The Sunny Side: Photovoltaics

solar panels are in a class of their own for electricity generation. Most power sources turn a turbine and create current from a moving magnetic field. Solar panels go straight from light to electricity. When the high-energy light from the Sun hits certain materials in the panels called , some unusual properties of light allow electrons to jump around and conduct, forming an electric current.

Mysteries of quantum physics

Solar panels rely on the wave-particle duality of light, a mysterious aspect of modern physics first uncovered in Einstein’s Nobel Prize-winning explanation of the photoelectric effect. The photoelectric effect is important in modern physics because it is clear indication of the fact that light is both a wave and a particle. Waves imply continuity, swells rising and falling in time as they move. Particles, on the other hand, are discrete. They are entities that you can count. It’s impossible for something to be both, right? But both interpretations are consistent with the behavior of light, both are needed to explain what we see. Quantum mechanical phenomena like wave-particle duality suggest that there’s much more beneath the surface, that our limited day-to-day experiences on Earth are so incomplete to understand the universe.

Further investigations in modern physics uncovered more mind-bending phenomena: that time slows down when you approach the speed of light, that apparent barriers can be tunneled through in impossible ways. It might be a hopelessly poetic interpretation, but to me the findings of modern physics show us that the world of our everyday lives is not the way things have to be. Seemingly impossible barriers like the power of the 1%, and centuries-old systems of oppression can be broken down. Huge social transformations can unfold in the blink of an eye. As Frederic Jameson said, “it may be easier to imagine the end of the world than the end of capitalism,” but the world of today isn’t what it seems when you look closely. Science says so! It seems appropriate that wave-particle duality is behind the technology that is most likely to revolutionize our relationship with power: .

Electricity from the Sun

Let’s discuss how solar panels work. High energy photons from sunlight cause electrons in semiconductor material like silicon to jump across a forbidden band into an outer conduction band. Since light energy is proportional to its frequency, or inversely proportional to wavelength, there are specific colors of light that allow a band gap to be exceeded, depending on the substance. Sunlight can excite electrons within materials like silicon, cadmium-telluride, copper-indium-gallium-selenium, and gallium arsenide.

The most common panels made today use silicon as the semiconductor material. Many clean energy options require rare materials, but silicon is not rare by any stretch of the imagination. Silicon is the 2nd most abundant element in the Earth’s crust, following only oxygen. Much is mined in the form of quartz or sand.

In order to herd the electrons that are dislodged into the conduction bands in semiconductors, impurities are added. Within silicon, boron, an element with a spare “hole” for another electron, is added on one side (the positive, or p-type wafer), while phosphorus, an element with an extra roaming electron, is added to the other side (the negative, or n-type wafer). This sets up a voltage difference which allows the sunlight to generate a steady current.

Because of the properties of the materials, there are maximum efficiencies of the conversion from sunlight into electricity. Typical single-crystal silicon panels of today have efficiencies around 20%. Efficiencies are measured under standard test conditions, which assumes 1000 W/m2 of sunlight (typical  for a bright summer day), temperature of 25o C and a standard sun angles, to standardize the amount of the atmosphere that the sunlight shines through. Since the efficiency of panels is defined based on 1000 W/m2 of power, the nameplate capacity of panels only depends on their area and the efficiency.

Nameplate capacity in watts = 1000 * efficiency * area in m2
A typical size of a silicon cell is 0.156 m by 0.156 m. These are often arranged in modules with 60 or 72 cells, and dimensions 1 m by 1.6 m, or 1 m x 2 m. A 72 cell panel with area 2 m2 and efficiency 20% has nameplate capacity 400 W. The nameplate capacity of 60 cell modules is often around 300 W. Multiple modules are installed into a rooftop array, depending on the size of the roof or the desired electricity to be generated.

How sunny is your home?

Nameplate capacity, of course, is not the power that’s actually generated. One must take into account the capacity factor, which depends strongly on location for solar. The map below shows that it’s a function of latitude and cloudiness. Solar panels still work when it’s cloudy, but have less output.

Since there are 24 hours in a day, a PVOUT value of 2.4 kWh/kWp implies a 10% capacity factor. Much of the world is above 20% capacity factor (4.8 kWh/kWp), and parts are above 25%. Although high latitude regions are not marked on the graphic above, high latitude regions can generate a large amount of solar power in the summer, if panels are arranged to capture sunlight throughout the day, with a tracking system or within a circular arrangement.

Sunlight is not uniform throughout the day, of course. Below are some observations taken from the rooftop of the Atmospheric Sciences and Geophysics building at University of Washington over 4 sunny days in April 2021. The sunlight peaks at around 850 W/m2 at local noon, and the average sunlight is 290 W/m2, averaged over day and night. These days, the capacity factor would average over 29% (higher than 290/1000 because the tilt of the panels would lead to more absorbed sunlight), higher than anywhere in the world on an annual average! Seattle’s average capacity factor for a properly tilted solar panel is only 12.5% though (3 kWh/kWp), due to its short winter days and often cloudy weather conditions.

In order to build a solar arrays it is necessary to add a few components to the modules. First, an AC inverter must be connected in order to transfer the direct current (DC) generated from the panel into alternating current (AC). There also needs to be wiring, mounting on the rooftop, and one should expect losses from coverage, temperature, shading, and normal aging. All in all, 9-15% losses compared with the nameplate capacity are not unusual. Often rooftop solar systems are connected to the grid, so they can provide others with electricity when generation exceeds demand. If a residence is to generate its own power when the local grid is down, one needs a battery. This allows power to be synchronized with demand, as is needed to run electrical devices. Each of these components adds to cost, both in terms of money and embedded energy.

