Institut für angewandte Physik, Universität Karlsruhe
D-76128 Karlsruhe
The efficiency of solar cells is limited by non- radiative recombination
occurring in
the bulk via defect states or at the surfaces via surface states. In present day high
efficiency
silicon solar cells bulk recombination is reduced so far that surface recombination
is an
important loss mechanism. Surface states are caused by the interruption of the
periodical
arrangement of the atoms and by the deposition of impurities at the surface. Their
density
can be as large as 1015 / cm3, resulting in surface
recombination
velocities vs of up to 107 cm/s. The surface
recombination
velocity vs defines the surface recombination rate of
electrons
re,s by
re,s =
vsne,s =
senss,hve,thne,s
(1)
where ne,s is the density of electrons at the surface,
se is the capture cross section for electrons of the
nss,h surface states occupied by a hole and
ve,th is
the thermal velocity of the electrons. The strategy forreducing surface
recombination would
either call for a reduction of the density nss of surface
states by altering
the environment of the surface atoms or a reduction of the density
nss,h
of holes in surface states or a reduction of the density ne,s
of free
electrons at the surface. The most effective way of reducing the density of surface
states on
a silicon surface is by covering it with a carefully grown layer of SiO2.
In the following, we describe how the remaining surface states at a
Si/SiO2
interface are distributed over energy between the conduction and valence band
edges and
how the surface recombination rate is determined by the position of the
Fermi-energies
within this distribution of surface states.
The distribution of the surface states over energy shows up in a measurement
of the
electrical capacitance between a contact on the SiO2 and a contact
on the Si
back side. Due to the depletion of holes in a space charge layer of the p-type Si by a
positive voltage on the SiO2 , a decrease of the capacitance with
increasing
voltage is expected as shown by the theoretical curve in Fig.1.
The experimental curve is displaced with respect to the theoretical curve to
more
negative voltages, indicating the presence of a positive sheet charge at the
Si/SiO2 -interface for zero applied voltage. Also the slope of the
experimental
curve is different from the slope of the theoretical curve, which is due to a change of
the
sheet charge with applied bias. If we neglect the contribution of a change of the
surface
charge to the capacitance itself due to the large frequency at which it is measured
and
consider only variations with the slowly varying bias voltage, then the band bending
follows
directly from the value of the measured capacitance according to the Mott-Schottky
formula.
The change of the interface sheet charge with respect to the band bending translates
into a
change of the occupation of the surface states with respect to the position of the
Fermi-energy at the surface. The resulting distribution of the density of surface states
which
is a typical example for Si/SiO2-interfaces is shown in Fig.2.
The contact which was applied to the SiO2 was made by a
transparent
paste used in electrocardiography. It can be removed without any damage of the
SiO2. Since it is transparent we could illuminate the contact area
simultaneously with the electrical measurements and observe the intensity of the
photoluminescence which is shown in Fig.3 as a function of the band bending at the
surface.
Fig.2 | Fig.3 |