« Your Tax Money at Work. | Main | Carnival of the Capitalists »

August 16, 2004

Solid Oxide Fuel Cells

FuturePundit prompted me to speak to Professor Ignatiev about the new advance in fuel cells.
I hope you'll excuse me using industry insider knowledge to provide a little perspective on this announcement. There aren't that many who work in this field and I may be the only one with a blog.
My work in this field is from the day job, the delightfully named "The Low Hanging Fruit Company Ltd", where we supply much of the world's usage of scandium. (Don't worry, it's just a metal and yes, very few people have heard of it.) /advertising off.
Over the past few years we've been in contact with a number of fuel cell people as scandium seems to solve some of the problems people were having with Solid Oxide Fuel Cells (SOFCs). Over the years the specific problem has changed. First it was making sure that the SOFC actually produced electricity at something close to the theoretical limit for the technology, some 60 - 65 %. This was achieved by using Yttria Stabilised Zirconia (YSZ) as the electrolyte. Don't worry too much about what this is but yttria is yttrium oxide, (about $30 a kg in bulk) and zirconia is zirconium oxide, a cheaper material. While there are some costs in making the compound, it isn't, as these things go, all that expensive.
However, operating temperatures of 900 - 1,000 oC mean that all of the supporting structure needs to be made out of comparatively expensive materials.
The next stage was to see if a slightly different electrolyte could be in itself more efficient, allowing the cell stack to run at a lower temperature. Scandia Stabilised Zirconia (ScSZ) allows this, operating at about 800oC. This seemingly small change means that much cheaper materials, simpler steels, can be used as the supports and surrounds. Great, except scandium oxide is much more expensive than yttria, about $500 a kg in bulk. Still the first large scale manufacturer of SOFC power plants went the scandia route : Toho Gas. There is also a problem with ScSZ, in that it can crack during the heat up/cool down cycle. Toho fixed this with an addition of ceria (cerium oxide) and Professor John Irvine in Scotland (who is the head of the EU investigation into all of this) used a ScYSZ electrolyte.
OK, so far so good, we've got a lower operating temperature, thus lower component costs, still at or near theoretical limits of efficiency, but at the expense of a much dearer electrolyte.
(If you want to see other players, have a look here. There's all sorts of interesting ideas out there, like Rolls Royce intending to run turbines off the waste heat from large plants to people who will sell you an operating plant today.)
The next stage, about 18 months ago as I recall, was to try and thin the electrolyte so as to reduce costs. The first group to leap up and down in excitement were these guys at Lawrence Livermore. Their aim was to get material costs down below $400 per kilowatt, something they could achieve with their process for 500 nanometre layers of electrolyte (that figure came from Professor Ignatiev this afternoon, sorry, I've forgotten what Steve Visco told me). Thinner electrolyte also meant less heat to disperse, further lowering temperatures and the cost of alloys needed for all the surrounds and supports.
All of this leads us to Prof Ignatiev and his team. Their electrolyte layers are down to 100 nanometre. This they do by a one stage lithography process, something that is cheaper, way cheaper, than the manufacture of silicon chips. Again, the thinning of the electrolyte layers has led to a reduction in operating temperatures, meaning that again all of the surrounding materials can be of yet cheaper manufacture. This is a bit of a guess but I would be willing to put money on the idea that you would not even need stainless steel, not even 18/8 at this point. Good old carbon steel at $400 a tonne should do it.
I think you can see that this isn't some leap in the dark, some out of the blue occurence. It's good science and engineering being done well, getting to one point, evaluating the problems then solving those, then around for another iteration.
This latest development means, from my worm's eye view, that the cost constraints on SOFCs are really and truly gone. We know how to do one stage lithography (although currently Prof I is using crystal nickel upon which to deposit, still distressingly expensive) cheaply and easily and all the materials come in well under the $400 per kWatt level.
Great, we've solved all our problems?
Not so fast. SOFCs are not, for technical reasons, well suited to use in transport. These new lower operating temperatures might make them so but I'm not at present in a position to say. Local generating plants? Oh yes. Combined Heat and Power? For a theoretical efficiency above 80%? Bring It On!! Transport? Someone else will have to determine that.
There is also the problem of fuel. SOFCs can use methane, methanol, ethanol without too many problems. They can also use hydrogen without it having to be excessively pure (this is what "internal reformer" means, I think), yet we still need to work out where we're going to get any of these things from. Given the greater efficiency of the unit over simply burning the stuff, natural gas usage would reduce emissions. Yet we've only really solved the problem when we have a non fossil fuel method of creating any of the above and hydrogen would be best.
Cheap efficient SOFCs are, from what I can see, now only an engineering/manufacturing problem. How about we run the windmills and the solar cells to electrolyse water (getting round their problem of inconsistent output) and use hydrogen as the battery, the SOFC as the means to turn the hydrogen back into power. Yes, inefficient, but the cumulative inefficiencies seem to be less than those of the fossil fuels we use right now.
Here's an interesting calculation:
Insolation is at roughly one horsepower per hour per square yard for seven equator equivalent hours per day just about anywhere with extensive human habitation. That's 5 or so KWatts per sq yard per day. What's the size of the average American houses' roof? 1,000 sq foot? 100 or so sq yards? 500 Kwatts a day? Solar cells with 30% efficiencies are out there (Berkeley, GaAs/GaN/InN). 150 KWatts. As I don't know the efficiency of a process to separate the hydrogen from the water, I'll assume 50%. OK, we've got 75 KWatts of usable power now. Our SOFC produces electricity at 60% efficiency: 45 KWatts per day of storable power. From land that's already in use for something else. Average US household daily electricity usage? 30 KWatts.
OK, OK, there's a number of leaps in those numbers but we are getting there, we really just on the cusp of being able to power a household from the ground it already occupies.Anyone who wants to correct me on the above please do.
Now all I have to do is to convert Professor Ignatiev from his use of yttria stabilised zirconia to scandia stabilised. I did plant the idea this afternoon and he says it is on his list of things to do. Just how cool would it be to be part, in however a small way, of the solution to global warming?
(If it's happening of course, and if it's not the sun itself. I do read TCS after all.)

