2

Photovoltaics

Photovoltaics (PV ) is the direct conversion of radiation into electricity.

W hile electricity is to be increasingly used as a source of energy,
photovoltaics w ill play an im portant role in the field of renew able
energies. PV technology is m odular (i.e., existing system s are expandable),
has a long lifetim e (m anufactures give guarantees of up to 25 years), is
silent and em issions-free during use. There is a considerable
potential for cost reduction due to know n sem iconductor technology;
in addition, present production processes can be m ade m ore efficient
and cheaper developing m ass production techniques. Although at
present (2005), prices at the factories are stabilized in the vicinity of 3
€/W p , due to a lack of silicon as a
raw m aterial. It is expected after new production facilities for silicon are into
operation in 20 06 prices w ill fall again. Latest inform ation indicates
that production costs of 1–1.6 €/W p are feasible.

2.1 Brief History (Green 2000)

Solar cells derive their origin from som e of the m ost im portant
scientific developm ents of the 20th century, com bining the N obel prize w
inning w ork of several o f the m ost im portant scientists of that century.
The G erm an scientist, M ax Planck, began the century engrossed in the
problem of trying to explain the nature of light em itted by hot bod ies, such
as the sun. H e had to m ake assum ptions about energy being restricted to
discrete levels to m atch theory and observations. This stim ulated Albert
Einstein, in his “m iraculous year” of 1905 to postulate that light w as
made of sm all “particles,” later called photons, each w ith a tiny am
ount of energy that depends on the photon's color. B lue photons have
about tw ice the energy of red photons.
20 Solar Electric Pow er G
eneration
Infrared photons, invisible to the eye have even less energy. U ltraviolet
photons, the cause of sunburn and skin cancer, are also invisible but carry
even m ore energy than the blue ones, accounting for the dam age they can
do. Einstein's radical suggestion led to the form ulation and developm
ent of quantum m echanics, culm inating in 1926 in Erw in
Schrödinger’s w ave equation. W ilson solved this equation for m aterial
in solid form in 1930. T his allow ed him to explain the difference betw een
m etals, good conductors of electricity and insu lators; also the properties of
sem iconductors w ith their interm ediate electrical properties. E lectrons, the
carriers of electrical charge, are free to move around in metals, allowing
electrical currents to flow readily. In insulators, electrons are locked into
the bonds holding the atom s of the insulator together. They need a jolt of
energy to free them from these bonds, so they can becom e m obile. The
sam e applies to sem iconductors, except a sm aller jolt is needed – even
the red photons in sunlight have enough energy to free an electron in the
archetypical sem iconductor, silicon. R ussel O hl discovered the first silicon
solar cell by accident in 19 40. H e w as surprised to m easure a large
electrical vo ltage from w hat he thought w as a pure rod of silicon w hen
he shone a flashlight on it. C loser investigation show ed that sm all
concentrations of im purities w ere giving portions of the silicon properties
du bbed “negative” (n-type). These properties are now know n to b e due
to a surplus of m obile electrons w ith their negative charge. O ther regions
had “positive” (p-type) properties, now know n to be due to a deficiency o f
electrons, causing an effect sim ilar to a surplus of positive charge (som
ething close to a physical dem onstration of the m athem atical adage that
tw o negatives m ake a positive).
W illiam S ho ckley w orked out the theory of the devices form ed
from junctions betw een “positive” and “negative” regions (p-n junctions) in
1949 and soon used this theory to design the first practical transistors.
The sem iconductor revolution of the 1950s follow ed, w hich also resulted
in the first efficient solar cells in 1 95 4. This caused enorm ous excitem
ent and attracted front-page headlines at the tim e.
T he first co m m ercial use of the new solar cells w as on
spacecraft, beginning in 1958. T his w as the m ajor co m m ercial ap
plication until the early
1970s, w hen oil em bargoes of that period stim ulated a re-exam ination of
the cells' potential clo ser to hom e. From sm all beginnings, a terrestrial
solar cell industry took root at this tim e and has grow n rapidly, particularly
over recent
 Photovoltaics 21

years. Increasing international resolve to reduce carbon d ioxide em issions

as a first step to reigning in the “G reenhouse E ffect,” com bined w ith
decreasing cell costs, sees the industry poised to m ake increasing im pact
over the first tw o decades of the new m illennium .

2.2 Photovoltaic Effect

A solar cell is a large-area sem iconductor diode. It consists of a p-n junction

created by an im purity addition (doping) into the sem iconductor
crystal (consisting of four covalent bonds to the neighboring atom s for
the m ost com m only used silicon solar cells). If im purities are phosph
orus-atom s, w hich have five outer electrons, only four electrons are
required to fit the atom into the silicon crystal structure, the fifth electron
is m obile and free. So in this region of the crystal there are m any (a m
ajority) free negative charges, therefore it is called n-region. V ice versa
for the p-region: B y doping the crystal w ith boron atom s, w hich have
only three outer electrons, one electron is alw ays m issing for a com
plete binding into the crystal structure. This electron could be “borrow
ed“ from neighboring atom s, so the place of the m issing electron is shifted.
This m issing electron could also be seen as a “hole“ w ith a positive charge
that is m obile and w andering. T here are m uch m ore free holes than
free electrons in the p-regions, so the electrons are called m inority
charge carriers there.
D ue to the differences in concentration at the “frontier“ betw een the tw
o regions, electrons diffuse into the p-regions and “holes“ into n-
regions, therefore an electrical field in the form erly electrical neutral
junction com es into existence (see Figure 2.1): T he buildup of the space-
charge-region. It increases until a fu rther practical diffusion of carriers is
avoided by it.
Light (or sun radiation) falling into the sem iconductor generates electron-
hole-pairs, causing an increase in the concentration of the m inority charge
carriers by several orders of m agnitude. These charge carriers diffuse to the
space charge zone and are divided by the electric field there. B etw een
the contacts of the n-side and p-side a tension V could be detected, as show
n in Figure 2.1. W hen a load resistor R is applied, a current I flows through
it, and electrical pow er is dissipated.
22 Solar Electric Pow er G
eneration

