U.S. patent application number 14/949602 was filed with the patent office on 2016-11-17 for high-efficiency photovoltaic back-contact solar cell structures and manufacturing methods.
The applicant listed for this patent is SOLEXEL, INC.. Invention is credited to Anthony Calcaterra, Pawan Kapur, Karl-Josef Kramer, Mehrdad M. Moslehi, Virendra V. Rana, Sean M. Seutter, Emmanuel Van Kerschaver, David Xuan-Qi Wang.
Application Number | 20160336465 14/949602 |
Document ID | / |
Family ID | 44146184 |
Filed Date | 2016-11-17 |
United States Patent
Application |
20160336465 |
Kind Code |
A1 |
Moslehi; Mehrdad M. ; et
al. |
November 17, 2016 |
HIGH-EFFICIENCY PHOTOVOLTAIC BACK-CONTACT SOLAR CELL STRUCTURES AND
MANUFACTURING METHODS
Abstract
Back contact back junction solar cell and methods for
manufacturing are provided. The back contact back junction solar
cell comprises a substrate having a light capturing frontside
surface with a passivation layer, a doped base region, and a doped
backside emitter region with a polarity opposite the doped base
region. A backside passivation layer and patterned reflective layer
on the emitter form a light trapping backside mirror. An
interdigitated metallization pattern is positioned on the backside
of the solar cell and a permanent reinforcement provides support to
the cell.
Inventors: |
Moslehi; Mehrdad M.; (Los
Altos, CA) ; Kapur; Pawan; (Palo Alto, CA) ;
Kramer; Karl-Josef; (San Jose, CA) ; Wang; David
Xuan-Qi; (Fremont, CA) ; Seutter; Sean M.;
(San Jose, CA) ; Rana; Virendra V.; (Los Gatos,
CA) ; Calcaterra; Anthony; (Milpitas, CA) ;
Van Kerschaver; Emmanuel; (Los Altos, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
SOLEXEL, INC. |
Milpitas |
CA |
US |
|
|
Family ID: |
44146184 |
Appl. No.: |
14/949602 |
Filed: |
November 23, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
14629273 |
Feb 23, 2015 |
9196759 |
|
|
14949602 |
|
|
|
|
13057104 |
Aug 13, 2012 |
8962380 |
|
|
PCT/US10/59759 |
Dec 9, 2010 |
|
|
|
14629273 |
|
|
|
|
61285140 |
Dec 9, 2009 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H01L 31/18 20130101;
H01L 31/03529 20130101; H01L 31/1804 20130101; Y02P 70/521
20151101; H01L 31/02167 20130101; H01L 31/0201 20130101; Y02P 70/50
20151101; H01L 31/03921 20130101; H01L 31/056 20141201; H01L
31/1892 20130101; H01L 31/022458 20130101; H01L 31/02363 20130101;
Y02E 10/52 20130101; Y02E 10/547 20130101; H01L 31/0682 20130101;
H01L 31/068 20130101; H01L 31/022441 20130101; H01L 31/0236
20130101; H01L 31/035281 20130101 |
International
Class: |
H01L 31/0224 20060101
H01L031/0224; H01L 31/0216 20060101 H01L031/0216; H01L 31/0236
20060101 H01L031/0236; H01L 31/18 20060101 H01L031/18; H01L 31/0392
20060101 H01L031/0392; H01L 31/056 20060101 H01L031/056; H01L
31/068 20060101 H01L031/068; H01L 31/02 20060101 H01L031/02; H01L
31/0352 20060101 H01L031/0352 |
Claims
1. A back contact back junction solar cell, comprising: a
semiconductor layer, comprising: a light capturing frontside
surface with a passivation layer, a doped base region, and a doped
backside emitter region with a polarity opposite said doped base
region; a backside passivation dielectric layer on said backside
emitter region, backside emitter metallization contacting said
backside emitter region and backside base metallization contacting
said doped base region, wherein said backside passivation
dielectric layer and said backside emitter metallization and said
backside base metallization form a light trapping backside mirror;
and, a backside reinforcement on the backside of said back contact
back junction solar cell, said backside reinforcement having access
openings providing access to said backside emitter metallization
and said backside base metallization.
2. The back contact back junction solar cell of claim 1, wherein
said semiconductor layer is an epitaxial silicon layer with a
thickness in the range of 15 to 50 microns.
3. The back contact back junction solar cell of claim 1, wherein
said semiconductor layer is a planar epitaxial silicon layer.
4. The back contact back junction solar cell of claim 1, wherein
said light capturing frontside surface with a passivation layer
serves as an anti-reflection coating.
5. The back contact back junction solar cell of claim 1, wherein
said light capturing frontside surface with a passivation layer
provides field assisted passivation.
6. The back contact back junction solar cell of claim 1, wherein
said doped backside emitter region is an in-situ doped epitaxial
emitter region with an emitter junction thickness of less than 3
microns.
7. The back contact back junction solar cell of claim 1, wherein
said backside emitter metallization and said backside base
metallization is an interdigitated metallization pattern of a
distributed array of interdigitated fingers and busbars.
8. The back contact back junction solar cell of claim 1, wherein
higher concentration base doping regions under said backside base
metallization are separated from emitter regions, thereby forming
separated junctions.
9. The back contact back junction solar cell of claim 1, wherein
higher concentration base doping under said backside base
metallization abut emitter regions, thereby forming abutted
junctions.
10. The back contact back junction solar cell of claim 1, wherein
said mirror is a lambertian mirror.
11. The back contact back junction solar cell of claim 1, wherein
the localized doping concentrations under said backside emitter
metallization is higher than said doped backside emitter region,
thereby forming selective emitter contacts.
12. The back contact back junction solar cell of claim 1, wherein
said backside reinforcement is a permanent support reinforcement
plate and said access openings are through-hole openings.
13. The back contact back junction solar cell of claim 1, wherein
said backside passivation dielectric layer is aluminum oxide.
14. The back contact back junction solar cell of claim 1, wherein
said backside reinforcement is a permanent support reinforcement
plate and said access openings are through-hole openings.
15. The back contact back junction solar cell of claim 1, wherein
said backside reinforcement is a backside grid-shaped support
reinforcement.
16. The back contact back junction solar cell of claim 1, wherein
said dopes backside emitter region is a doped backside epitaxial
region.
Description
RELATED APPLICATIONS
[0001] This application is a continuation of U.S. application Ser.
No. 14/629,273 filed Feb. 23, 2015 which is a continuation of U.S.
application Ser. No. 13/057,104 filed Aug. 13, 2012 now U.S. Pat.
No. 8,962,380 issued Feb. 24, 2015 which is a National Stage of
International Application No. PCT/US2010/059759 filed Dec. 9, 2010
which claims the benefit of U.S. Provisional Application No.
61/285,140 filed on Dec. 9, 2009, all of which are hereby
incorporated by reference in their entirety for all purposes.
FIELD
[0002] This disclosure relates in general to the field of
photovoltaics and solar cells, and more particularly to back
contact back junction thin solar cells and methods for
manufacturing.
BACKGROUND
[0003] Currently, crystalline silicon has the largest market share
in the photovoltaics (PV) industry, accounting for over 80% of the
overall PV market share. And although moving to thinner crystalline
silicon solar cells is long understood to be one of the most potent
and effective strategy for PV cost reduction (because of the
relatively high material cost of crystalline silicon wafers used in
solar cells as a fraction of the total PV module cost), utilizing
thinner crystalline is fraught with the problem of mechanical
breakage caused by thin and often large substrate sizes. Other
problems include inadequate light trapping in the thin structure
because silicon is an indirect bandgap semiconductor material.
Further, it is difficult balance the requirement of high mechanical
yield and reduced wafer breakage rate with high manufacturing
yields in PV factories in a cost effective manner.
[0004] On a standalone crystalline silicon solar cell without
support, moving even slightly thinner than the current thickness
range of 140 .mu.m-250 .mu.m starts to severely compromise
mechanical yield during manufacturing. Thin film silicon is
particularly mechanically fragile causing manufacturing and
processing difficulties. Thus, solutions directed to process very
thin solar cell structures may utilize a cell process during which
the cell is fully supported by a host carrier throughout, or a cell
process which utilizes a novel self-supporting, standalone,
substrate with an accompanying structural innovation.
[0005] Although, in the past, there have been attempts in solar
industry to use carriers such as glass for thin substrates, these
carriers have suffered from serious limitations including low
maximum processing temperatures (in the case of glass) which
potentially compromises the solar cell efficiency. There have also
been attempts to make small area thin cells which do not have
serious breakage concerns; however, large cell areas are required
for commercial viability.
[0006] Achieving high cell and module efficiency with a low
fabrication cost is critical in solar cell development and
manufacturing. Back junction/back contacted cell architecture is
capable of very high efficiency--primarily because there is no
metal shading on the front side and no emitter on the front which
helps result in a high blue response, and also because of the
potentially low metal resistance on the backside. It is known to
those versed in the field that back contacted cell demands a very
high minority carrier diffusion length to substrate thickness ratio
(while a good criteria to have for any solar cell architecture
including front contact cells, this is especially important for
back contact cells). The ratio should typically be greater than
five.
[0007] Because cell thickness cannot be reduced easily without
compromising mechanical yield, for current back contact back
junction solar cells the emphasis is to use a very high lifetime
material. And while this may result in a larger diffusion length,
using a high lifetime material also increases the substrate cost.
However, by using thin cells, the diffusion length does not have to
be as high, resulting in an ease in the material quality
requirements and thus the cost of the cell. This cost reduction is
in addition to the obvious cost reduction of using less silicon.
Thus, a back contact/back junction cell on a very thin crystalline
silicon substrate has both a large cost and performance
advantage.
SUMMARY
[0008] In accordance with the disclosed subject matter, innovative
structures and methods for manufacturing very thin crystalline
silicon, large area (suitable for commercial application), back
contact/back junction solar cells are provided.
