GEOS24705

/ Solar Photovoltaics
EJM May 2011

The solar PV panel is a semiconductor device that converts radiation energy to electrical energy
through the photoelectric effect. When light strikes the PV material, it rips off electrons from
the atoms they were bound to, leaving positive ions behind and allowing some electrons to
flow, producing both an electrical current and a voltage drop and therefore electrical power.

The importance of semiconductors
All photovoltaics are made of semiconductors. Semiconductor devices have transformed
modern life, because their properties allow very targeted control of flow of electrical currents,
in turn enabling all of modern electronics. Semiconductor properties allow them act electrically
as

One-way valves (electrons flow in only one direction the diode)
Switches (small signal opens or closes valve)
Throttle valves (small signal controls how much opens)
Amplifiers (small signal stimulates large one)
Photoelectrics (light -> electrical energy)
Electroluminescers (electrical energy -> light)

Semiconductors play less of a role in the electric power sector, but are a factor in almost
everything that is new in the energy world. Semiconductor-based power electronics make
possible the power electronics that allow DC-DC voltage transformation and the new grid
control methods (FACTS). Light-responsive semiconductor materials make up solar
photovoltaics and the light-emitting diodes that may cut energy use in lighting. And of course
modern semiconductor-based digital communication is integral for the sensing and signaling
that is collectively lumped into the term smart grid.

Semiconductor physics
As the name implies, a semiconductor is a material that is neither a good conductor (charge
flows freely) nor a good insulator (charge does not flow at all). In a semiconductor, charge flows
under certain conditions.

In a good conductor like a metal, some electrons flow so freely that they can hardly be assigned
to any particular atoms. These are called the conduction electrons and they have essentially
no impediments to their motion: apply a voltage and they will flow. In a semiconductor like
silicon or germanium there are no naturally occurring conduction electrons all electrons are
all bound to particular atoms. But if those electrons are given just the right kick of extra
energy they can escape their tight bonds with individual atoms and become free-moving
conduction electrons.

The process of freeing electrons to move is typically described as having the electrons jump
across a band gap up to the conduction band. This metaphor is a little deceptive because
the directionality in up is not a physical direction the electrons dont literally get further
from their atoms. Up instead refers to rising up in energy level - when electrons have higher
energies they can conduct freely. Still, it can be a useful mental image to think of the
semiconductor electrons being kicked up into a region where they can flow freely.

The conduction electrons are not the only factors in making current flow in semiconductors. As
soon as electrons are lifted into the conduction band, it also becomes possible for current to
flow in the valence band. Once an electron has been ejected from its tight bond with an
individual atom, that atom is left with a hole, a place where an electron should be, and the
atom now carries a local positive charge because of its missing negative electron. While other
electrons in the valence band cannot flow freely, they can make small jumps from being bound
to one atom to changing allegiance to a neighboring atom, if a suitable bonding site (a hole) is
there to receive them. If you apply a voltage across a semiconductor, then, and some electrons
have been freed to the conduction band, what results is a net current comprised of two
components: a flow of negative free electrons and a simultaneous flow of positive holes in
the opposite direction, as the valence electrons reshuffle themselves.

Semiconductor junctions
Uniform semiconductor materials alone are not particularly useful. What makes
semiconductors so useful is when two materials of slightly different properties are placed
adjacent to each other. These semiconductor junctions are the basis of modern electronics.

The components of a semiconductor junction are not pure materials but are intrinsic
semiconductors that have been doped with certain impurities that make it easier for
particular kinds of charge to flow. In an n-type (n for negative) semiconductor, the dopant is
an element that likes to contribute free electrons but does not like to accept electrons in the
spot they are freed from. (For people with chemistry background, n-type dopants are
pentavalent they have five electrons in their outermost shell, so that the loss of one would
leave a complete 4-electron shell). The n-type dopant therefore gives a soure of free electrons
without corresponding holes. In a p-type semiconductor (p for positive), the dopant is the
reverse, an element that wants to grab free electrons, and so can be considered to contribute
extra holes. (Again for people with chemistry background, p-type dopants are trivalent). The
simplest semiconductor junction is the union of a p-type and an n-type semiconduction and is
termed (creatively) the p-n junction.