Fig. 2.1. Principle of photovoltaic energy conversion in a n-p-do ped

sem iconductor. G enerated pow er is supplied to an ohm ic load R (schem e).

Fig. 2.2. Front view of a square m ulti-crystalline silicon (m c-Si) solar cell at a size of
10 cm x 10 cm .
 Photovoltaics 23

T he characteristic of a solar cell w ithout any irradiance (dark

8
characteristic) corresponds to a diode characteristic , as show n in F igure
2.3 below . W hen the solar cell is illum inated, this characteristic shifts by
the am ount of the
photo current I phot in blocking direction (light characteristic). This solar cell
characteristic is determ ined by connecting a variable load resistor to it (see
Figure 2.1) and plotting the resulting currents and voltages at different loads.

Fig. 2.3. C urrent-voltage characteristics of a diode (dark solar cell) and a irradiated
solar cell w ith a short circuit current I sc and open circuit voltage V oc.

T he short-circuit cu rrent Isc is one of the m ost essential characteristics of

a so lar cell. It occurs in an illum inated, short-circuited solar cell. O pen-
circuit voltage V oc: O ne describes the ten sion b etw een the contacts if no
current is
taken (open circuit). The theoretically attainable (optim al) pow er w
hich
could be tak en from the term inal, P opt, is the product of short-circuit
current
Isc and open-circuit voltage V oc :

(1)

8
To m easu re the dark characteristic , an external variable pow er supply
is necessary.
24 Solar Electric Pow er G
eneration
T he attainable pow er P max is defined by the greatest possible product of V
and I at an operating point:

(2)

The so called “M axim um Pow er Point“ (M PP) is given by I mp and V mp.

 T he ratio of P m to P opt is ca lled th e fill fac tor FF. It describes
ax
the “rectangularness” of the trace of the characteristic. Figure 2.4 show s
the pow er P as a function of the tension V and the corresponding
Photovoltaics I-V
25
characteristic.

Fig. 2.4. C urren t-V oltag e characteristics and Pow er-V oltag e characteristics
of a silicon solar cell. A lso show n is the M axim um Pow er Point (M PP) at V mp
and I mp.

The ph otovoltaic conversion efficiency PV is defined by the ratio of the

electrical pow er output to the irradiated p ow er on a solar cell. PV
depends on irradiance and spectrum . The conversion efficiency is determ
2
ined under standard test conditions (ST C ): an irradiance o f 1,000 W /m
perpendicular onto the front surface, a cell tem perature of 25°C and a
spectral distribution according to solar irradiance passing at an elevation
angle of 41.8° through the atm osp here (air m ass 1.5). For physical
reaso ns the photovoltaic conversion efficiency has a theoretical
upper lim it. This am ounts to approxim ately 28 % for crystalline silicon
and has three m ain causes:
1. Silicon is a so called indirect sem iconductor: T his m akes the absorption
of a photon dependent on the occurrence of a phonon (lattice
vibration) – w hile this happens relatively seldom , its abso rption
coefficient is com paratively low .
2. The band gap of silicon am ounts to 1.1 eV : Photons w ith low er
energy aren't absorbed at all, while photons with higher energy transfer
the surplus energy to photons as lattice vibrations, i.e., as heat. This, and
further losses, define the so called “spectral sensitivity” or “spectral
response” of a solar cell (see below ).
3. T he m axim um voltage (open-circuit vo ltage) Voc depends on
the difference of the potentials given by the p-n-transition and am
ounts to approxim ately 0.7 V for silicon.
T his theoretical conversion efficiency is reduced in reality by different loss
m echanism s:
O ptical losses, as reflection and shadow ing losses caused by the
front contacts and also losses by non-absorbed (transm itted) irradiance.
O hm ic losses from series resistors (by contacts and sheet resistance)
and by parasitic parallel resistors (see Figure 2.12).
R ecom bination
losses
T he ability of a solar cell to convert an incom ing photon o f a
specific w avelength into an electron-hole pair is called “quantum
efficiency.” The “internal quantum efficiency” neglects reflection losses on
the surface of the solar cell, w hile the “external quantum efficiency”
includes them .
W hile the energy of the p hoto ns (or “quanta” of energy) increases w
ith their frequency (according to the law of M ax Planck), each is usu
ally creating one electron-hole-pair only w ith a constant en ergy
potential. Therefore, spectral efficiency, defined by electrical energy output
in relation to the irradiated energy, decreases as w avelengths becom e
sm aller. The spectral efficiency is best w hen the am ount of energy from
the incom ing photon is just sufficient to create one electron-hole pair. If the
energy of the photon is not sufficient to create an electron-hole pair, the
photovoltaic effect is nil – this happens at w aveleng ths greater than 1,100
nm for silicon solar cells. D ue to im purities in the silicon crystal the real
spectral characteristic is som ew hat different from the ideal described
above, so it has to be m easured. T he so called “spectral sensivity“ or
“spectral response” S( ) as show n in Fig. 2.5, 2.7, is defined by the photon
density j phot( ) div id ed by the inten sity
of incom ing irradiation (irradiance) E( ) or G ( ). Spectral efficiency
is
determ ined by m easuring the response of a solar cell to m odulation at
a specific w avelength. The m easurem ent is carried out by irradiating a
bias
spectrum (e.g., A M 1.5) in o rder to avoid disturbing effects caused by
the irradiance level. Such a device is show n in Figure 2.6.

