Content uploaded by Natalia Palina
Author content
All content in this area was uploaded by Natalia Palina on Jan 18, 2017
Content may be subject to copyright.
IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 4, NO. 6, NOVEMBER 2014 1445
Laser Chemical Processing of n-Type Emitters
for Solid-Phase Crystallized Polysilicon
Thin-Film Solar Cells
S. Virasawmy, P. I. Widenborg, N. Palina, C. Ke, J. Wong, S. Varlamov, A. A. O. Tay, and B. Hoex
Abstract—We report on the application of laser chemical pro-
cessing (LCP) to fabricate n-type emitters for polysilicon thin-film
solar cells on glass. Sheet resistance values of 2–5 kΩ/with a
peak phosphorus doping concentration in the range 8×1018–
1×1019 cm−3at a shallow doping depth of less than 350 nm are
achieved. After dopant activation and a hydrogenation process, the
best cell has an average Voc of (446 ±7) mV and a pseudofill fac-
tor (pFF) of (68.3 ±0.9)%. This paper demonstrates that LCP can
be successfully applied to fabricate an active layer for polysilicon
thin-film solar cells on glass. Further improvement in the Voc and
the pFF may be possible by optimizing the post-LCP annealing and
hydrogenation process, as well as using a poly-Si film of superior
material quality.
Index Terms—Laser chemical processing (LCP), laser doping,
Nd:YAG, open-circuit voltage, polysilicon thin film, Suns-Voc.
I. INTRODUCTION
AMONG the numerous techniques to improve solar cell ef-
ficiency and to reduce the cost of cell fabrication, lasers are
promising tools for the semiconductor and photovoltaic indus-
try. Lasers are fast, versatile, capable of spatial patterning, and
can compete against other traditional forms of processing such
as tube diffusion, photolithographic patterning, and so forth. For
instance, lasers can be applied either at an early stage during the
Manuscript received April 2, 2014; revised June 3, 2014 and August 10,
2014; accepted August 13, 2014. Date of publication September 4, 2014; date
of current version October 17, 2014. SERIS is a research institute at the
National University of Singapore and is supported by the National University of
Singapore and Singapore’s National Research Foundationthrough the Singapore
Economic Development Board. This work was also supported by the National
Research Foundation, Prime Minister’s Office, Singapore, under its Energy
Innovation Research Programme (EIRP Award NRF2009EWT-CERP001-046
and NRF2009EWT-CERP001-056). The work of C. Ke was supported by the
Clean Energy Program Office (CEPO) Ph.D. scholarship from the Economic
Development Board (EDB).
S. Virasawmy is with SERIS, National University of Singapore, Singa-
pore 117574, and is also with the Department of Mechanical Engineer-
ing, National University of Singapore, Singapore 117575 (e-mail: selven.
virasawmy@nus.edu.sg).
P. I. Widenborg, C. Ke, J. Wong, and B. Hoex are with SERIS, Na-
tional University of Singapore, Singapore 117574 (e-mail: per.widenborg@nus.
edu.sg; ke.cangming@nus.edu.sg; johnson.wong@nus.edu.sg; bram.hoex@
nus.edu.sg).
N. Palina is with the Singapore Synchrotron Light Source, National University
of Singapore, Singapore 117603 (e-mail: natalie.mueller@nus.edu.sg).
S. Varlamov is with the Photovoltaics Centre of Excellence, the University
of New South Wales, Sydney, NSW 2052, Australia (e-mail: s.varlamov@
unsw.edu.au).
A. A. O. Tay is with the Department of Mechanical Engineering, National
University of Singapore, Singapore 117575 (e-mail: mpetayao@nus.edu.sg).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JPHOTOV.2014.2349654
cell fabrication process (e.g., laser crystallization [1] and laser
annealing [2]) or toward the end of the metallization/module
fabrication process (e.g., cell isolation [3], etc.). Recently, liq-
uid phase crystallization (LPC) of poly-Si thin film on glass
fabricated by a continuous wave diode laser [4] or an electron
beam [5] has produced superior open-circuit voltages (585 and
656 mV, respectively). During LPC, the entire silicon stack is
molten and subsequently crystallizes into large grain poly-Si
films. Poly-Si made by LPC has a low defect density and pos-
sesses an excellent electronic quality.
