Getting better and cheaper all the time, and even more promising varieties on the horizon; it will keep the lights on instead of nuclear power Prof Peter Saunders
When solar cells were first developed, just over half a century ago, they were expensive and not very efficient at converting light into electricity. Their most significant application was in satellites, for which they were the only feasible energy source. Since then, the price per watt has dropped so much that solar photovoltaic (PV) is very close to achieving grid parity, i.e. producing energy at the same cost that utility companies pay for their current mix of sources. Furthermore, this price drop shows every sign of continuing; the cost is now only 10% of what it was five years ago [1]. Countries like Germany and the UK, which have been offering subsidies to help solar power get off the ground, are now phasing them out as the need has disappeared.
It’s not surprising that this is happening. Solar cells are based on semiconductors and the physics that underpins them, and every year, just about all such devices become better, smaller and cheaper. If you own a smart phone, you are carrying in your pocket or handbag the computing power that once would have occupied a large room and cost millions of dollars. As with other semiconductor-based devices, we can be confident there is more to come.
When a photon hits an atom, it can raise the energy of one of the electrons and allow it to become free, i.e. no longer attached to that atom. By itself, this isn’t very useful because the electron soon gives up its extra energy (as heat or light), falls back into orbit around a nucleus, and we are back to where we started. To get useful energy out of a system, we need an electric current, i.e. there must be some way to make the free electrons move off in the same direction.
The most common solar cells in use today consist essentially of two thin crystals of silicon, one doped with phosphorus (i.e. one with some of the silicon atoms replaced by phosphorus) and the other doped with boron. A silicon atom has four electrons in its outer shell whereas phosphorus has five and boron only three. This difference results in an electric field being established at the interface between the two pieces, and this drives the electrons in the direction of the phosphorus-doped wafer.
If the circuit is completed by a wire connecting the two sides of the cell, a current will flow from the phosphorus-doped wafer to the boron-doped wafer and energy can be extracted from the system.
Silicon cells have proved their worth and continue to improve, but they have some drawbacks. Although silicon is easily obtained because it is a constituent of common types of sand, to be used in solar cells it has to be refined to a purity of 99.9999%. Silicon crystals are hard to work with and cells made of them are relatively bulky.
So even though the tremendous improvement in solar PV has almost all been accomplished using silicon wafers, there is a great deal of research into thin film solar cells made of other materials. The layers are typically about 1μ (one micron, i.e one thousandth of a millimetre) thick whereas silicon wafers are about 350μ.
The thin film cells you are most likely to have seen are made from amorphous (i.e. non-crystalline) silicon and work in much the same way as the more common silicon wafer cells. Amorphous silicon is not very efficient at converting solar energy and is mostly found in devices such as calculators where size is an issue and only a very small amount of energy is required.
Two other common types are based on cadmium telluride (CdTe) and copper indium gallium selenide (CIGS). These can reach about the same efficiency as silicon wafers, but present technical challenges in design and manufacture, though these are being overcome at an impressive rate. There are also problems with the raw materials they require. Cadmium is highly toxic and so CdTe solar cells must be strictly regulated both in use and in recycling (see [2] How Green is Solar? SiS 49). Tellurium is common on some undersea ridges and it is relatively abundant in the universe as a whole, but few accessible sources are known, though that may be simply because it is only recently that it has become commercially important and geologists have started looking for it [3] .
Indium, which is also used in screens for televisions and mobile phones, is expensive and in relatively short supply. It may be possible to increase the amount obtained as a by-product of the refining of zinc, with which it generally occurs. Alternatively, a group at Oxford University are developing cells in which indium tin oxide is replaced by silicon doped zinc, which has a lower conductivity but is much cheaper and easier to source [4].
Dye sensitive solar cells (DSSCs) are also thin film but they operate differently [5, 6]. The basic idea is actually quite simple; you can even find on the web instructions for making one on the kitchen table [7]. A thin layer of an organic dye (blackberry juice will do) is immersed in an electrolyte, i.e. a liquid conductor, above which is a transparent electrode. The electrolyte is a solution with a lower redox potential than the dye, which means that electrons that are freed by photons striking the dye will move into the titanium dioxide rather than into the electrolyte. Again, connecting the two sides of the cell with a wire will cause a current to flow, in this case from the titanium dioxide to the upper electrode.
As this example shows, DSSCs can be made from materials that are neither rare nor toxic. The materials do not have to be highly purified and the actual manufacture is quite straightforward: it can be done commercially by printing onto foil. The challenge is the practical one of designing DSSCs that are efficient enough to be competitive with other solar cells while retaining the advantages of relative ease of manufacture and low cost of materials [8].