Fig. 2.5. A ctual spectral response of a m ulti-crystalline silico n solar cell (A SE).

Fig. 2.6. D evice to m easure I-V characteristics, efficiency and spectral

response. Show n w ith a high efficiency solar cell at U N SW .
T he specific spectral distribu tion of the S un’s radiation on the
Earth’s surface is dependent on the thick ness and condition of the atm
osp here through w hich it has to pass. To som e extend the efficiency
and pow er output of solar cells are also dependent these conditions (see
Figure 2.7, further inform ation in chapter “irradiance m odeling ”).

Fig. 2.7. A ctual spectral response of a m ulti-crystalline (m c) silicon solar cell (A S

E) together w ith the solar spectra for different elevation angles of the sun, equivalent
air m asses (A M ), and tim e at the equator for Equinox.
2.3 Photovoltaic Generator

T he internal electrical field in a solar cell is relatively w eak and only sm

all differences in electrical potential can be achieved (0.3 V for germ anium
and
0.7 V for silicon). The actual open circuit voltage that co uld be achieved is
slightly less than these values. T o get higher voltages, solar cells are
connected in series – so called “strings” of solar cells.
B ecause these string s are very fragile, they are usually em bedded in a
soft plastic and glass sandw ich creating a so called “solar m odule” or
“PV m odule.” The transparent soft plastic above and beneath the cell
strings is usually the copolym er EV A (ethylene-vinyl-acetate), but PV B ,
silicones or T PU (therm oplastic polyurethane, a very recent dev elopm
ent) are also possible options. In order to stiffen the com pound and to
m ake it m ore durable, a tem pered front glass is added. If the b
ackside consists of a com posite foil (w hich consists of com
bination of layers such as PV F–alum inum –P V F or PV F–
polyester–P V F), w e call it a ”lam inated
m odule”; if the backside is glass, w e call it an “encapsulated m odule” (see
Figure 2 .8).
T he output voltage at open circuit conditions of su ch a PV -m odule
is usu ally set to 17 to 35 V for off-grid applications, w hich allow s to
fully charge a 12 V (resp ectively a 24 V ) battery, so 36 to 72 cells are req
uired in a series connection. The m odule is com pleted by attaching a sm all
term inal box w hich contains the electric term inals and a m etal or plastic
fram e w hich helps to m ount the m odule and provides additional stiffness
to it.

Fig. 2.8. C ross section of an encapsulated PV -m odule.

Fig. 2.9. C ross section of a lam inated PV -m odule.

2.3.1 Electrical Properties

E lectrical characteristics of a PV -m odule are indicated by som

e current-voltage characteristics, equivalent to the one of solar cells (see
Figs.
2.3, 2.4). B y connecting a variable ohm ic load to the term inals of the
irradiated PV m odule, com binations of current and voltage can be
recorded w hich result in an I-V-curve w hen the load is varied.
M ultiplication of the I-V pairs leads to output pow er w hich reaches
its highest value at the so called "M axim um Pow er Point" (M PP). To m
ake the data com parable, "Standard Test C ondition s" (ST C ) have been
set up: The spectrum is fixed and related to a sun spectrum at A ir M
2
ass 1.5, the irradiance is 1,000 W /m and the cell operating tem perature
is set to 25 C. (IE C 904 -1 and IEC 891 , resp. D IN EN 609 04-1 and D
IN E N 60891, see also A nnex). The variation o f the irradiance E (or
G ) has only a sm all influence on the open-circuit voltage of a PV
-m odule in a range of
2
350–1,000 W /m . A t low er irradiance levels, the voltage
decreases
lo garithm ically (see Fig. 2.10). The short-circuit cu rrent is
directly proportional to the irradiance because the current is equivalent to the
num ber of electron-hole pairs generated by the absorbed photons.
Consequently; the possible output pow er of a P V -generator is
2
proportional to the irradiance from 350–1,00 0 W /m (constant conversion
efficiency).

Fig. 2.10. C urrent-voltage characteristics of a m ulti-crystalline silicon PV

- m odule for variations of irradiance (tem perature and spectrum are
kept constant at 25 C resp. A M 1.5)

For low er irradiance lev els the conversio n efficiency decreases due
to voltage losses, w hich depend on the internal shunt resistance. Solar
cells w ith a high shunt resistance are m ore suitable for low irradiance
levels than
cells w ith lo w shunt resistance (m ainly caused by im purities in the
cell m aterial).