Within the realm of laser doping, laser chemical processing
(LCP), developed from the patented LaserMicroJet technology
by Synova S.A, is a practical technique for doping [6] and
microstructuring applications [7]. By using a highly pressur-
ized doping medium as the optical waveguide, laser light is
coupled inside the liquid jet by total internal reflection. As a
result, a practically infinite supply of dopant precursors is con-
stantly available from the liquid jet for doping. Recently, we
reported on a novel application of LCP for doping polysilicon
thin film. It was shown that relatively high doping levels (close
to 1019 cm−3) at moderate doping depths (within 350 nm) were
routinely achievable by using a Q-switched frequency-doubled
(532 nm) Nd:YAG laser coupled inside a 42.5% phosphoric acid
jet (doping medium was pressurized at 130 bars). A p−/SiNx
layer structure on borosilicate glass was used as the substrate
material. The material quality of the LCP-doped areas was as-
sessed by ultraviolet (UV) reflectance and transmission electron
microscopy (TEM). The study showed that the LCP-doped lay-
ers were of suitable material quality. More details about this
paper can be found elsewhere [8].
In this study, we performed n-type doping on a p−/p+/SiNx
layer structure on borosilicate glass. By carrying out n-type
doping on the surface, a solar cell was made with an n-type
emitter at the surface (air side) of the device structure. The
open-circuit voltages (Voc) of those nonmetallized solar cells
were measured with a customized thin-film Suns-Voc tester
[9]. A few doped samples were annealed in a nitrogen-purged
oven and then hydrogenated in a low-pressure chemical vapor
deposition (LPCVD) tool with an inductively coupled remote
plasma source [10]. A major improvement in Voc (reaching val-
ues above 400 mV) and pseudofill factor (pF F > 65%)was
realized through those two process steps. Even though the Voc
values are not yet state of the art, this paper shows that LCP can
be applied successfully to fabricate an active layer (e.g., an emit-
ter or back surface field) for poly-Si thin-film solar cells. The
sheet resistances and peak dopant concentration are also suitable
2156-3381 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
1446 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 4, NO. 6, NOVEMBER 2014
Fig. 1. Cross-sectional view of sample structure used in LCP doping (not
drawn to scale).
for an active layer in poly-Si thin-film solar cells. Additionally, it
is found that the LCP-doped emitters exhibit less dopant smear-
ing than conventional emitters in SPC poly-Si thin-film solar
cells on glass, which is beneficial for the performance of the
devices. Overall, we are optimistic that further enhancement in
Voc and pFF may be possible by optimizing the hydrogenation
process, the post-LCP anneal conditions, and by using a poly-Si
absorber of better material quality (e.g., poly-Si made by LPC
such as in [4] and [5]).
II. EXPERIMENT
The details about the sample preparation were similar to those
described in [8] except that the deposited layer structure in this
study was 1.9 μmp
−/100 nm p+/70 nm SiNxon a 3.3–mm-
thick borosilicate glass, as shown schematically in Fig. 1. Fig. 1
shows a cross-sectional view of the sample structure used in
LCP doping (not drawn to scale).
LCP doping was carried out using process parameters sum-
marized in Table I. After LCP doping and sample cleaning,
samplesof4cm×2 cm, enclosing the strip-like cells (active
area 40 mm ×7 mm), were cut out from the main sample and
subjected to different thermal anneals. The sheet resistance of
the LCP-doped layers was determined by a manual four-point
probe (Jandel Engineering Limited, Bedford, U.K.). Based on
the geometrical parameters of the sample, a correction factor
of 0.755 was applied to the measured sheet resistances [11].
The active dopant concentration in the samples was determined
by electrochemical capacitance–voltage measurements using
a commercial system (CVP21 ECV Profiler, WEP Control,
Germany) [12].
The electronic quality of the LCP-doped nonmetallized so-
lar cells was assessed by quasi-steady-state open-circuit voltage
(Suns-Voc) measurements. Additional parameters such as the
pseudofill factor, pFF (the upper limit of the fill factor with-
out the effect of series resistance), and the effective ideality
factor (neff )were also extracted from the Suns-Voc data [13],
[14]. More details about the setup of the customized Suns-Voc
system can be found in [9]. Prior to Suns-Voc measurements,
the samples were wet etched (corner etched) in a mixture of
nitric acid (70%)/hydrofluoric acid (49%)/deionized water in
a volume ratio of 1:1:1 to reveal the buried p+layer for the
probe contacts.
Finally, to further improve the device quality of the films,
a few representative samples from each batch were manually
cleaved into two and one sample from each pair was subjected
to a standard hydrogenation process at 600 °C for 30 min in a
LPCVD tool with an inductively coupled remote plasma source.