August 16, 2004 in The Low Hanging Fruit Company (BVI) Limited | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Solid Oxide Fuel Cells:

» Thin Film Fuel Cells May Obsolesce Large Electric Plants from FuturePundit
Small thin film fuel cells may obsolesce large electric power generating plants based on fossil fuels. "By using materials science concepts developed... [Read More]

Tracked on Aug 16, 2004 9:08:04 PM

» Thin Film Fuel Cells May Obsolesce Large Electric Plants from FuturePundit
Small thin film fuel cells may obsolesce large electric power generating plants based on fossil fuels. "By using materials science concepts developed... [Read More]

Tracked on Aug 17, 2004 1:03:53 AM

Lynne Kiesling OK, I said we were about more than electricity here, but this is cool: Randall Parker's post on thin-film superconductor technology that may, in time, make large-scale centeral fossil-fuel generation of electric power obsolete. I also th... [Read More]

Tracked on Aug 17, 2004 3:49:07 PM


I read your page http://timworstall.typepad.com/timworstall/2004/08/solid_oxide_fue.html
(trackback'ed from Futurepundit) and I was thinking very well of you until I found this:

>Insolation is at roughly one horsepower per hour per square yard for seven equator
>equivalent hours per day just about anywhere with extensive human habitation. That's
>5 or so KWatts per sq yard per day. What's the size of the average American houses'
>roof? 1,000 sq foot? 100 or so sq yards? 500 Kwatts a day? Solar cells with 30%
>efficiencies are out there (Berkeley, GaAs/GaN/InN). 150 KWatts. As I don't know the
>efficiency of a process to separate the hydrogen from the water, I'll assume 50%. OK,
>we've got 75 KWatts of usable power now. Our SOFC produces electricity at 60%
>efficiency: 45 KWatts per day of storable power. From land that's already in use
>for something else. Average US household daily electricity usage? 30 KWatts.

How, HOW can you expect anything you write to be taken seriously when you cannot
even make the proper distinction between kilowatts and kilowatt-HOURS?

When you say meaningless things like "Kwatts a day" and "horsepower per hour"?

When you hold up the entire field for ridicule? See http://www.denbeste.nu if you
haven't yet.

Please. If you can't get it right, GET AN EDITOR who can catch the errors for you.
As it is I cannot even think about citing your information about SOFC materials
because it's obvious that you spout off about things you do not understand, and you
may be no more accurate about the use of scandium oxide than you are about kilowatts.

