If you're going to measure the "clean" of solar power, why would you neglect the production of all of the minerals that are used in the process? These minerals include arsenic, bauxite, boron, cadmium, coal, copper, gallium, indium, iron ore, molybdenum, lead, phosphate, selenium, silica, tellurium, and titanium dioxide. Some of these minerals are difficult to source and mine, and almost always create a large degree of environmental damage in their wake. SourceThere are also the environmental costs of providing backup (for when the sun isn't shining) in the form of batteries, or other storage, and the costs of disposing of the panels after their lifetime. As the Economist article says, "The consequence of all this number-crunching is not as clear-cut as environmentalists might hope."
This underlines what I have said in my previous post when it comes to making policy: rather than try government try to identify all the environmental implications of any policy with inescapably limited knowledge at fixed point in time, we should rather be identifying the outcomes we want to see and rewarding their achievement, however that is done. This would be more practical than attempting to conduct entire life-cycle analyses over all possible policy choices - which, even if it were possible, would be instantly made obsolete by new technology and our expanding scientific knowledge. It would also cohere more closely with goals that can be clearly articulated and that are meaningful to ordinary people: those that specify desirable levels of plant, animal and human health.
For more on applying the Social Policy Bond concept to the environment see here.