It is this doping that makes semiconductors in general and solar photovoltaics in particular
dependant on exotic materials. Typical pentavalent (extra electron) n-type dopants include
antimony, arsenic, and phosphorus. Typical trivalent (needing just one more electron) p-type
dopants include boron, aluminum, gallium, and indium. These are marked on the periodic table
below. Note that not only the dopants but also the most common intrinsic semiconductor
materials themselves (silicon, germanium) all fall along the border between metals and non-
metals in that confusing regime where elements have some metal-like properties (electrons can
flow, but not easily).
intrinsic semiconductors
p-type dopants n-type dopants

(Note: most solar PV to date has been made of doped silicon, but new designs are using more
exotic semiconductor material, such as CdS and CdTe).

When semiconductors of the two types are mated together, the resulting junction acts
somewhat as a one-way valve. First, remember that the dopant holes cannot flow across the
junction they are property of the atoms that are structurally bound to the material. The only

charge that can actually cross the junction is the electrons. And the p side doesnt have any
readily free-able electrons to contribute to flow to the n side. Unless you somehow rip
electrons away from the p-side semiconductor material (usually a destructive process that
causes irreversible damage), you cant move them to the n side.

It is easier to push electrons the other way, from n to p, since they are already free in the
conduction band. They do need a bit of help, however. To get charge to flow you have to
overcome a certain natural resistance to flow that occurs because at the vicinity of the junction,
some free electrons do diffuse across and combine with holes on the other side, leaving a slight
excess of negative charge on the p side and inhibiting any further flow. In the figure above, that
effect is indicated by rise on the p side of the energy levels of electrons - to flow to the p side,
electrons would have to be given some energy in
order to make it over this energy barrier. That is,
you would have to apply an external voltage to
allow more charge to flow. For most common
semiconductor junctions, that voltage is quite
low, ~ 0.6 V. Once that barrier is overcome,
though, charge can flow freely: electrons from
the n-side cross the p-n junction and travel
through the p-side, hopping from hole to hole.

The p-n junction therefore acts normally as a ~ 0.6 V
one-way valve: highly resistant to flow in one
direction, but if only a small voltage is applied in
the other direction, current flows unimpeded. It
is this property that makes the junction so breakdown
incredibly useful. voltage

(One note: be careful when drawing intuition from the p-n junction figure, because it is
representing energy levels of electrons, whereas voltage is defined inversely, in terms of the
energies of hypothetical positive charges ... If you remember back to the history of electric
motors, this occurred because of early confusion over charges and currents, which ended up
with the moving charge, the electron, unfortunately labeled as the negative one. To push
electrons uphill to the p side, the normal diode direction, you need to apply your 0.6V push as a
positive voltage on the p side).

Light-emitting diodes and photovoltaics
Light-emitting diodes (LEDs) and photovoltaics (PVs) are both specialized semiconductor diodes
in which electrical current is associated with light, either with losing some electrical energy as
radiation (in the LEDs) or with the current and electrical energy being driven by radiation
energy (in the PVs). In an LED, electrons that are pushed across the p-n junction in the normal
diode direction (n->p) release energy as light. In a PV, light absorbed at the p-n junction breaks
apart some of the electron-hole pairs that have combined there, letting electrons flow back in
the p->n direction. Think of the LED/PV pair just as you do the motor/generator pair, as the
same device that can be run in either direction. Much as a generator is a motor run backwards
(mechanical-> electrical energy instead of electrical-> mechanical), a PV is an LED run
backwards (light -> electrical energy instead of electrical->light).

Note that neither the LED nor the PV is compatible with AC power a one-way device can
neither be driven by alternating current nor can produce alternating current. The LED requires a
DC power source (or a rectifier) and the solar photovoltaic produces DC power (which must be
converted to AC with an inverter if it is to be put on the grid).