[0009] In one embodiment, the back contact back junction solar cell
comprises a substrate having a light capturing frontside surface
with a passivation layer, a doped base region, and a doped backside
emitter region with a polarity opposite the doped base region. A
backside passivation layer and patterned reflective layer on the
emitter form a light trapping backside mirror. An interdigitated
metallization pattern is positioned on the backside of the solar
cell and a permanent reinforcement provides support to the
cell.
[0010] The disclosed subject matter, as well as additional novel
features, will be apparent from the description provided herein.
The intent of this summary is not to be a comprehensive description
of the claimed subject matter, but rather to provide a short
overview of some of the subject matter's functionality. Other
systems, methods, features and advantages here provided will become
apparent to one with skill in the art upon examination of the
following FIGURES and detailed description. It is intended that all
such additional systems, methods, features and advantages included
within this description, be within the scope of the accompanying
claims.
BRIEF DESCRIPTIONS OF THE DRAWINGS
[0011] For a more complete understanding of the disclosed subject
matter and advantages thereof, reference is now made to the
following description taken in conjunction with the accompanying
drawings in which like reference numbers indicate like features and
wherein:
[0012] FIG. 1 is a graph illustrating a simulation result for
determining the optimum thin film silicon substrate (TFSS)
thickness for maximizing solar cell efficiency;
[0013] FIG. 2 is a cross sectional diagram of a back contact back
junction thin film solar cell;
[0014] FIG. 3 is a process flow showing a fabrication process for
making a back contact back junction thin film solar cell;
[0015] FIGS. 4A through 4k are cross sectional diagrams of a solar
cell after key fabrication process steps as it is manufactured
according to the fabrication process of FIG. 3;
[0016] FIG. 5 is a diagram highlighting back contact interdigitated
emitter and base metal fingers and busbars of the solar cell;
[0017] FIGS. 6A through 6D are diagrams showing the backside of a
back contact solar cell after key fabrication steps;
[0018] FIGS. 7A through 7D are cross sectional diagrams of the
solar cell corresponding to the embodiment of FIGS. 6A through
6D;
[0019] FIGS. 8A through 8C are cross-sectional diagrams of a solar
cell showing the formation of a Lambertian mirror;
[0020] FIGS. 9A and 9B are diagrams illustrating two busbar designs
in accordance with the disclosed subject matter;
[0021] FIG. 10 is a cross sectional diagram of a back contact back
junction thin film solar cell with front side reinforcement and
abutted junctions;
[0022] FIG. 11 is a process flow showing a fabrication process for
making the cell of FIG. 10;
[0023] FIGS. 12A through 12J are cross sectional diagrams of the
solar cell after key fabrication process steps of FIG. 11;
[0024] FIG. 13 is a cross sectional diagram of a back contact back
junction thin film solar cell with backside reinforcement and
separated junctions;
[0025] FIG. 14 is a diagram showing a backside schematic view of a
grid-shaped backside reinforcement;
[0026] FIG. 15 is a process flow showing a fabrication process for
making the back contact back junction thin film solar cell of FIG.
13;
[0027] FIGS. 16A through 16J are cross sectional diagrams of the
solar cell after key fabrication process of FIG. 15;
[0028] FIG. 17 is a cross sectional diagram of a back contact back
junction thin film solar cell with backside reinforcement and
abutted junctions;
[0029] FIG. 18 is a process flow showing a fabrication process for
making the back contact back junction thin film solar cell of FIG.
17;
[0030] FIGS. 19A through 19I are cross sectional diagrams of the
solar cell after key fabrication process steps of FIG. 18;
[0031] FIG. 20 is a cross sectional diagram of a back contact back
junction thin film solar cell with frontside reinforcement, abutted
junctions, and non-selective emitters;
[0032] FIG. 21 is a process flow showing a fabrication process for
making the back contact back junction thin film solar cell of FIG.
20; and
[0033] FIGS. 22A through 22J are cross sectional diagrams of the
solar cell after key fabrication process steps of FIG. 21.
DETAILED DESCRIPTION OF THE SPECIFIC EMBODIMENTS
[0034] The following description is not to be taken in a limiting
sense, but is made for the purpose of describing the general
principles of the present disclosure. The scope of the present
disclosure should be determined with reference to the claims.
Exemplary embodiments of the present disclosure are illustrated in
the drawings, like numbers being used to refer to like and
corresponding parts of the various drawings.
[0035] FIG. 1 is a graph illustrating a simulation result for
determining the optimum thin film silicon substrate (TFSS)
thickness for maximizing solar cell efficiency by taking the
minority carrier bulk silicon lifetime and back mirror reflector
quality (lambertian and specular) into account. A key parameter in
designing a back contact back junction solar cell is to determine
the TFSS thickness. Typically, there exists an optimum thickness of
the silicon layer which maximizes cell efficiency. Optimum
thickness depends on several parameters including the bulk lifetime
of the silicon material and the degree of light trapping enabled by
the back surface diffuse mirror properties. The optimum thickness
is due to the following tradeoff. The necessity toward minimizing
minority carrier recombination drives substrate thickness lower
while requirement of more efficiency light trapping drives the
thickness higher. FIG. 1 shows a simulation of cell efficiency as a
function of TFSS thickness, minority carrier bulk lifetime, and
specular and lambertian mirrors. It is evident that thinner film
thicknesses and improved light trapping allows one to use lower
lifetime material. Based on economic and other advantages of using
thinner films, a TFSS thickness optimum in the range of between 15
um to 30 um may be found. However, this simulation only represents
a specific example and in general substrate thickness may be
dictated by several considerations.
[0036] The present disclosure provides structural solutions and
fabrication process solutions for back contact back junction thin
semiconductor solar cells. And although described with reference to
silicon, other semiconductor materials such as germanium or gallium
arsenide may also be used without departing from the scope of the
disclosed structures and methods. Heterojunctions and multijunction
solar cells using silicon or other semiconductor materials are also
within the scope of the disclosed subject matter.
[0037] In operation, large area (in the range of 156 mm.times.156
mm), thin solar cell substrates with a general thickness of less
than 100 um (more specifically in 15 um to 50 um range) are first
manufactured using epitaxial growth on top of a reusable template,
and are subsequently dislodged. The reusable template may be
substantially planar, or in another embodiment be have
three-dimensional features. It is reused several times for epi
growth, which amortizes template cost. The TFSS is released from
the template using a sacrificial layer which not only is able to
transfer the crystallinity from template to the TFSS, but is also
easily removed selectively compared to the TFSS and the reusable
template. One example of the sacrificial layer is porous silicon,
whose porosity can be modulated or graded to achieve both the
aforementioned critical functions. After the thin solar cell
substrates are manufactured using above means, a key challenge is
to handle these TFSS during fabrication of the solar cell without
breakage and cracking. The following structures and methods address
TFSS handling problems as well as provide increased overall
efficiency.
[0038] Further, the disclosed subject matter provides solutions of
the thin planar/substantially planar/or three dimensional TFSS
handling by using temporary and/or permanent carriers for the TFSS
during solar cell manufacturing process. Carrier is a loose term
used to describe a robust material which is able to support the
thin film solar substrate (TFSS). A key requirement for the
successful fabrication of the cells described is that the TFSS
should be supported during all process steps.
[0039] Because both sides of the solar cell need to be processed (a
frontside and a backside) two carriers are usually required: one
for support during the processing of each face of the solar cell.
The carriers should satisfy several criteria: First, they should be
cost-effective. The carrier cost should be less than the thickness
of the silicon that they save. Second, at least one of the carriers
must be able to withstand high temperature processing required in
manufacturing of typical solar cells. In addition, if only one of
the carriers is able to support high temperature cell processing,
the process flow needs to be tailored to ensure that all high
temperature processing steps are on this carrier. Third, at least
one of the carriers must be able to withstand wet processing
conditions required in manufacturing of the solar cell. An example
of the wet processing steps includes silicon front surface
texturing in diluted and heated KOH solutions. Fourth, once one
side is partially or fully processed, the TFSS should be easily
detached from the preferably reusable carrier (for high volume
manufacturing) and transferred to the permanently attached end
carrier (usually attached to the side which was processed first),
in preparation for processing of the other side. Subsequently, in a
case where the first side was only partially processed, the
remaining steps should be able to be completed.
[0040] In the disclosed processes, the first carrier is a thick
semiconductor (e.g., preferably crystalline silicon for crystalline
silicon solar cells) wafer, such as a reusable template. The second
carrier is a low cost material that may be capable of supporting
TFSS permanently. The reusable template carrier, because it is
standard thick silicon wafer, is capable of withstanding high
temperature. The template can be in various sizes such as 200 mm or
300 mm, shapes, such as round or square, and thicknesses capable of
going through full solar cell process without breaking (with
thicknesses of 200 .mu.m upwards). The cost of this carrier is
brought down significantly by reusing and amortizing it over a
plurality of TFSS fabrication cycles. Finally, this carrier also
satisfies the aforementioned carrier criteria of being conducive to
the detachment of the TFSS with high yield which is accomplished
using a porous silicon layer between the template and the TFSS.
Release processes include mechanical release (MR) or sonicated
mechanical release in a liquid (SMR).
[0041] Several choices of second carriers are possible, forming
different classes of manufacturing methods. All disclosed processes
herein are characterized by the criteria that at least partial
solar cell processing steps are performed on the TFSS while it is
still on the first carrier--the template. The choice of the second
carrier is closely linked with whether partial or all process steps
are completed on the template.
[0042] In the case of partial backside processing on the template,
where upon after TFSS release both front-side cell processes as
well as some back-side cell processes remain, several second
carrier embodiments are possible.
[0043] In an embodiment referred to as Frontside reinforcement
(FSR), the TFSS is released from the template using a temporary
carrier attached on the partially processed backside. The temporary
carrier is highlighted by its ease of release using means such as
electricity (ex. mobile electrostatic chuck) or a temporary
adhesive which is released upon heating at high temperature.