Posted by: Engineer-Poet | Aug 17, 2004 6:20:03 PM

I do ask in the post for anyone who wants to correct me to do so. My knowledge of scandium I will put up against anybody on the planet, given that that is what I do for a living, and being one of the very few who do.
My knowledge of units of power? Obviously lacking. Straighten me out why don't you?
I would also be interested to know whether the mistakes actually change the conclusion: If we had a method of storing the the energy in sunlight falling on the roof of a house, given the above efficiencies of each individual stage, would we have enough energy at the end to cover the demand of the average houshold?
I think that's an interesting question and would love someone who knew more about the various units involved to help me out with that.

Posted by: Tim Worstall | Aug 17, 2004 6:50:44 PM

"Horsepower per hour" is meaningless.  You mean just plain horsepower.

At 1 HP/yd^2 and 111 square yards of roof, you get 111 horsepower.  Over 7 equivalent hours/day, that is 777 horsepower-hours or (at 746 W/HP) 580 kilowatt-hours per day or about 24 KW average.  At 15% conversion efficiency you'd get 87 KWH/day output; at 30% you'd get 174 KWH/day.

A 30%-efficient electrolysis/fuel cell cycle would cut that to 26.1 and 52.2 KWH/day, respectively.

Unless you had an application that could only be served by hydrogen you'd probably reduce your capital costs considerably by dispensing with it and using batteries instead.

Posted by: Engineer-Poet | Aug 18, 2004 4:01:14 AM

"SOFCs can use methane, methanol, ethanol without too many problems."

If this is so, why worry about Hydrogen which is really nasty stuff to store and transport. There is lots of methane and more can be had from deep sea deposits and even garbage dumps. Methanol can be manufacured by cooking almost any carbon containing compound and ethanol can be diverted from a higher and better use as vodka.

Posted by: Robert Schwartz | Aug 18, 2004 6:02:47 AM

Why hydrogen? Partly becasue it sidesteps the carbon issue altogether, partly because I think that electrolysis is more efficient that , say, growing plants to make ethanol. Both are means of using sunlight to power the fuel cell.
You're right, it's not a necessary step.

And to engineer-poet, thanks for the corrections. My piddling about with units that I don't understand very well obviously makes me look stupid but I'm glad to see that the conclusion stil stands, that we do have a set of technologies that would allow a house to be powered by the sunlight falling upon it.
Why hydrogen and not batteries? I have a feeling, when these various things are more technologically mature, that the capital costs will be cheaper, no proof, just a feeling that things ultimately based upon chip making technology will turn out to be cheaper than lumps of lead, lithium or nickel. There is also the existence of another type of fuel cell, the PEM, musch more suited transportation uses. Excess hydrogen could be used to fuel them.

None of the above addresses the capital costs of the process, which at present would be absolutely horrendous. This isn't going to happen soon, it doesn't help with large users of power and it will require at least a decade of further engineering and cost reductions to even get close to being actually usable.
That wasn't however my point. Only that, to power households, we don't have to pave Ohio.

Posted by: Tim Worstall | Aug 18, 2004 8:42:34 AM

Randall Parker has noted that we have already covered an area the size of Ohio (see http://www.futurepundit.com/archives/002185.html).  Given that, why not make better use of all that covering?

(Idiot blog deletes properly-formatted HTML links.  There should be at least a note about this.)

If we get something like Nanosolar's TiO2-based solar cells and they get down to $30/m^2, we'll see daytime power at as little as one cent per KWH.  The same advance will not make storage any cheaper, so the economic incentive will be to make devices and industrial processes which can exploit this cheap, intermittent power.

Posted by: Engineer-Poet | Aug 18, 2004 12:44:58 PM

Interesting point about the cheap solar cells. I was looking in hte other direction, at more efficient ones as I semi-deal with the metals they are made from.
The Ohio thing I lifted from Randall .

Posted by: Tim Worstall | Aug 18, 2004 1:37:05 PM

According to NREL, the amount of solar insolation varies by location. For my location (western Missouri), it's about 4.8 kWh/m^2/day (assuming the PV is fixed in its alignment, and angled at lattitude). That's the annual average; higher in the summer months, lower in the winter months. I think you get the idea. Places like southern Arizona are more like 7-8 kWh/m^2/day, which would get your closer to your figures.

