neither of these cases hold, then it is very difficult to say what is efficient, and the world is in quite a bind.

Does anyone know if there has been work done on creating marginal cost estimates for CO2 emissions? or if there is research that suggests an 'emissions limit'?