Oh - and here's something else. Let's say the graphite dome is 25cm in diameter, so about 0.195m2 in surface area. Let's say it's heated to 3500K to avoid it sublimating away. The the Stefan-Boltzmann radiation formula gives a total radiated power of 1.7MW, all of which heat will have to be dissipated somehow. Because you're running the dome at a reddish temperature, the radiance will peak in the infrared and only about 14% of the power will be in the visible. The mean total radiance on the PV cells will be 1.7MW/m2, assuming an array with 1 m2 area (over 1,000 times the solar case). For a passively cooled system, assuming the cell array is itself radiating as a blackbody, the equilibrium temperature of the array would be about 2350K. That's not going to be good. For a 1000 sun system, you then have a peak power density on the silicon of 1.7GW/m2. Ouch, ouch, ouch!
If the reaction doesn't produce 1.67MW of power then a dome of that size will not get to 3500K.
You can reduce the power output by making the dome smaller, but then you have to engineer all the gubbins inside the dome, molten silver and all into a smaller dome. Or you can run the dome at a lower temperature where the PV cells are even less sensitive and you throw away even more heat.
The steam engine is looking ever so attractive.