|
Post by matterbeam on May 12, 2017 13:24:24 GMT
Hi! I've been looking at nuclear-heated thermophotovoltaics. The current concept is the following: a nuclear source of heat raises the temperature of an intermediate element to about 1200K. This element emits infrared light. The light is absorbed by a thermophotovoltaic cell and converted into electricity. The high operating temperature of the thermovoltaic cell and the long wavelengths (low photon energy) combine to produce a system with very low efficiency. It is difficult to get rid of the heat produced, as it comes out at 600-800K (massive radiators). The total efficiency gain when combined with traditional thermoelectrics is 8%... Another version tries to overcome the Shockley-Queisser Limit, which is the maximum energy a solar cell can capture from sunlight. For silicon-based n-p junction gaps, this is about 32%. Using thermophotovoltaics allows the capture of the spectrum wasted by a solar cell, but n-p junction gap efficiency fall sharply with higher temperatures and real-world performance is less than 10% efficiency, compared to the 80%+ theoretical maximum. [ gcep.stanford.edu/pdfs/CTlSZRP4nww_77Shjd-A2g/ShanhuiFan_Symposium2009.pdf] I was wondering if it were possible to improve upon this concept. Shorter wavelength from higher temperature emitters allow for more efficient thermophotovoltaic cells to be used. A 6000K emitter can replicate the full spectrum of sunlight. A 25000K emitter emits the vast majority of its energy in a very narrow ultraviolet band (https://astrogeology.usgs.gov/tools/thermal-radiance-calculator/). It is not a coincidence that 25000K is the temperature of uranium gas in a nuclear lightbulb. We can do away with intermediate emitters and work directly with the high-energy photons emitted by the uranium gas. The photovoltaic cells can be tuned to work specifically at this wavelength, but I need examples of such materials. Maximum Carnot efficiency between a 25000K heat source and a 300K PV cell is 96%, so the efficiency of the system is simply how effective the PV cell is at absorbing and converting ultraviolet light. Temperatures from the PV cells absorbing the incident ultraviolet light must be dealt with by either active cooling, distance between the emitter and the cells or a combination of both. Here is a simple schematic for a distance-reliant no-active-cooling reactor. It has a massive weight. If a cooling system is used to remove 1kW/m^2 from the solar cells without raising their temperature, then the radius can be decreased to 470m. At 2kW/m^2, it can be reduced to 332m and so on.
|
|
|
Post by leerooooooy on May 12, 2017 16:39:08 GMT
>literally a nuclear lightbulb
I love this
|
|
|
Post by matterbeam on May 12, 2017 16:50:56 GMT
>literally a nuclear lightbulb I love this I hadn't realized the irony...
|
|
|
Post by n2maniac on May 13, 2017 19:13:22 GMT
There are a few things that can be done. Yes, increasing the hot side temperature gets you past some of those pesky limits of thermodynamics, but I'll mention that current solar cells are not terribly limited by that theoretical issue with a 6000K hot side (95% theoretical with 300K cold side, yet records are closer to 40-50%). For p-n junctions, it would be wise to stick with available materials. UV LEDs are difficult due to the very high bandgaps needed (3V for a bit of 400nm), and solar cells would see similar technological difficulty. Worse, there are chemical/physical limits you hit trying to go higher ( see bandgap ranges: they top out below 6V). A dichroic mirror surrounding the hot source may increase efficiency greatly if configured correctly. Have photons that aren't useful to your process? Just build it to reflect those back at the heat source to reduce the number transmitted! In theory, you could use this to go from a blackbody source to a narrow band source. Now, just adjust the narrow band to the peak efficiency band of the photovoltaic and we have increased efficiency! (Wait, this sounds like an actual business idea. Damn, someone else beat me to it) Please note that, in a CDE context with a cheap heat source but expensive radiator rejection area, operating with lower cold side temperatures than about 75% of the hot side hits diminishing returns based on electricity per radiator area.
|
|
|
Post by apophys on May 13, 2017 21:55:41 GMT
|
|
|
Post by matterbeam on May 14, 2017 12:01:38 GMT
There are a few things that can be done. Yes, increasing the hot side temperature gets you past some of those pesky limits of thermodynamics, but I'll mention that current solar cells are not terribly limited by that theoretical issue with a 6000K hot side (95% theoretical with 300K cold side, yet records are closer to 40-50%). For p-n junctions, it would be wise to stick with available materials. UV LEDs are difficult due to the very high bandgaps needed (3V for a bit of 400nm), and solar cells would see similar technological difficulty. Worse, there are chemical/physical limits you hit trying to go higher (see bandgap ranges: they top out below 6V).A dichroic mirror surrounding the hot source may increase efficiency greatly if configured correctly. Have photons that aren't useful to your process? Just build it to reflect those back at the heat source to reduce the number transmitted! In theory, you could use this to go from a blackbody source to a narrow band source. Now, just adjust the narrow band to the peak efficiency band of the photovoltaic and we have increased efficiency! (Wait, this sounds like an actual business idea. Damn, someone else beat me to it) Please note that, in a CDE context with a cheap heat source but expensive radiator rejection area, operating with lower cold side temperatures than about 75% of the hot side hits diminishing returns based on electricity per radiator area. Part in bold is very interesting. How do I understand how the band gap ranges relate to the wavelengths accepted?
