10
m minority-carrier diffusion lengths in Si wires synthesized
by Cu-catalyzed vapor-liquid-solid growth
Morgan C. Putnam,
a
Daniel B. Turner-Evans, Michael D. Kelzenberg,
Shannon W. Boettcher, Nathan S. Lewis, and Harry A. Atwater
California Institute of Technology, 1200 E. California Blvd., Pasadena, California 91125, USA
Received 5 August 2009; accepted 18 September 2009; published online 23 October 2009
The effective electron minority-carrier diffusion length,
L
n
,eff
, for 2.0
m diameter Si wires that
were synthesized by Cu-catalyzed vapor-liquid-solid growth was measured by scanning
photocurrent microscopy. In dark, ambient conditions,
L
n
,eff
was limited by surface recombination to
a value of
0.7
m. However, a value of
L
n
,eff
=10.5
1
m was measured under broad-area
illumination in low-level injection. The relatively long minority-carrier diffusion length observed
under illumination is consistent with an increased surface passivation resulting from filling of the
surface states of the Si wires by photogenerated carriers. These relatively large
L
n
,eff
values have
important implications for the design of high-efficiency, radial-junction photovoltaic cells from
arrays of Si wires synthesized by metal-catalyzed growth processes. ©
2009 American Institute of
Physics
.
doi:
10.1063/1.3247969
Photovoltaic cells based on Si wire arrays offer a number
of features that are compatible with obtaining high device
performance through the use of low-cost fabrication steps.
Some of these features include the opportunity to optimize
carrier collection through formation of either radial or axial
p
-
n
junctions, the ability to directly synthesize large-area ar-
rays of Si wires through the use of chlorosilane precursors,
and the ability to embed the resulting Si wire arrays in flex-
ible polymeric sheets; thus enabling the facile removal of the
Si wire arrays and concomitant reuse of the templating
growth substrate.
1
–
4
The vapor-liquid-solid
VLS
growth mechanism pro-
vides a demonstrated pathway to the fabrication of arrays of
one-dimensional Si wires.
1
However, VLS growth processes
that employ metallic precursors unavoidably result in metal
incorporation into the Si wires. The presence of metallic im-
purities has the potential to degrade the minority-carrier life-
time and diffusion length in the Si wires. Hence, measure-
ments of the minority-carrier diffusion length are critical for
assessing the potential of VLS-grown Si wire arrays as com-
ponents in the fabrication of efficient photovoltaics.
Minority-carrier diffusion lengths up to 4
m have been
reported for Au-catalyzed, VLS-grown Si wires.
3
,
5
These val-
ues are considerably smaller than the diffusion length in
high-purity bulk Si, but are in good agreement with the value
expected based on the measured concentration of the known
deep trap, Au, in the Si wires.
6
Cu-catalyzed, VLS-grown Si
wires are expected to have longer diffusion lengths than
wires grown from Au catalysts, because Cu is not as detri-
mental as Au to the minority-carrier lifetime of Si.
7
,
8
We
report herein, through use of scanning photocurrent micros-
copy
SPCM
, an effective electron minority-carrier diffu-
sion length,
L
n
,eff
, of 10.5
1
m for Cu-catalyzed VLS-
grown Si wires.
Cu-catalyzed Si wires were grown at 1000 °C and 1 atm
from a 500:10 standard cubic centimeters per minute
SCCM
gaseous H
2
:SiCl
4
stream, in a process similar to
that reported previously
1
see supporting information
9
. The
as-grown undoped wires were 2.0
m in diameter and
100
m in length. These wires were highly resistive, so in
subsequent growths 0.06 SCCM BCl
3
5% in H
2
was added
to the reactant stream. Four-point probe measurements of
wires that were removed from the Si growth substrate indi-
cated that the wires were
p
-type with a resistivity of
0.19
0.02
cm. This value corresponds to an acceptor
concentration
N
A
of
1.05
0.15
10
17
cm
−3
, assuming a
bulk hole mobility
3.1
10
2
cm
2
V
−1
s
−1
for Si. Prior to
SPCM characterization, a 5:1:1
by volume
mixture of
de-ionized H
2
O
18 M
cm resistivity
:NH
4
OH
29
%
:
H
2
O
2
30
%
RCA1
was used to remove any organic con-
tamination, and a 6:1:1
by volume
mixture of de-ionized
H
2
O:HCl
37
%
:H
2
O
2
RCA2
was used to remove residual
Cu. The chemical/native oxide was then removed by etching
the wires for 15 s in buffer HF Improved
Transene, Inc.
.
The wires were then exposed to air for 3.5 days to grow a
native oxide on the wires.
To obtain diffusion lengths from the SPCM method, both
an Ohmic and a rectifying contact were required on the Si
wires.
3
,
5
The Si wires were first removed from the growth
substrate and dispersed onto a Si
3
N
4
-coated Si
100
wafer.
Photolithography was then used to pattern the contacts onto
individual Si wires. Following a 15 s etch in Buffer HF Im-
proved to remove the native oxide, Ohmic contacts were
formed by sputtering Al
with 1% Si
onto the
p
-Si wires.
Rectifying contacts were then formed by sputtering Al onto
the native oxide of the
p
-Si wires. This metal-insulator-
semiconductor
MIS
contact between Al and
p
-Si produces
a rectifying contact.
10
After deposition of metal, a contact
anneal was performed at 300 °C for 10 min in forming gas
5% H
2
in N
2
.
