100,000 deaths from 500,000 cars? That's hard to believe.
The average diesel passenger car emitted about 8kg of NOx in 2008 and a third less in 2015 according to [1] and [2], so on average that's 6.6kg per car per year. If VW emitted 20 times that much since 2009, as alleged, that's 6.6 * 20 * 6 years = 792kg. An average diesel engine would have emitted 40kg, so that's 752kg extra for VW.
Multiply by 500,000 cars = 376,000 tons of extra emissions. [2] claims that cutting emissions by 187,000 tons saves 4000 lives. Hence, 376,000 tons should cause about 8000 deaths. That's assuming NOx levels and deaths have a linear relationship, which is highly doubtful.
So it appears that my feeling/belief that 500,000 cars could not have caused 100,000 deaths is correct. But if all 11 million VW cars of that type worldwide have that same "defect", then of course the number of premature deaths rises to 176,000.
All of this is most likely complete bullshit, but you are right, it makes sense to at least try to run some numbers ;-)
[1] http://www3.epa.gov/otaq/consumer/420f08024.pdf
[2] http://www.arb.ca.gov/msprog/ordiesel/documents/OFRDDIESELhe...
[Edit] I already found a mistake in my estimate. The NOx numbers in [1] don't seem to be for diesel or not only for diesel engines. But other sources give similar averages for diesel engines (after the catalytic converter does its job of course)
Nothing at all.
First CO2 does not cause any deaths, unlike pollution.
Second the amount of CO2 is very small compared to global emissions, so it's quite irrelevant.