Fig. 2.11. Current-voltage characteristics of a m ulti-crystalline silicon PV m odule at

variations of tem perature (irradiance and spectrum kept constant at 1,000 W /m
² resp. A M 1.5)

A n increasing cell tem perature at constant irradiance causes a reduction of

the open-circuit voltage and consequently of the output pow er by -0.4% /K
to
-0.5% /K for crystalline silicon solar cells (see F ig.
2.12).
2.3.1.1 Equivalent Electrical Circuit

Fig. 2.12. Equivalent electrical netw ork diagram of a solar cell according to the “one
diode m odel”.
The current-voltage characteristic of a solar cell approached by a “one diode
m odel” (equivalent electrical netw ork diagram see Fig. 2.12) could
be described as follow s:

(3)

(4)

(5)

T he current-voltage-characteristics of a PV m odule could be described as:

(6)
w ith:
-19
a avalanche factor q elem ent. charge (1.602 10 A s)
E irradiance in W /m Rp parallel (or shunt) resistor in
²
I0 diode saturation current in A Rs series resistor in (0.05-0.5 )
Iphoto photo current in A T absolute tem perature in K
I Rp current of parallel resistor in A V br breakdow n voltage in V
-23
k B oltzm ann const. (1.381 10 J/K ) coefficient for dependence of
m avalanche exponent irradiance in R p, in m ²/W

U sually, the parallel resistor is considered to stay constant. H ow ev

-3 2 -1
er, Z im m erm ann 1995 gives a typical = 1.69 10 m W , so the
-2
parallel resistor
P R is 350 in the dark, at an irradiance of 1,000 W m it
decreases to 70 . M easured tem perature coefficients for silicon PV m
odules could be found in Table 2.1.

2.3.1.2 B ypass Diodes

A s w ith all series connections (e.g., at batteries or at P V ), the elem ent w

ith the low est current defines the to tal current. In order to avoid losses,
only cells w ith an equivalent current at operation voltage are selected for
series connection. Also the current of a cell m ay be reduced by local
shadow ing (e.g., due to dirt on the surface of the m odule) w hich
therefore lim its the total current an d p ow er output. If the string is large
enough, the (reverse) voltage at the shadow ed cell can surpass the negative
breakthrough voltage and could lead to a local pow er dissipation that co
uld even destroy the cell. T o overcom e th is problem “bypass diodes”
are sw itched in parallel (in
opposite direction of the solar cell diode, often also called “antiparallel”) to
the solar cells or a sm all string of solar cells (see Figure 2.13). W hen a cell’s
current is reduced by shadow ing, a reverse voltage builds up in the cell
(assum ing a load is connected) until it surpasses the breakthrough voltage of
the bypass diode and thus part of the total current flow s through the bypass
diode, w hile the rem aining I rest passes through the cell.
O ther w ays of overcom ing this problem are to use a low reverse
breakthrough vo ltage of the solar cell diodes (e.g., at m ulti-crystalline
solar cells) or a cell integrated bypass-diode (G reen 1980: “Integrated Solar
C ells and Shunting D iodes,” A ustralian Patent 524,51 9; U .S. Paten t
4,323,71 9).

Fig. 2.13. Schem e of operation of a bypass diode: solar cell w ith partial shading in
a string of 3 cells in series connection.

2.3.1.3 Electrical Term

inals

T he w ires from the cell strings pass through the m odule lam inate and
are fixed by a pull relief. Then the w ires are connected to an extension or
end in a plastic box linked w ith a plugable or screw able term inal. Latest
m odules are pre-equipped w ith external cables w ith attached w
eatherproof plugs for m odule interconnection to reduce tim e for
installation. B ypass diodes are integrated in the fram e or are m ounted in
the term inal box.
2.3.1.4 Parallel Strings

A s the efficiency of the pow er conditioning com ponents (as

inverters) increases w ith increasing voltage, m odules are sw itched in
series. In some countries (e.g., the US) safety regulation limits maximum
voltage to 600 V (500 V incl. a security factor). For the m ost com m on size
of cells (10 cm by
10 cm to 15 cm by 15 cm ) this means a maxim al power output of 2 to 3 kW p
.
To achieve higher pow er outputs w ithout exceeding the m axim um
voltage
of 600 V (or to get higher currents), m odules (or strings of m odules) are
sw itched in parallel. In case of failure, such as voltage reduction in a string,
e.g., caused by a higher tem perature or by a shadow ed cell (voltage is also
low er w hen a bypass diode is applied), the rem aining strings try to “feed”
the defected string and m ay destroy it. T herefore, in order to protect the
string, “string”-diodes are connected in series to each string of so lar cells
to avoid reverse currents (Fig. 2.14). If then the vo ltage of a string is
reduced by a failure, the “string” diode is in blocking state and the rem
aining strings deliver their pow er to the desired load. A disadvantage of
this configuration is perm anent voltage loss at the string diode. Y et voltage
loss can b e reduced by using a low breakthrough vo ltage diode as germ
9
anium or Scho ttky barrier types . A lso, m agnetic field senso rs m ay be
used to detect reverse currents
and trigger an off-sw itch.

Fig. 2.14. Parallel connection of PV strings via string diodes for protection.