In the LED, when you apply In the photovoltaic cell, light

voltage across junction and push absorbed near the junction frees
electrons through from n to p, electrons and lets them flow to the
energy from that push is released n side. (Meanwhile electrons from
as light. Electrons continue through the p-side atoms move into the
the p side as normal for diodes. now vacant holes at the junction).

A light-emitting diode is a specialized variant of the diode whose current flow produces light.
As in all diodes, electrons can be pushed by an applied voltage from the n to the p side, where
they can flow by hopping from hole to hole. (That movement is also referred to as an effective
flow of holes). In an LED, when the flowing electrons first recombine with a p-side hole, they
lose significant energy and that energy is emitted in the form of radiation. (This is termed
radiative recombination).

The wavelength (color) of the emitted light is a function of the energy change and so of the
material properties. The LED therefore allows a direct conversion of electrical energy radiation
energy, with color choice provided by careful engineering of materials. Each LED can produce
only a single wavelength, but is possible to reproduce the broadband spectrum of the sun that
people find pleasant with several different strategies either combining several LEDs of
different wavelengths, or coating the LED with a phosphorescent material that absorbs and
then re-emits the light at different wavelengths. Most white LEDS sold for consumer use the
phosphorescence approach (as you saw in the lighting lab, when the LED produced a
broadband output). Expensive theatrical or commercial lighting will use the multi-color LED
approach, often with LEDs separately controlled for computer-controllable output color.

Radiative recombination in LEDs is highly energy efficient a very large fraction of the applied
energy comes out as light making the LED a significantly energy-saving alternative to the ~2%
efficiency incandescent bulb, in which electrical energy is used to heat a filament that then
glows and emits light. The newest LEDs should also beat fluorescents, but the phosphorescent
coating does degrade their efficiency somewhat its another energy conversion step (radiation
-> chemical -> radiation).

A photovoltaic, as stated above, is essentially an LED run backwards. While in an LED an
external energy source pushes electrons uphill from n to p, emitting light in the process, in a
PV, light incident on the semiconductor junction breaks apart existing electron-hole pairs in the
junction region, lifting the electrons to the conduction band and letting them flow downhill to
the n side. A PV kept in the dark acts just like a normal diode. if it is illuminated, though, a
current spontaneously flows in the opposite direction from the normal diode current. Think of
an illuminated photovoltaic as one where
youve managed cleverly to use light to
push current in the wrong direction
without the destructive means youd
usually have to resort to (imposing a
voltage so high you rip off electrons from
the p side but damage the material in the
process). Instead the light non-
destructively provides those free
electrons for you.

The figure to the right show the I-V open circuit
diagram for a normal diode (in red), voltage
where no current can flow in the reverse (current is zero)
direction, but current can flow freely if
you overcome the diode voltage drop of short circuit
~0.6 V. In an iiluminated photovoltaic, current
current flows spontaneously in the anti- (voltage is zero)
diode direction.
(Image: Alan Doolittle, Georgia Tech)

PV behavior
The short summary of how to think about photovoltaic behavior is that everything is backwards
from the industrial AC generators we discussed earlier. The industrial generator wants to
maintain a constant voltage (or at least, its operator tries to run his turbine so as to keep
voltage constant). Because of that, the current put out by the generator varies with the
resistance of the load, with lower resistances drawing more current and therefore more power.
A photovoltaic, by contrast, wants to maintain a constant current. Because of that, the
voltage across the photovoltaic varies with the resistance of the load, with higher resistances
producing higher voltages and therefore drawing more power.

PV behavior: a solar photovoltaic acts like a current source under low load
The key to understanding photovoltaic behavior is to realize that for most normal PV operation,
the current that is produced is proportional to light level: every electron that is released at the
diode junction by the incident light keeps flowing. This means though that current is constant
regardless of what the PV is hooked up to a very strange concept, that current should be
independent of load! Thats not at all what we found for ordinary industrial generators that
convert mechanical motion to electrical energy. (Those were operated at constant voltage, and
the current depended on what the load was). In a photovoltaic in normal operation, the current
is constant, since it is determined by the light, and the voltage then varies according to the
load.