A big step forward has been the discovery that DSSCs can be made using perovskite, a mineral that is much cheaper and more abundant than the rare earths used in other thin film cells. Perovskite was first identified in 1839, but it is only recently that Michael Grätzel and colleagues in Lausanne began experimenting with it [9].
The first perovskite cells were unpromising because the material was rapidly degraded by the electrolyte. Even while they lasted, they were very inefficient, converting only about 3.5 % of the energy from sunlight into electricity as opposed to about 15 % from silicon wafers [9]. But researchers have since discovered how to replace the liquid electrolyte by a solid material and this solved the problem of rapid degradation. The efficiency has now been raised to 15% and from their successes so far and their understanding of the physical processes involved, researchers expect it could equal that of silicon wafers, which is over 20 % under laboratory conditions.
Another possibility is to use graphene, either in all-carbon cells or in combination with silicon. These are still in the experimental stage, but they may prove to have high efficiencies because while most cells only operate over a comparatively narrow range of frequencies (i.e. they collect energy only from light within a small part of the spectrum), these are not limited to the same extent. ([10]. Graphene and Solar Power for the Masses SiS 59).
In the meantime, improvements continue to be made in silicon wafer cells, and because they are the established technology, they may well remain the most widely used, at least for the next while. It has also been suggested that perovskite could be used to augment silicon cells rather than independently [9].
While the improvement in solar cells over the past decades has been an impressive achievement, it doesn’t tell us what we can expect from them. How much of our energy will PV be able to provide and when, and what are the issues involved in reaching that proportion? In 2012, for example, the UK total electricity production was 363 TWh, approximately 6 000 kWh for every person in the country, so we are talking about very large amounts.
We can get an idea of how much solar energy is available from a simple calculation. The sun provides about 10 000 times the total commercial energy that humans use [11]. With an efficiency of 20%, we would need to cover 0.05% of the Earth with solar panels. The surface area is roughly 5× 108 km2, so that would mean about 250 000 km2, roughly twelve times the size of Wales.
In principle, therefore, we could actually get all the energy we need from PV. That’s not really practicable, of course. There is a limit to how much of the Earth’s surface we want to cover with solar panels, not forgetting that less than 30 % of the area is land and much of that is at relatively high northern latitudes. And even if we had solar cells with efficiencies of 20 %, that wouldn’t mean we could harvest a fifth of all the sunlight that fell on the panels. All the same, PV can clearly make a significant contribution.
The European Photovoltaic Industry Association estimates that solar PV could be providing as much as a quarter of the EU’s electricity demand (electricity is about a fifth of the total energy) by 2030, with the obstacles they envisage being political and organisational more than technological [12]. Manufacturers’ organisations have a tendency to be a bit optimistic about what their industry can achieve, but the above calculation tells us they are at least in the right ball park. And considering how much solar cells have improved over the past 10 or 20 years, this is one area where a bit of optimism may be justified.
Solar cells generate electricity only during daylight hours and more in the summer when the sun is high in the sky than in the winter. There therefore has to be either some means of storing energy or an alternative source, or both.
While progress has been made, the problem of storage has not yet been solved. Batteries are improving, but they are still too expensive and too big for the amount of electricity they can store; this is of course also a major obstacle to the electric car. There are other ways of storing energy, for example by pumping water uphill when more electricity is being generated than is needed and letting it flow down again and drive turbines when there is a shortfall. We can heat molten salts or even water and recover the energy when it is required. We can even store energy in heavy flywheels. All of these are already in use, some in connection with solar heating rather than PV, but mostly at relatively small scales.
Another alternative is to use the energy of sunlight to split water into hydrogen and oxygen, replicating the natural process of photosynthesis ([13, 14] Splitting Water with Ease, Harvesting Energy from Sunlight with Artificial Photosynthesis, SiS 43 ). The hydrogen can then be used whenever and wherever it is required. There has been research in this area since the 1974 oil crisis, but there are serious technical problems that have not yet been overcome.
For the foreseeable future, the sun will not be the only source of electricity for most consumers, and the obvious solution will be to use the alternatives when more energy is needed. In developed countries, users will continue to remain connected to a grid, feeding electricity in or drawing it out as required.
The sort of grid that will be needed will be quite different from the present ones; in particular there will be far less electricity being transmitted over long distances, with the large loss of energy that entails [15] (Renewable Ousting Fossil Energy, SiS 60), [16]. This is already creating problems for grid operators not only in areas of high insolation like the southwestern US but even in countries as far north as Germany. Moody’s investors’ service states [17]: "Large increases in renewables have had a profound negative impact on power prices and the competitiveness of thermal generation companies in Europe.” A problem for those companies but an advantage for almost everyone else is that both PV generation and the demand for power peak at midday.