Figure
1
illustrates the observed rectifying behavior for a
device with an ideality factor of 1.8 and an effective barrier
height of 0.6 eV. Ideality factors ranged from 1.4 to 3.1, and
the effective barrier height ranged from 0.4 to 0.7 eV. The
inset of Fig.
1
displays an optical microscope image of the
device. The bright spot near the center of the wire corre-
a
Electronic mail: putnam@caltech.edu.
APPLIED PHYSICS LETTERS
95
, 163116
2009
0003-6951/2009/95
16
/163116/3/$25.00
© 2009 American Institute of Physics
95
, 163116-1
Downloaded 13 Nov 2009 to 131.215.193.211. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
sponds to the area illuminated by the laser, while the other
two spots are artifacts that arose from reflections in the mi-
croscope optics.
SPCM measurements were made using a WiTec scan-
ning near-field optical microscope in confocal mode. Local
illumination was provided by a 650 nm laser, chopped at
30 Hz, with a diffraction limited spot size of 0.4
m.
Broad-area illumination was provided by the microscope
light
color temperature=3200 K
. Using a photodiode to
measure the illumination power density, the optical carrier
generation density for both sources was calculated to be a
factor of 40 less than the equilibrium hole concentration of
1
10
17
cm
−3
, thus satisfying the requirement for low-level-
injection illumination conditions
see supporting informa-
tion
. The photocurrent from the MIS rectifying contact was
detected by a preamplifier connected to a lock-in amplifier.
3
For measurements at an applied bias, the Ohmic contact was
connected to a dc voltage source.
In SCPM, a two-dimensional map of the photocurrent is
generated by recording the photocurrent while scanning the
sample underneath a local illumination source
the 650 nm
laser
. Figure
2
depicts the zero-bias SPCM images of a Cu-
catalyzed Si wire measured in the dark
Fig.
2
b
and in the
presence of broad-area illumination
Fig.
2
c
. In the dark, a
small photocurrent
note respective scale bars
was observed
along the wire, as well as on the MIS rectifying contact in
regions immediately above and below the wire. The photo-
current above and below the wire on the MIS contact is
believed to arise from an optically thin coating of Al that
formed, as a result of the directional nature of sputtering and
the large ratio of the wire diameter to the thickness of the
contact, along the sidewall of the wire. Under broad-area
illumination, a much larger signal, which extended greater
than half the length of the device, was observed. Figure
2
a
depicts a scanning electron microscope
SEM
image to fa-
cilitate correlation of the observed photocurrent response
with the geometry of the device.
The larger amplitude of the photocurrent and the in-
creased distance over which photocurrent was observed are
indicative of a larger effective electron minority-carrier dif-
fusion length,
L
n
,eff
, under broad-area illumination than in the
dark. Because of the observed long time decay in
L
n
,eff
time
scale of minutes
after removal of the broad-area illumina-
tion, as shown by the dark photocurrent response after expo-
sure to broad-area illumination
Fig.
2
d
, Fig. S1
9
,weas-
sume that an improvement in surface passivation produces
the increase in
L
n
,eff
with broad-area illumination. An in-
crease in
L
n
,eff
, as a result of an increase in the bulk lifetime,
would be expected to decay on the time scale of the
minority-carrier lifetime
10–100 ns
.
The photoinjection of electrons into the oxide
oxide
trapped charge
11
could produce a decrease in the surface
recombination velocity
S
and is consistent with the long time
decay in
L
n
,eff
. Figure
2
e
depicts the expected band struc-
ture for
p
-Si coated with a native oxide that contains positive
fixed oxide charge. The presence of positive fixed oxide
charge, which is well known to exist in SiO
2
,
12
,
13
will intro-
duce negative surface band bending. This band bending will
result in a large minority-carrier surface concentration and
hence produce a high surface-recombination velocity. How-
ever, the photoinjection of electrons into the oxide could bal-
ance the positive fixed oxide charge, thereby reducing the
negative surface band bending and decreasing
S
Fig.
2
f
.
A long time scale
time scale of minutes
for electrons to
tunnel out of an oxide
14
is consistent with the observed long
time decay in
L
n
,eff
under the proposed surface passivation
mechanism.
Figure
3
displays photocurrent cross sections along the
length of the wire, taken from SPCM images, as a function
of the bias voltage applied to the Ohmic contact. These data
allowed calculation of values of
L
n
,eff
based upon the ex-
FIG. 1.
Color online
Current-voltage sweep of a 2.0
m diameter Si wire
device. Inset: optical microscope image of the measured device; the Ohmic
contact is on the left and the MIS contact is on the right. The laser illumi-
nation spot can be seen in the center of the wire.
FIG. 2.
Color online
a
SEM image of the device of Fig.
1
.
b
SPCM
image of the device measured in the dark.
c
SPCM image of the device
measured under low-level-injection, broad-area illumination.
d
SPCM im-
age of the device measured in the dark for a measurement started within 5 s
after exposure to broad-area illumination. All SPCM images are at zero bias
and took 4 min and 14 s to acquire.
e
Schematic illustration of the pro-
posed band diagram in the dark.
f
Schematic diagram of the proposed band
diagram under broad-area illumination.
163116-2
Putnam
etal.
Appl. Phys. Lett.
95
, 163116
2009
Downloaded 13 Nov 2009 to 131.215.193.211. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp