Summary: This in-depth article explains
- the working principle of photovoltaic cells,
- important performance parameters,
- different generations based on different semiconductor material systems and fabrication techniques,
- special PV cell types such as multi-junction and bifacial cells,
- and various technical details such as surface passivation and texturing techniques.
Photovoltaic cells are semiconductor devices that can generate electrical energy based on energy of light that they absorb. They are also often called solar cells because their primary use is to generate electricity specifically from sunlight, but there are few applications where other light is used; for example, for power over fiber one usually uses laser light.
For solar power generation, one uses solar power modules containing multiple cells, well encapsulated for protection against various environmental influences such as humidity, dirt or hail. Conversion efficiencies well above 20% are routinely achieved with modern technology, resulting in about 200 W of electric power per square meter for full sun illumination. Due to dramatically decreasing fabrication cost, combined with increasing conversion efficiencies, photovoltaics is already one of the cheapest options for power generation.
Working Principle of Photovoltaic Cells
A photovoltaic cell essentially consists of a large planar p–n junction, i.e., a region of contact between layers of n- and p-doped semiconductor material, where both layers are electrically contacted (see below). The junction extends over the entire active area of the device. One exploits the photovoltaic effect, which is closely related to the internal photoelectric effect. The following is a qualitative explanation of how this works, which should be understandable even without detailed knowledge of semiconductor physics:
- When light with a photon energy above the band gap energy is absorbed in a semiconductor, free carriers can be generated: electrons in the conduction band and holes in the valence band. While this is simple, we need to understand how one can utilize that effect to supply an external current in combination with a voltage, such that electric power is provided by the cell.
- We first consider the situation without any light input, where we obtain an equilibrium state with zero voltage across the connected wires:
- Directed diffusion of electrons and holes is driven by the carrier density gradient at the junction (and also sensitive to an internal electric field, as will be discussed shortly). In particular, the high density of conduction band electrons in the n layer drives electron diffusion toward the p side, which has almost no conduction band electrons intrinsically. As a result, the p layer becomes negatively charged, while the n layer becomes positively charged. These charges build up until the resulting electric field, which pushes the carriers in opposite directions, balances further carrier migration.
- Despite the electric field around the junction (in the resulting depletion region), there is still zero voltage across the connected wires. This is because the different semiconductors at the two metal–semiconductor interfaces have different work functions. (Without this effect, you would get a voltage, and the device would be a perpetuum mobile). We do not need to consider these contact potentials here.
- There is also a certain rate of thermally induced carrier generation. Due to the internal electric field, the generated carriers are separated: electrons are driven towards the n-layer and holes in the opposite direction. However, at zero voltage across the wires, the resulting generation current is just compensated by the diffusion current. (Note that it is common to consider two hypothetical currents that cancel each other out.
- Now consider the situation with incident light, much of which is absorbed in the depletion region. We also assume that no current is allowed to flow through the connected wires (open circuit):
- Each absorbed photon generates an additional electron–hole pair, and this effect adds to the thermally induced generation current. This unbalances the diffusion current, causing a current to flow for a short time, which modifies the electric charges and thus the internal electric field, until the two currents cancel each other out again. The result is a non-zero voltage between the wires: the p-contact becomes positive. For strong illumination of a silicon-based solar cell, this voltage is a little more than 0.7 V. (For other solar cell materials, it can be different, mainly due to different band gap energies.)
- So far, no electric current is generated because we do not allow an external current to flow. The carriers produced can only diffuse back and eventually recombine, producing luminescent radiation which partly leaves the cell, and some heat.
- If we now connect a load to the wires of the cell, the generated voltage drives a current. We first consider the most extreme case of zero load resistance – effectively a short circuit –, where we get the maximum current (the short-circuit current), with almost one electron per absorbed photon (since in high quality devices only a small fraction of the carriers are lost), which means a high quantum efficiency. Recombination of carriers is now largely suppressed. However, we still get no delivered power, since that power is the product of voltage and current, and the former is zero. Instead, the cell is just heated.
- With a finite load resistance, we obtain a lower current, but now in combination with a non-zero voltage, resulting in some electrical power delivered to the load. Even if the voltage is dropping only around 10 to 20% below the open-circuit voltage, carrier recombination is strongly reduced, and the current is not far below the short-circuit current.
- For optimum power generation, i.e., the maximum possible product of voltage and current, the load must be chosen suitably, resulting in a current just a little less than the short-circuit current and a voltage which is somewhat less than the open-circuit voltage. The coordinates of that maximum power point (MPP) depend on the intensity of the illumination. In practice, one often uses a special electronic device (an MPP tracker) to accurately stay at that point under all conditions – usually, not for a single photovoltaic cell, but for one or several modules, each one containing many cells.
The diagram above shows the resulting I/U characteristics of an example case of a silicon PV cell. Several details can be seen:
- The open-circuit voltage (zero current, i.e., on the horizontal coordinate axis) is slightly above 0.7 V. (Typical values are between 0.6 V and 0.7 V.)
- The short-circuit current (at zero voltage) reaches up to 9.75 A. With increasing voltage, this current decreases only slightly at first, but then decreases more rapidly.
- The maximum power point corresponds to a significantly reduced voltage, but a current not much lower than the short-circuit current.
- For a sufficiently high external voltage, the current would go into negative territory. (Such a current would normally be called positive for an ordinary rectifier diode). This could happen, for example, if several cells with very different light levels are connected in parallel and no external current is allowed to flow: the strongly lit cells could then drive a current through the weakly lit cells. This is unlikely to happen when a module is operated at its maximum power point, which has a slightly lower voltage.