Before the hydrogenation process, the samples underwent a HF
dip to remove any native oxide layer. After the hydrogenation
process, the samples were again analyzed by Suns-Voc and four
point probe measurements.
III. RESULTS AND DISCUSSION
A. Sheet Resistance Measurements
The dopants incorporated by LCP are activated by a thermal
anneal, either in a rapid thermal processing (RTP) tool or in
a nitrogen-purged oven. Table I summarizes the average sheet
resistances of the annealed samples measured at different loca-
tions. The annealed samples yielded sheet resistances around
2–5 kΩ/. Typically, for the same dopant concentration, the
sheet resistance of poly-Si films is much higher than that of
bulk silicon (kΩ/range as compared with Ω/range). This is
due to lower dopant activation and lower carrier mobility as a
result of the high defect density in poly-Si thin films. A sheet
resistance of 2kΩ/is desirable for poly-Si-based solar cells,
where the doped poly-Si layers must carry current laterally.
Sample S6 was processed with higher laser fluence and, conse-
quently, suffered from significant ablation. Therefore, most of
the n+layer was removed. Comparing S1 and S5, it is observed
that a slightly higher sheet resistance is obtained for the same
laser fluence, similar pulse overlap, and a longer pulse length.
Since the sheet resistance is related to the active dopant den-
sity within the samples, the higher sheet resistance is due to the
decreased amount of dopants from using a longer pulse length
[refer to the active doping profiles in Fig. 2(c)]. In contrast with
our previous findings [8], prolonged annealing at the same tem-
perature (e.g., 610 °C for 2 h) for this sample structure resulted
in a higher sheet resistance. This could be the result of dopant
redistribution across the grains and grain boundaries which then
lead to an increased sheet resistance.
B. Diffusion Profiles (Electrochemical Capacitance–Voltage
Measurements)
The doping concentration and diffusion depth of the LCP-
doped layer are critical for minimizing carrier recombination
within the solar cell. If the p-n junction is located too close to
the air side, minority carriers that are predominantly generated
close to the glass side of the solar cell are more susceptible to
recombination before they are collected at the space charge re-
gion. On the other hand, a highly doped emitter layer decreases
the short wavelength light response in a solar cell. To assess
the active dopant concentration within the samples, ECV mea-
surements were performed on a batch of LCP samples annealed
at 610 °C for 30 min. Fig. 2(a)–(d) displays the active dopant
profiles of the samples processed using the LCP parameters
VIRASAWMY et al.: LASER CHEMICAL PROCESSING OF n-TYPE EMITTERS FOR SOLID-PHASE CRYSTALLIZED POLYSILICON 1447
TAB LE I
LCP PARAMETERS USED FOR THE EXPERIMENTS IN THIS WORK (PULSE SHAPE AND JET PRESSURE WERE SET TO SQUARE-SHAPED AND 130 BAR,RESPECTIVELY)
Sheet Resistance Measurements
LCP +oven LCP +oven LCP +oven
Pulse anneal at 610 anneal at 610 anneal at 700
Sample Fluence Overlap °C for 30 min °Cfor2hrs °C for 30 min
Parameter Optimization Number Laser Parameters (J/cm2(%) (kΩ/)(kΩ/)(kΩ/)
Influence of pulse energy S1 [14 μJ, 100 kHz, 20 ns]2.0 80 1.9 4.2 ±0.1 1.8
S2 [12 μJ, 100 kHz, 20 ns]1.7 80 2.1 3.2 ±0.1 2.0 ±0.1
Influence of pulse-to-pulse S3 [14 μJ, 100 kHz, 20 ns]2.0 90 2.0 ±0.1 3.1 ±0.1 1.9 ±0.1
overlap
(as compared to S1and S2) S4 [12 μJ, 100 kHz, 20 ns]1.7 90 2.1 ±0.2 3.4 ±0.2 1.9
Influence of pulse length S5 [14 μJ, 100 kHz, 40 ns]2.0 80 2.5 ±0.1 4.5 ±0.2 2.9 ±0.2
S6 [24 μJ, 100 kHz, 60 ns]3.4 80 5.0 ±0.5 10.0 ±1.1 7.1 ±0.4
Influence of repetition rate S8 [16 μJ, 150 kHz, 20 ns]2.3 87 2.5 ±0.1 5.1 ±0.5 2.7 ±0.3
S9 [12 μJ, 200 kHz, 20 ns]1.7 90 2.1 3.6 ±0.4 2.5 ±0.1
The measurement uncertainty reflects the standard deviation in the measurements.