A team at Imperial College London has estimated the extra costs involved in integrating PV into the electricity supply system in Europe. They showed that to install up to 480 GW by 2030 is both technically feasible and not very expensive. The largest extra cost would be back-up capacity which they estimate as €14.5/MWh in northern Europe though considerably less in the south. Most of the extra costs would be relatively smaller when PV is providing less than 10 % of the electricity supply. This means there is a relatively small expenditure in the early stages and consequently little wasted investment if the technology improves over the next 15 or 20 years [18] -- yet another contrast with nuclear energy where we are locked into yesterday’s technology by the very high up-front cost.
In parts of Africa, many communities are not connected to a grid at all and much of the electricity is provided by small generators, often using kerosene as fuel. When solar panels are installed, the generators typically remain in place, but are now required only for backup. In these areas there may eventually be local grids but none on a regional scale, just as there is no need to go the great expense of building a national landline telephone network when so many people are already using mobile phones. Artificial photosynthesis could be especially important in these areas, which also have the greatest insolation and consequently the greatest capacity for solar power of all kinds. Other renewable sources, such as anaerobic digesters, could be also used to supplement solar PV.
In the 2008 White Paper on nuclear energy [19], the UK government pointed out that almost all the existing nuclear power stations are scheduled to stop producing electricity by 2023. Nuclear currently provides about a sixth of the UK’s electricity [20]. About a third of the coal fired stations are due to close in the next few years as well. Much of this capacity will have to be replaced, and building a new fleet of nuclear power stations will enable us to do this without increasing CO2 emissions.
Now Chernobyl and Fukushima have demonstrated clearly the hazards of nuclear power (See e.g. [21] The Truth about Chernobyl; [22] Lessons of Fukushima and Chernobyl; [23] The World Must Take Charge at Fukushima). It is also not an economical option; the government is expected to guarantee a price of just under £100/MWh for the next 40 years and take over responsibility for the long term disposal of the waste and almost all the liability if there is a serious accident.
Even if we were willing to accept the risk of a serious accident and to pay such a high price for electricity (and commit our children and grandchildren to doing the same), there is another issue to consider. Nuclear plants take a very long time to build. The earliest the first new one could be in operation is 2020, and it would probably be later than that. A new nuclear fleet could contribute nothing towards keeping the lights on in the short term and not very much in the medium term.
In contrast, solar PV can make a major contribution and it can do it now. The Department of Energy and Climate Change (DECC) estimates that 11.3 GW or more of solar PV could be available in the UK by 2017; even more could be available, but the present grid would have great difficulty coping [24]. The problem therefore appears to primarily one of management and storage rather than generation; if those can be overcome, as much as 22GW could be up and running before the first 1.6 GW of nuclear could come into operation.
Article first published 09/10/13
Got something to say about this page? Comment
There are 3 comments on this article so far. Add your comment above.
James Cooley Comment left 10th October 2013 02:02:28
"For the foreseeable future, the sun will not be the only source of electricity for most consumers"
Of course the sun is the source of almost all electricity used by consumers today, either from fossil fuels such as coal and oil, falling water from solar evaporation, wind from solar derived temperature gradients, or nuclear fuel from solar fusion in the distant past. Even Geo-thermal energy is derived from solar heat trapped in the earth's core, supplemented by radioactive decay of solar forged fusion elements such as uranium and radium. Our electricity is almost entirely solar based, but it's not very renewable with a few important exceptions.
Alan Bangay Comment left 16th October 2013 16:04:24
As far back as the 1960;s Thorium was seriously considered as the fuel for so-called nuclear power plants. I suspect that we went for enriched uranium instead because it can produce weapons grade forms of the original uranium as the nuclear plant is functioning. Apparently if we were to use thorium as fuel we would have very safe plants with no risk of meltdown. I do not know how the cost of oonstruction and decommisioning actually compares with enriched uranium plants but it is bound to be considerably less and a whole lot safer. It might be a better option than solar because it could be done on a very small scale - for use in cars and would work 24 hours a day with little need for batteries or sunshine. Food for thought.
Peter Saunders Comment left 31st October 2013 02:02:08
We discussed thorium reactors a couple of years ago, in SiS 52. They do seem to be safer than pressurised water reactors (PWRs). They would probably produce electricity at about the same cost as PWRs, but that's still a lot more than we expect to pay for energy from renewable sources by the time any thorium reactors could be in operation.
Neither the Liquid Fluorine Thermal Reactor (LIFTR) nor the Accelerator-driven System (ADS) is really suitable for small scale use. Actually, one of the reasons the Americans went with PWRs was that they were considered better for powering nuclear submarines. I certainly wouldn't want any kind of nuclear reactor in something that's going to be driven on public roads!