Useful photon absorption is only possible for photons with energy above the band gap energy; longer wavelength light cannot be absorbed and therefore cannot be used. On the other hand, light with a photon energy far above the band gap energy can be efficiently absorbed, but still not very efficiently utilized, since much of the absorbed energy is lost through thermalization: while a high-energy photon is still likely to contribute one electron to the current generated, that electron cannot contribute more to the electrical power generated than the lower-energy photons, since the cell voltage (quantifying the energy per electron) is limited by the band gap energy. The most efficient light has a photon energy slightly above the band gap energy, allowing efficient absorption in the junction.
It can be seen that efficient utilization of sunlight, which is inherently polychromatic, is more difficult than for narrowband (quasi-monochromatic light) light. There is a trade-off regarding the band gap energy: it should be small enough to allow absorption of a substantial fraction of sunlight, but large enough to allow a reasonably high cell voltage. For any given band gap energy of a single-junction photovoltaic cell (and for a standardized sunlight spectrum after transmission through the atmosphere), one can calculate the Shockley–Queisser limit for the theoretically achievable conversion efficiency , which is e.g. about 30% for 1.1 eV, the value of silicon. The optimal trade-off could be achieved for a band gap energy of 1.34 eV, which allows a conversion efficiency of 33.7%. However, this trade-off can be avoided by using a multi-junction cell (see below), where different parts of the spectrum are used with different junctions.
For several reasons, photovoltaic cells operate less efficiently at high temperatures:
- The band gap energy is reduced. While this can lead to more efficient light absorption, it also reduces the cell voltage and thus the energy delivered per electron.
- Carrier lifetime can be reduced, and this reduces the current obtained, as more carriers are lost.
- The increased thermal generation of carriers alters the electric field profile in a detrimental way.
- The electrical resistance, e.g. of metal contacts, may increase, leading to a further drop in cell voltage.
In addition, cell aging can be accelerated at high temperatures. However, the severity of all these effects is highly dependent on the semiconductor material used and other details.
Photodiodes have a similar semiconductor structure and can, in principle, also generate power through the photovoltaic effect. However, they are not optimized for efficient power generation. Some of them are used at least in photovoltaic mode for detection purposes.
Important Performance Parameters of PV Cells
The following are the most important performance parameters of a photovoltaic cell:
- The open-circuit voltage for a given material system and standard illumination conditions (see below) can be an indication of cell quality.
- The short circuit current gives an indication of the carrier collection efficiency (for a given cell area and illumination level).
- The generated electric power at the optimum power point, again for standard illumination conditions, is the most relevant parameter and determines the power conversion efficiency of the cell.
- This power divided by the product of the open-circuit voltage and the short-circuit current is called the fill factor. Depending on the material system, it can be around 0.85 or closer to 0.9.
- The temperature coefficient of conversion efficiency quantifies the loss of output power due to increased temperature. It is typically around −0.3%/K to −0.5%/K for crystalline silicon cells, which means a significant loss of efficiency when the temperature rises by several tens of Kelvin, e.g. under intense illumination in a hot environment. (Note that the coefficient actually applies to the output power, not the conversion efficiency; for example, an initial conversion efficiency of 20% might be reduced to 20% · (1 − 0.4%) = 19.92% for a 1 K temperature rise, or actually drop to 19.6%). Some thin-film cells have a much lower temperature coefficient, such as −0.2%/K.
Standard illumination conditions for solar cell testing include
- homogeneous irradiation with an irradiance level of 1000 W/m2, which approximates peak sunlight conditions in many countries, and a standardized optical spectrum ranging from the infrared to the ultraviolet region
- clear atmospheric conditions
- an “air mass” of 1.5, which means a 1.5 times longer path length of sunlight through the atmosphere compared to overhead sunlight
- 25 °C cell temperature
The stated peak powers of PV cells or modules apply to those standard conditions, but may sometimes be significantly exceeded under favorable conditions, such as bright direct sunlight combined with reflected light from snow.
In addition, measurements can be made under low light conditions. The generated power is in most cases roughly proportional to the irradiance between 100 and over 1000 W/m2, i.e., the conversion efficiency drops only moderately for decreasing illumination. For very weak illumination, the efficiency can be reduced more rapidly: the voltage then drops substantially. Additional (often more substantial) losses in performance result in that regime from a reduced efficiency of the inverter used (see below).
Under very intense irradiation, especially in the context of concentrated photovoltaics (see below), efficiency may also decrease due to increased cell temperature unless very effective cooling is used.
Standardized test equipment is available to accurately measure cell performance, which is particularly important for evaluating the results of modified cell designs. Of course, real-world performance can vary significantly from standard conditions.
Note that durability of performance is another very important parameter for practical applications. While mature cell technologies routinely achieve very high durability, with performance still quite good after 20 years of operation, some new technologies produce much less durable results and thus need to be improved. For some cell types, durability requires protection from external influences such as moisture.
Theoretical Performance Limits
Although photovoltaics does not rely on a Carnot process, it is of course also fundamentally limited in efficiency by thermodynamics. Note that, in principle, any PV cell could not work if it were in thermal equilibrium with the incoming radiation – which in the case of sunlight means an operating temperature of thousands of Kelvin. Under such conditions (far from real operating conditions), power generation would violate the second law of thermodynamics: one would have a perpetual motion machine of the second kind. For power generation, i.e. the generation of energy with zero entropy based on incoming radiation with finite entropy, it is inevitable that additional entropy is generated, and this happens through the generation of waste heat at a lower operating temperature. (Note that the entropy generated there is the amount of heat energy divided by the temperature, and this amount must be greater than the entropy loss of the sunlight).