Fig. 2. (a) Active dopant profiles throughout the cell structure. The back-
ground p-type dopant concentration was about 2×1017cm−3and the peak
phosphorus (blue symbols) doping concentration was 1019 cm−3.(b)Influ-
ence of pulse energy/overlap ratio over the doping depth. (c) Influence of pulse
length over the doping depth. (d) Influence of repetition rate and pulse overlap
over the doping depth.
shown in Table I. The blue symbols represent the n-type dopant
(phosphorus), while the corresponding red symbols refer to the
p-type dopant (boron). From Fig. 2(a), the background p-type
dopant concentration was about 2 ×1017 cm−3and the peak
doping concentration of the p+layer was about 2×1018 cm−3.
The peak phosphorus doping concentration was close to 1019
cm−3. In Fig. 2(b), an increase in pulse energy resulted in a
deeper junction depth because a higher amount of energy was
available, and thus, the melt front moved deeper into the polysil-
icon [8] [refer to samples S1 and S2 processed with a laser
pulse energy of 14 and 12 μJ, respectively]. On the other hand,
an increase in pulse overlap leads to a deeper junction depth
due to a higher number of melt cycles per unit area [refer to
samples S2 and S4 processed with a pulse overlap ratio of 80%
and 90%, respectively]. However, an increase in pulse energy
has a more significant influence over the junction depth, as
shown by samples S2 & S1 and S4 & S3. Fig. 2(c) shows how
the pulse length affects the active dopant profile. Using a con-
stant pulse energy and a longer pulse length, the peak power
decreases and melting is achieved at a lower energy threshold
[15]. Therefore, the doping depth is shallower (refer to sam-
ples S1 and S5 processed with a pulse length of 20 and 40 ns,
respectively). Even though melting was achieved at a lower en-
ergy for longer pulse lengths, S6 was processed with a much
higher fluence resulting in significant material damage. This is
also reflected in the irregular doping profile. Fig. 2(d) illustrates
the doping depth as a function of the repetition rate. In this case,
the repetition rate affects both the incident energy and the pulse
overlap. It is shown that for the same incident laser energy, a
change in pulse overlap [achieved by altering either the chuck
speed (e.g., S4) or the repetition rate (e.g., S9)]affects the num-
ber of pulses per unit area and the doping depth. As discussed
above, S8 has a much deeper doping depth because the pulse
energy has a more prominent influence over the doping depth
as compared with the pulse overlap.
C. Suns-Voc Measurements
The Voc of a polysilicon thin-film solar cell is influenced
by numerous factors such as the doping concentrations/profiles
within the active layers, the width of the space charge region,
the location of the p-n junction, and the defect density within
the material (intragrain and grain boundary defects) [14]. Since
minority carrier transport is dominated by diffusion under low
injection conditions, the diffusion length of these carriers has
a significant effect on the Voc of the solar cell. A short diffu-
sion length increases the diode’s dark saturation current density,
which then lowers the Voc [16]. The second important parame-
ter that can be extracted from the Suns-Voc measurements is the
pFF, which is given by [17]
pF F =Voc(MPP)×Jnorm (MPP)
Voc(1 Sun)(1)
where Voc(1 Sun) is the open-circuit voltage at 1 Sun, and
Voc(MPP)and Jnorm(MPP)are the open-circuit voltage and
normalized current density at the maximum power point (MPP),
1448 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 4, NO. 6, NOVEMBER 2014
Fig. 3. Average Voc of the oven-annealed samples in superstrate configura-
tion. The error bars represent the standard deviation in the Voc measurements.
respectively. Typically, a decrease in Voc is also accompanied
by a decrease in pFF.
To model the recombination behavior in a polysilicon thin-
film solar cell, it is common to consider a two-diode model,
whereby each diode corresponds to the recombination behavior
of the solar cell. J01 and J02 are the diode saturation current
densities with ideality factor 1 and 2, respectively. J01 represents
recombination in the quasi-neutral bulk regions and at the sur-
faces, whereas J02 is influenced by recombination in the space
charge region or at grain boundaries. A commonly used param-
eter to evaluate the material quality of a solar cell is the effective
ideality factor (neff ), which is calculated from the slope of the
Suns-Voc curve. The effective ideality factor can be calculated
as [13], [14]
neff =Voc(MPP)−Voc(1 Sun)
VT×ln( Suns(MPP)) (2)
where Voc(MPP) and Voc(1Sun)denote the open-circuit voltage
at maximum power point (MPP) and at 1 Sun, respectively, VT
is the thermal voltage (0.0257 V at 25 °C) and Suns(MPP) is the
illumination intensity in Suns at MPP of the pseudo-I–Vcurve.
