of 6
High-performance Si microwire photovoltaics
Michael D. Kelzenberg,
a
Daniel B. Turner-Evans,
a
Morgan C. Putnam,
a
Shannon W. Boettcher,
b
Ryan M. Briggs,
a
Jae Yeon Baek,
a
Nathan S. Lewis
*
bc
and Harry A. Atwater
*
ac
Received 12th October 2010, Accepted 24th November 2010
DOI: 10.1039/c0ee00549e
Crystalline Si wires, grown by the vapor–liquid–solid (VLS)
process, have emerged as promising candidate materials for low-
cost, thin-film photovoltaics. Here, we demonstrate VLS-grown Si
microwires that have suitable electrical properties for high-perfor-
mance photovoltaic applications, including long minority-carrier
diffusion lengths (
L
n
[
30
m
m) and low surface recombination
velocities (
S

70 cm
$
s

1
). Single-wire radial p–n junction solar
cells were fabricated with amorphous silicon and silicon nitride
surface coatings, achieving up to 9.0% apparent photovoltaic effi-
ciency, and exhibiting up to

600 mV open-circuit voltage with over
80% fill factor. Projective single-wire measurements and optoelec-
tronic simulations suggest that large-area Si wire-array solar cells
have the potential to exceed 17% energy-conversion efficiency,
offering a promising route toward cost-effective crystalline Si
photovoltaics.
The favorable band gap and natural abundance of Si, combined with
the large expertise base for semiconductor wafer processing, have led
to the use of wafer-based crystalline Si in the vast majority of
photovoltaic cells and modules produced worldwide. However, the
high cost of purifying, crystallizing, and sawing Si wafers has
inhibited these photovoltaic energy sources from approaching cost
parity with fossil fuels, thus motivating the development of thin-film
solar cells that can be deposited onto inexpensive or reusable
substrates.
1,2
Crystalline Si wires, grown by the catalytic vapor–
liquid–solid (VLS) chemical vapor deposition (CVD) process,
3
have
recently emerged as promising candidate materials for thin-film
photovoltaics that seek to combine low-cost Si deposition techniques
with mechanically flexible, high-performance solar cell geometries.
4–20
Unlike conventional planar CVD techniques, VLS growth enables
the precise control of the crystal grain size (
i.e.
, the wire dimensions)
16
as well as the physical access to modify or passivate the crystal
surfaces, allowing the fabrication of novel solar cell geometries that
optimally balance optical absorption with charge-carrier collection,
such as radial p–n junctions.
4
Furthermore, arrays of VLS-grown
wires afford numerous advantages for thin-film photovoltaic
processes, including high absorption and effective optical concen-
tration,
8,9
ease of layer-transfer by polymer encapsulation and
mechanical peel-off,
17
and compatibility with either re-usable
18
or
low-cost (
e.g.
, glass or foil) growth substrates.
7,8,19
For these reasons,
VLS-grown Si-wire solar cells have attracted substantial interest,
resulting in up to

8% efficient wire-array devices,
5–8

3.5% efficient
single-wire devices;
10–12
and

3% efficient photoelectrochemical
devices.
13–15
Reaching the >17% efficiency and >600 mV open-circuit
voltage predicted for VLS-grown Si-wire solar cells,
20
however,
a
Thomas J. Watson Laboratories of Applied Physics, California Institute of
Technology, 1200 E. California Blvd, Pasadena, CA, 91125, USA. E-mail:
haa@caltech.edu; Fax: +1 626 844-9320; Tel: +1 626-395-2197
b
Division of Chemistry and Chemical Engineering, California Institute of
Technology, 1200 E. California Blvd, Pasadena, CA, 91125, USA.
E-mail: nslewis@caltech.edu; Fax: +1 626 395-8867; Tel: +1 626 395-6335
c
Member, Kavli Nanoscience Institute, California Institute of Technology,
USA
† Electronic supplementary information (ESI) available: Experimental
methods including fabrication procedures, details of the SPCM
technique, spectral response measurements, numerical simulations, and
optical
characterization
of
the
PECVD
films.
See
DOI:
10.1039/c0ee00549e
‡ These authors contributed equally to this work.
Broadercontext
Throughout decades of double-digit growth of the photovoltaics (PV) industry, crystalline Si wafer-based solar cells have remained
the predominant worldwide PV technology due in part to high module efficiencies (15–21%) and the tremendous scale at which they
are manufactured. However, producing Si wafers is a costly and energy-intensive process, which has enabled lower-cost, less-efficient
(8–12%)
thin-film
PV technologies such as CdTe to gain significant market share in recent years. Technologies that combine the high
efficiency, abundance, and non-toxicity of crystalline Si with the low cost and light weight of thin-film PV have the potential to
significantly accelerate the adoption of PV energy. In this report, we investigate crystalline Si microwires grown by the vapor–liquid–
solid (VLS) process, a thin-film vapor deposition technique. By fabricating and measuring single-wire test structures, we find that
these wires possess remarkably high material quality, potentially enabling Si microwire-array solar cells to reach efficiencies that
rival those of many wafer-based crystalline Si technologies (>17%). These results suggest that VLS-grown microwires may offer
a viable alternative to wafers for the production of cost-effective crystalline Si photovoltaics.
866 |
Energy Environ. Sci.
, 2011,
4
, 866–871
This journal is
ª
The Royal Society of Chemistry 2011
Dynamic Article Links
C
<
Energy &
Environmental Science
Cite this:
Energy Environ. Sci.
, 2011,
4
, 866
www.rsc.org/ees
COMMUNICATION
Downloaded by California Institute of Technology on 18 March 2011
Published on 07 January 2011 on http://pubs.rsc.org | doi:10.1039/C0EE00549E
View Online
requires substantial improvements in junction quality, surface
passivation, and minority-carrier diffusion length (which thus far has
been reported as high as