A detailed theoretical analysis is rather involved. Early work e.g. by L. Landau  already clarified the fundamental thermodynamics of radiation. Some later core results on energy conversion, taken from a book chapter by A. Luque and A. Martí , are briefly summarized in the following:
- The ultimate limit is called the Carnot efficiency based on the temperature of the Sun's surface (determining the spectrum of solar backblody radiation), and would require a completely reversible process, i.e., without any net entropy generation; that would lead to conversion efficiencies around 95%. This is result is not based on the analysis of any concrete device design; it is obtained with purely thermodynamic considerations.
- A first caveat is that Carnot efficiency does not take into account that some blackbody radiation is inevitably emitted by the photovoltaic device; only the net received radiation power is considered, while for practical efficiency considerations the total received power from the sun (which is somewhat higher) is relevant. Taking into account this problem, one arrives at the Landsberg efficiency  which is a bit lower, around 93%. Still, not even a theoretical cell design is known which (with ideal materials) could reach that thermodynamically possible performance.
- Theoretically conceivable photovoltaic converters (multi-junction cells based on ideal materials, e.g. with zero light reflection, complete light absorption, zero conduction losses, etc.) could reach about 87%. The utilized physical processes are not completely reversible, i.e., generate some net entropy – although of course far less than e.g. if the solar radiation would simply be absorbed to generate low-temperature heat.
In practice, that latter limit is far from being reached so far; experimentally achieved power conversion efficiencies have always remained at least somewhat below 50%. A multitude of practical problems, including light reflection and incomplete absorption, finite minority carrier lifetimes, resistive losses and others, explain that discrepancy. (Refined numerical models can quite accurately quantify the different loss mechanisms.) It is hard to predict to which extent practical devices will in the future reach efficiencies approaching the theoretical limits.
Generations of Photovoltaic Cells
Various types of photovoltaic cells have been intensively developed over many decades. Before discussing some of the technologies in more detail, we will give a brief overview of the so-called generations of photovoltaic technology:
The development started with silicon cells because silicon is by far the most abundant semiconductor material and has a suitable (albeit somewhat low) band gap energy. Both monocrystalline and polycrystalline silicon have been used (and are still dominating today).
Early on, gallium arsenide (GaAs) was also used. With its higher band gap energy than silicon, it can be more efficient at converting sunlight into electricity. However, the high cost of manufacturing made this technology viable only in certain applications, such as space satellites.
One attempt to significantly reduce manufacturing costs has focused on amorphous silicon (a-Si) cells, in which a thin film of amorphous silicon is deposited on a surface. This avoids the high cost of growing large monocrystalline silicon wafers and also requires less material. Less degradation of performance at low light levels and/or high operating temperatures are additional advantages. However, the lower efficiency is a serious problem: it increases the cost of additional components due to the larger area that must be covered for a given annual energy production.
Another strategy being pursued for second-generation cells is the manufacture of thin-film cadmium telluride (CdTe) and cadmium sulfide (CdS) cells. Due to the high absorption coefficient, very thin films can be used, leading to very low material and energy consumption in production. However, the toxicity of cadmium is a problem, and the performance is significantly lower than that of silicon cells.
Copper Indium Gallium Selenide (CIGS) is another material system of this generation that achieves similar performance to silicon while using less toxic materials.
The third generation targeted several new material systems:
- organic and polymeric materials (OSC)
- dye-sensitized solar cells (DSSC)
- perovskite cells
- quantum dots cells
In addition, multi-junction solar cells have been developed to achieve significantly higher efficiencies than silicon cells.
Another approach is cells in which the semiconductor material has an additional intermediate band, allowing the use of longer wavelength radiation despite a high band gap energy.
The fourth generation focuses on innovative materials and concepts such as
- metal oxides
- metal nanoparticles
- nanostructured surfaces for increased absorption efficiency
- organic-based nanomaterials like graphene, carbon nanotubes, and graphene derivatives
- bio-hybrid material combinations
- dye-sensitized cells
- quantum dots
- multi-junction cells
- flexible cells for new application areas
Overview on Photovoltaic Material Systems
For a variety of reasons, silicon cells have a clearly dominant market share in photovoltaics:
- Silicon is one of the most abundant elements on Earth.
- It is non-toxic.
- There is a huge body of technological experience from microelectronics technology.
- Silicon cell technology has been highly optimized over many years with large investments. This has led to a very high level of maturity, with not only good efficiencies (now often well over 20%) and long lifetimes (well over 20 years), but also amazing progress towards lower and lower manufacturing costs, despite the sophisticated technology required. The key to success is to build production facilities that achieve very high annual throughput of optimized cells.
However, silicon has some non-ideal properties. While the band gap energy is somewhat lower than ideal for solar conversion, it can be regarded as reasonably suitable. The primary problem is that the indirect band gap of silicon leads to much weaker absorption (except for photon energies far above the band gap energy) because additional phonons must be involved for momentum conservation. That has some negative consequences:
- Substantially thicker layers of silicon are required for efficient light absorption compared to direct band gap materials such as gallium arsenide.