Suns-Voc measurements were subsequently carried out on a
few representative samples from the as-doped and the annealed
batches of samples. The as-doped samples showed very low Voc
between 30 and 60 mV and pFF of only 31%. This is because
the low fraction of active dopants from the n-type emitter layer
gives rise to a low built-in potential at the p-n junction. Fig. 3
displays the average Voc of a few oven-annealed samples in
superstrate configuration that is when light is shone through the
supporting material.
The annealed samples displayed a moderately high Voc
(177–225 mV) and pFF (48–58%). It is also observed that
the samples annealed at 610 °C for 2 h displayed the lowest
Voc, possibly due to dopant redistribution across the layers that
may shift the p-n junction slightly. In contrast, the samples an-
nealed for shorter durations (e.g., 610 °C for 30 min) or at higher
temperature (e.g., 700 °C for 30 min) showed higher Voc.
D. Hydrogenation
Polysilicon is an inherently defective material with intragrain
and grain boundary defects such as dangling bonds, disloca-
TAB L E I I
AVERAGE SHEET RESISTANCES OF THE LCP-DOPED SAMPLES AFTER
HYDROGENATION AT 600 °CFOR 30 MININALPCVD REACTOR WITH
AN INDUCTIVELY COUPLED PLASMA SOURCE
Sample number LCP +
hydro-
genation
(kΩ/)
LCP +
anneal at
610 °Cfor
30 min +
hydrogena-
tion
kΩ/)
LCP +
anneal at
610 °Cfor
2h+hy-
drogenation
kΩ/)
LCP +
anneal at
700 °Cfor
30 min +
hydrogena-
tion
kΩ/)
S1
[14 μJ, 100 kHz,
20 ns, 80%
overlap]
1.5 ±0.1 1.1 ±0.1 1.5 ±0.1 1.2
S2
[12 μJ, 100 kHz,
20 ns, 80%
overlap]
1.6 ±0.1 1.1 1.5 ±0.1 1.3
S3
[14 μJ, 100 kHz,
20 ns, 90%
overlap]
1.6 – – –
S4
[12 μJ, 100 kHz,
20 ns, 90%
overlap]
1.5 – – –
S5
[14 μJ, 100 kHz,
40 ns, 80%
overlap]
1.7 ±0.1 – – –
S6
[24 μJ, 100 kHz,
60 ns, 80%
overlap]
2.2 ±0.1 – – –
S8
[16 μJ, 150 kHz,
20 ns, 87%
overlap]
1.6 ±0.1 1.0 1.5 ±0.1 1.3
S9
[12 μJ, 200 kHz,
20 ns, 90%
overlap]
1.5 1.0 1.7 ±0.1 1.1
tions, oxygen-related defects, etc. [18], [19]. A common way
to passivate electronic defects in poly-Si is by exposure to hy-
drogen plasma, commonly known as a hydrogenation process.
Exposure of poly-Si thin film to hydrogen plasma is well estab-
lished in the thin-film transistor [20] and the photovoltaic field.
Table II lists the average sheet resistances of the LCP-doped
samples after hydrogenation at 600 °C for 30 min in an LPCVD
reactor with an inductively coupled plasma source.
From Table II, it is observed that a hydrogenation process
can further reduce the sheet resistance of the samples to values
less than 2 kΩ/(see Table I for the previous sheet resistance
values) due to improved defect annealing and carrier mobility
[refer to sample “S2_after hyd” as compared with sample “S2”
in Fig. 2(a)]. These values are promising for decreasing the
resistive losses of a poly-Si thin-film solar cell. It is known that
hydrogenation improves the carrier mobility as it reduces the
grain boundary trap state density in poly-Si [19]. From Fig. 2(a),
it is observed that sample “S2_after hyd” is characterized by
a flat-top doping profile and shows less dopant smearing than
conventional nonmetallized solar cells on glass [12]. This makes
LCP an attractive technique to fabricate an active layer (either an
emitter or a back surface field) for poly-Si thin-film solar cells.
VIRASAWMY et al.: LASER CHEMICAL PROCESSING OF n-TYPE EMITTERS FOR SOLID-PHASE CRYSTALLIZED POLYSILICON 1449
Fig. 4. (a) AverageVoc , (b) average pFF, and (c) extracted neff of the samples
after a hydrogenation process at 600 °C for 30 min in a LPCVD tool with an
inductively coupled plasma. The measurement uncertainty reflects the standard
deviation in the measurements. The best Voc (>400 mV) and pFF (>65%)
were achieved for the samples that were annealed at 700 °C for 30 min prior to
the hydrogenation process.