10
m
m).
21,22
Herein, we present VLS-grown
Si single-microwire solar cells that combine these properties to realize
high photovoltaic performance.
The steps used to fabricate the Si wire solar cells are depicted
in Fig. 1, and are described in detail in the ESI†. Ordered arrays of
p-type crystalline Si microwires (1.2–1.8
m
m diameter) were grown
from gaseous SiCl
4
/BCl
3
on Si(111) wafers using a patterned VLS
CVD process (Fig. 1a).
16
Whereas most prior reports of VLS-
grown Si-wire solar cells have employed Au as the growth-cata-
lyst metal,
6–8,10–14
the growth of our wires was catalyzed by Cu,
which requires higher temperatures for VLS growth, but which is
less deleterious to the performance of crystalline Si photovoltaics
as an impurity.
23
Deposition was performed at 1000

C at 1 atm
total pressure, yielding growth rates of

5
m
m min

1
. After
growth, the Cu catalyst was chemically removed. Radial p–n
junctions were then selectively formed within the upper portion
of each wire by thermal phosphorus diffusion at 850

C, using
a polymer-infill (PDMS) etch-mask to define a SiO
2
diffusion
barrier over the lower portion of each wire (Fig. 1b–e).
5
Single-
wire devices were fabricated from wires that were removed from
the growth wafer, deposited onto transparent (sapphire) or
reflective (SiN
x
-coated Ag) insulating substrates, and then
patterned with Ag-capped Al contacts.
12
Typical devices had
a single contact to either end of the wire, however, four (Fig. 2a)
or eight contacts were patterned to some wires, to allow full
characterization of the effective base doping (

1

10
17
cm

3
)
and the sheet resistance of the emitter (

6k
U
,

1
), as well as to
confirm that the contacts to both regions were ohmic and low-
resistance (<5% of device resistance). Fig. 2b shows the dark
current–voltage (
I
V
) characteristics of a typical single-wire
device. Ideality factors,
n
, of 1.0–1.2 were observed for most
devices, indicative of high-quality, low-recombination p–n
junctions.
Due to the high surface-area-to-volume ratio, Si wire solar cells are
very sensitive to surface recombination, particularly within the
bottom segment of each wire where no radial p–n junction is present
(henceforth referred to as the
axial
region of the device). In this
region, minority carriers must di
ffuse axially to reach the junction in
order to be collected (Fig. 3a, top). The effective distance that carriers
can travel before recombining,
L
eff
, can be greatly reduced by surface
recombination, and surface-limited values of
L
eff
as low as

20 nm
have been reported for VLS-grown Si nanowires.
22
To determine the
L
eff
of our Si microwires, scanning photocurrent microscopy (SPCM)
was performed to obtain spatially resolved maps of minority-carrier
collection within the single-wire solar cells.
12,21
AsshowninFig.3a
(bottom), relatively uniform carrier collection was observed
throughout the radial portion of the wires, but no carrier collection
was observed from the axial portion. In fact, the abrupt spatial
transition between the two collection regimes could not be resolved
by the

0.5
m
mdiameterbeamspotofthe
l
¼
650 nm illumination
Fig. 1
Fabrication of radial p–n junction Si microwire arrays. Schematic
diagrams depict: (a) VLS-growth of p-type Si microwire arrays; (b)
catalyst removal and growth of a thermal-oxide diffusion-barrier; (c)
selective removal of the oxide barrier using a polymer-infill etch mask;
and (d) thermal diffusion of radial p–n junctions. (e) SEM images of
a microwire array following the fabrication step depicted in (d), viewed at

