@ Development of defect
passivation etch-less cleaning method and its application to Si solar cells
Figure 1 Structure of Si solar cells and mechanism of losses.
For improvement of energy conversion efficiency of Si solar cells, various losses shown in Figure 1 should be prevented. We have developed a fabrication method of
ultra-low reflectivity Si surfaces to avoid the reflection low. For preventing losses due to recombination of
photo-generated electron-hole pairs, our laboratory has developed “Defect
Passivation Etch-Less (DPEL) cleaning method” and ‘Nitric Acid Oxidation of Si
(NAOS) method”.
Figure 2 Separation of photo-generated electrons and
holes by Si solar cells; (a) in the presence of defect states and/or metal
contaminants; (b) in the absence of defect states and metal contaminants.
When light is
incident to semiconductors, its energy is absorbed by the semiconductors,
resulting in excitation of an electron in the valence band to the conduction
band, resulting in a hole in the valence band and an electron in the conduction
band. Solar cells are the apparatuses
which separate photo-generated electrons and holes. In the absence of defect states in the
semiconductor band-gap, electrons and holes can be separated smoothly, i.e.,
electrons move toward the lower potential energy region while holes move
towards the higher (Figure 2a). In the presence of defect states or metal
contaminants, however, energy levels are induced in the semiconductor band-gap,
and electrons and holes are captured in the gap-states, resulting in the
formation of flat-band region in the semiconductor (Figure 2b). In the flat-band region, electrons and holes
are difficult to be separated, and defect states act as recombination centers
of electrons and holes. In Kobayashi
laboratory, we have developed a method of elimination of semiconductor defect
states, which method can also remove metal contaminants simultaneously. We are aiming to improve Si solar cell
efficiencies using this DPEL method.
Figure 3 Comparison between (a) conventional
semiconductor cleaning method and (b) DPEL cleaning method.
Figure 3 compares the DPEL method
and the conventional semiconductor cleaning method. The conventional semiconductor cleaning
method, i.e., RCA method, uses NH3 plus H2O2
aqueous solutions and HCl plus H2O2 aqueous
solutions. The conventional method
removes metal and particle contaminants by slightly etching semiconductor
surfaces. Etching often roughens the
surfaces and generate defect states, leading to degradation of semiconductor
device characteristics. Because the
metal removal activity of the cleaning solutions is not high, the cleaning
solutions with considerably high concentrations, i.e., % order concentration,
are required to be used at elevated temperatures in the range 50~80°C. Removed metal contaminants are present in the
form of bare ions in the cleaning solutions, and re-adsorption proceeds. Therefore, the cleaning solutions cannot be
used repeatedly, and thus a large amount of the cleaning solution is required
for semiconductor cleaning.
In the case of the DPEL cleaning method
developed in Kobayashi laboratory, on the other hand, low concentration CN− ions in the solutions directly react
with metals, leading to formation of complex ions which have high stability in
aqueous solutions, resulting in avoidance of re-adsorption. Therefore, DPEL cleaning does not cause
etching. The reactivity of CN− ions with metals is so high that
cleaning can be performed using very dilute (e.g., ppm order concentration)
DPEL solutions at room temperature.
Moreover, CN− ions are selectively adsorbed on defect states such as Si dangling
bonds, leading to elimination of defect states. Therefore, electrical
characteristics of semiconductor products such as TFT, LSI, and solar cells can
be improved due to these two effects, i.e., removal of metal contaminants and
elimination of defect states.
Figure 4 Total reflection X-ray fluorescence spectra of the Si wafer observed before
and after DPEL cleaning.
Figure 4 shows the X-ray fluorescence spectra of metal-contaminated Si wafers measured before and after DPEL cleaning. Before
cleaning, typical metal contaminants such as Fe, Zn, Ni, Cr, Cu, Mn, etc. are
present. After cleaning, the
concentrations of all the metal contaminants become below the detection limit
(i.e., ~3×109 atoms/cm2 which corresponds to ~1/300,000
monolayer).