Additionally, the batch “LCP +hydrogenation” was made to
study the feasibility of omitting the annealing process in between
the LCP and hydrogenation process.
After corner etching to reveal the p+layer, Suns-Voc
measurements were performed on the hydrogenated samples.
Fig. 4(a)–(c) shows the measured Voc and pFF in superstrate
configuration, as well as the extracted neff from the Suns-Voc
measurements. Table III displays the measured and extracted
Suns-Voc parameters from the batch “LCP +hydrogenation.”
The best Voc (>400 mV) and pFF (>65%) were achieved
for the samples that were annealed at 700 °C for 30 min prior
to the hydrogenation process except for S2. These Voc values
are relatively close to those of nonmetallized poly-Si thin-film
solar cells fabricated on nontextured glass (i.e., planar glass)
which range between 435 and 475 mV [12]. For some sam-
ples, the large variation in the Voc values could be the result
of nonhomogenous doping due to laser/jet instability from the
LCP process or from defects within the poly-Si. In contrast, the
batch of samples annealed at 610 °C for 2 h or at 610 °Cfor
TABLE III
MEASURED AND EXTRACTED SUNS-Voc PARAMETERS FROM TH E BATC H
“LCP+HYDROGENATION”
Sample number Voc (mV) pFF (%) neff
S1
[14 μJ, 100 kHz, 20 ns, 80% overlap]348 ±14 59 ±52.1±0.4
S2
[12 μJ, 100 kHz, 20 ns, 80% overlap]364 ±7 65.7 ±21.6±0.2
S3
[14 μJ, 100 kHz, 20 ns, 90% overlap]365 ±8 65.4 ±31.6±0.2
S4
[12 μJ, 100 kHz, 20 ns, 90% overlap]376 ±5 66.5 ±21.6±0.2
S5
[14 μJ, 100 kHz, 40 ns, 80% overlap]353 ±9 67.5 ±0.9 1.4 ±0.1
S6
[24 μJ, 100 kHz, 60 ns, 80% overlap]370 ±20 68.2 ±0.1 1.5 ±0.1
S8
[16 μJ, 150 kHz, 20 ns, 87% overlap]344 ±12 63.4 ±0.6 1.7 ±0.1
S9
[12 μJ, 200 kHz, 20 ns, 90% overlap]369 ±4 63.9 ±11.7±0.1
The measurement uncertainty reflects the standard deviation in the measurements.
30 min showed slightly lower average Voc and pFF, as compared
with those annealed at 700 °C for 30 min. The samples with the
lowest Voc and pFF were those that were directly hydrogenated
after the LCP process (see Table III). This could be the result
of an incomplete passivation of defects from the hydrogenation
process or lower dopant activation.
Considering that the same hydrogenation parameters were
applied to all of the samples, it appears that the intermediate
oven anneal is the limiting factor in order to achieve a high Voc
and pFF. From Fig. 4(a) and (b), it is observed that a higher
temperature (i.e., 700 °C for 30 min) results in higher aver-
age Voc and pFF values except for S2, which could be due to
nonhomogenous doping or defects within the poly-Si. Hence,
it is likely that a high-temperature anneal such as a RTP before
the hydrogenation process can lead to much higher Voc values
than the ones obtained in this study. Prior work by Rau et al.
[21] demonstrated that the increase in open-circuit voltages af-
ter hydrogenation depended strongly on the RTP step applied
to the samples. They also reported that the Voc increased al-
most linearly with the RTP plateau temperature. Additionally,
the hydrogenation parameters used in the current investigation
could be further optimized for the LCP-doped samples. Other
studies showed that hydrogenation can also lead to detrimen-
tal effects such as deactivation of dopants, particularly p-type
dopants [22], platelets [23], etc. For instance, a recent study by
Qiu et al. [19] has revealed that depending upon the hydrogena-
tion temperature, platelets can be localized at different depths
within the poly-Si thin-film solar cells. Such hydrogen-induced
defects have localized states within the bandgap and can de-
crease Voc and pFF significantly. Furthermore, in this study, the
Voc and pFF were also limited by the material quality of the
poly-Si absorber, which was fabricated by SPC. More details
about the different recombination mechanisms affecting SPC
poly-Si thin-film cells can be found in [24]. It is believed that
LCP is transferrable from SPC to higher quality poly-Si films
(e.g., poly-Si made by the LPC approach) resulting in similar
1450 IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 4, NO. 6, NOVEMBER 2014
Fig. 5. Effective ideality factor of the hydrogenated and nonhydrogenated
solar cells after the LCP process. The effective ideality factor of the nonhy-
drogenated as-doped samples was not included because the dopants were not
activated.
cell structures as described in [4]. Therefore, further improve-
ment in the Voc and pFF is possible.