45

tilt (left) and

90

tilt (right).
Fig. 2
Single-Si-microwire device. (a) SEM image, and (b) dark
I
V
behavior of a typical single-wire device. The arrow indicates the start of
the radial p–n junction, which extends to the left. Nearly identical
I
V
behavior was observed between the inner
vs.
the outer contacts. The
junction current density and the dark saturation current density (
J
0
) were
both normalized to the estimated surface area (

3

10

6
cm
2
) of the
radial p–n junction, which was calculated assuming a 100 nm junction
depth.
This journal is
ª
The Royal Society of Chemistry 2011
Energy Environ. Sci.
, 2011,
4
, 866–871 | 867
Downloaded by California Institute of Technology on 18 March 2011
Published on 07 January 2011 on http://pubs.rsc.org | doi:10.1039/C0EE00549E
View Online
source, indicating that the effective
axial minority-carrier collection
length,
L
eff
,was
#
0.5
m
m for the as-fabricated Si wires.
To determine whether
L
eff
was limited by bulk or surface recom-
bination, the SPCM measurements were repeated on single-wire
devices that were fully immersed in pH
¼
5.0 buffered hydrofluoric
acid (BHF). Hydrofluoric acid removes oxides from crystalline Si
surfaces
and
provides
hydroge
n-termination
of
dangling
bonds, resulting in unusually low surface recombination velocities
(
S
<1cms

1
).
24
With the single-wire devices immersed in BHF
(waiting

3 min for the SiO
2
diffusion-barrier to be removed),
uniform carrier collection was observed from both the axial and
the radial portions of the wire (Fig. 3b), with no apparent spatial
decay throughout the

20
m
m length of the axial wire segment
(
L
eff
[
20
m
m). This behavior implies that the bulk minority-
carrier diffusion length,
L
n
,was
[
20
m
m, and that the relatively
short
L
eff
observed prior to BHF immersion was thus due to a high
surface recombination velocity of
S
>4

10
5
cm s

1
(see ESI† for
details of SPCM measurements and determination of
L
eff
and
S
).
To effect passivation of the Si mi
crowire solar cells with an air-
stable solid-state surface termination, we used plasma-enhanced
CVD (PECVD) to coat the wires with amorphous hydrogenated thin
films of either Si (a-Si:H) or silicon nitride (a-SiN
x
:H). PECVD a-
Si:H and a-SiN
x
:H coatings are commonly used in solar cell fabri-
cation due to their stability and ease of growth (
T
< 400

C).
25,26
Low
surface recombination velocities (
S
<10cms

1
) have been reported
for Si surfaces coated with either material, and each offers additional
beneficial properties for photovoltaic applications: a-SiN
x
:H also
functions as a versatile anti-reflect
ive (AR) coating for crystalline Si,
27
while a-Si:H forms a heterojuncti
on to crystalline Si, enabling
extremely high open-circuit voltages (up to 743 mV) for wafer-based
Si solar cells.
28
After diffusing p–n junctions into a Si wire-array as
described above, we removed the remaining surface oxide, performed
a standard clean, and then split the array into several pieces. One
array was coated with

10 nm of undoped PECVD a-Si:H that was
grown from SiH
4
(5% in Ar) at 240

C. Another was coated with low-
stress PECVD a-SiN
x
:H that was grown from SiH
4
(5% in N
2
)and
NH
3
at 350

C, the thickness of which tapered from

120 nm at the
wire tips to

60 nm at the wire bases (see ESI, Fig. S3†). Both
PECVD films were also deposited onto planar p-type control wafers
(400
m
m float-zone Si, double-side-polished,
r
>4k
U
cm) to enable
optical characterization by spectroscopic ellipsometry (see ESI,
Fig. S1 and S2†), as well as microw
ave-frequency photoconductivity
decay measurements
29
which indicated
S
<20cms

1
for a-Si:H
passivation and
S
<10cms

1
for a-SiN
x
:H passivation. The a-Si:H-
coated single-wire devices were annealed at 275

Cfor30minin
forming gas (5% H
2
in N
2
) to produce ohmic contacts through the a-
Si:H (Fig. 3c). To enable electrical contact to the a-SiN
x
:H-coated
wires, prior to removal from the growth substrate, the nitride was
removed from the wire tips using an infill-masked chemical etch
(Fig. 3d).
Fig. 3
SPCM characterization of minority-carrier recombination in Si
microwire radial p–n junction solar cells. (a) Top: schematic diagram of
SPCM measurements performed on the single-wire test structures of the
type depicted in Fig. 2b, which had SiO
2
-coated axial-region sidewalls
and bare (native-oxide-coated) radial-region sidewalls. Below: SEM
image (center) and SPCM image (bottom) of a typical device. Schematic
diagrams (top) and SPCM images (bottom) for wires with surfaces
passivated by buffered HF (b), a-Si:H (c), and a-SiN
x
:H (d). All SPCM
measurements were normalized to the incident beam photocurrent and
are reported in terms of external quantum efficiency (EQE). In all SPCM
images, the white arrow indicates the start of the radial p–n junction
(which extends to the left) and the hashed white areas indicate the loca-
tion of the metal contacts.
868 |
Energy Environ. Sci.
, 2011,
4
, 866–871
This journal is
ª
The Royal Society of Chemistry 2011
Downloaded by California Institute of Technology on 18 March 2011
Published on 07 January 2011 on http://pubs.rsc.org | doi:10.1039/C0EE00549E
View Online
Shown in Fig. 3c, the SPCM profile of a typical a-Si:H-coated
single-wire solar cell indicated axial-region carrier collection with
a characteristic decay length of
L
eff
z
10
m
m, indicating a surface
recombination velocity of
S
z
450 cm s

1
(see ESI, Fig. S7†). In
contrast, the SPCM profile of a typical a-SiN
x
:H-coated single-wire
solar cell (Fig. 3d) exhibited hi
gh carrier collection efficiency
throughout the entire axial portion of the wire, with no apparent
decay length (
L
eff
[
30
m
m). Furthermore, the EQE of the a-
SiN
x
:H-coated devices was markedly higher than that of the non-
coated devices, due to the anti-reflective nature of the nitride coating.
In fact, the EQE was usually greatest within the axial portion of these
wires, because the tapering thickness of the a-SiN
x
:H in this region
yielded a nearly optimal antirefl
ective coating at the excitation
wavelength (see ESI, Fig. S9–S11†). These observations imply a low
surface recombination velocity (
S

70 cm s

1
)aswellasthelongest
minority-carrier diffusion length reported to date for VLS-grown Si
wires (
L
n
[
30
m
m,
s
n
[
500 ns). Because our observation of
L
n
was limited by the

30
m
m axial length of the longest a-SiN
x
:H-
coated single-wire devices, the uniform SPCM profiles suggest not
only that
L
n
was many times this value, but also indicate an extremely
low effective surface recombination velocity at the Al:p–Si interface.
We attribute this behavior, observed under both a-SiN
x
:H and BHF
passivation, to the presence of a p
+
layer beneath the contact (
i.e.
,
a back surface field) diffused from the degenerately doped growth
wafer during growth and oxidation.
The above measurements reveal that, although the growth of the Si
wires was catalyzed by Cu (one of the more soluble, mobile, and
prevalent impurities in Si microelectronics fabrication),
30
it was
nonetheless possible to synthesize long-diffusion-length material with
well-passivated surfaces. These ac
hievements enabled us to fabricate
single-wire solar cell test structures that exhibited the highest open-
circuit voltages (
V
OC
), fill factors (
FF
), and apparent
photovoltaic
efficiencies (
h
) reported to date for VLS-grown Si wire solar cells, as
summarized in Table 1. To improve the absorption of incident
sunlight, all devices were fabricated on reflective substrates consisting
of Si wafers that had been coated with 100 nm of evaporated Ag (to
provide high reflectivity) and

300 nm of PECVD SiN
x
(to prevent
shorting the contacts). Simulations suggest that using reflective Ag
substrates enabled 17–22% greater
J
SC
than would be possible using
the Si
3
N
4
-coated Si substrates of prior studies
12
(see ESI†). Fig. 4
plots the current–density
vs.
voltage (
J
V
) behavior of the most-
efficient device of each surface coating type.
Following the convention of prior single-wire solar cell studies,
10–12
current density was determined by normalizing the device current by
the total non-shaded physical area of each wire (including both the
axial and radial regions and the surf
ace coating thickness). We note,
however, that the wave nature of light and the photonic dimensions
of micro- and nanowires enable them to interact with (and potentially
absorb) more sunlight than predicted by their physical area, from
a classical ray-optics perspective. This ill-defined absorption area
prevents a true definition of photovoltaic efficiency for single-wire
devices, and resulted in apparent EQE exceeding 100% at certain
wavelengths for some of our devices (see ESI, Fig. S13†). Thus, it is
understood that the apparent photovoltaic efficiency of single-wire
solar cells does not necessarily represent that which could be achieved
over macroscopic device areas, for example, by densely tiling the
horizontal wires to occupy the optical plane. Nonetheless, for
microwires of the diameter range studied herein (1.2–1.8
m
m),
numerical simulations suggest th
at minimal systematic error (<4%
relative overstatement of
J
SC
) is introduced by normalizing the
photovoltaic performance of our champion devices to their physical
area (see ESI†).
Comparing the PECVD coatings, we see that the long collection
length and reduced reflectivity of the a-SiN
x
:H-coated devices
consistently yielded the highest short-circuit current densities (up to
26 mA cm

2
), and resulted in the device with the greatest apparent
photovoltaic efficiency (
h
¼
9.0%). Interestingly, the a-Si:H-coated
devices consistently produced the highest open-circuit voltages (up to
595 mV), despite having many times greater
S
within the axial region.
This behavior indicates that, although the a-SiN
x
:H provided the best
passivation of the p-type (base) surfaces, the a-Si:H provided superior
passivation of the n-type (emitter)
surfaces or the Al:n-Si contacts,
and that the latter recombination sources are more detrimental to the
V
OC
of the Si microwire solar cells. It is well known that the wide
Table 1
Measured properties of single-wire solar cell test structures, including the observed minority-carrier collection lengths (
L
eff
), inferred axial-
region surface recombination velocities (
S
), and measured photovoltaic properties (AM 1.5G, 100 mW cm

2
) for each surface coating type. Upper values
(bold) represent the champion cell of each group; lower values represent the measurement range within each group (
N
denotes number of samples)
Wire coating
L
eff
/
m
m
S
/cm s

1
h
(%)
V
OC
/mV
J
SC
/mA cm

2
FF
(%)
Original
a
(
N
¼
12)
<0.5
>4

10
5
4.6
451
13
77
1.5–4.6
390–496
6.9–16
58–81
a-Si:H (
N
¼
20)
5–10
450–600
7.4
564
16
81
3.6–7.4
561–595
7.8–17
77–82
a-SiN
x
:H (
N
¼
13)
[
30

70
9.0
535
23
75
4.8–9.0
462–543
17–26
56–78
a
‘‘Original’’ wires had a native oxide coating over the radial portion and a thermal oxide coating over the axial portion, as depicted in Fig. 3a.
Fig. 4
Photovoltaic
J
V
characteristics of the champion single-wire test
structures of each surface passivation type (left) and a schematic diagram
of the illumination configuration (right). The current density of each
device was normalized to the total non-shaded wire area (determined by
SEM) to determine apparent photovoltaic efficiency.
This journal is
ª
The Royal Society of Chemistry 2011
Energy Environ. Sci.
, 2011,
4
, 866–871 | 869
Downloaded by California Institute of Technology on 18 March 2011
Published on 07 January 2011 on http://pubs.rsc.org | doi:10.1039/C0EE00549E
View Online
band gap of a-Si:H provides effective passivation of metallic contacts
to crystalline Si surfaces (forming a minority-carrier mirror).
31
Furthermore, studies have shown that a-SiN
x
:H provides less-effec-
tive passivation on phosphorus-diffused (n-type) emitters than on
mildly doped p-type Si surfaces.
26
These observations thus suggest
that efficiency gains for Si microwire photovoltaics could result from
improving the axial-region surface-passivation efficacy of the a-Si:H
coating (noting that we observed
S
<20cms

1
for our planar control
wafers), then coating this layer with an antireflective a-SiN
x
:H film. A
device combining the highest
V
OC
,
FF
,and
J
SC
values reported herein
would achieve apparent efficiencies in excess of 12%.
In seeking to further improve the efficiency of Si microwire
photovoltaics, however, the perfor
mance of our horizontally oriented
microwire devices may ultimately be limited by incomplete optical
absorption, as evidenced by the low
J
SC
and poor infrared spectral
response (see ESI, Fig. S13†) of eve
n our best test structures despite
our use of back-reflecting substrates and antireflective coatings. The
optimal thickness of planar crystalline Si solar cells is theoretically
much thicker (

100
m
m) even with ideal light trapping.
32
Thus, solar
cells based on horizontally oriented crystalline Si microwires would
likely benefit from the use of larger-diameter wires than studied
herein, or advanced absorption enhancement techniques such as
dielectric resonance engineering
33
or integrated optical micro-
concentrators (
e.g.
, those demonstrated on Si microcells).
34
Alternatively, recent optical absorption studies have shown that
large-area solar cells made from mechanically flexible arrays of
vertically oriented
, polymer-embedded Si microwires, utilizing the
light-trapping elements depicted in Fig. 5a, can absorb up to 85% of
above-band gap sunlight, and thus offer a high photovoltaic
performance potential for microwires of the dimensions studied
herein.
9
This absorption study was performed on wire arrays of
nearly identical dimensions as those produced here (Fig. 1e), and
recently, such solar cells have demonstrated photovoltaic efficiencies
approaching 8% over