Subsequently, the front side cell processes--such as texturization
and passivation--are carried out with the temporary backside
carrier supporting the TFSS. Finally the remaining backside steps
are performed by transferring the TFSS from the temporary backside
support to permanent front side reinforcement (ex. EVA/glass
combination), thus freeing up the backside for processing. A
specific requirement on the front side reinforcement being that it
does not degrade light coupling beyond the degradation usually
incurred due to module level packaging--making it possible to
integrate the reinforcement into the solar module and thus
permanent.
[0044] In a second embodiment referred to as backside reinforcement
(BSR), the TFSS is released from the template (the first carrier)
using a permanent backside reinforcement. The permanent backside
reinforcement only partially covers the backside which allows
processing on the backside through the open, uncovered backside
areas after front side processes are completed. An example of this
BSR is a grid design with a substantially large open area between
the grids providing access to the backside for last several
processing steps. Another example of this BSR is a backside
reinforcement with holes which provide access to the underlying
metal which was deposited or patterned while TFSS was on the
template.
[0045] A third embodiment is similar to the BSR in that it also is
a permanent backside reinforcement, but the reinforcement has
integrated structures which obviate the need for substantial
further backside processing. Thus, nearly all process steps on the
backside were finished, while the TFSS was on the template.
[0046] FIG. 2 is a cross sectional diagram of a back contact back
junction thin film solar cell with front side reinforcement and
separated junctions (this specific embodiment hereinafter referred
to as an FSR-SJ cell) with an integrated lambertian mirror. The
back contact back junction structure of this cell allows for
increased conversion efficiency. Throughout this document,
reference will be made to terms abutted (AJ) and separated (SJ)
junctions. Abutted junction refers to the solar cell design where
the local heavy base doping in silicon under the base contact
directly abuts the emitter. Separated junction refers to the cases
where the local heavy base region in silicon under the base contact
is substantially isolated from the heavily doped emitter by the
lightly doped base region. For the AJ and SJ classification only,
the term `junction` refers to the metallurgical junction formed
when a certain concentration of doped region is next to a region of
a different dopant concentration of same or different type.
However, when reference is made to `back junction`, it is
understood that this is an electrical junction formed by adjacent
p+/n or n+/p regions.
[0047] As shown, this FSR-SJ cell utilizes epitaxial silicon
substrate 2 comprised of emitter layer 4, base layer 6, and front
surface field 8. The solar cell backside (or contact side)
structure comprises: backside passivation dielectric 10, back
mirror 12, selective emitter contact 14, base contact 16, seed
metal deposition 18 (such as Al/Ag) on the base and emitter
contacts, and plated metal 20 (such as Ni/Cu/Ni) on the Ag
deposition. The solar cell frontside (or sunny side) structure
comprises: textured front layer 22, and frontside reinforcement 26
attached by adhesive layer 24.
[0048] For example, in one specific embodiment, epitaxial silicon
substrate 2 has n-type (phosphorous-based) base layer 6 and a
p-type (boron-based) emitter layer 4--in practice, these polarities
may be reverse. The epitaxial silicon substrate thickness has a
specific range of values between 15 .mu.m to 30 .mu.m, but in
general may be less than 100 .mu.m. Front surface field (FSF) 8 is
optional depending on the quality of the front surface
recombination, base resistance, and the amount of Auger
recombination.
[0049] The frontside (sunny side) of the fabricated solar cell is
reinforced with a permanent planar carrier (frontside reinforcement
26). Because the carrier is permanent the material must be
transparent, such as PV-grade glass or polymer which are preferred
but not required. Shown, the thickness of the carrier is in the
range of 0.2 mm to 2 mm. The front side reinforcement, which may be
called frontside reinforcement/plate/carrier is mounted on the
solar cell substrate using a thin layer of adhesive (shown as
adhesive layer 24) such as PV-grade EVA, Z68, or silicone.
[0050] The front side silicon surface is textured with the texture
size much smaller than the epitaxial thickness (shown as textured
front layer 22), the texturing in this case leaving the silicon
substrate substantially planar. The textured front side also has an
anti-reflection coating such as PECVD SiN. PECVD SiHxNy:H may be on
top of a thermal oxide or directly on top of the silicon substrate
(shown as epitaxial silicon substrate 2). SiHxNy:H also serves the
very important function of providing high positive fixed charge
density--which may be in the range of 4.times.10.sup.12 cm.sup.-2.
For an n-type base and p-type emitter solar cell, the positive
fixed charge reflects the minority carrier (holes) away from the
surface and prevents them from recombining at the surface which
improves efficiency.
[0051] This diagram shows a back contact back junction solar cell
structure where both contact polarities of metal (to n-type and to
p-type regions) representing the base and the emitter of the solar
cell for extracting electrical current or power from the solar cell
are on the back side. Advantages of a back contact back junction
solar cell have been detailed by others and includes: a) having no
metal on the sunny side thereby eliminating metal
reflection/shading induced efficiency loss, b) the potentially
lower resistance of the metal interconnect lines as they may be
made wider without a metal reflection penalty, c) better blue
response because there is no high doping emitter area in the
frontside of the cell where the blue light is absorbed, d) easier
to connect and put together the completed cell in a module, and e)
better aesthetics--for example the cell may be engineered to be
completely black on the front side (Sunnyside).
[0052] The solar structure shown in FIG. 2 is also characterized by
selective base contacts, shown as base contacts 16. Selective base
contact refers to a solar cell design where base contacts are made
to local, heavily diffused base doping areas in silicon. This is in
contrast with non-selective base contacts, where base contacts are
made to lightly doped (local or non-local) base diffusions.
Selective base contacts, along with minimum contact areas, are an
important efficiency booster because they help minimize the
recombination of the minority carrier at the silicon to metal
interface, which increases open circuit voltage. In addition to
minimizing the base contact, the emitter contact is also minimized
(shown as selective emitter contact 14). The smaller emitter
contact areas also help minimize contact recombination at the metal
to silicon interface, thus improving efficiency through improvement
in the open circuit voltage (Voc).
[0053] Back mirror 12 shown as a patterned reflective layer as an
example, is another feature advantage of this cell. Optionally, the
back mirror may also be formed by the presence of metal on top of
backside passivation dielectric 10. Thermal oxide is an example of
a passivation dielectric which may be used in the present
embodiment. However, other good passivating dielectrics such as,
but not limited to, SiHxNy:H and Al.sub.2O.sub.3 may be used.
Because of interference effects, a controlled passivation layer
thickness under the back metal significantly improves the net
reflectivity of the back mirror stack. The effectiveness of the
back mirror may be defined by two attributes: its net reflectivity
and the degree to which it is capable of diffusing the reflected
light--high reflectivity and high diffusivity are desirable. Metal
such as Al or Ag on top of a dielectric of optimum thickness
(.about.1000 A oxide) can give high reflectivity in the range of
95% while surface roughness and other techniques may be used to
create a diffuse (lambertian) mirror. It is potentially
advantageous to have a high metal coverage on the cell backside
(greater than 85% as an example) to ensure most of the light falls
on the back mirror. Because a larger part of the solar spectrum and
wavelengths fall on the back mirror as the solar cell gets thinner,
reliance of high efficiency on a good quality back mirror is even
more critical for thin silicon. Process flows in this disclosure
outline manufacturing methods which yield high metal coverage with
desirable mirror properties.
[0054] The two features of the cell shown in FIG. 2 which
distinguish it from the other embodiments of back contact back
junction thin film silicon solar cells in this disclosure are
separated junctions and front side reinforcement. The separated
junction attribute refers to the fact the n.sup.++ region where the
base contact is made is substantially isolated from the p.sup.+
emitter layer. This separation mitigates the risk of shunting in
the solar cell and allows high fill factor (FF), which in turn
helps achieve highest possible efficiencies. Additionally,
separated junctions allow high reverse breakdown voltage.
[0055] The thin dimensions of the back contact back junction solar
cell structures of this disclosure requires that the cell is never
processed in the manufacturing line without reinforcement.
Frontside reinforcement 26 in FIG. 2 is an example of a permanent
frontside reinforcement that is part of the manufacturing process
and also may be later utilized in a solar module. A key advantage
of a permanent transparent frontside reinforcement is that it does
not need to be removed because it is made of index matched material
(such as PV grade EVA and glass) which does not compromise light
coupling from the front side. Often, these materials are put on top
of the cell during module assembly.
[0056] FIG. 3 is a process flow showing a fabrication process for
making a back contact back junction thin film solar cell FSR-SJ.
FIGS. 4A through 4k are cross sectional diagrams of the solar cell
after key fabrication process steps as it is manufactured according
to the fabrication process of FIG. 3. The structural features
depicted in the cross sectional diagrams of FIGS. 4A through 4K are
consistent unless otherwise noted. In FIGS. 4A through 4G the
cross-sectional diagrams of the solar cell show the cell with the
frontside (sunnyside) facing downwards and backside
(non-sunny/contact side) facing upwards to better illustrate
processing steps. The cross sectional diagram orientation is
adjusted in FIGS. 4H through 4k.
[0057] As shown in FIG. 3, the fabrication process starts at step
30 with a cleaning of a re-usable silicon template. In general, the
starting template may be any shaped wafer, for example 8 inch round
or 156 mm.times.156 mm pseudo square or full square, with doping
concentration and type (n-type vs. p-type) such that it is
conducive to forming porous silicon. The template may be the same
area as the desired final thin film solar substrate (TFSS) or it
may be oversized allowing for the suppression of detrimental edge
effects on the performance of the solar cell. Further, it may be
any combination of rough or polished on its two sides. In a
specific embodiment, this may be an 8 inch round wafer with p-type
doping in excess of 3.times.10.sup.18 cm.sup.-3 (which is known to
form porous silicon). The re-usable template goes through a general
cleaning process to take out any organic or metal impurities. In a
specific case this can be achieved using the standard RCA clean
consisting of an organic clean (referred to as SC1) and metal
contaminations clean (referred to as SC2 clean). In another
specific embodiment, the cleaning sequence employs an alkaline
etching step which has a finite etch rate of silicon and which is
used to remove surface contaminants--by etching or undercutting.