9
V oltage loss of 0.3 V instead of 0.7 V at silicon diodes.
 A nother m ethod is the use of relatively sm all “string”-inverters at
the pow er conditioning side. The (synchronized) AC power output of each
string is then paralleled.
2.3.2 M echanical Properties

2.3.2.1 Sandwich Lam ination

A lam ination consisting of glass–plastic–solar cell–plastic–glass is called an

“encapsulated” PV m odule, w hile a lam ination of glass–plastic–
solar cell–plastic is called a “lam in ated” PV m odule. The plastic is m
ainly a foil (thickness 0.5 to 0.7 m m ) of E V A (ethylene-vinyl-
acetate), w hich is processed at 150 C in a vacuum lam inator: T he
copolym er E V A “cures” at that tem perature and m akes the lam ination
process non-reversible. The vacuum avoids air bubbles inside the lam
inate. To avoid degradation by ultraviolet sun radiation (“yellow ing”
or “brow ning”), a specific U V absorber is added to the resign. The
foil on the backside used for the “lam inated PV m odule” instead of
® ®
glass is usually a com pound of foils of Tedlar –Polyester–Tedlar or
® ®
Tedlar –A lum inum –Tedlar at a thickness of
®
0.5 m m . T edlar is a polyvinyl fluoride film (P V F). T he glass sh eets of
both
types are m ade of iron free and therm al tem pered glass at a thickness of 2
to
4 m m , to achieve a high optical transm ission and to fulfill the ISO
10
203 regulations . Further m aterial properties are given in the A nnex, T
ables A 10 to A 14.

2.3.2.2 Fram
ing

Fram es of PV m odules are m ainly m ade of alum inum profiles, w hich

hold the lam inate onto the fram e. The fram e’s corners are secured by
stainless steel screw s or saw tooth inlays. Sm all tube profiles m ade out
of silicon rubber squeezed betw een lam inate and alum inum fram ing, keep
the lam inate fixed, but how ever are tolerant to tem perature expansion and
to m echanical stress. A nother, “cleaner” option is the use of a self-adhesive
seal tape (e.g., B utyl) applied on the rim of the lam inate, preceding the
clenching of the fram e. To reduce costs, but also for a better self-cleaning
effect and a low er energy consu m ption for production, fram eless m
odules are used in increasing num bers. The m ounting of such m odules
to the structure is done by fixture com pounds such as show n in Figure
2.15.

10
Resistance for hail up to a diam eter of 25 m m , torosion stability of the PV
m odule for w indspeeds up to 200 km /h.
2.3.2.3 Fixing and M ounting

M odules w ith fram es are equipped w ith screw -threads or holes for
fixing. Due to the required resistance to corrosion resistivity all screws and
threads have to be made out of stainless steel in V2A to V4A quality
(for maritime applications). For fram eless PV m odules, fixing is done by
fixture com pounds (Figures 2.15 and 2 .16).

Fig. 2.15. C ross-section of a fixture-com pou nd for fram eless PV m odules

(after
Sch m id 19
88).

Fig. 2.16. Roof-m ounted PV generator with fram eless m odules and fixture com pounds.
A s w ell as the need for less m aterial, handling is also advantageous
since fixing and screw ing is now at the front of the panel. A positive side
effect is that less dust and dirt accum ulation occurs at the edges of the
m odule. L ayers of dirt are w ash ed aw ay at rainfall, w hile fram ed m
odules tend to cause an accum ulation of dirt w hich grow s from the
edges (at the fram e boundary) to the center of the front surface of the m
odule. A dditionally, architects appreciate the m ore hom ogeneous
visual ap pearance that is presented by the fram eless PV m odules.
A relatively new developm ent tow ard cost-effective and visually
appealing m ounting of PV m odules is the adhesion technology.
First
experiences, for exam ple w ith the PH A L K -M ont S oleil 560 kW p in
Sw itzerland, look very prom ising – the installation w as faster and
corrosion
problem s in the m ounting com pound consisting alum inum fram ing,
stainless steel screw s and steel support structures w ere elim inated. O n the
other hand, nondestructive dism antling of the m odule is not
possible, yet in consideration of the extensive lifetim e and reliability of
PV m odules, this factor is less critical. In the future rising costs of hum an
labor and sinking m odule prices w ill m ake this approach even m ore
favorable. H ow ever, w hile the therm al expansion coefficient of the
plastics involved is larger than of that of glass, the m odule tends to be
stressed considerably by bending during processing, so m odule size is lim
ited.
2.4 Properties of PV Generators in Operation Conditions

T he electrical pow er output of silicon solar cells decreases as

cell tem perature increases due to voltage-losses. For frequently used single-
and m ulti-crystalline silicon solar cells, the voltage and pow er losses am
ount betw een 12% and 15% for an increase in tem perature of 30 K (see
Table 2.1). W ith conventional m ounting or roof integration of th e P V g
enerators, cell tem peratures of 30 K and m ore above am bient tem
peratures can be reached during a sunny day. Especially around noon, w
hen irradiance reaches its peak value, the conversion efficiency of the solar
cells becom es poorest due to the tem perature effect.