Considering the fact that most PV available today are in the 17-20% efficient range, 100 m^2 of PV would give about 17-20 kW output, or an average of about 89 kWh/day (that's splitting the difference on 17 and 20%, multiplying by 4.8 hours and 100 m^2). Yes, the higher efficiency cells ARE available, but they are exceedingly difficult to manufacture, and are literally worth more than their weight in gold. The cheaper stuff is on the order of $4-5/watt. 18.5 kW worth of such PV would cost about $83k, but would give about 89 kWh/day.

So, in my location, I'd average about 89 kWh/day. I haven't been able to find any better figures on the electrolyzer, so I'll go with your 50% efficiency figure. That makes about 44 kWh worth of hydrogen (36 kWh of hydrogen = approx. 1 kg of hydrogen, so about 1.22 kg H2/day). Even if you get 60% efficiency on the fuel cell, you're still getting about 26 kWh of stored energy out of the system, at best. I mean, we're putting electricity through a 50% efficient process, then putting its output through a 60% efficient process to get back to electricity. 0.50 x 0.60 = 0.30, so 30% of what goes into the system comes back out (ignoring energy losses compressing the hydrogen, etc.).

My household typically uses about 2000 kWh / month, or about 67 kWh / day average (it's an old, inefficient house). If we did the PV -> electric -> hydrogen -> electric route, we still need to buy electricity or hydrogen at the end of the day. If we just keep the 89 kWh as electricity, pumping the excess into the grid when we don't need it and drawing on it when we do (net metering), we've got a net energy surplus.

Don't even think about a fuel cell vehicle. They get, at best, about 50 miles/kilogram of hydrogen. That's about 61 miles or so worth of driving, per day, from $83k worth of photovoltaics. Not a very good trade-off. You'd be better off with an electric vehicle. Some of the newer ones being demonstrated have over 200 miles range on a single charge, and get about 5 miles / kWh of electricity.

Posted by: Tony | Aug 19, 2004 3:16:25 AM

Tony how can use 67 kwh a day!??? We have solar PV fitted in Scotland under the UK PV demonstration scheme and so can provide some real data. The first thing I'd like to say is that fitting microgeneration aids energy efficiency. We have cut our total yearly consumption from 3000kw/h to a predicted less than 2000kw/h 2 years ago before the system was fitted. This is particularly impressive since we export 60% to the grid. We have notionally 1kwp fitted (really .75kwh since that is the maximum output of the invertor). The only month when we generate almost nothing is December. In March we can generate 6kw/h per day the same as we do in summer. We generate 70-80kwh in march 120KW in july. I'm interested in the fuel cell H2 route as a long term means of heating and lighting my house. I know there are efficiency losses at both parts of the process but then are huge losses in a centralised grid, I realy don't like nuclear power and some day the gas is going to run out before that becoming very very expensive. Of course the ineffecincies on the fuel cell side are less important since they can be recovered as heat. I intend to add more PV capacity both a new type of slate coming onto the market and more modules. We are preparing to submit for planning permission for this and roof mounted wind turbine. If Ceres bring out their domestic CHP SOFC cell we would be interested in replacing our elderly boiler with it. Initally we would use mains gas but then look to produce our own hydrogen. I found this site whilst looking for a way to calculate how much 13000 kwh of gas would produce in electricity terms. I did read that one kg of H2 would produce 52KWh of electricity at plausible effeciencies but despite my chemistry background I cannot get this answer. I'm actually a mature phd student looking at a certain type of fuel cell but its not for domestic use.

Payback time of the PV system looking about 10 years at the moment (with grant).

Posted by: Neil | Dec 1, 2005 1:24:42 PM

Why spend money on a very expensive PV system when you can capture the suns energy as biomass?

The biomass can be liquified at 85% effieciency by weight. The resulting bio-oil has an energy value of about 60% of feul oil by volume.

It is very easy to transport with conventional infrastructure. It can be stored in a tank and ran into a SOFC. It can also be burned in a conventional heater to heat the house.

About 10 tons per acre dry weight switchgrass can be grown in the US. This would produce about 8.5 tons of bio-oil, a very convenient fuel for the SOFC in the basement.

This doesn't sell as much metal though.

Posted by: don | Mar 24, 2006 9:21:01 PM