Potassium-, sodium-, or other hydroxides (KOH, NaOH or other) may
be used in this cleaning step. For metal removal, the use of HCl,
HF or a combination thereof is effective and cost effective due to
bath lifetimes over chemistries that employ peroxides.
[0058] Step 32 involves porous silicon formation by anodic etching.
The template is subsequently put in a process tool which is capable
of forming porous silicon on the flat surface in the front of the
wafer. There are several ways of forming porous silicon, a specific
process consists of using an anodic etch process with Hydrofluoric
acid (HF) and Isopropyl alcohol (IPA) mixture. The anodic etch
process chemically starts to etch away the silicon sporadically
thereby forming a porous silicon at the surface. The reaction
proceeds from the front surface and the thickness of the porous
silicon increases with the time that the wafer is exposed to the
process. FIG. 4A illustrates bi-layer porous silicon 72 formed on
the top surface of re-usable silicon template 70. The
characterization metric for this layer is its porosity. The
porosity at the surface has to be low enough such that after a
thermal reflow step immediately prior to epitaxial deposition, a
single crystal silicon may be epitaxially grown on top of the
porous material. On the other hand, the porosity has to be high
enough such that the layer may be selectively removed while leaving
the template and the epitaxial substrate unharmed. In one specific
implementation, the aforementioned balance is achieved by tuning
the porosity of a single layer. In a different implementation a
multilayer stack with different porosities is used. In particular,
a bilayer stack with the top layer on which silicon epitaxy is done
has a low porosity--for example, but not limited to, 10-35%
range--and the bottom layer in contact with the reusable template
has a high porosity--for example, >40% --is desirable.
[0059] Step 34 entails the growth of the silicon layer on top of
the porous silicon surface by an expitaxial process. A single
crystal epitaxy (epi) is desirable. However, in general, the grown
material may be multi or polycrystalline. An important figure of
merit for this step is the quality of the material as measured by
its minority carrier diffusion length. In general, the merit is
dictated by the crystallinity (single crystal having the highest
lifetime) and the contamination level (no metal, organic, and/or
oxygen contamination desirable) in the film. The advantage of this
approach is that while growing the epi, various diffusions and
dopings integral to a solar cell may be done in-situ. Specific
examples of these doping are base, emitter, and front and back
surface fields. Depending on whether the base doping is n-type
(such as phosphorous based) or p-type (such as boron based), the
solar cell is referred to as NBLAC (n-type) or PBLAC (p-type),
respectively. And the corresponding emitter will have an opposite
polarity. Hence for NBLAC and PBLAC, the emitter will be p-type
(such as boron based) and n-type (such as phosphorous based),
respectively. Although NBLAC is used for descriptive purposes
herein, PBLAC is an alternative embodiment fully within the scope
of the disclosed process and structures. The base doping may be
constant, or continuously varying or graded throughout the
substrate thickness depending on what is needed to drive the
efficiency up. One potential advantage of graded doping is that it
produces a built-in electric field which in turn, allows minority
carriers to drift quickly to the emitter where they are away from
the danger of getting recombined--as opposed to relying entirely on
the random diffusion process. In essence, this increases the
effective diffusion length for a given lifetime of the material.
The impact of an enhanced effective diffusion length is an increase
in the short circuit current density (Jsc).
[0060] The solar cell designs presented in this document are back
contacted back junction cells with the emitter on the backside.
Backside (also the contact side or non-sunnyside) is defined as the
side opposite to where the sun comes in--frontside (sunnyside) is
opposite the backside and defined as the side where the sun comes
in. Thus, the two defining features of back contact back junction
solar cells are: 1) all metal connections are on the non-sunny
side, and 2) the emitter is also on the non-sunnyside. In the
embodiment described, the emitter is grown last in the epitaxial
process to prevent its dopant profile from being affected by
excessive thermal budget. This embodiment allows the backside
processing to be done on the template as this side is the exposed
side while on the template. For NBLAC the base doping may be
achieved using phosphine (phosphorous imparting) gas in addition to
silicon imparting (trichlorosilane--TCS) and other necessary gases,
and when emitter growth is done toward the end of the epitaxial
process the reactor may be programmed to switch to diborane (boron
imparting) instead of phosphine. The thicknesses of the base and
the emitter region should be optimized to give the best solar cell
performance. In one embodiment, a base thickness less than 100
.mu.m with doping between 5e14 and 1e17 cm-3 and an emitter
thickness of less than 3 .mu.m with doping between 1e18 and 3e20
cm-3 are preferred (the base thickness may be as thin as 15 to 30
.mu.m). The emitter may also be multi-step with each step resulting
in different concentration. This epitaxial structure can facilitate
a high open circuit voltage (Voc) of the solar cell and thus a
higher efficiency.
[0061] In another embodiment, the emitter is not grown in-situ as
part of the epitaxial process, but rather ex-situ after the
epitaxial deposition process. This can be accomplished among others
by the deposition of a boron containing pre-cursor such as boron
doped oxide, (BSG), deposited using atmospheric pressure chemical
vapor deposition (APCVD), followed immediately or later in the
process flow by a subsequent anneal that drives the emitter dopant
in.
[0062] A front surface field (FSF) is optional. This consists of a
heavier doping of the same kind as the base, grown epitaxially
first on porous silicon before the base. For NBLAC this may be
achieved using phosphorous doping. A front surface field,
generally, has two advantages and a disadvantage. The first
advantage is that an FSF shields the photogenerated, minority
carriers from recombining at the front surface (where there is a
lot of recombination) by reflecting them away from that surface
through an electric field. The second advantage is that it helps
mitigate the base resistance by providing a path for electrons to
travel through a lower resistance area. The disadvantage stems from
a larger recombination in the FSF area itself due to Auger
recombination. In general, a good front passivation will obviate
and a bad front passivation will necessitate the use of FSF.
[0063] FIG. 4B illustrates epitaxial silicon substrate 74 comprised
of n+ front surface field layer 76, n-type base 78, and in-situ
doped thin p+ emitter layer 80. The n-type base epitaxial silicon
layer is grown on the porous silicon surface of the re-usable
silicon template.
[0064] As shown in FIG. 3, the step 36 is to form shallow trenches
on the silicon surface by direct-write silicon etching or ablation.
The shallow silicon trenches are henceforth referred to as trench
isolation. The purpose of this etch is to nest the base finger and
busbar areas and to isolate the base region from the emitter
region, as is the defining characteristic of this particular
separated junction (SJ) design, and to provide access to the base
for base metal contact. In a different embodiment, the base metal
fingers and/or the bus bars may not be nested. As shown in trench
82 in FIG. 4C, the depth of shallow trench should be such that it
is deeper than the emitter layer thickness, such that the emitter
is locally completely removed (shown as emitter layer 80 in FIG.
4B). In a specific embodiment if the emitter region is shallower
than 0.5 m, the silicon etch depth may be between 0.5 to 1 .mu.m.
This etched trench pattern follows the same pattern as the base
metal fingers and busbar, except it is wider to nest the actual
metal finger and busbar areas, if nesting is required. In this
embodiment, the base metal fingers and busbar are deposited within
the shallow trenches without overlapping to the non-trenched
emitter layer. The shallow trench width is preferred to be as small
as possible. The motivation for keeping the trench widths small is
that this width constitutes the extent of interruption in the
emitter. Thus, wider trenches increase the chance of minority
carriers to recombine under the trench area,--a phenomenon known as
electrical shading.
[0065] In a preferred implementation, trench isolation may be
performed using a direct write laser ablation of silicon or a
direct etch using laser assisted halogen-based chemicals. Utilizing
processes such as these is important because it uses all dry,
in-line, non-contact processing, which avoids TFSS damage during
on-template thermal processes. Further, a direct write laser etch
or ablation process allows formation of the SJ architecture with
only one additional process step compared to the abutted junction
(AJ) architecture. For achieving the highest cell efficiency, it is
desired that the silicon be ablated without any thermal damage to
the silicon substrate, the so-called `cold ablation` process. The
`cold ablation` of silicon is possible using low picoseconds or
femtoseconds pulse width lasers. In cases where wet processing is
acceptable, the trench isolation can also be done using patterning,
followed by wet silicon etchants such as KOH. In another
implementation the recess in silicon or the trench isolation may be
done by screen printing etch paste which etches silicon. This
implementation requires extra steps of firing the etch paste and
cleaning the residue. However, a cleaning process using wet
cleaning is less desirable on the TFSS, but may be made
advantageous. In yet another implementation scheme, forming the
silicon shallow trench may be achieved by first patterning a
masking layer then performing silicon plasma or reactive ion
etching (RIE). After the shallow trench etching, the patterned
masking layer is removed and followed by a corresponding substrate
cleaning step.
[0066] As shown in FIG. 3, the next step, step 38, is to deposit a
protective dielectric layer. Dielectric layers include, but are not
limited to, thermally grown SiO.sub.2 and SiN. A general property
of the dielectric layer is that it should be a good passivation
layer. Further, the physical thickness of the dielectric layer is
governed by a balance between two factors: (1) the layer should be
thin enough so that it may easily patterned by the subsequent laser
ablation step, and at the same time (2) the layer should be thick
enough so that it is capable of blocking diffusion of gas phase
dopant where it is not opened if such a dopant is used to form the
junctions (as is the case for a non-selective emitter structure
which is detailed later). Gas phase dopant diffusion is a
subsequent fabrication step and will be detailed later. In a
specific embodiment, this protective layer is thermally grown oxide
with thickness in the range of 100 nm to 250 nm. The dielectric
layer growth/deposition temperature and time may be selected to
target an appropriate thickness. One consideration in picking the
oxide growth conditions is to minimize the segregation of the boron
emitter into the grown oxide. This requires minimizing the thermal
budget of oxidation (favors a wet ambient process) and growing at
an elevated temperature where less boron segregation is
achieved.