- This also implies longer path lengths for carriers, and thus a need to have a very low density of defects and impurities for having a long enough minority carrier lifetime. However, highly purified crystalline material is expensive to grow.
- Obviously, the combination of requiring relatively large amounts of material which at the same time needs to be highly purified is a problem concerning low-cost production.
In the early years, solar cell production at least benefited from the fact that significant amounts of high-purity silicon were available as rejected (not sufficiently high quality) material from the microelectronics industry. However, roughly from 1995 on, this resource was eventually exhausted as production volumes increased, and solar-grade silicon (with significantly lower quality than electronic-grade silicon) had to be produced specifically for photovoltaics. The remaining options for cost reduction were (a) to improve solar-grade silicon production methods and (b) to minimize the amount of silicon required. Both approaches have been pursued quite successfully.
Monocrystalline and Polycrystalline Silicon Cells
Silicon is used in both monocrystalline and polycrystalline forms, and in this section we concentrate on silicon in bulk form, produced either as wafers (for monocrystalline material) or polycrystalline ingots. The former has a uniform crystal lattice extending over an entire PV cell, while the latter has domains of quite limited size. Various detrimental processes at the (easily visible) domain boundaries and in particular at crystal dislocations cause a loss of efficiency. In any case, the material must be of high chemical purity.
Polycrystalline material is often produced in the form of square cast ingots, as opposed to cylindrical monocrystals. Wafers can be obtained from ingots by sawing, which unfortunately results in a significant loss of material. To solve this problem, there are also attempts to make ribbon silicon by carefully drawing flat thin films from molten silicon, which does not require sawing, but the efficiencies achieved are lower.
While the highest efficiencies and best durability are achieved with monocrystalline silicon, it is more expensive to produce, mainly due to the slow growth of large monocrystalline silicon wafers. As a result, polycrystalline silicon has been the most widely used material for some time, and efforts are still underway to achieve higher efficiencies with polycrystalline silicon despite lossy processes at the domain boundaries. However, the market share of monocrystalline cells has been growing strongly since around 2017 due to further advances in low-cost production.
It is also possible to introduce a monocrystalline seed crystal into a polycrystalline casting chamber. With this approach, one can obtain a large monocrystalline region around the seed crystal, while the outer regions remain polycrystalline. By slicing the material, one obtains both a higher-grade monocrystalline material and a lower-grade polycrystalline output. The latter can also be recycled to produce only monocrystalline material.
The amount of silicon required for the cells has been substantially reduced over the years. For example, thinner wafers are now used (typically below 200 μm) and less material is lost through optimized diamond wire sawing.
Thin-film Cells with Amorphous Silicon
Traditional silicon cells require a relatively thick semiconductor layer, since the absorption length in silicon is quite long (tens to hundreds of microns) for the longer wavelengths. This is a consequence of the indirect band gap band structure, which is also relatively small (1.1 eV), and has the obvious disadvantage of requiring large amounts of the expensive material.
However, the second generation of solar cells introduced thin-film cells based on amorphous silicon (a-Si), which has a much higher light absorption due to its more favorable electronic band structure with a direct band gap. Although the band gap energy is quite large (about 1.6 to 1.8 eV), the absorption coefficient increases more rapidly with increasing photon energy, and efficient absorption is achieved even within one or two micrometers over a substantial spectral range. It can be further improved by reducing the band gap energy through adding some germanium (obtaining a-SiGe). It is even possible to make multi-junction cells based on a-Si for one junction and a-SiGe for another one or two junctions.
Pure amorphous silicon would have an extremely low carrier lifetime due to dangling bonds. This problem can be greatly mitigated by hydrogenating the material (resulting in so-called a-Si:H, with e.g. 10 to 15% hydrogen content in terms of atomic density). The defect density in this material is still much higher than in crystalline silicon, but is now manageable. The cell efficiency with sunlight is much lower, e.g. 14%, compared to about 22 to 26% with (about 100 times thicker) crystalline silicon. The efficiency drops somewhat during the first few hundred hours of intense illumination (Staebler–Wronski effect), but is less reduced by higher temperatures than with crystalline silicon.
A crucial advantage is that thin amorphous silicon films can be produced with deposition methods, performed at moderate temperatures. Production can thus potentially be far cheaper than with wafer-based methods.
Amorphous cells typically do not have a simple p–n junction, but rather a p–i–n structure containing an intrinsic (undoped) region between p- and n-doped layers. Preferably, the overall thickness is reduced to well below a micron and the absorption efficiency is improved with trapping schemes based on surface texturing. In addition, up to three junctions can be stacked, with the lower ones using the residual light transmitted by the upper ones. Electrically, the junctions are connected in series, resulting in a higher output voltage.
The open-circuit voltage of single-junction a-Si cells is slightly higher for amorphous silicon, but this is of little practical concern.
The aesthetic appearance of thin-film cells, with their typical anthracite (dark gray to black) color, is often perceived as better than that of traditional silicon cells, especially polycrystalline ones. This can be relevant for building integration, for example.
Despite the mentioned advantages, the already small market share of amorphous silicon cells has in recent years declined as more progress has been made with crystalline cells.