It may help to think about this behavior starting with the endmember case, thinking of a
photovoltaic exposed to a constant light level but with no resistive load at all. If you wire the
one side of a PV to the other with a simple wire (resistance effectively zero) and expose it to
light, every electron released by the incident light will happily flow, while doing essentially no
work. Unlike the generators youve seen so far, you can short-circuit a photovoltaic with no ill
effects you wont blow out the PV. Since P=IV and I is fixed but no work is done (P=0), the
voltage across the short circuit photovoltaic must be zero. If you add a resistance (like a
lightbulb) to the circuit, the photovoltaic will still want to push a constant current across it.
Since by Ohms law V=IR, and the photovoltaic is keeping I constant, the voltage across the
photovoltaic is directly proportional to R. That is, the PV panel adjusts its voltage to match
whatever resistance is attached to keep a constant current flowing.

How much power is delivered to the load? Since P=IV, and I is fixed in this regime, the power
transferred to the load is a function of the load itself (P=I2R). Unlike what youve seen before,
the larger the resistance, the larger the output power. In the low load range, increasing
resistance increases the power output.

PV behavior: a solar photovoltaic acts like a voltage source under high load.
Of course, a PV panel cant put out constant current regardless of how big a resistance you
attach to it, if power delivered goes up with resistance. The sun only has so much power, and as
you increased the demand for power, eventually the PV would no longer be able to keep up.
That is, it could not physically continue to put out constant current anymore something must
give. In fact, in the high-load range when the photovoltaic is hooked up to too large a
resistance - the PV panel acts much like the electrical generators we discussed earlier. That is, it
acts like a constant voltage source, wanting to keep a constant voltage difference across its
terminals, and reducing current accordingly. Since V = IR by Ohms law, if V is constant we get
that current is proportional to 1/R. Now as we increase the resistance past the point where the
photovoltaic cant keep up, less current flows. Since power P=IV, if V ~ constant then P~ V2/R,
and the bigger the resistor the lower the power. In the high load range, adding resistance
reduces the power output.

If you combine those behaviors youll see that if you gradually increase the resistance of the
load on a photovoltaic, for awhile current is constant and the voltage across the PV panels
gradually increases, which in turn means that power delivered increases. That works until you
hit the peak power that the photovolataic can put out (which happens at a voltage close to the
diode voltage drop of ~0.6 V). From that point on, the PV is at its maximum voltage increased
any more and the current would start flowing backwards. From here, continuing to increase the
resistance of the load results in a very swift crash in current. When resistance is infinitely high
no current can flow and the photovoltaic is effectively open circuit . At this point all the
released electrons have nowhere to go they just stack up until the voltage induced by that
charge exactly cancels the 0.6 V diode drop and no current flows.

The practical consequence of this behavior is that you cant think of the power produced by the
photovoltaic separately from the load it is connected to, and you had better match your load
very closely to the panel if you want optimal power generation.

P = I*V

This figure shows the same I-V diagram as the previous figure, only now flipped
over. (We just redefined the direction of the current to be positive thats normal
for thinking about photovoltaics). Since P = I*V, the maximum power point is that
point where the arrow from the origin to the I-V curve is longest.

Photovoltaic specifications:
The specifications given on a photovoltaic are its rated (i.e. maximum) power, its open-circuit
(no current) voltage, and its short circuit (no voltage) current. If the I-V curve were a perfect
rectangle, the rated power would be the short-circuit I x the open-circuit V, but in actuality is a
bit less. Panel numbers are given calibrated to an illumination level of 1000 W/m2.