0.1 mm
2
areas.
5
Here, we present single-wire
measurements and simulations to pr
edict the efficiency potential of
this three-dimensional photovoltaic microstructure.
Prior optical absorption measurem
ents indicate that square-tiled
vertical Si microwire arrays could produce up to
J
SC
¼
35.9 mA cm

2
under normal-incidence, one-sun (AM 1.5G) illumination (assuming
that all absorbed photons result in charge-carrier collection).
Considering the 7

7
m
m unit cell of the wire array (Fig. 5b, left), this
value corresponds to
I
SC
¼
17.6 nA per wire. To emulate these optical
excitation conditions, we measured the
I
V
behavior of the champion
a-Si:H-coated single-wire test structure (oriented horizontally) under

2.3 sun illumination intensity, which produced the desired
I
SC
¼
17.6 nA (Fig. 5b). If each wire within a vertical array exhibited this
identical
I
V
behavior, the device would have an open-circuit voltage
of 614 mV, a fill factor of 80%, and photovoltaic efficiency exceeding
17%.
Three-dimensional optoelectroni
c simulations were also performed
to predict the behavior of the wire-array solar cell proposed in Fig. 5a,
taking into account the experimentally observed wire dimensions,
doping profiles, and recombination parameters presented herein, and
also considering the optical effect
s of a 60 nm indium tin oxide (ITO)
transparent top contact on the cell performance. Adapted from our
prior modeling work,
20
these simulations comb
ined full-field optical
absorption simulations (Lumeri
cal FDTD) with finite-element
device-physics simulations (Synopsys Sentaurus Device) to simulate
one unit cell of a periodic wire-array solar cell under one-sun normal-
incidence illumination (see ESI† for simulation details). The results
(Fig. 5b, blue) predict that, although the reflection and absorption
losses of an ITO top contact produce a lower
J
SC
(32.4 mA cm

2
), the
reduced area of the electrical contacts to vertically oriented Si wires
produces a higher
V
OC
(649 mV), yielding similar overall efficiency
(
h
¼
17.4%) as predicted by measurements on single-wire test
structures. Although these project
ive measurements and simulations
do not account for many challenges facing real wire-array solar cells
(such as contact-grid shading, resistive losses, or variations in wire
size or quality), the calculations are based on experimentally
measured quantities, and use a phys
ically meaningful device area for
the definition of photovoltaic efficiency (rather than the ill-defined
effective area of the single-wire devices). The results suggest the
potential of VLS-grown Si wire-array solar cells to compete, on an
efficiency basis, with commercial polycrystalline wafer-based Si
technology.
The dramatic improvements in the bulk, surface, and junction
properties of the Si microwire devices reported above have enabled us
to overcome a fundamental challenge facing the development of
efficient thin-film solar cells from VLS-grown Si wires: achieving high
operating voltage. Despite the inh
erent contamination of VLS-grown
Si by catalyst metal impurities, our best single-wire test structures
exhibited reasonably high open-circuit voltages (

600 mV) and fill
Fig. 5
Projected Si wire-array solar cell performance. (a) Proposed
large-area solar cell geometry consisting of vertically aligned Si wires
embedded within a transparent polymer, employing light-scattering
particles and a transparent top contact. (b) Predicted efficiency and
J
V
characteristics of the wire-array solar cell of configuration (a). Shown in
red: the
I
V
behavior of a horizontal single-wire test structure (left axis)
was measured under

2.3 sun illumination to achieve
I
SC
¼
17.6 nA, the
per-wire photocurrent corresponding to prior optical absorption
measurements on vertical Si microwire arrays.
9
The
J
V
curve (right axis)
has been normalized to the 49
m
m
2
area of the wire-array unit cell. Shown
in blue: an optoelectronic device-physics model
20
was employed to
simulate the
J
V
behavior of the depicted wire-array unit cell based on
measured values of
L
n
¼
30
m
m and
S
¼
70 cm
$
s

1
.
870 |
Energy Environ. Sci.
, 2011,
4
, 866–871
This journal is
ª
The Royal Society of Chemistry 2011
Downloaded by California Institute of Technology on 18 March 2011
Published on 07 January 2011 on http://pubs.rsc.org | doi:10.1039/C0EE00549E
View Online
factors (