T able 2.1. M easured tem perature coefficients (TC ) for Silicon PV m odules

Type of PV TC (V OC) TC (I SC) TC (FF) TC (P mp)

m odule in % /K in % /K in % /K in % /K

Single-Si # 1 -0.2817 0.0411 -0.1265 -0.3619

Single-Si # 2 -0.3413 0.0130 -0.1642 -0.5035

M ulti-Si # 1 -0.2632 0.0435 -0.1172 -0.3318

M ulti-Si # 2 -0.3675 0.0675 -0.1732 -0.4690

M ulti-Si # 3 -0.2925 0.0407 -0.1556 -0.3996

A SE 300-D G -0.3726 0.1097 TC (V mp) = -0.4752 -0.4397

/50 (m ulti-Si) TC (I mp) = + 0.0372

a-Si min -0.0393

a-Si max -0.2045

R eferences: Em ery et al. 1996; for A SE -300-D G /50: K ing et al. 1996

U ntil now the reference conditions used for the classification of PV m odules
11 12
(ST C : Standard Test C onditions , SO C : Standard O perating C onditions)

11
STC : cell tem perature 25°C , irradiance 1000 W /m ² (perpendicular),
sun
spectrum equivalent to A ir M ass 1.5 (see also IE C 60904-1, IE C 62145 and IE
C
61215).
12
SO C : as STC , but is using an actual m easured cell tem perature, occurring at
an irrad ian ce of 800 W /m ², an am bien t tem perature of 20°C and a w ind
velocity of 1 m /s. C om m on values are betw een ca. 42°C to 57 °C . SO C are ach
iev ing m ore
(continu ed...)
only give perform ances for one specific operating point (specific spectrum
, perpendicular incidence, constant cell tem perature and constant air
speed). For the user, the knowledge of the yield in a certain period of tim e
(including all occurring operation conditions) is m ore im portant.
Therefore, exam ination and forecast of the actual daily tem perature,
efficiency profile and actual electricity generation is crucial, especially
for the econom ic evaluation of a PV pow er plant. C onsiderable
differences in estim ations based on SO C , and especially of ST C, could
be observed. This is due not only to elevated tem peratures, bu t also to op
tical reflection losses.

2.5 Mounting of PV Modules

W hile continuous R & D into m ore advanced production m ethods resu lted
in low er costs of solar cells and PV m odules (see Fig. 2.17), expenses
for in stallation and m ounting rem ained constant or even increased
due to elevated lab our costs. The share of expenses fo r installation already
am ounts to 21% –53% (average 40% , see W ilk 19 94, Strippel 1994).
C onventional m ounting in the open field consists of a concrete
foundation, m etal tubes or profiles w hich are often even tailor-m ade to fit
the size of the m odules. Such a specific construction involves high m
aterials and labor costs for as w ell as high m aintenance costs due to
corrosion susceptibility. O ften, m aterials that require a lot of energy to
produce such, as alum inum , lead to elevated energy-payback-tim es for the
PV generating system .
In addition to im proved m odule fixing m etho ds such as the
glueing technology (see chapter 2.2.2.3), w hich allow s fo r cost
reductions in m ounting, innovations at the support-structure and the
foundation could also occur (see chapters 10.2 to 10.4), resulting in faster
installation by screw -less and foundation-less construction.

12
(...continued)
realistic values for operation than STC (see also IEC 60891, IEC 61853 (draft) and
IEC 61 721).
Fig. 2.17. Distribution of costs at a sm all PV pow er plant installation (w ithout pow
er co nditioning, by G oetzberger 1994).

C urrently, photovoltaics are m ost cost-effective in rem ote areas such as

13
in alpine regions . There, the transportation of m aterials and tools as
steel bearers and concrete-m ixer m ay cause difficulties how ever.

13
In rem ote areas electrical grid connection is expensive, conventionally pow ered
generators such as diesel generators require additional transp ortation costs for fuel,
oil and spare parts. In these rem ote areas renew able energy generating sets often
are the m ost cost-effective, even in todays econom ical conditions (K ayne
1992, V allvé 1994).
Fig. 2.18. C onventional construction of
concrete and alum inum for foundation,
support structure and m ounting (CEPEL,
R io de Jan eiro).

Fig. 2.19. Conventional, expensive

m ounting construction in the A
lps (from W ilk 19 94).
2.6 Future Development of Photovoltaics