Figure 5 Cu concentration on Si wafers vs. the period
of cleaning using the DPEL solution with 2.7 ppm concentration at room
temperature.
Figure 5
shows the Cu concentration of Si wafers vs. the cleaning period using the 2.7
ppm DPEL solutions at room temperature.
It can be seen that all the copper contaminants are removed in two
minutes using such a low concentration DPEL solution.
Figure 6 Improvement in characteristics of poly-crystalline Si pn-junction solar
cells by the DPEL method.
Figure 6
shows the photocurrent density vs. photovoltage curves for the poly-crystalline
Si (poly-Si) pn-junction solar cells.
High density defect states are present mainly in the grain boundary
regions of poly-Si. Consequently, the
conversion efficiency is not high, i.e., ~10 %.
With the DPEL treatment for ~2 min, the conversion efficiency is
improved to 13 %, indicating effective elimination of defect states. With anti-reflection coating, the conversion
efficiency will be increased to more than 16 %.
The DPEL cleaning solutions contain a low
concentration of poisonous CN− ions. In Kobayashi
laboratory, methods of synthesis of the DPEL solutions from non-poisonous
compounds such as methane and methanol plus ammonia have been developed. We have also developed a method of
decomposition of the DPEL solutions using UV plus ozone method, which can
decompose CN species to nitrogen and carbon dioxide. We are performing cooperative research with
solar cell producers and semiconductor device producers in order to apply the
developed method to fabrication of semiconductor products.
AFormation of ultra-low reflectivity Si surfaces by use of chemical
surface structure transfer method
Figure 7 Mat-textured surface produced on single
crystalline Si by the conventional isotropic alkaline etching method.
Figure 8 Reflectivity of mat-textured surfaces on
single crystalline Si.
Flat Si surfaces possess a high
reflectivity in the range between 30 and 50 %.
To decrease the surface reflectivity, a method of formation of pyramidal
structure on Si surfaces (cf. Figure 7) is
generally employed in Si solar cell fabrication. The pyramidal structure can be produced by
isotropic chemical wet etching using e.g., NaOH solutions containing
isopropanol. However, the reflectivity
of the mat-textured surfaces with the pyramidal structure is not sufficiently
low as shown in Figure 8. In Kobayashi laboratory, the surface
structure chemical transfer method has been developed to form ultra-low
reflectivity Si surfaces.
Figure 9 Principle of the surface structure chemical transfer method.
Figure 9
shows the developed method for surface structure chemical transfer. Si is immersed in hydrogen peroxide (H2O2)
plus hydrofluoric acid (HF) aqueous solutions.
Without metal catalysts, no reaction proceeds on the Si surfaces, but
when Si surfaces are contacted with metal catalysts such as Pt, atomic oxygen
is generated by decomposition of H2O2, resulting in the
formation of SiO2. The
produced SiO2 is immediately etched away by HF. Consequently, the inverse structure of the
metal catalyst surfaces (e.g., inverted pyramidal structure) is formed on Si
surfaces. Using this method, ultra-low
reflectivity surface structure can be formed even on poly-Si surfaces, on which
uniform mat-textured surfaces cannot be fabricated.
Figure 10 Reflectivity of poly-Si surfaces after surface structure chemical transfer.
Figure 10
shows the reflectivity of poly-Si surfaces formed by the surface structure
chemical transfer method. The
reflectivity in the wavelength region between 300 and 800 nm is as low as ~5%,
i.e., much lower than that of conventional mat-textured Si surfaces. When this method is applied to Si solar
cells, the energy conversion efficiency will be greatly increased because of an
increase in the photocurrent density due to a decrease in reflectivity.
Conventional mat-textured surfaces contain
high density defect states at the surfaces (i.e., surface states), leading to a
decrease in the photovoltage. The defect
density on the Si surfaces formed by the surface structure chemical transfer
method, on the other hand, is very low, resulting in high photovoltage.
|