[0067] Then, in step 40, interdigitated contact openings in the
aforementioned dielectric layer exposing the underlying silicon are
formed. A pattern in which the dielectric will be opened is the
inter-digitated fingers and bus bar, where the base and the emitter
lines are separated and continuous. The specifics of the pattern as
defined by standard dual bus bar or distributed bus bars will be
detailed in ensuing discussion. The purpose of the base and emitter
contact openings is for subsequent selective doping. In an NBLAC
embodiment, base contact openings will be doped heavily with n-type
phosphorous material and emitter contact openings will be doped
with p-type boron. Both base and emitter opening regions are opened
simultaneously in this step. A specific implementation of this step
may be carried out using a direct laser ablation of the oxide
layer. A pulsed picosecond laser in visible or UV wavelength is
conducive to ablating an oxide layer. FIG. 4D illustrates emitter
contact opening 84, base contact opening 86, and front and back
side SiO.sub.2 layer 86.
[0068] Step 42 involves applying both n-type and p-type dopants
selectively over the base and the emitter contact open
areas--following interdigitated pattern previously defined. The
dopants should cover the openings and may have a slight overlap
with the dielectric layer and/or go on top of it. For the NBLAC
specific embodiment: on the emitter area this dopant has to be
p.sup.++ type (for instance boron) and on the base contact area it
has to be n.sup.++ type (phosphorous based). A specific method of
implementation of the dopants is using the Inkjet printing
technique. In addition, specific examples of the inks that may be
deposited are silicon nano-particle based phosphorous and boron
inks. This step is followed by an optional step of using inkjet
printer to print all cell areas (or areas excluding laser-ablated
contacts) with undoped Si (or glass) nano-particle ink. This is
followed by sintering the ink as required by specific ink handling
instructions. The purpose of the undoped ink is to use it to
randomly texture the oxide surface--random texturing improves the
Lambertian properties of the back mirror and will enhance cell
efficiency (discussed subsequently). FIG. 4E illustrates
inkjet-printed emitter dopant 88, inkjet-printed base dopant 90
(optionally sintered), and undoped inkjet-printed material 92
(nano-particles that form a blanket textured surface layer).
[0069] Then in step 44 the inkjet-printed boron, phosphorous, and
undoped ink is annealed to form n++ and P++ emitter contact
regions. Optionally, the annealing step may either be followed by
or be integrated with another anneal in a low O.sub.2 or steam
environment, which serves to oxidize the undoped silicon particles
and create randomly textured oxide surface. FIG. 4F illustrates P++
selective emitter region 94, n++ selective base region 96, and
surface-textured silicon oxide layer (SiO.sub.2) 98.
[0070] Then in step 46 the TFSS is separated from the reusable
template using dry or wet separation or a combination of
techniques. At this step the TFSS--with the aforementioned base and
emitter patterns and doping--is separated from the reusable
template. The removal occurs along the interface formed by the
sacrificial porous silicon. In a specific embodiment, this is
accomplished using a shallow trench laser scribe in a pattern
conducive to the final shape of the solar cell. Amongst several
possibilities this laser scribe may be 125 mm.times.125 mm or 156
mm.times.156 mm pseudo squares or oversized with respect to the
final solar cell size. The laser scribe is subsequently followed by
clamping on the base/emitter contact side and a mechanical release
where it is pulled away from the template. The clamping may be
performed using vacuum forces and assisted or accomplished by means
of an electrostatic chuck (ESC) or a mobile electrostatic carrier
(MESC). For descriptive purposes, an MESC is cited wherever a
mobile carrier is employed. The separation of the TFSS from the
template typically occurs at the highest porosity portion of the
porous silicon layer. Optionally, after the separation, a final
cutting procedure to size the TFSS may be performed. FIG. 4G
illustrates a backside supported TFSS with the released process
TFSS 100 separated from reusable template 102 and supported by
temporary TFSS carrier (such as an MESC) 104.
[0071] Step 60 shows the template reconditioning and cleaning in
preparation for another use. The epitaxially grown silicon outside
the TFSS area on the template is removed and the template is sent
back for another cycle of reuse defined by template clean, porous
silicon, and epi growth. In a specific implementation the removal
of the extraneous silicon may be performed using mechanical
grinding/lapping or polishing--which may be performed on the top,
along the edges, and on the backside of the template and serves to
remove extraneous silicon where needed. These steps can be carried
out with every re-use or once every several reuses.
[0072] At this point in the process the separating TFSS may be
self-supporting if the thickness of the TFSS is greater than 75
.mu.m. However, when the TFSS is thin--such as thinner than 50
.mu.m--it should be supported. In a specific implementation, the
mechanical clamping during the release step or after the release
step is performed by a mobile ESC (MESC). The MESC subsequently
latches on to the thin TFSS and, importantly, supports the thin
TFSS as a temporary carrier on the cell backside (the non-sunny
side where the base and emitter patterns exist). The frontside
(sunny side) is now exposed.
[0073] Then in step 48, the frontside of the cell is cleaned and
any remaining debris of porous silicon (henceforth referred to as
quasi monocrystalline silicon or QMS) is removed. The clean may be
performed using a single sided wet clean where the cleaning etchant
only touches the TFSS frontside while TFSS is being held on the
backside by the MESC. The backside of TFSS is protected from
etching through the MESC, or equipment suitable for the application
of single side etch processes are used. Amongst various wet
cleaning options, HF/nitric acid, HNA, TMAH, and KOH based silicon
etch may be used.
[0074] The QMS clean is followed by texturing of the front side
surface. The texturing wet chemistry may be the same as QMS removal
chemistry (for example, but not limited to, KOH), resulting in both
tasks being accomplished in a single step. Alternatively, there may
be a series of chemical treatments resulting in QMS removal and
texturing. After texturing, the frontside surface of the substrate
is treated. In one embodiment, the substrate undergoes several
steps including a metal removal clean, typically using qualified
acids or acid combinations such as hydrofluoric acid (HF) or
hydrochloric acid (HCl). Optionally, an organic removal such as
ozone is applied. In other embodiments, chemical oxides of good
quality to enable a low density of surface states are employed as a
last wet process step prior to deposition of the passivation
layer--which in the case of silicon nitride may also be the
anti-reflection layer (ARC).
[0075] In FIGS. 4H through 4K the cross-sectional diagrams of the
solar cell have been adjusted (turned 180.degree.) to better
illustrate subsequent processing steps. Accordingly, the cell is
now depicted with the frontside (sunnyside) facing upwards and
backside (non-sunny/contact side) facing downwards. FIG. 4H
illustrates the backside supported TFSS with textured frontside
surface 104.
[0076] One alternative embodiment which includes a slight variation
of the process flow shown in FIGS. 4A through 4H occurs when the
template is pre-textured. In this case, the substrate TFSS is still
substantially planar, however, the released TFSS is already
textured which obviates the need to perform wet texturing of the
TFSS after QMS removal. The remaining of the process flow remains
the same as discussed in FIG. 3 and all the design and process
variations discussed in the context of previous flow remain
applicable to the pre-textured TFSS embodiment.
[0077] Next, in step 50 single-sided surface passivation and
anti-reflection coating (ARC) is performed. In a specific
embodiment this may be achieved using microwave plasma enhanced
chemical growth of silicon nitride. In another embodiment, this may
be a silicon dioxide layer followed by PECVD SiH.sub.xN.sub.y:H
layer. In yet another embodiment, the passivation structure may be
an amorphous silicon layer followed by silicon nitride. In each of
the above embodiments, the deposition temperature may be tailored
between 100 and 400 deg C. to match the need and capability of the
material already in the cell stack. This passivation serves not
only the critical function of reducing the surface recombination
velocity on the cell frontside (which helps significantly boost
efficiency of a back contacted cell), but also serves as an
anti-reflection coating (ARC) which enables strong coupling of
light into the solar cell. FIG. 4I illustrates the backside
supported TFSS with frontside surface passivation and
anti-reflective coating (ARC) layer 106.
[0078] As shown in FIG. 3, in step 52 the temporary backside
support is converted to a permanent front side reinforcement. With
the passivation/ARC layer created on the frontside (sunnyside),
only backside processing of the back contact back junction cell
remains. A specific implementation of this frontside reinforcement
is to use thin PV glass attached by PV Silicone, Z68 or EVA
material. An advantage of reinforcing the front side with this
material is that this reinforcement can be permanent because when
the solar cell is packaged into a module, this same material often
goes on the frontside of the cell. Thus, this type of reinforcement
ensures that there is no degradation of optical performance of the
finished module. Once this front side reinforcement is complete,
the temporary backside support, such as the MESC, is released
through electrostatic discharge action and the cell backside
(non-sunny side) is available for final metallization processes. It
is important to note that at no point during the support conversion
is the TFSS not supported/reinforced. FIG. 4J illustrates the
permanent frontside reinforcement 108 (using a material such as
glass) and reinforcement attachment layer 110 (using materials such
as PV-grade EVA or silicon to attach the frontside
reinforcement).
[0079] In step 54 the frontside of the reinforced TFSS is cleaned
to remove the dopant residue (such as from the Inkjet) and clean
the contact areas for a good metal adhesion and electrical contact.
The cleaning etchant may be HF to remove the dopant residue and
surface oxide and/or a mild selective silicon etch to clean the
area. In another embodiment, the process residue from the inkjet
can be removed by dry techniques such as pulsed laser etching,
thereby exposing the underlying silicon for contact formation. This
laser process can also be performed right after the dopant
diffusion when the TFSS is still on the template carrier.
[0080] Step 56 involves metallization. And although there are
several ways to perform metallization, a specific implementation
with a few variations is described. After cleaning, Silver (Ag) or
aluminum (Al) nano particle ink is selectively deposited on top of
both the base and the emitter contact areas. The ink is deposited
such that it follows the shape of the emitter and the base fingers
and busbars. However, the ink deposition is optionally much wider
than the respective finger and busbar widths to create a full area
back mirror. This metal coverage may be >85% to ensure that most
of the light on the back falls on metal and is then reflected back
into the structure for more absorption passes. Care must be taken
to ensure that metals from the opposite polarity do not touch each
other. In an alternative embodiment, the same final structure may
be achieved using blanket Al deposition (by using evaporation or
PVD, for example) and laser ablating the Al deposition to form the
gap areas. In yet another alternative embodiment, metallization
occurs by depositing Ag or Al ink to both the base and emitter
contact areas and forming continuous metal fingers and busbars.