Varying the light level.
Increasing the illumination on a
photovoltaic doesnt change the I*V
behavior pattern. More light does free
more electrons, though, so the
photovoltaic puts out more current in its
constant-current mode, and therefore
more total power. The short-circuit
current is approximately linear with
illumination twice the light = twice the
current, which yields four times the
power since P = I2R. The voltage at which
current must drop necessarily remains
the same at the intrinsic diode drop.
Photovoltaic materials and construction
Most commercial solar panels are made of polycrystalline silicon.
As of 2008, only twelve factories in the world produced this
material, and more than half the poly silicon of sufficiently high
grade was going to the production of solar cells. All the panels
that we we have to play with are polycrystalline silicon, and you
can see the grain boundaries between distinct silicon crystals.

Commercial photovoltaics are increasingly made using
semiconductor materials deposited in thin films rather than
grown as crystals, of which the most common are cadmium
telluride / cadmium sulfide and amorphous silicon. Thin-film
cells are lower in efficiency, but the hope is that the
manufacturing can become cheaper than for crystalline silicon,
making them the more economical choice. Thin films are also
less subject to breakage.

PV panel arrangements
Individual solar photovoltaic cells are quite small (1-10 cm on a side). The larger solar panels
that you see as self-contained units are actually many individual PV cells wired together. Since
(as mentioned above) each cell can only produce a very small voltage (~0.6 V), it is normal to
wire many cells in series one after another like beads on a string to form a module so that
the overall voltage of the panel is the sum of the voltage drops across all the individual cells.
Individual modules are themselves wired together in a single panel, either in series or in
parallel. Finally, because individual solar panels produce relatively little energy (even at peak
insolation of ~ 1000 W/m2 a 15% efficiency panel of 1m2 will produce only 150 W, enough for
perhaps two lightbulbs), any practical solar PV system consists of many panels. The panels are
not separately wired to the inverter but again are connected, again often in series.

Single photovoltaic cells (1)

connected in series form a
photovoltaic module (2).
Several modules assembled
together create a photovoltaic
panel (3). Image from scienza
giovane (Univ. Bologna).


Partial shading
One very large drawback of the series wiring of PV modules, panels, and systems is that the
entire panel/module/system is then electrically connected, and problems in any part affect the
overall power level. Series-wired panels show a strong and very negative partial shading
response, drops in power when the panel receives uneven sunlight. The partial shading
response is strongly nonlinear: if you shade a tenth of your panel, your power may go down by
50%. This make panel installation problematic in locations where there are any potential shade
obstacles (shadows from trees or buildings, falling leaves, passing clouds). PV panels really work
best in in the desert, in bright cloudless places with uniform illumination, little water vapor to
make clouds and no trees to drop leaves or make shadows.

The key issue for why partial shading affects a solar cell so dramatically is that series wiring
means that if any current is to flow, it has to flow through them all. Its obvious that localized
damage to one cell could have big consequences. If you broke one cell altogether, for example,
so that nothing could flow through it, you'd have a total panel failure even though the damage
was in one small area, because you'd have prevented all current flow. If you completely shade
one solar cell, you remove its ability to pass current, almost as though you had damaged it.
Now the cell is a diode, and you are trying to push current the wrong way through a diode. The
cell becomes a bottleneck.

Partial shading or even just non-uniform illumination thus produces a conflict between the
current that different cells want to pass, a disagreement that has to be remedied (since the
same current does have to pass through all the cells). In this adjustment, the whole panel
adjusts to a non-optimal operation to push any current at all through it. From the I-V curve for
a photovoltaic you can see power drops dramatically if the photovoltaic is overloaded. For a
grid-connected panel, the inverter (DC-> AC converter) to which the panel is connected
essentially puts a fixed load on the system, optimized for the illumination that youd expect to
have. The response to overload is very sharp though: power falls steeply if the load is just a bit
too high. If some of the cells are knocked out, you basically put the whole panel in this overload
state and see a very sharp reduction in total power.

Another nasty side effect of partial shading is that pushing current through a solar cell that's
not illuminated takes power, rather than consuming power. This is a relatively minor effect on
total panel performance but does mean that the individual shaded solar cell will heat up, which
can in turn lead to actual permanent failure. Shade one cell and its neighbors can fry it, ruining
the entire panel forever. This is yet another reason why residential solar or solar in
intermittently cloudy place is problematic.