Thin-film Cells with Crystalline Silicon
There are also various developments aiming at semiconductor structures with a thickness between few micrometers and a few tens of micrometers, usually made of polycrystalline silicon with a wide range of grain sizes. In this context, no wafer-based methods are meant, but those using deposition of silicon on some substrate. Different substrate materials are used, which require different deposition temperatures. (For example, excessive diffusion of substrate material into the silicon must be avoided.) Relatively low temperatures (few hundreds dC) would be preferable in terms of cost and a greater choice of substrate materials, but larger grain sizes are achieved at higher temperatures, and those lead to higher cell efficiencies. For those, it is also vital to minimize surface recombination losses.
The low absorption of crystalline silicon (resulting from its indirect band gap) would naturally lead to poor absorption efficiency in thin-film cells, but the problem can be mitigated with various means, including front surface texturing and also diffuse reflection of light at the back side. Note that light scattered back from there at high enough angles against the surface normal can be trapped quite well and has a good chance of being eventually absorbed.
Compared with bulk crystalline silicon cells, the reached efficiencies have so far remained quite low, making it difficult for thin-film cells to compete.
Cells with Other Inorganic Semiconductors
Due to some non-ideal properties of silicon – in particular, its indirect band gap with larger than ideal band gap energy –, several other inorganic semiconductors have also been extensively studied. These are briefly discussed below.
Gallium arsenide (GaAs), a III–V semiconductor well known in electronics, has long been used in photovoltaic cells. With its direct band gap of moderate size (1.42 eV), it allows cell efficiencies above 30%. In addition, it is quite durable and can withstand high operating temperatures, low light conditions and irradiation. These characteristics make it ideal for space applications.
On the other hand, the manufacturing cost is quite high due to the expensive material and manufacturing details, and because much less resources have been invested compared to silicon cell technology. Note also that gallium resources are limited.
In addition to pure GaAs cells, there are multi-junction cells (below) that use GaAs for one of the junctions.
Cadmium tellurite is well suited for thin-film cell fabrication because its light absorption is quite efficient. Its band gap energy of 1.45 eV is quite suitable. However, the high toxicity of cadmium is definitely a drawback, although this is mitigated by the relatively small film thickness required and the ability to recycle or at least safely dispose of the material after use. Another challenge is the scarcity of tellurium.
The manufacturing process can be relatively simple and easily scaled to large production volumes, resulting in low production costs.
Cadmium is also often used for a thin cadmium sulfide buffer layer, reducing surface recombination loss.
Achievable cell efficiencies can be a little over 20%, which is slightly lower than advanced silicon cells, but reasonably good.
Durability can be quite high, with good performance even after 20 years of operation, and typical warranty periods of that order. High temperatures, humidity and UV radiation can accelerate degradation.
Copper Indium Gallium Selenide
Copper indium gallium selenide (CIGS) is an interesting semiconductor material that is also highly absorbent, making it suitable for thin-film cells, although its band gap energy of 1.15 eV is somewhat low.
CIGS cells can even be made flexible, which expands the range of possible applications.
The durability of CIGS cells can also be quite high, similar to that of silicon cells.
Quantum Dot Cells
There is ongoing research and development to realize photovoltaic cells based on quantum dots, i.e. nanoscopic structures with special electronic properties. These can be made from a wide range of inorganic semiconductors. Quantum dots have some interesting properties, such as the tunability of their band gap energy and the potential for multiple exciton generation, which could hopefully increase the conversion efficiency. In addition, some types of colloidal quantum dots can be fabricated using inexpensive solution processing methods.
In theory, quantum dot cells could be highly efficient, especially for use in concentrated photovoltaics. In practice, however, the efficiencies of conventional PV cells have not yet been achieved, in part because of the challenge of achieving effective charge transport in optically thick arrays of quantum dots. Further research is needed.
Organic photovoltaics (OPV) is based on the idea of using organic materials – either small molecules or organic polymers – that can be processed quickly using inexpensive (fast) techniques. In contrast, inorganic semiconductors are usually expensive to obtain and process. An interesting aspect is that thin, lightweight and flexible organic PV cells can be made, which can be integrated into building materials, for example. Such cells can also be semi-transparent (e.g. for shaded windows) and colored.
Several such technologies have been developed and are briefly described below. However, none has yet reached a level of maturity sufficient for large-scale use in photovoltaic power generation. A common problem is too fast degradation, e.g. under the influence of oxygen, whose diffusion into the devices through thin protective layers is difficult to prevent.
Some typical organic materials are briefly discussed below:
Small Molecule Organic Photovoltaics
Some small organic molecules, such as phthalocyanines and fullerene derivatives like PCBM, can be processed using vacuum deposition techniques. Cells based on this approach can be quite efficient, but production using vacuum deposition remains costly.
Polymer Organic Photovoltaics
The active layer of a PV cell can be made of a conductive organic polymer. Such materials can be subjected to a potentially low-cost solution-based process such as spin coating or printing, and can be used to produce flexible and/or printable solar cells. The efficiencies achieved are still modest, but can be improved with further development.
Dye-sensitized solar cells (DSSC) are a type of thin-film cell in which the semiconductor structure contains a photo-sensitized anode, a cathode, and an electrolyte between them. This configuration is effectively an electrochemical cell. Here, the organic dye, typically in contact with an n-doped titanium dioxide layer, absorbs sunlight and transfers electrons to the titanium dioxide. (Carrier generation and collection thus take place in two separate materials, unlike in a conventional solar cell). After passing through a transparent electrode and the connected load, the electrons are transferred back to the dye through the counter electrode and the electrolyte. For sufficiently efficient light absorption, the surface of the titanium dioxide layer to which the dye is attached is nanostructured to obtain a large surface area. Due to the significant role of Michael Grätzel in the development of this technology, these cells are often referred to as Grätzel cells.