Since the p-n junctions in this study were formed predom-
inantly at the surface (air side) of the solar cells, the light-
generated electron–hole pairs have to diffuse a relatively long
distance before their separation by the p-n junction. As a result,
the light-generated current and the Voc were relatively low. In
order to predict the gain in Voc for a p-n junction located close to
the glass side, a simplified model of a 2-μm-thick poly-silicon
thin-film solar cell on glass was implemented into the solar cell
modeling program PC1D [25]. By using the ECV data from
Fig. 2 as the peak doping concentration/profiles of the active
layers (i.e., the bulk layer, the emitter, and the back surface
field), the model showed that a shift in the location of the p-n
junction from the air side to the glass side could result in a
Voc gain of about 7 mV. In principle, for a superstrate device,
such a p-n junction cannot be fabricated directly by LCP due
to the device architecture of polysilicon thin-film solar cells on
glass, which comprises an emitter, an absorber, and a back sur-
face field. However, for superstrate architecture, LCP could be
used to potentially fabricate an n-type back surface field on an
n−/p+/SiNxglass. Another way is to fabricate a p-type back sur-
face field on a p−/n+/SiNxglass. From Fig. 4(c), the effective
ideality factor of the best hydrogenated cells (Voc >400 mV)
is close to 2, suggesting that the dominant recombination mech-
anism in the LCP-doped hydrogenated samples may either be
localized within the space charge region, such as the p-n junc-
tion, or due to extended defects, such as dislocations and grain
boundaries.
Fig. 5 compares the effective ideality factor of the hy-
drogenated solar cells before and after the hydrogenation
process. The effective ideality factors of the as-doped nonhy-
drogenated samples are not included because the dopants were
not activated, and thus, Suns-Voc measurements could not be
conducted.
From Fig. 5, there is an increase in the effective ideality factor
of most hydrogenated samples (in particular those annealed at
610 °C for 2 h and 700 °C for 30 min) as compared with that
of their nonhydrogenated counterparts. It would be expected
that the neff would decrease upon hydrogenation due to passi-
vation of grain boundary defects such as dangling bonds, etc.
Instead, these results suggest that hydrogenation-related defects
such as platelets may be the likely cause of the higher neff .
Deep-level platelets could be localized within the space-charge
region that could yield higher saturation currents [19]. A possi-
ble solution could be to hydrogenate the samples at lower tem-
peratures. Sample S1 from the as-doped hydrogenated batch
showed a much higher effective ideality factor and lower pFF
(see Table III). This could be due to jet instability during dop-
ing. As a result, the sample can be treated as an outlier in this
experiment.
Overall, it is probable that sample S5 (see Table III) may yield
the best Voc and pFF after a similar hydrogenation process. A
supporting argument is that the effective ideality factor is quite
low (about 1.4). Even though S6 displays a similar effective
ideality factor, the high fluence during the LCP process causes
simultaneous removal of the doped material and introduces sig-
nificant material damage. This also resulted in an irregular dop-
ing profile [see Fig. 2(c)]. Future work will consist of optimizing
the hydrogenation process and measuring the external quantum
efficiency and light current–voltage (I–V) for a complete assess-
ment of the device properties and performance. Additionally,
future studies will identify the dominant structural [e.g., using
Raman spectroscopy, TEM, etc.]and electrically active defects
[e.g., by electron beam-induced current]in the LCP-doped films.
The idea behind such study would be to investigate whether the
recombination is mostly linked to shallow band or deep-level re-
combination. The former arises mostly from clean dislocations,
while the latter is influenced by charged dislocations or grain
boundaries [24].
IV. CONCLUSION
In this paper, we have investigated the suitability of LCP in
fabricating n-type emitters for polysilicon thin-film solar cells
on glass. The measured sheet resistances of the annealed sam-
ples were about 2–5 kΩ/, and the dopant concentration was
about 8×1018–1×1019 cm−3at a doping depth of less than
350 nm (as measured by electrochemical capacitance–voltage).