In the past, these kind of partial shading problems often arose for entire solar installations
whose separate panels were wired in series. Partial shade on one panel would mess up that
panel, and the messed-up panel would then degrade the whole solar PV installation. Modern
installations now often have bypass systems included to let current keep flowing in case one of
the panels malfunctions. Smarter inverters also adjust the load as if panel performance is
dropping to at least minimize problems and keep the system near its optimal performance.

Performance in cloudy conditions
Solar panel output power drop sharply on a cloudy day, even when there is enough light to see
well by, for much the same reason as the drop for partial shading. Again, if your system no
longer has enough push from the sun but the load on it is fixed, it falls into the overload state
and current and power drops sharply. A bit of drop in light level can compromise panel output
severely.

This effect can be more severe than your eye would expect, because your eye adjusts so well to
changing light levels that you often dont notice how much light may be changing. Illumination
on a day of thick clouds may be 10 x less than on a clear sunny day. (And light levels indoors are
typically also 10 x less than those outdoors). You wouldnt expect the solar panel to produce
power in response to indoor lighting. Nor can it produce power on deeply cloudy days. The eye
is a poor guide to incident power has evolved to allow you to see in as wide a range of light as
possible: it is basically logarithmically sensitive, and you have a variable aperture in the front
(your pupil expands and contracts) which further increases your dynamic range.

The solution for the cloudy-day problem is to have a smarter inverter than can adjust the load
as needed, so that it can keep the system from going into its "not enough push" shutdown
mode. With a smart inverter, you get less power out than in the optimal illumination case, but
at least you'd get some power. In a real solar array, electronics in the inverter the connects a
solar array to the grid can be designed to do just that, to compensate for changing illumination
by effectively asking for less power when less power is available. You'll still get losses on a
cloudy day, but you minimize them by not asking the system to give you something it just can't
do.

Lab activities

In this lab you can play with some solar panels and explore their characteristics. Below are
some suggestions for activities that illustrate the performance of solar panels.

Use a panel to turn an electric motor or motors. Observe how the motor speed varies
with illumination.
Verify that panels behave as specified measure the open-circuit voltage and short-
circuit current and compare to the specs on the back of the panel.
Directly observe the panels changing current output with illumination. The easiest way
to do this is to short-circuit the panel through a current meter (a multimeter in current
mode) so that youre measuring the short-circuit current, and then change the panel
illumination either by tilting it toward and away from the sun, or using a filter to cut
down the sunlight, or (with some care) even concentrating light on the panel with a
lens. (The last is hard to do because it takes some finesse to get even illumination across
the panel). Measure solar radiation with the solar power meter and verify that the
current response is linear with power from the sun.
Test and observe the non-linear power dropoff that results from partial shading. For this
you have to have some load on the panel you need to hook it up to a resistor or a
motor. You can measure both current and voltage to get a direct measure of power, or
you can get a qualitative sense by watching a motor slow down noticeably when the
panel is partially shaded.
Test the efficiency of the panel by comparing the solar insolation (in W/m2, from which
you know the total W striking the panel) to its output power when driving a load. You
should get an efficiency somewhere around 15% for a optimized load.
Map out the I-V curve of the panel measure I and V simultaneously for different loads.
Using colored filters, test the response of the PV panel power to different frequencies
(colors) of light - is it equally sensitive to the red, green, and blue parts of the
spectrum?
 Verify viscerally the suns power by using a Fresnel lens to concentrate power on a
thermometer and heat it, or (even more viscerally) light something on fire with it. The
Fresnel lenses inside old projection TVs, if taken out and put in the sun, can generate
temperatures high enough to melt a penny. (You know the mirrors in a solar PV
installation can produce temperatures of over 500 C, as hot as the metal of the boiler
system will stand; the melting point of copper is 1000 C).

E. Moyer 2011