After sintering the ink, electroplating steps are conducted by
using the metal layer as the electroplating seed layer. For the
case of Al ink, a second ink can optionally be applied in similar
fashion prior to electroplating. The purpose of the added
electroplating steps is to provide increased electrical
conductivity in a cost-effective manner. As an example, a metal
stack of nickel/copper/nickel (Ni/Cu/Ni) with Cu thickness in the
range of 10 to 50 .mu.m may be electroplated. The thin Ni layer
under the Cu serves as a Cu barrier to prevent its diffusion into
silicon, while the Ni layer on top of the Cu layer serves as a
passivation layer to prevent Cu surface oxidation and corrosion. In
yet another alternative embodiment, instead of forming the back
mirror by using the described Ag ink metallization or through an
evaporated/PVD Al layer, a reflective insulating layer such as
inkjet-printed may be deposited to serve as the back mirror. Using
this approach, metal on the base side and on the emitter side
remain isolated from each other and potential electrical shunting
through the underlying dielectric layer is avoided. FIG. 4K
illustrates the fabricated back contact back junction solar cell
with selective emitter contact 112, base contact 114,
inkjet-printed patterned reflective layer 116, plated metal 120
(such as Ni/Cu/Ni), and seed metal deposition 118 (such as
inkjet-printed Ag and depicted in the diagram as the black region
positioned between plated metal 120 and P++ selective emitter
region 94).
[0081] FIG. 5 is a diagram showing the backside (non-sunnyside) of
a back contact solar cell (NBLAC or PBLAC) highlighting the back
contact interdigitated emitter and base metal fingers and busbars
of the solar cell. The inter-digitated finger and busbar pattern is
shown as an example and other patterns are possible. The shallow
silicon trench region is positioned under the base metal fingers
and busbar. As previously described, the trench depth is slightly
deeper than the emitter layer thickness. The trench width should be
minimized so that electrical shading may be kept at a minimum. In
this particular embodiment, the metal patterns are formed by inkjet
printing, such as inkjet printing of Ag or Al inks followed by a
sintering process at an elevated temperature. Also in the
embodiment shown, the back mirror is formed by a reflective
dielectric layer. As shown in FIG. 2 and FIG. 4K the base metal
layer does not overlap with the emitter regions and as a result,
potential electrical shunting across the backside surface
passivation layer is avoided. However, as a tradeoff, the thickness
of the metal layer should be thick (thicker than 50 .mu.m in some
instances) so that the metal layer could handle the electrical
current and power extraction requirement of the solar cell. For the
particular interdigitated fingers and busbar design shown in FIG.
5, the length of the interdigitated fingers may equivalent to the
width of the solar cells themselves, close to 125 mm or 156 mm long
in some instances. The width of the metal fingers may be in the
range of 100 .mu.m to 500 .mu.m and the pitch between adjacent
metal fingers in the range of 0.5 mm to 3 mm.
[0082] FIGS. 6A through 6D are diagrams showing the backside of a
back contact BLAC solar cell highlighting an alternative
metallization pattern of base and emitter contact openings after
key fabrication steps. FIGS. 7A through 7D are cross sectional
diagrams of the solar cell illustrating the metallization
embodiment of FIGS. 6A through 6D. In this embodiment, the starting
TFSS (a front side reinforcement with separated junctions cell) for
metallization corresponds to the cell shown in FIG. 4J.
[0083] FIG. 6A illustrates the top view of the emitter and base
contact openings with the base contact opening nested in a shallow
silicon trench. Corresponding FIG. 7A illustrates the
cross-sectional view of the base contact region (emitter region
omitted to simplify the drawing). The emitter contact is similar
except that it is positioned on the emitter surface rather than in
the shallow silicon trench where the base contact is
positioned.
[0084] FIG. 6B and corresponding FIG. 7B illustrate at least one
deposited thin metal layer with a narrow isolation gap to separate
base and emitter metal regions. The thin metal layer serves two
purposes: 1) to provide an electroplating seed layer for subsequent
thick metal stack plating, and 2) to provide the back mirror. The
thin metal layer may be deposited by metal inkjet printing (such as
Ag or Al ink) followed by sintering. In this direct-write process,
the metal isolation gap is formed as printed therefore no extra
thin metal patterning step is needed. In an alternative method, the
thin metal is blanket deposited by metal (such as Al or a stack of
Al/NiV/Sn) evaporation or PVD. Next, forming the metal gap may be
achieved by direct laser ablation with controlled ablation depth so
that the laser power does not damage the silicon surface under the
narrow isolation gap. Alternatively, a masking layer may be screen
printed on the Al back mirror surface followed by a chemical
etching of the exposed thin Al layer to create the narrow isolation
gap. The etching masking layer is then removed after Al
etching.
[0085] As shown in FIG. 6C and corresponding FIG. 7C, the next step
is to deposit an insulator material (a dielectric layer) covering
the gap areas. The dielectric material may overlap slightly with
the thin metal on both sides (as depicted in FIG. 6C). The utility
of this dielectric layer includes acting as a protection layer from
the subsequent plating step--if there are any cracks, scratches or
undesirable defects in the underlying oxide layer between the metal
lines (these can be created by handling and the process steps post
oxidation), plating being a wet process will plate in these defects
thus connecting and shunting the two metal lines. Depositing a
dielectric layer in the gaps ensures that any defects are plugged
and there is no undesirable plating. The dielectric insulating
layer may be applied through an inkjet print, screen print, or
blanket deposition.
[0086] As shown in FIG. 6D and corresponding FIG. 7D, the next step
in metallization is to plate metal selectively on top of the
patterned thin metal layer. A specific embodiment includes a
Ni/Cu/Ni stack. Ni serves as a barrier to Cu, while Cu may be as
thick as needed to ensure a low resistance and a good Fill-Factor.
The metal layer's thickness depends on the busbar design. Other
metals such as Ag (although expensive) are also possible. A
preferred method of plating is electroplating which ensures
selective plating only on the underlying conductive areas--thus
preventing shunt. The metallization sequence shown here completes
the entire cell flow, but may be followed by an optional forming
gas anneal to improve open circuit voltage (Voc).
[0087] FIGS. 8A through 8C are cross-sectional diagrams of the
solar cell showing the formation of a Lambertian mirror after key
fabrication steps. The term "Lambertian" refers to the quality of
the back mirror by which it is able to diffuse the reflected light
in all directions as opposed to specular reflection where light
retains the memory of its incidence angle and is reflected at an
angle equal to the incidence angle. The diffuse or Lambertian
reflector has the advantage that it increases the path length for
light giving light more interaction distance with silicon for
absorption before the light reaches the opposite side of the cell.
This attribute is especially critical for very thin solar cells
ranging from 15-30 .mu.m because for thin solar cells even shorter
wavelengths of light will hit the back mirror. In addition, it has
been shown with simulations that using a Lambertian mirror gives
the solar cell a high efficiency which is more immune to lifetime
variations--an attribute that provides a major binning advantage in
manufacturing. Two methods of making a Lambertian mirror integrated
with the previously disclosed process flow are provided. In the
first implementation, the cell is textured on the backside prior to
oxidation, using silicon etch texturing for example. This is
followed by deposition of rear metallic reflector (such as Al or
Ag) using inkjet printing of metallic nano-particle ink or PVD of
metal (Al or Ag). A potential consideration of this method includes
possible emitter shorts due to texturing etch punching through the
emitter junction area. In addition, it might potentially increase
the backside surface recombination velocity resulting in a higher
emitter dark current density and poor Voc.
[0088] Another method of making a Lambertian mirror (diffuse
mirror) was implicitly disclosed in the above process flow and
shown in FIGS. 8A through 8C. The method is to texture the backside
passivation dielectric (e.g. thermal oxide) by inkjet printing of,
preferably undoped, silicon nano-particle ink followed by sintering
and oxidizing to form a textured, rough backside surface. FIG. 8A
shows the deposited silicon nano-particles on the rear cell
passivation dielectric layer. The passivation dielectric may be a
thin layer of thermal oxide with a thickness in the range of 100 nm
to 250 nm. FIG. 8B shows the cell after the thermal oxidation of
the silicon nano-particles. Alternatively, the backside may be
coated by inkjetting glass (quartz) nano-particles as well.
Subsequently, as shown in FIG. 8C, the backside diffuse mirror
formation is completed by inkjet printing of metallic nano-particle
ink or PVD of metal (Al or Ag). This method of forming a Lambertian
(diffuse mirror) occurs before the TFSS release.
[0089] FIGS. 9A and 9B are diagrams illustrating two metal busbar
designs in accordance with the disclosed subject matter. An
important attribute of a thin, yet high efficiency cell design is
the busbar design. The standard busbar design, shown in FIG. 9A, is
a dual bus bar design with inter-digitated metal pattern and may be
used with the disclosed back contact cells. However, this design
may require thick metal in the back because the current has to be
carried by the fingers all the way from one edge of the TFSS to the
other. The line presents a large resistive loss of power. A thicker
metal typically in the less 30 .mu.m range is applicable for a
standard silicon cell which is greater than 150 .mu.m thick.
However, thin silicon (15 .mu.m to 50 .mu.m) back contacted solar
cells, such as those in this document may not be able to withstand
the stresses of less than 30 .mu.m thick Cu metal lines.