|
|
|
Post by n2maniac on May 14, 2017 19:16:52 GMT
There are a few things that can be done. Yes, increasing the hot side temperature gets you past some of those pesky limits of thermodynamics, but I'll mention that current solar cells are not terribly limited by that theoretical issue with a 6000K hot side (95% theoretical with 300K cold side, yet records are closer to 40-50%). For p-n junctions, it would be wise to stick with available materials. UV LEDs are difficult due to the very high bandgaps needed (3V for a bit of 400nm), and solar cells would see similar technological difficulty. Worse, there are chemical/physical limits you hit trying to go higher (see bandgap ranges: they top out below 6V).A dichroic mirror surrounding the hot source may increase efficiency greatly if configured correctly. Have photons that aren't useful to your process? Just build it to reflect those back at the heat source to reduce the number transmitted! In theory, you could use this to go from a blackbody source to a narrow band source. Now, just adjust the narrow band to the peak efficiency band of the photovoltaic and we have increased efficiency! (Wait, this sounds like an actual business idea. Damn, someone else beat me to it) Please note that, in a CDE context with a cheap heat source but expensive radiator rejection area, operating with lower cold side temperatures than about 75% of the hot side hits diminishing returns based on electricity per radiator area. Part in bold is very interesting. How do I understand how the band gap ranges relate to the wavelengths accepted? Plank's constant. In units relevant to this (with a factor of c multiplied into it), it is 1240 nm*eV. Basically, a 1eV photon has a 1240nm wavelength, and they are inversely proportional.
|
|
|
Post by matterbeam on May 17, 2017 14:37:27 GMT
Part in bold is very interesting. How do I understand how the band gap ranges relate to the wavelengths accepted? Plank's constant. In units relevant to this (with a factor of c multiplied into it), it is 1240 nm*eV. Basically, a 1eV photon has a 1240nm wavelength, and they are inversely proportional. So I guess the only solutions are to use lower temperatures or an intermediate emitter and accept the efficiency losses?
|
|
|
Post by n2maniac on May 18, 2017 7:38:34 GMT
Plank's constant. In units relevant to this (with a factor of c multiplied into it), it is 1240 nm*eV. Basically, a 1eV photon has a 1240nm wavelength, and they are inversely proportional. So I guess the only solutions are to use lower temperatures or an intermediate emitter and accept the efficiency losses? Either way leading to efficiency losses, yes. Might be able to do something more exotic in theory, but in practice when you start getting below 200nm you find that just about everything tries to absorb it (it is called vacuum UV at that point, because contraptions utilizing it have to evacuate air, which strongly absorbs it, from the beam path). Could use a dichroic mirror to only use the longer wavelengths, but you are 1) not really any better than a cooler blackbody at that point, and 2) going to have trouble making a mirror withstand the higher energy UV. For a on-paper design, I would suggest to look around for charts showing quantum efficiency versus wavelength on solar cells (such as this one on pg 6), choose a blackbody and cell temperature (and hope you know the effects from the cell not operating at designed temperature), use a dichroic mirror allowing no less than 40% of the radiation through, and start looking at the characteristics of such a design. For one, I would suggest that the power density you can achieve may be an important factor, as most of the ingame reactors have exceptional power throughputs (just checked one of mine: it is sitting at 40MW/m2 of thermoelectric area, or about 40,000 suns!). Meanwhile, typical solar cells under concentrators don't typically get tested much above 1000 suns. Food for thought.
|
|
|
Post by ash19256 on May 19, 2017 18:56:52 GMT
You know, with how much thermal energy is being output, I wonder how easy it would be to make this system use a two-stage cooling loop, with there being a thermocouple in between the two loops to put some of that wasted thermal energy to use. Possibly also having the outer loop run a turboelectric set-up for even more efficiency?
|
|
|
Post by RiftandRend on May 19, 2017 19:43:45 GMT
You might be able to direct the light coming out into a gain medium and create a direct nuclear laser.
|
|
|
Post by ash19256 on May 19, 2017 20:54:44 GMT
You might be able to direct the light coming out into a gain medium and create a direct nuclear laser. ... Wouldn't that basically be a fission pumped version of the normal lasers we use these days in CoaDE?
|
|
|
Post by newageofpower on May 20, 2017 4:48:00 GMT
You might be able to direct the light coming out into a gain medium and create a direct nuclear laser. Didn't I post an article on this?
|
|
|
Post by RiftandRend on May 20, 2017 9:56:54 GMT
You might be able to direct the light coming out into a gain medium and create a direct nuclear laser. Didn't I post an article on this? The lasers described in the article were most definitely not nuclear lightbulb lasers, which is what I was suggesting.
|
|
|
Post by matterbeam on May 20, 2017 12:29:04 GMT
You might be able to direct the light coming out into a gain medium and create a direct nuclear laser. Didn't I post an article on this? Oh, where? Also, relevant to topic: Carbon nanotube 3D rectennas.
Potential for 90% single wavelength efficiency.
|
|