Generating a tera

How many solar panels would it take to generate the decent living energy power in locations around the world? We calculated 66-136 W of local generation needed, depending on the location. Assuming capacity factor of 8-25% worldwide, 20% efficiency of panels, and 14% losses, the lower limit can be generated with 1.5-4.8 m2 of panels and the upper with 3.2-10.1 m2 of panel area. Given that the decent living space requirement works out to 15 m2 per person, the average power needed can be easily supplied anywhere on Earth with solar panels over only a little over 2/3rds of the living space. In most locations, significantly less area is needed, implying that even multi-story residences can provide all the electricity needed for their residents with rooftop solar arrays. Try it out yourself on the neighborhood scale in the interactive app below.

Solar power interactive tool by EarthGames

This, of course, assumes that power can be used when the Sun is shining, which is not always the case. But the ease of satisfying the decent energy need shows that the average generation picture is one of solar energy abundance, not energy scarcity. In the next chapter we’ll consider how sunlight varies seasonally and on cloudy versus clear days, and we’ll cover methods of storing energy later in the book.

How many panels would it take to generate a terawatt of nameplate capacity? At 20% efficiency, the area required seems huge: 5 trillion square meters. But this is only 5000 square kilometers, which can be contained in an area of 71 km x 71 km.

To generate a terawatt of useful electricity, one needs to increase the area due to the capacity factor and the typical losses. Assuming 20% efficiency and 14% losses, 30 trillion square meters or 30,000 square kilometers are enough to produce 1 terawatt on average. This can be contained in a square with dimensions 173 km x 173 km, or a little more than 100 miles x 100 miles. Not much!

Energy costs

Certain types of solar panels require a large amount of energy to produce. They all pay for themselves eventually, within a fraction of their lifetime, but the initial energy cost is quite substantial. There are two quantities that we’ll consider for the energy cost. First is (EROI), which is the amount of energy that the panel generates over its lifetime divided by the initial energy investment. The second measure is the (EPBT), which is the amount of time it takes for the panel to recover the initial energy investment. Despite the usage of the words “investment” and “payback,” these calculations only depend on energy, not monetary investments.

Single crystal silicon, quickly becoming the industry standard, is highly energy intensive. One must first turn sand into metal-grade silicon, then create polysilicon using chemicals and heat, then growth of a large single crystal using high temperature and precise gradients. The formation of polycrystalline panels requires these same steps except a lower temperature casting and crystallization process replacing the last step. Finally wafers are cut to precise thicknesses. It was estimated in 2015 that a typical square meter of single crystal panel takes between 3-9 GJ/m2 (833-2500 kWh/m2), with an average of around 6 GJ/m2 (1667 kWh/m2), while polycrystalline silicon panels take 1.7-6 GJ/m2 (472-1667 kWh/m2) with an average of 3.9 GJ/m2 (1083 kWh/m2). These values have substantial difference across regions and production facilities, and are decreasing.

Another highly problematic aspect of polysilicon production in the current supply chain is that a large amount of the factories that create polysilicon are in the Xinjiang province of China, where allegations of forced labor among Uyghur Muslim and other minority communities are frequent. It is extremely important that any part of the transition to renewable energy be free of atrocities like forced labor. Close attention should be paid to all aspects of the supply chain. In addition to ensuring ethical treatment of workers, some have suggested that a global minimum wage could be an additional safeguard against exploitative labor practices.

Thin-film technologies require much less energy to make, because less material is involved, and sometimes less processing. Cadmium telluride has been calculated to have less than one third of the embedded energy than single crystal silicon panels, on a per square meter basis. Amorphous silicon panels are also more energy efficient.

Thin film panels have not been part of the rapid expansion of solar power in the last decade; the most growth has by far been in single crystal silicon. Production is currently 95% single- or poly-crystalline silicon, with single crystal silicon amounting to 84% of the total production in 2020, up from 66% in 2019. As the world turns to cleaner electricity, this will be less of an issue, but currently the frequent use of coal power for energy to make the panels is problematic.

Energy payback times depend to some extent on the calculation method. If the time to replace the primary energy in production is used, and the electricity generated is translated back into a primary energy assuming the typical grid efficiency in the area of installation (meaning shorter payback times when panels are installed in areas with low grid efficiency), the EPBT varies from as little as 0.44 years in India to 1.42 years in Canada. Payback times have been decreasing in recent years, in part because less material is being used in the panels, with wafer thicknesses dropping from 400 microns in 1990 to 175 microns in the 2010s, and because efficiencies have been rising, from 15% in 2006 to 18% in 2018.

Monetary costs

The price of solar cells has dropped precipitously over the last 10 years. The install costs for rooftop solar in the US is currently around $2.70 per watt of nameplate capacity. This doesn’t include any subsidies, which are often present; the current federal tax credit knocks 26% off the price. The “soft costs” which refer to the labor, permitting, and other aspects that do not include the panels or rigging themselves, are now larger than the hard costs. This is reflective of the fact that there is a large potential for job creation within the solar industry. We need a lot of workers to install all those panels.

The installation costs for utility scale systems are significantly less, around $1/watt, but there are many advantages to local power generation which we’ll discuss in the chapter after next, including the potential for microgrids, which build resilience into the electrical system.


Solar United Neighbors is a non-profit that helps to organize communities into local cooperatives that allow more residents to install solar energy in their neighborhood.