80%). Furthermore, simulations indicate that even higher
open-circuit voltages should be p
ossible by improving the device
structure. The surprising material quality of VLS-grown Si wires,
combined with the advantageous o
ptical and mechanical properties
of wire arrays, offers a promising route for the development of effi-
cient, thin-film crystalline Si solar cells.
Acknowledgements
This work was supported by BP and in part by the Department of
Energy, Basic Energy Sciences Energy Frontier Research Center
program under grant DE-SC0001293 as well as the Department of
Energy under grant DE-FG02-
07ER46405. The work made use of
facilities supported by the Caltech Center for Sustainable Energy
Research and by the Center for Science and Engineering of Materials,
an NSF Materials Research Science and Engineering Center at
Caltech. Research was in part carried out at the Kavli Nanoscience
Institute (KNI) and at the Molecular Materials Research Center of
the Beckman Institute at Caltech. S.W.B. acknowledges the KNI for
fellowship support. D.B.T-E. acknowledges the NSF for fellowship
support. The authors acknowledge the assistance of Bruce
Brunschwig, Joshua Spurgeon, Emily Warren and the KNI staff.
References
1 A. Shah, P. Torres, R. Tscharner, N. Wyrsch and H. Keppner,
Science
, 1999,
285
, 692–698.
2 R. Brendel,
Thin-Film Crystalline Silicon Solar Cells—Physics and
Technology
, Wiley, 2003.
3 R. S. Wagner and W. C. Ellis,
Appl. Phys. Lett.
, 1964,
4
, 89–90.
4 B. M. Kayes, H. A. Atwater and N. S. Lewis,
J. Appl. Phys.
, 2005,
97
,
114302.
5 M. C. Putnam, S. W. Boettcher, M. D. Kelzenberg, D. B. Turner-
Evans, J. M. Spurgeon, E. L. Warren, R. M. Briggs, N. S. Lewis
and H. A. Atwater,
Energy Environ. Sci.
, 2010,
3
, 1037–1041.
6 O. Gunawan and S. Guha,
Sol. Energy Mater. Sol. Cells
, 2009,
93
,
1388–1393.
7 T. Stelzner, M. Pietsch, G. Andra, F. Falk, E. Ose and
S. Christiansen,
Nanotechnology
, 2008,
19
, 295203.
8 L. Tsakalakos, J. Balch, J. Fronheiser, B. A. Korevaar, O. Sulima and
J. Rand,
Appl. Phys. Lett.
, 2007,
91
, 233117.
9 M. D. Kelzenberg, S. W. Boettcher, J. A. Petykiewicz, D. B. Turner-
Evans, M. C. Putnam, E. L. Warren, J. M. Spurgeon, R. M. Briggs,
N. S. Lewis and H. A. Atwater,
Nat. Mater.
, 2010,
9
, 239–244.
10 B. Tian, X. Zheng, T. J. Kempa, Y. Fang, N. Yu, G. Yu, J. Huang
and C. M. Lieber,
Nature
, 2007,
449
, 885–889.
11 T. J. Kempa, B. Tian, D. R. Kim, J. Hu, X. Zheng and C. M. Lieber,
Nano Lett.
, 2008,
8
, 3456–3460.
12 M. D. Kelzenberg, D. B. Turner-Evans, B. M. Kayes, M. A. Filler,
M. C. Putnam, N. S. Lewis and H. A. Atwater,
Nano Lett.
, 2008,
8
,
710–714.
13 A. P. Goodey, S. M. Eichfeld, K.-K. Lew, J. M. Redwing and
T. E. Mallouk,
J. Am. Chem. Soc.
, 2007,
129
, 12344–12345.
14 J. R. I. Maiolo, B. M. Kayes, M. A. Filler, M. C. Putnam,
M. D. Kelzenberg, H. A. Atwater and N. S. Lewis,
J. Am. Chem.
Soc.
, 2007,
129
, 12346–12347.
15 S. W. Boettcher, J. M. Spurgeon, M. C. Putnam, E. L. Warren,
D.
B.
Turner-Evans,
M.
D.
Kelzenberg,
J.
R.
Maiolo,
H. A. Atwater and N. S. Lewis,
Science
, 2010,
327
, 185–187.
16 B. M. Kayes, M. A. Filler, M. C. Putnam, M. D. Kelzenberg,