In previous decades, consid erable advances have been m ade both

in photovoltaic conversion efficiencies within the laboratory (G reen 1995)
and in reduction of production costs (see Fig. 2 .22). N evertheless,
photovoltaic energy generation is co m petitive only in off-grid applications
if so cial co sts of fossil fuels are not considered.
Potential for developm ent of single system com ponents is not very
far aw ay from the theoretical optim um . W hile solar cells m ade out of m
aterials other than silicon could achieve higher efficiencies, these m aterials
do have low availability, are m ore expensive, and very often are less
environm entally sound than silicon. There is a need for further developm
ent in the area of cost-effective production, com position of PV system s
and installation (thin film technology, integrated pow er conditioning,
applications). In o rder to m aintain low production costs, details applied to
laboratory sam ples could not be transferred into m ass production. A s a
result, solar cells available on the m arket reach m axim um conversion
efficiencies of 20% to 21% (e.g., Sunpow er), w hile laboratory sam ples
reach up to 25% (e.g., U N SW , ISE ). D ue to necessary series connection
of cells additional ohm ic losses occur and current possible is lim ited by the
w eakest link, the w orst cell. A dditionally, the w hole m odule area
cannot be covered by solar cells due to space requirem ents fo r
electrical insu lation and therm al ex pansion of the cells. T his necessity lim
its the m odule area that can be covered by solar cells and
consequently the overall efficiency reaches only 15% to 17 % . U nder
real operating conditions the follow ing additional losses could be
observed:
optical reflection losses due to non -perpendicular irradiance (K
rauter
1994a),
losses due to low irradiance levels (reduction of form factor and voltage)
therm al losses as voltage reduction du e to elevated cell tem
peratures
(K rauter 199 3c)
R eduction of output current for irradiance sun sp ectra w ith an air m
ass low er than A M 1.5.
S hadow ing: If a cell is sh adow ed in a serial string, the output cu rrent
is lim ited by the reduced current of the shadow ed cell. B ypass diodes
can avoid this effect to som e extent. The leakage current of byp ass diodes
m ay provoke som e perform ance losses also. If parallel strings are used
in the so lar generator, strings w ith low voltage due to shadow ing w ould
operate as load. Therefore, serial diodes are used to avoid load condition
on the string. O n the other hand voltage losses of 0.3-0.7 V (depending
on the type of diode) at the serial diodes occu r.
 Pow er conditioning units are very often located in a sm all building som
e distance aw ay from the generator. According to the literature, the w
iring losses from the generator to the converters are in the vicinity of
3% for m ost applications.
T he inverters often have a high conversion efficiency at the rated pow
er input, but for low irradiance levels and low pow er input the
conversion efficiency decreases. Therefore, the average conversion
efficien cy over a w hole day could be considerably low er than the rated
one.
In considering all these losses, the relation of irradiance to generated
electrical energy ov er a year is o nly 10 :1 to 12 :1. To take account of
these losses and the consequent differences w ith the expected yields,
literature uses a vague correction factor, “Perfo rm ance R atio” (accuracy
10-20% ), to m akeup the difference betw een the expected pow er output
rating and real perform ance. R ecently, this expression is also used for
inverters. D etailed analyses on the causes of perform ance losses are rare,
as are constructive im provem ents to avoid them . Som e approaches to com
pensate these deficits are presented in this w ork.
U nder real operating conditions the yields achieved are 23-45% low
er than the ones in the laboratories or under Standard T est C onditions (ST
C ). Investigations of that effect have been carried out within I-M A P
(Intensive M onitoring and A nalysis Program ) as part of the G erm an
“1,000-R oofs-PV - Program ” by K iefer 1994 and at the PV power plant
“N eurather See” by V oerm an 1984.
B y im proving the optical an d therm al properties of PV m
odule installation, a gain of electrical yield of 12% could b e achieved
under real operating conditions, w hich brings operational efficiency m
uch closer to nom inal efficiency (K rauter 1993c, 1994b, 1996a).
A s a result of greater efficiencies and low er costs of PV system
com ponents (solar m odules and inverters), the im portance of installation
and m ounting issues is grow ing. C osts, especially for m ounting the m
odule, are disproportionally high, as could be seen in Fig. 2.17. How ever,
there is a huge potential to reduce the costs of these cost sectors and ultim
ately lead to a
faster sp read of photovoltaics in the energy m
arket.
B y expanding the know ledge of the interface of the PV m odule and
real environm ent, efficiency gains could be expected that exceed the
ones achievable in solar cell technology.
2.7 Research Funding for Photovoltaics

Public funding on research of terrestrial photovoltaics started after the 1973

oil crises. Figure 2.20 charts funding from 1974 to 1996 (resp. to 1992 in the
U S). C learly visible, until 198 0, is the intense funding by the U .S. C
arter A dm inistration and the significant decrease by Reagan. In G erm any
a drop in direct funding during 1990 to 199 4 could be observed in favor of
indirect funding w ithin the “1,000-P V -R oofs-P rogram ” (w hich
then actually supported 2,250 PV roofs). D uring that tim e the direct
funding budget in the U S increased constantly.

Fig. 2.20. Direct governm ental funding for PV R & D in Japan, U SA and Germ
any. R eferences: Sissine 1994, Sandter 1993 resp. B rauch 1997, U S D epartm
ent of Energy; and International Energy A gency (IEA -PV PS), based on
2003 U S$ ex ch an ge rates. Exch an ge rates betw een U S D ollars, G erm an M
arks resp. Euros are available in Table A 16 in the A nnex.

Funding for photovoltaics and other renew able energies is at a factor of

eight below the equivalent expenses for the research of nuclear energy. O
nly 9.2% of the G erm an energy R & D budget w as spent on solar energy
in the years
1981 to 1990, while for exam ple Denm ark was spending 26.9% of
the equivalen t budget on renew able energies (Scheer 1993).
In G erm any a hundred tim es the ren ew able energy budget is sp ent
to support uncom petitive coal m ining (in 1997: 4.57 billion €; in
1998:
4.74 billion € – see Table A 6). The subvention for each m iner and
other personnel of the m ining industry am ounts to 51,000 €/a. Th is w
ould be sufficien t to create tw ice the num ber of jobs in the solar energy
sector.
2.8 Market Development of Photovoltaics

Developm ent of PV production w orldwide is shown in Figure 2.21.

Production of PV m odules in term s of output pow er under Standard Test
C onditions (indicated by the index “p,” deriving from “peak”) increased
from 3.3 M W p
in 1980 to 1,000 M W p in 2 004. This corresp onds to an average yearly
grow th of ca. 14% . The total installed pow er at the end of 2004 w as
4,000
M W p . This developm ent ap pears quite satisfactory, but in com parison
to conventional pow er plant capacity the value is still sm all.

Fig. 2.21. M arket developm ent of photovoltaics: Installed capacity and yearly
production (R eferences: PV -N ew s, PV IR (Photovoltaic Insiders R eport) 1982–
2000, W orldw atch Institute 2004, IT Pow er 2005).