[0090] FIG. 9B illustrates an alternative distributed busbar design
for a very high efficiency, thin, back contacted cell. In this
embodiment, there are N bus bars for emitter regions and the same
number for the base regions. An advantage of this design is that
the thinner fingers are responsible for carrying the current for a
much shorter distance--dramatically mitigating the resistive
losses. All the emitter busbars are connected together and the base
busbars are connected together. Compared to the standard N=1 (dual
busbar), for N pairs of busbars the busbar current is reduced by a
factor of N. This allows the Cu thickness to be reduced by a factor
of N without compromising the resistive losses--enabling Cu
thickness between 5-10 .mu.m for N=3 (an N=3 embodiment is shown in
FIG. 9B) and N=4. For thin silicon cells this is a major advantage.
However, one potential issue with distributed bus bars may be
increased contact recombination and electrical shading because of a
larger thickness metal. This may be mitigated by making a slotted
busbar design in which the contact to underlying silicon is in
slots, but the overhanging metal joins together to form a
continuous line. This method requires that the spacing between the
slots is no more than twice the thickness of the metal. Note that
the busbar design is decoupled from the process flow discussed
above as it only dictates the pattern in which the laser ablates
the dielectric and the thin metal layer.
[0091] FIG. 10 is a cross sectional diagram of a back contact back
junction thin film solar cell with front side reinforcement and
abutted junctions (this specific embodiment hereinafter referred to
as an FSR-AJ cell) and integrated lambertian mirror in accordance
with the disclosed subject matter. In this embodiment, the base and
emitter junctions are abutted because there was no a pre-fabricated
shallow silicon trench to nest the base junctions and contacts. As
a result, the base contact doping has to be strong enough such that
it overcomes and counter-dopes the underlying and already existing
emitter layer under it.
[0092] FIG. 11 is a process flow showing a fabrication process for
making a back contact back junction thin film solar cell FSR-AJ.
FIGS. 12A through 12J are cross sectional diagrams of the solar
cell after key fabrication process steps as it is manufactured
according to the fabrication process of FIG. 11. The structural
features depicted in the cross sectional diagrams of FIGS. 12A
through 12J are consistent unless otherwise noted. In FIGS. 12A
through 12G the cross-sectional diagrams of the solar cell show the
cell with the frontside (sunnyside) facing downwards and backside
(non-sunny/contact side) facing upwards to better illustrate
processing steps. The cross sectional diagram orientation is
adjusted in FIGS. 12H through 12J.
[0093] The described process flow of the FSR-AJ is identical to the
FSR-SJ embodiment described in FIG. 3 and FIGS. 4A through 4K with
one important difference: there is no trench isolation step to
separate the emitter and the base junctions in the case of FSR-AJ.
In this case the p.sup.+ and the n.sup.+ regions are abutted. Thus,
after opening contact areas the emitter region is exposed to the
base area. So when the dopant phosphorous dopant is applied using
inkjet, the drive-in has to be strong enough such that it overcomes
and counterdopes the underlying and already existing boron based
emitter. To facilitate this, the emitter will have a restriction of
being fairly shallow. All variations and nuances discussed above
for FSR-SJ flow are equally applicable to the FSR-AJ device
including distributed busbar design, MESC approach, lambertian
mirror strategy and various metallization strategies.
[0094] FIG. 13 is a cross sectional diagram of a back contact back
junction thin film solar cell with backside reinforcement and
separated junctions (this specific embodiment hereinafter referred
to as an BSR-SJ cell) and integrated lambertian mirror in
accordance with the disclosed subject matter. Importantly, the
BSR-SJ finished cell does not have the front side reinforcement in
the form of EVA/glass material stack as in the frontside
reinforcement embodiments previously disclosed. Rather,
reinforcement is on the backside, which may be in a shape conducive
to providing mechanical strength to the standalone TFSS.
[0095] FIG. 14 is a diagram showing a backside schematic view of a
grid-shaped backside reinforcement that may be utilized with the
BSR-SJ cell. The grid-shaped backside reinforcement provides
mechanical support for the thin TFSS substrate and is formed of
intersecting grid lines. The width of the grid lines may be in the
range of 0.3 mm to 1 mm and the thickness of the grid line can be
in the range of 5 .mu.m to 300 .mu.m. The opening shapes--occupying
the space between grid lines--may be square, rectangular, circular,
or an alternative shape. In the case of square shape openings as
shown in FIG. 14, the size of the squares may be in the range of 5
mm.times.5 mm to 50 mm.times.50 mm.
[0096] FIG. 15 is a process flow showing a fabrication process for
making a back contact back junction thin film solar cell BSR-SJ.
FIGS. 16A through 16J are cross sectional diagrams of the solar
cell after key fabrication process steps as it is manufactured
according to the fabrication process of FIG. 15. The structural
features depicted in the cross sectional diagrams of FIGS. 16A
through 16J are consistent unless otherwise noted. In FIGS. 16A
through 16G the cross-sectional diagrams of the solar cell show the
cell with the frontside (sunnyside) facing downwards and backside
(non-sunny/contact side) facing upwards to better illustrate
processing steps. The cross sectional diagram orientation is
adjusted in FIGS. 16H through 16J.
[0097] The process flow for making BSR-SJ is also similar to that
for manufacturing FSR-SJ (FIG. 3), except for a few key exceptions.
First, after the n.sup.++ and p.sup.++ base and contact region
furnace anneal step on the template, FSR-SJ goes through a laser
cut and release, while in the BSR-SJ fabrication flow a patterned
grid shaped permanent reinforcement is attached on the cell
backside. As shown in FIG. 16G, the reinforcement material should
be capable of allowing subsequent processing up to 450.degree. C.,
or at least 250.degree. C., and have minimal optical absorption
with very low loss. In a specific instance, the reinforcement
material can be may of hard plastic, PTFE, or PV-grade silicone
glue. The grid-shaped reinforcement material may be applied by
screen printing, inkjet printing, or laser stereolithography rapid
prototyping tool. Alternatively, a prefabricated grid structure
from a suitable material such as PTFE or other high temperature
polymeric or fiber/polymer composite materials may be laminated and
thermally fused to the TFSS backside surface prior to the TFSS
releasing. Since the next step involves the release TFSS, this
reinforcement step allows further processing on a thin TFSS while
it is free standing. This reinforcement may also be permanent
because it only occurs on the backside (the nonsunny side). The
frontside is ready for processing after release and can go through
TFSS clean, texturing, and dielectric (e.g. SiHxNy:H) passivation
deposition supported by the backside reinforcement. And the
backside metallization may be processed with the reinforcement in
place. The cell metallization distributed busbars and
interdigitated fingers may run vertically and horizontally with
respect to the Grid-Shaped patterns--preferably in an N.times.N
distributed busbar pattern. Note that FIG. 9B shows a specific
example of Nxl distributed busbar design. An N.times.N design
allows the backside reinforcement to be placed between the bus
bars.
[0098] The FSR designs have to use a temporary (such as an MESC)
carrier on the backside before the front side reinforcement is put
in place--this reinforcement is temporary and detaches after the
frontside is reinforced. In contrast, in BSR designs the
reinforcement on the backside is permanent so the fabrication
process does not require the extra step of frontside reinforcement.
Except for the aforementioned differences, all pre-release steps
and their sequence in BSR-SJ such as, template clean, porous
silicon formation, silicon trench recess, thermal oxidation,
interdigitated pattern and inkjet print/anneal are common may be
identical to an FSR-SJ process such as that described in FIG. 3. In
addition, several post-release steps of TFSS clean, texturing, and
SiHxNy:H deposition are common between FSR-SJ and BSR-SJ as well.
The BSR-SJ fabrication flow may also share all the variations and
nuances described for the case of FSR-SJ fabrication flow
above.
[0099] FIG. 17 is a cross sectional diagram of a back contact back
junction thin film solar cell with backside reinforcement and
abutted junctions (this specific embodiment hereinafter referred to
as an BSR-AJ cell) and integrated lambertian mirror in accordance
with the disclosed subject matter. Importantly, the BSR-AJ finished
cell does not have the front side reinforcement in the form of
EVA/glass material stack as in the frontside reinforcement
embodiments previously disclosed. Rather, reinforcement is on the
backside, which may be in a shape conducive to providing mechanical
strength to the standalone TFSS.
[0100] FIG. 18 is a process flow showing a fabrication process for
making a back contact back junction thin film solar cell BSR-AJ.
FIGS. 19A through 19I are cross sectional diagrams of the solar
cell after key fabrication process steps as it is manufactured
according to the fabrication process of FIG. 18. The structural
features depicted in the cross sectional diagrams of FIGS. 19A
through 19I are consistent unless otherwise noted. In FIGS. 19A
through 19F the cross-sectional diagrams of the solar cell show the
cell with the frontside (sunnyside) facing downwards and backside
(non-sunny/contact side) facing upwards to better illustrate
processing steps. The cross sectional diagram orientation is
adjusted in FIGS. 19G through 19I.
[0101] The process flow for making BSR-AJ is also similar to that
for manufacturing FSR-AJ (FIG. 11), except for a few key exceptions
discussed above in the case of the BSR-SJ relating to backside
reinforcement.
[0102] All of the described cell embodiments (FSR-SJ, FSR-AJ,
BSR-SJ, and BSR-AJ) share a common feature of selective emitters.
This refers to an attribute whereby the doping concentration of the
p-type material using boron (for NBLAC rendition) under the contact
area is higher than the boron doping elsewhere (where there is no
contact) in the boron doped thin layer. An advantage of this
structure is that it may yield a higher Voc, and thus higher cell
efficiencies. However, in an alternative embodiment of the
disclosed subject matter, the solar cells may not have selective
emitters. Thus, the doping of the emitter (in the NBLAC case, boron
doping) is the same under the contact and elsewhere. While this
variation in the emitter attribute may be applied to all four of
the aforementioned structures (FSR-SJ-NSE, FSR-AJ-NSE, BSR-AJ-NSE,
BSR-SJ-NSE), the following describes an FSR-AJ-NSE as an example to
disclose the non-selective emitter design and method of
fabrication.
[0103] FIG. 20 is a cross sectional diagram of a back contact back
junction thin film solar cell with frontside reinforcement, abutted
junctions, non-selective emitters (this specific embodiment
hereinafter referred to as an FSR-AJ-NS cell) and integrated
lambertian mirror in accordance with the disclosed subject matter.