This technology has exciting aspects such as potentially cheap mass production without the need for expensive, scarce or toxic materials. An interesting detail is that a thin layer of conductive plastic can be used on the front instead of a glass plate, as long as it provides sufficient mechanical protection. However, even after several decades of development, significant challenges remain. The conversion efficiency is limited to about 10–15% and, more importantly, the degradation during operation is relatively fast.
Organic–inorganic Hybrid Cells
Perovskites are materials with a perovskite crystal structure. For photovoltaics, mostly some organic-inorganic hybrid compounds like CH3NH3PbI3 are investigated (although there are also purely inorganic compounds like CsPbI3), which have several promising aspects:
- The high absorption coefficients allow the realization of thin-film designs that require relatively little semiconductor material.
- Conversion efficiencies have improved rapidly and in some cases are already above 25%. Theoretically, around 30% may be possible.
- Production can be relatively inexpensive, potentially much lower than for silicon cells. A wide range of substrates are suitable, including curved and flexible ones.
Problems arise from toxicity (e.g. the compounds often contain lead) and limited durability, with adverse effects from moisture and high temperatures. Research is ongoing to find less toxic compounds that are still efficient and hopefully more durable.
Perovskites can also be used in multi-junction cells (see below).
Multi-junction (Tandem) Cells
As explained above, for a single-junction photovoltaic cell, there is a fundamental trade-off between efficient light absorption (requiring a small band gap energy) and high cell voltage (requiring a larger band gap). This problem can be solved with the principle of the multi-junction cell. Here, two or more junctions with different band gap energies are stacked on top of each other, with the incident light first entering the junction with the highest band gap energy. The photons with the highest energy are largely absorbed and contribute most to the total cell voltage. Lower energy photons are largely transmitted to the next lower band gap energy junction, allowing absorption at longer wavelengths. Even with only two junctions, the energy of the broadband sunlight can be harvested with much higher efficiency. With three or more junctions, the theoretical efficiency limit is pushed even higher, but with diminishing returns. It could also be helpful to have spectrally optimized reflectors between the different junctions, although that is probably difficult to realize.
Of course, this operating principle would not work for narrow-band light, as commonly used in power over fiber.
Alternative names for multi-junction cells are tandem cells and multi-gap systems.
The challenge is to design and manufacture multi-junction cells that actually achieve much higher efficiencies without being much more expensive to manufacture than the simpler single-junction cells. At the same time, other properties such as long lifetime should be maintained.
The choice of material combinations must take into account multiple aspects:
- Appropriate band gap energies are desirable, and the internal efficiencies of all junctions should be high.
- There must be ways to grow the films for the different junctions on top of each other while avoiding seriously increased defect densities due to the lattice mismatch: generally, different semiconductor materials have different lattice constants, and a significant lattice mismatch easily leads to stronger growth imperfections. Apart from choosing material combinations with small lattice mismatch, one may work with buffer layers mitigating the problem.
- The different junctions are connected in series (often using lattice-matched tunnel layers between them), so that the same current must flow through all of them. (Separately connecting the junctions would hardly be practical.) That reduces efficiency if one cell cannot provide as much current as the other one(s), e.g. if a lower-lying cell receives too little light. Indeed, this aspect introduces further constraints on the choice of materials concerning their band gap energies. For example, it would not work well to have an only slightly lower band gap energy for the lowest junction.
A typical material choice is indium gallium phosphide (InGa, 1.8–1.9 eV band gap) for the top layer, followed by indium gallium arsenide (InGaAs, 1.4 eV) and germanium (Ge, 0.67 eV).
Monocrystalline layers are typically grown by metal organic chemical vapor deposition (MOCVD). The low growth rate is acceptable given the thin films required. There are also multi-junction cells using polycrystalline or amorphous materials and faster methods such as sputtering, chemical vapor deposition or electrodeposition.
It is also possible, for example, to use two separate single-junction cells based on different materials and illuminate one with the light transmitted by the other. This is technically simpler than integrating two junctions into one cell and avoids, for example, lattice mismatch problems. However, the overall manufacturing process can still be more expensive, and additional light is lost at the additional optical interfaces between the cells. For these reasons, monolithically integrated multi-junction cells are the preferred solution.
Bifacial cells are designed to harvest sunlight from both sides, whereas the usual situation is that only light from the top can be used. Bifacial cells must have a transparent back layer with a contacting method that blocks as little light as possible. They are expected to be slightly more expensive to manufacture.
Solar modules based on bifacial cells are suitable for mounting in a way that allows reflected or backscattered sunlight to reach the back surface. This is possible in ground-mounted solar farms, especially if the ground has a high albedo, but also for some panels mounted on buildings. Another possibility is to use them in high latitude regions with frequent snow cover.
Cells for Concentrated Photovoltaics
For operation with direct sunlight, the power density of the incident light is limited to the order of 1 kW/m2, resulting in an output power of a few hundred watts per square meter. One way to make better use of expensive semiconductor materials is to operate cells with concentrated sunlight, obtained with suitable optics, typically curved mirrors or plastic Fresnel lenses. Depending on the cycle type, concentration factors of up to several hundred, perhaps even of the order of 1000, are realistic. It is then worthwhile to use specially optimized cells:
- They should have a particularly high conversion efficiencies. Usually, one uses cells with III-V semiconductors, possibly with a multi-junction design for further increased performance.