N. S. Lewis and H. A. Atwater,
Appl. Phys. Lett.
, 2007,
91
, 103110.
17 K. E. Plass, M. A. Filler, J. M. Spurgeon, B. M. Kayes,
S. Maldonado, B. S. Brunschwig, H. A. Atwater and N. S. Lewis,
Adv. Mater.
, 2009,
21
, 325–328.
18 J. M. Spurgeon, K. E. Plass, B. M. Kayes, B. S. Brunschwig,
H. A. Atwater and N. S. Lewis,
Appl. Phys. Lett.
, 2008,
93
, 032112.
19 Z. Fan, H. Razavi, J.-W. Do, A. Moriwaki, O. Ergen, Y.-L. Chueh,
P. W. Leu, J. C. Ho, T. Takahashi, L. A. Reichertz, S. Neale,
K. Yu, M. Wu, J. W. Ager and A. Javey,
Nat. Mater.
, 2009,
8
,
648–653.
20 M. D. Kelzenberg, M. C. Putnam, D. B. Turner-Evans, N. S. Lewis
and H. A. Atwater,
Predicted Efficiency of Si Wire Array Solar
Cells, Conference Record of the 34th IEEE PVSC, Philadelphia, PA
,
2009.
21 M. C. Putnam, D. B. Turner-Evans, M. D. Kelzenberg,
S. W. Boettcher, N. S. Lewis and H. A. Atwater,
Appl. Phys. Lett.
,
2009,
95
, 163116.
22 J. E. Allen, E. R. Hemesath, D. E. Perea, J. L. Lensch-Falk, Z. Y. Li,
F. Yin, M. H. Gass, P. Wang, A. L. Bleloch, R. E. Palmer and
L. J. Lauhon,
Nat. Nanotechnol.
, 2008,
3
, 168–173.
23 J. R. Davis, Jr, A. Rohatgi, R. H. Hopkins, P. D. Blais, P. Rai-
Choudhury, J. R. McCormick and H. C. Mollenkopf,
IEEE Trans.
Electron Devices
, 1980,
27
, 677–687.
24 E. Yablonovitch, D. L. Allara, C. C. Chang, T. Gmitter and
T. B. Bright,
Phys. Rev. Lett.
, 1986,
57
, 249–252.
25 S. Dauwe, J. Schmidt and R. Hezel,
Very Low Surface Recombination
Velocities on p- and n-Type Silicon Wafers Passivated with
Hydrogenated Amorphous Silicon Films, Conference Record of the
29th IEEE PVSC, New Orleans, LA
, 2002.
26 A. G. Aberle and R. Hezel,
Progr. Photovolt.: Res. Appl.
, 1997,
5
, 29–
50.
27 P. Doshi, G. E. Jellison and A. Rohatgi,
Appl. Opt.
, 1997,
36
, 7826–
7837.
28 E. Maruyama, A. Terakawa, M. Taguchi, Y. Yoshimine, D. Ide,
T. Baba, M. Shima, H. Sakata and M. Tanaka,
Sanyo’s Challenges
to the Development of High-Efficiency HIT Solar Cells and the
Expansion of HIT Business, Conference Record of the 4th IEEE
WCPEC
, 2006.
29 W. J. Royea, A. Juang and N. S. Lewis,
Appl. Phys. Lett.
, 2000,
77
,
1988–1990.
30 A. A. Istratov and E. R. Weber,
Appl. Phys. A: Mater. Sci. Process.
,
1998,
66
, 123–136.
31 R. M. Swanson,
Approaching the 29% Limit Efficiency of Silicon Solar
Cells, Conference Record of the 31st IEEE PVSC, Orlando, FL
, 2005.
32 T. Tiedje, E. Yablonovitch, G. D. Cody and B. G. Brooks,
IEEE
Trans. Electron Devices
, 1984,
31
, 711–716.
33 L. Cao, P. Fan, A. P. Vasudev, J. S. White, Z. Yu, W. Cai,
J. A. Schuller, S. Fan and M. L. Brongersma,
Nano Lett.
, 2010,
10
,
439–445.
34 J. Yoon, A. J. Baca, S. I. Park, P. Elvikis, J. B. Geddes, L. F. Li,
R. H. Kim, J. L. Xiao, S. D. Wang, T. H. Kim, M. J. Motala,
B. Y. Ahn, E. B. Duoss, J. A. Lewis, R. G. Nuzzo, P. M. Ferreira,
Y. G. Huang, A. Rockett and J. A. Rogers,
Nat. Mater.
, 2008,
7
,
907–915.
This journal is
ª
The Royal Society of Chemistry 2011
Energy Environ. Sci.
, 2011,
4
, 866–871 | 871
Downloaded by California Institute of Technology on 18 March 2011
Published on 07 January 2011 on http://pubs.rsc.org | doi:10.1039/C0EE00549E
View Online