The only difference compared to the FSR-AJ embodiment of FIG. 10 is
that the emitter contacts of FSR-AJ-NSE are directly made on the p+
emitter surface.
[0104] FIG. 21 is a process flow showing a fabrication process for
making a back contact back junction thin film solar cell FSR-AJ-NS.
FIGS. 22A through 22J are cross sectional diagrams of the solar
cell after key fabrication process steps as it is manufactured
according to the fabrication process of FIG. 21. The structural
features depicted in the cross sectional diagrams of FIGS. 22A
through 22J are consistent unless otherwise noted. In FIGS. 22A
through 22F the cross-sectional diagrams of the solar cell show the
cell with the frontside (sunnyside) facing downwards and backside
(non-sunny/contact side) facing upwards to better illustrate
processing steps. The cross sectional diagram orientation is
adjusted in FIGS. 22G through 22J.
[0105] In comparison to the FSR-AJ case of FIGS. 10 and 11, only
the inkjet printing of phosphorous is performed to form the
selective base contacts while the inkjet printing of boron step
shown in previously described process flows is omitted. Therefore,
the emitter metal layer is directly applied onto the p.sup.+
silicon thin layer. In an alternative embodiment, the inkjet
printing of phosphorous is replaced by using POCl doping. An
advantage of using POCl is that it is a mature process and has been
in used production for a long time. FIG. 21 and FIG. 22D illustrate
and describe the POCL-based process flow of making the FSR-AJ-NS.
The template clean, porous silicon, Epi, and thermal oxidation
steps are same as described previously with inkjet dopant based
selective emitters. Note all variations of that process are equally
applicable to this process including, but not limited to using an
MESC as a temporary carrier. Post oxidation, however, the two flows
(FSR-AJ-NS and FSR-AJ) diverge. In the POCl (non-selective emitter)
flow describes and shown in FIG. 21 and FIGS. 22A through 22J,
oxidation is followed by laser ablation only of the base area as
opposed to both base and emitter at the same time. Subsequently, a
POCL furnace based contact doping is done in the ablated areas. In
this FSR-AJ-NSE embodiment, the POCl condition has to be strong
enough to counter-dope the emitter area. And the oxide thickness
has to be thick enough such that the phosphorous is blocked
everywhere except where the oxide is ablated by the previous laser
step. This step is subsequently followed by the oxide ablation
using the laser to open emitter and base contact areas. A reason
both base and emitter areas may not be ablated at the same time
prior to the POCl process is because POCl is gas phase doping and
would have also gone in the emitter area while intended to
counter-dope the base area only. In the FSR-AJ-NE fabrication
embodiment, subsequent to the emitter ablation by laser the TFSS
will be released with a mobile carrier attached on the ablated side
(non-sunnyside). Then the following fabrication steps including
TFSS clean, texturing, SiHxNy:H deposition, front side
reinforcement, and finally metallization are described earlier in
context of FIG. 3 are applicable to this flow with all their
variations already discussed.
[0106] In a different embodiment of the POCl based fabrication
flow, to create the FSR-SJ-NS version, a silicon etch of the base
nesting area will be added after thermal oxidation and before the
base ablation using the laser. All methods of silicon etching to
create the trench isolation (recess) described in the context of
FIG. 3 are equally applicable to this structure.
[0107] In different embodiments of the POCl based fabrication flow
to create the BSR-AJ-NS and BSR-SJ-NS versions, the flows are
similar to FSR-AJ-NS and FSR-SJ-NS, respectively, except, in the
BSR version before release and after the emitter contact openings
are formed there is a permanent reinforcement, using for example a
hard plastic material and silicone glue, on the cell backside. As
described, in one embodiment the shape of this reinforcement may be
grid-shape. Another difference of the BSR version compared to FSR
in the nonselective emitter case, is that the BSR versions do not
use a temporary mobile carrier for reinforcement on the
non-sunnyside as there is already a permanent reinforcement in
place. This eliminates the step of attaching the front side
reinforcement before the final step of metallization is required.
In the BSR case the metallization is done with the backside
reinforcement in place working around the reinforcement pattern. As
described in the context of the inkjet selective emitter flows,
this may make the N.times.N busbar design more conducive to
metallization for the both BSR designs.
[0108] In operation, the disclosed subject matter provides for the
following:
[0109] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is front side reinforced
(FSR) by temporary and/or permanent carriers. The solar cell has
back contacts, separated junctions (SJ), selective emitters, and
integrated Lambertian mirror. This type of solar cells is referred
to as FSR-SJ cells.
[0110] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is front side reinforced
(FSR) by temporary and/or permanent carriers. The solar cell has
back contacts, abutted junctions (AJ), selective emitters, and
integrated Lambertian mirror. This type of solar cells is referred
to as FSR-AJ cells.
[0111] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is backside reinforced
(BSR) by temporary and/or permanent carriers. The solar cell has
back contacts, separated junctions (SJ), selective emitters, and
integrated Lambertian mirror. This type of solar cells is referred
to as BSR-SJ cells.
[0112] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is backside reinforced
(BSR) by temporary and/or permanent carriers. The solar cell has
back contacts, abutted junctions (AJ), selective emitters, and
integrated Lambertian mirror. This type of solar cells is referred
to as BSR-AJ cells.
[0113] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is front side reinforced
(FSR) by temporary and/or permanent carriers. The solar cell has
back contacts, separated junctions (SJ), none-selective emitters
(NSE), and integrated Lambertian mirror. This type of solar cells
is referred to as FSR-SJ-NSE cells.
[0114] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is front side reinforced
(FSR) by temporary and/or permanent carriers. The solar cell has
back contacts, abutted junctions (AJ), none-selective emitters
(NS), and integrated Lambertian mirror. This type of solar cells is
referred to as FSR-AJ-NSE cells.
[0115] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is backside reinforced
(BSR) by temporary and/or permanent carriers. The solar cell has
back contacts, separated junctions (SJ), none-selective emitters
(NS), and integrated Lambertian mirror. This type of solar cells is
referred to as BSR-SJ-NSE cells.
[0116] A solar cell made on a released thin film silicon substrate
(TFSS) that is substantially planar and is backside reinforced
(BSR) by temporary and/or permanent carriers. The said solar cell
has back contacts, abutted junctions (AJ), none-selective emitters
(NSE), and integrated Lambertian mirror. This type of solar cells
is referred to as BSR-AJ-NSE cells.
[0117] Variations among the selection of types of substrate
reinforcement (FSR or BSR, temporary or permanent), junctions (SJ
or AJ), emitters (SE or NSE), and Lambertian mirrors (integrated or
separated), are all considered in the scope of claims of the
present invention.
[0118] Further, fabrication aspects include, a first carrier is a
thick semiconductor (e.g., preferably crystalline silicon for
crystalline silicon solar cells) wafer--reusable template--while
the second carrier is a low cost material capable of supporting
TFSS--a permanent backside reinforcement (which is permanently
attached to the cell). The reusable template carrier, because it is
standard thick silicon wafer, is capable of withstanding high
temperature. This host wafer may be in various sizes such as 200 mm
or 300 mm round or square, thicknesses capable of going through
full solar cell process without breaking such as 200 .mu.m upwards
(including semiconductor standard 300 mm diameter, 750 .mu.m thick
wafers), and may also take square or other geometrical form factor
of any size. The amortized cost of this carrier is brought down
significantly by reusing it over again and amortizing it over a
plurality of TFSS fabrication cycles. This carrier must also
satisfy the aforementioned third criteria of being conducive for
high-yield detachment of TFSS or thin epitaxial substrate (thus
making it the reusable temporary carrier). This is accomplished
using a porous silicon layer (preferably a bilayer with two
porosities or a multilayer with multiple porosities) between the
template and the TFSS. This porous silicon thin layer is easily
formed on top of the carrier using subtractive electrochemical
process. The porosity of this layer is catered to strike the right
balance such that the layer is porous enough (and mechanical weak
enough) to detach with ease (using any of the release processes
such as mechanical release (MR) or sonicated mechanical release in
a liquid (SMR), while simultaneously not too porous such that it is
conducive for both growth of a high quality, low-defectivity thin
TFSS and allows for no premature release during on-carrier process.
One approach is to use a porous silicon multilayer with at least
two different porosities (i.e., a bilayer porous silicon), with the
bottom layer next to the template having a higher porosity value
and the top layer of porous silicon next to the epitaxial layer
having a lower porosity value. This results in specific doping
requirements for the template carrier.
[0119] Additional fabrication aspects include the second carrier,
and subsequent carriers, required for providing temporary
mechanical support during various processing steps in manufacturing
the solar cell from the said released TFSS. Examples of the
processing steps include MR, metallization, front surface texturing
and passivation. Examples of the carriers include mobile
electrostatic chucks (MESC), mobile vacuum chucks (MOVAC) and
mobile chucks with a removable adhesive layer.
[0120] Additional aspects of the present disclosure include the use
of short laser pulses with sub-nanosecond pulse duration is
described to accomplish the contact opening for at least one,
preferably both polarities of the contacts of the back junction
back contact solar cell. Also, the use of laser pulses is described
to accomplish the doping of the contact areas for at least one,
preferably both polarities of the contacts regions to the above
described back junction back contact solar cell.
[0121] Additional fabrication aspects a last carrier providing a
permanent reinforcement of the thin solar cell and supporting it in
a solar cell module. There are several choices of backside
supports/reinforcements including low cost soda lime glass and
various low cost plastic materials.
[0122] The foregoing description of the preferred embodiments is
provided to enable any person skilled in the art to make or use the
claimed subject matter. Various modifications to these embodiments
will be readily apparent to those skilled in the art, and the
generic principles defined herein may be applied to other
embodiments without the use of the innovative faculty. Thus, the
claimed subject matter is not intended to be limited to the
embodiments shown herein but is to be accorded the widest scope
consistent with the principles and novel features disclosed
herein.
* * * * *