- The cells must work well under high irradiance conditions, implying high electric current densities and relatively high operating temperatures.
On the other hand, cell product cost is less critical, as only small amounts are needed for a given generation power.
Depending on the degree of light concentration (in one or two dimensions), more or less sophisticated means for effective cooling of the cells are required to avoid overheating. Somewhat higher operation temperatures are hard to avoid, and these, together with increased resistive power losses, unfortunately tend to offset the theoretically higher cell efficiencies resulting from thermodynamic considerations.
During the day, the devices need to moved to achieve optimal focusing of the sunlight with its variable incidence direction. Dual-axis tracking systems are often required. It is challenge to realize that movement and tracking system such robust operation over many years is achieved in combination with reasonably low cost and maintenance.
There is an alternative technical approach to solar energy concentration not necessarily requiring moving parts: one can use luminescent solar concentrators. These contain a layer of a dye that can absorb sunlight and then generate fluorescent light that is largely guided (with a waveguide structure) to photovoltaic cells at the edges. Although theoretically this technology could work in a fairly efficient way,, the conversion efficiencies achieved so far remain quite low compared to conventional devices. Substantially better materials and perhaps also improved structures would need to be developed.
So far, concentrated photovoltaics has not gained much traction for large-scale application, despite impressive progress in usable cells. Record efficiencies of around 47% have been achieved with multi-junction designs, about twice that of conventional PV cells, but the more complicated application of concentrating optics and tracking motion seems to be hindering a market breakthrough, while luminescent solar concentrators simply do not reach sufficiently high efficiencies.
Cells for Power over Fiber
While most photovoltaic cells are used for solar power generation, some are used for Power over Fiber (PoF), i.e. to deliver power in the form of light through an optical fiber (typically a multimode fiber). The requirements for the cell are very different from those for solar power generation:
- An active area of a few square millimeters is sufficient, unless you want to use a larger area for easier heat dissipation.
- The delivered laser light is quite narrowband. If the laser wavelength is in a suitable range, a much higher conversion efficiency is possible than with sunlight. A suitable combination is a laser diode emitting around 750–980 nm and a photovoltaic cell based on silicon or GaAs, or possibly InGaAs. While GaAs-based cells are the most efficient, they are also much more expensive.
- If a long transmission distance (possibly several kilometers) is required, a relatively longer optical wavelength such as 1310 nm or 1550 nm is usually used; this drastically reduces losses due to Rayleigh scattering in the fiber. Very special cells based on SiGe or quantum dots can then be used.
Cells for Thermophotovoltaic Generators
A small niche application for special photovoltaic cells is the use in thermophotovoltaic generators, where instead of sunlight one uses thermal radiation from a hot body, typically with a temperature between 1000 °C and 2000 °C. In effect, this technology converts high temperature heat into electricity using a photovoltaic cell, which is operated at a much lower temperature (usually not too far above room temperature). The thermal radiation used is filtered to obtain a spectrum suitable for the photovoltaic cell. The thermal radiator is usually heated with a fuel to generate power on demand, or sometimes a radioisotope heat source for space missions. (In principle, concentrated solar radiation could be used, but this approach would be less attractive than traditional photovoltaics).
The requirements on such cells differ in various respects from those for using sunlight:
- As the average photon energy is substantially smaller, a small band gap energy of the order of 0.6 eV is appropriate. A common choice of semiconductor material is gallium antimonide (GaSb).
- The power density is much higher, similar as for concentrated photovoltaics, as one utilizes radiation from a very nearby emitter. This results in high current densities, requiring low resistance interconnects.
Effective cooling of the photovoltaic cells is also required to avoid a substantial reduction of conversion efficiency. So far, efficiencies of around 15% have been achieved. This is somewhat less than could be achieved with a combustion engine, for example, but thermophotovoltaic generators have various advantages over these, including very low levels of noise and vibrations, and good scalability, e.g. for converting relatively low powers of a few hundred watts, although multi-kilowatt output powers are also possible.
Further Technical Details
To achieve the highest efficiency, the recombination of carriers must be minimized, especially on the front surface. This is done by coating the surface with a suitable material:
- A traditional method is to use silicon oxide (SiO2, silica), which is obtained by thermal oxidation, i.e. heating the cell in an oxygen-rich atmosphere.
- Silicon nitride, deposited by plasma-enhanced chemical vapor deposition (PECVD), also provides effective passivation while reducing losses due to light reflection.
- Aluminum oxide (Al2O3) is another option. By introducing a fixed negative charge, it can repel electrons from the surface.
- Hafnium oxide (HfO2) and some other dielectrics with relatively high electrical susceptibility (high-k dielectrics) are also being explored.
Some advanced cell designs also use passivated back surfaces. Contact is then made through laser-drilled holes.
Any light reflected from the front surface is, of course, lost for power generation. One can reduce the reflection losses by microscopically texturing the surface so that incoming light is “trapped”. This texturing can be applied to the front surface of the passivation layer using a laser.
Textured back surfaces are also used. Especially if the incoming light has a nearly perpendicular incidence, which would normally minimize the path length in the cell, the average effective path length can be significantly increased by scattering the light in a wide range of directions.