In 1973, prices fo r PV m odules w ere about 700 $U S/W p and

consequently w ere only used for aerospace applications. Terrestrial PV m
odules had been offered for 70–80 $U S/W p in 1975, now adays (2000)
terrestrial m odules are already available for 2 –3 $U S/W p . Developm ent
of prices is plotted in Figure 2.22. An example of the current production
costs is given in Table 2.2.
Relative cost shares for com plete PV installation which consists of PV
m odules, installation, support structure, land, and ground preparation
(w ithout pow er conditioning and energy storage) are show n in Figure
2.17.
Fig. 2.22. D evelopm ent of prices of photovoltaic m odules (in € per Wp ) from 1975 to
2003 w ith an am plification of the recent period from 1992 to 2003 (data given in
U S$ has been referenced to 1998). Source: W orldw atch Institute 2003, IEA -PV PS
2005.

Terrestrial w orld m arket shares in PV system s in the early 1990's:

C om m unication (e.g., transm itters) 21 %

Solar-H om e-System s (1-2 m odules) 16 %
H igh-pow er consum er applications 15 %
W ater pum ping system s 11 %
G rid-connected system s 11 %
R em ote living (> 2 m odules) 7%
Sm all pow er consum er applications (w atches etc.) 6%
V illage pow er supply 4%
C athodic corrosion prevention 3%
M ilitary; signalization 3%
R em ote (other) 3%
D uring the last decade the grid-connected share grew from 11% to m ore
than
70% in 2 002. T his developm ent w as triggered by governm ental
support program s in the first world, while governm ental support for
off-grid PV system s, m ainly located in developing countries, w as less, nu
m bers of off- grid installations increased steadily, but their relative
m arket share decreased. W ithin the off-grid sector the relative share
of the different applications rem ained the sam e.

T able 2.2. Production costs of m ulti-crystalline silicon PV m odules in G erm any

 R eference M ertens 1992 and LB ST 1995 Frantzi et al. 2000
Staiß 1995 in D M /W in D M /W p in $/W p
p

Process low est average highest 5 20 in 2000 in 2010

prices price price MW /a MW /a 10 M W 100 M W /a
/a

Silicon (basic 0.51 0.58 0.50

m aterial)
0.82 1.11 1.37 2.25 1.79 0.82 -
C asting

W afer 0.80 1.13 1.00

production

Cell 0.70 1.48 2.38 2.46 1.32 0.44 -

fabrication

Production of 1.03 1.85 2.62 1.28 0.87 0.86 -

m odule

C ommon - - - 1.80 1.20 - -

costs, profit
3.90 6.20 7.80 7.79 5.18 2.11 1.15
Total

A reas of application for autonom ous P V -System s in com parison

to conventional energy supply technologies are presented in Figure 2.23.
Fig. 2.23. Costs of autonom ous PV applications as a function peak pow er of PV
generator and com pared w ith other m ethods of electrical energy supply (Schm id
-1 -2
1995). Light grey is for C entral European conditions (1,000 kW h a m ), dark is
-1 -2
for locations closer to the E quator (2,000 kW h a m ).

D istribution of Cell Production by Technology

(R eference: P.D . M aycock, PV m arket update, 2003)

• M ulti-crystalline Silicon: 58%

• Single-crystalline Silicon: 32%
• Thin-film technology: 7%
• O thers: 3%
G overnm ental Policies for M arket Introduction of PV - Germ any as
an
Exam
ple

T o accelerate m arket penetration of renew able energies, the G erm

an G overnm ent is paying a sustainable fee for each kW h of
photovoltaic electricity injected into the public grid since A pril 2000.
The paym ent of renew able electrical energy g eneration was already very
successful in the field of w ind converters w hich led to increm ents of
installed pow er by 30% per year over the last decade (the total installed
w ind pow er generation capacity in G erm any reached 16,629 M W in
200 4, by far w orld’s largest capacity).
W hile form erly entire PV -installations been financed, the
technology advanced (as for grid inverters during the “1,000 PV Roo
ftops Solar E lectricity Program ”) but the prices of PV -system co m
ponents rem ained stable or even rose slightly due to a stable dem and (as
for PV m odules). B y just paying for the energy generated, the plant
operator is now forced to set up and to keep his installation as effective as
possible. M onitoring the yield of the pow er plant over the entire lifetim e of
25–30 years is now done by the hom e ow ner him self. C om petition of
different m anufacturers led to cost and technology im provem ents. From
January 1999 until the end of 2003 the “100,000 R ooftops So lar
Electricity Program ” w as carried out, w ith a granted capacity of 345.5
M W and 65,700 system s built.
B efore that program s m entioned above, som e cities and
federal governm ents, forced by a lot of solar initiatives, paid up to 1 .1 €
for each kW h grid injected photovoltaic electricity. The general
“Electricity Feed L aw ” introduced in 1 991 w as replaced by the “Renew
able Energy Sources A ct (E E G )” in A pril 20 00 by the Federal G
overnm ent. In 20 04 im proved feed-in tariffs ranging from 0.454 to 0.574
€/kW h, depending on size and installation type (ground installation, roof
mounted installation above 30 kW p and roof m ounted system s below 30
kW p ) w ithout lim its for the total
num ber of installations. This regulation boosted the installation and
production in G erm any to 300 M W p per year. W ithin a short tim e G erm
any becam e a w orld lead er in term s of installed PV capacity per year, is
and is very likely to overtake the actual PV production leader Japan w ithin
2006. In order to stim ulate progress also for cost reduction m easures,
the PV com pensation is reduced gradually by 5% per year.