Contacting of Cells
Photovoltaic cells generate a voltage between their front and back sides. Both sides must be electrically contacted. At least for the front side (and for bifacial cells, the back side as well), this must be done in such a way that the light input is reduced as little as possible. The typical method for conventional silicon cells is to apply a grid of fine “finger” wires connected to larger “bus bars” by screen printing a silver paste onto the front surface. Silver is advantageous because of its high electrical conductivity, so a minimal width of fingers can be used, blocking only a small fraction of the incoming light. For the back side, you can usually cover the whole area, making conductivity less critical, so a cheaper aluminum paste can be used.
Another option is to use a transparent top electrode and sometimes a transparent bottom electrode. This is not common for crystalline silicon cells, but for several other cell types such as amorphous silicon, perovskite, and various organic solar cells. Possible transparent electrode materials include indium tin oxide (ITO), fluorine-doped tin oxide (FTO), and aluminum-doped zinc oxide (AZO). Such electrode layers should result in minimal light sources, but also minimal conduction losses; the latter would lead to a loss of cell voltage under load. Rare materials such as indium should also be avoided.
Any surface passivation layer must be removed before applying the electrodes – for example by laser ablation (see below).
After deposition, the paste must be heated to several hundred degrees Celsius to obtain metal electrodes with good ohmic contact to the silicon. An additional electroplating step may be applied to further improve quality.
Each solar cell then receives wires to connect multiple cells within a solar module (photovoltaic panel).
Use of Laser Material Processing
The use of laser material processing has become essential for cheap mass production of solar cells. It is used in various manufacturing steps such as the following:
- As mentioned above, the surface texturing of the passivation layer can be done with a laser. This is a method of laser surface modification.
- Prior to electrical contacting, the passivation layer can be ablated at the relevant locations.
- The edges of a cell must be isolated to avoid a short circuit. This is achieved by laser ablation of the conductive material around the edges.
- Laser drilling of tiny holes in the semiconductor material can also be used for doping.
- In the manufacture of thin-film cells, laser scribing is used to divide the cell into smaller units connected in series to achieve higher voltage and lower current output.
- Detected defects in a silicon wafer can be removed with a laser, reducing efficiency losses.
In fact, photovoltaic cell manufacturing is a good example of the versatility of laser material processing and its ability to significantly improve quality and productivity. Computer-controlled robotic applications offer great flexibility in process optimization.
While individual solar cells can be used directly in certain devices, solar power is usually generated using solar modules (also called solar panels or photovoltaic panels), which contain multiple photovoltaic cells. Such a module protects the cells, makes them easier to handle and install, and usually has a single electrical output.
Some modules have an integrated inverter, but most modules simply provide DC power, and a central inverter is connected to multiple modules.
|L. Landau, “On the thermodynamics of photoluminescence”, J. Phys. (Moscow) 10, 503 (1946)
|M. A. Weinstein, “Thermodynamic limitation on the conversion of heat into light”, J. Opt. Soc. Am. 50, 597 (1960); https://doi.org/10.1364/JOSA.50.000597
|W. Shockley and H. J. Queisser, “Detailed balance limit of efficiency of p–n junction solar cells”, J. Appl. Phys. 32 (3), 510 (1961); https://doi.org/10.1063/1.1736034
|P. T. Landsberg and G. Tonge, “Thermodynamic energy conversion efficiencies”, J. Appl. Phys. 51 (7), R1 (1980); https://doi.org/10.1063/1.328187
|J. E. Parrott, “Thermodynamics of solar cell efficiency”, Solar Energy Materials and Solar Cells 25 (1-2), 73 (1992); https://doi.org/10.1016/0927-0248(92)90017-J
|A. Martí and G. L. Araújo, “Limiting efficiencies for photovoltaic energy conversion in multigap systems”, Solar Energy Materials and Solar Cells 43 (2), 203 (1996); https://doi.org/10.1016/0927-0248(96)00015-3
|P. T. Landsberg and V. Badescu, “Solar energy conversion: list of efficiencies and some theoretical considerations Part I – Theoretical considerations”, Prog. Quantum Electron. 22 (4), 211 (1998); https://doi.org/10.1016/S0079-6727(98)00012-3
|A. Luque and A. Martí, “Theoretical limits of photovoltaic conversion”, book chapter in “Handbook of Photovoltaic Science and Engineering”, eds. A. Luque and S. Hegedus, Wiley (2003), ISBN 0-471-49196-9
|K. L. Chopra, P. D. Paulson, and V. Dutta, “Thin‐film solar cells: an overview”, Progress in Photovoltaics: Research and applications 12.2‐3, 69 (2004); https://doi.org/10.1002/pip.541
|P. V. Kamat, “Quantum dot solar cells. The next big thing in photovoltaics”, J. Phys. Chem. Lett. 4 (6), 980 (2013); https://doi.org/10.1021/jz400052e
|A. R. Kirmani et al., “Colloidal quantum dot photovoltaics: current progress and path to gigawatt scale enabled by smart manufacturing”, JACS Energy Lett. 5 (9), 3069 (2020); https://doi.org/10.1021/acsenergylett.0c01453
Questions and Comments from Users
Here you can submit questions and comments. As far as they get accepted by the author, they will appear above this paragraph together with the author’s answer. The author will decide on acceptance based on certain criteria. Essentially, the issue must be of sufficiently broad interest.
Please do not enter personal data here; we would otherwise delete it soon. (See also our privacy declaration.) If you wish to receive personal feedback or consultancy from the author, please contact him, e.g. via e-mail.
By submitting the information, you give your consent to the potential publication of your inputs on our website according to our rules. (If you later retract your consent, we will delete those inputs.) As your inputs are first reviewed by the author, they may be published with some delay.