Selected LCP-doped samples were then subjected to thermal
annealing and hydrogenation. The best cell had an average Voc
of (446 ±7) mV and a pFF of (68.3 ±0.9)%. It was also
observed that the annealing step was the limiting factor for a
higher Voc and pFF as demonstrated by the samples treated
at 700 °C for 30 min. Furthermore, the effective ideality fac-
tor of those cells were closer to 2 indicating that the dominant
form of recombination could result either from the space-charge
region or from extended defects along grain boundaries, dislo-
cations, and so forth. A comparison between the effective ide-
ality factor of the hydrogenated and nonhydrogenated samples
revealed that hydrogenation-induced defects such as platelets
may have been the cause. In summary, it is likely that a RTP
step combined with an optimized hydrogenation process may
further improve the Voc and the pFF in this study. Future work
will consist of identifying the dominant structural and electri-
cally active defects that limit the performance of the LCP-doped
solar cells.
VIRASAWMY et al.: LASER CHEMICAL PROCESSING OF n-TYPE EMITTERS FOR SOLID-PHASE CRYSTALLIZED POLYSILICON 1451
REFERENCES
[1] J. Dore, R. Evans, S. Ute, B. D. Eggleston, D. Ong, K. Kim, J. Huang,
O. Kunz, M. Keevers, R. Egan, S. Varlamov, and M. A Green, “Thin-film
polycrystalline silicon solar cells formed by diode laser crystallisation,”
Prog. Photovoltaics: Res. Appl., vol. 21, pp. 1377–1383, 2013.
[2] B. Eggleston, S. Varlamov, and M. Green, “Large-area diode laser defect
annealing of polycrystalline silicon solar cells,” IEEE Trans. Electron.
Devices, vol. 59, no. 10, pp. 2838–2841, Oct. 2012.
[3] P. J. Gress, P. I. Widenborg, S. Varlamov and A. G. Aberle, “Wire bonding
as a cell interconnection technique for polycrystalline silicon thin-film
solar cells on glass,” Prog. Photovoltaics: Res. Appl., vol. 3, pp. 221–228,
2010.
[4] J. Dore, D. Ong, S. Varlamov, R. Egan, and M. A. Green, “Progress in
laser-crystallized thin-film polycrystalline silicon solar cells: Intermediate
layers, light trapping, and metallization,” IEEE J. Photovoltaics,vol.4,
no. 1, pp. 33–39, Jan. 2014.
[5] J. Haschke, D. Amkreutz, L. Korte, F. Ruske, and B. Rech, “Towards
wafer quality crystalline silicon thin-film solar cells on glass,” Sol. Energy
Mat. Sol. Cells, vol. 128, pp. 190–197, 2014.
[6] F. Granek, C. Fleischmann, S. Kluska, D. Erath, and J. Rentsch, “Screen-
printed silicon solar cells with LCP selective emitters” in Proc. 25th
Eur. Photovoltaic Sol. Energy Conf. Exhib/5th World Conf. Photovoltaic
Energy Convers., Valencia, Spain, 2010, pp. 2573–2577.
[7] S. Hopman, A. Fell, K. Mayer, M. Mesec, A. Rodofili, and D. Kray,
“Comparison of laser chemical processing and laser microjet for structur-
ing and cutting silicon substrates,” J. Appl. Phys. A, vol. 95, pp. 857–866,
2009.
[8] S. Virasawmy, N. Palina, P.I. Widenborg, A. Kumar, G. K. Dalapati,
H. R. Tan, A. A. O. Tay, and B. Hoex, “Direct laser doping of poly-silicon
thin films via laser chemical processing,” IEEE J. Photovoltaics,vol.3,
no. 4, pp. 1259–1264, Oct. 2013.
[9] H. Hidayat, P. I. Widenborg, and A. G. Aberle, “Large-area Suns-Voc
tester for thin-film solar cells on glass superstrates,” Energy Procedia,
vol. 15, pp. 258–264, 2012.
[10] O. Kunz, Z. Ouyang, S. Varlamov, and A. G. Aberle, “5% Efficient evapo-
rated solid-phase crystallised polycrystalline silicon thin-film solar cells,”
Prog. Photovoltaics: Res. Appl., vol. 17, pp. 567–573, 2009.
[11] F. M. Smits, “Measurement of sheet resistivities with the four-pointprobe,”
Bell Syst. Tech. J., vol. 37, pp. 711–718, 1958.
[12] H. Hidayat, A. Kumar, F. Law, C. Ke, P. I. Widenborg, and A. G. Aberle,
“Impact of rapid thermal annealing temperature on non-metallised poly-
crystalline silicon thin-film diodes on glass,” Thin Solid Films, vol. 534,
pp. 629–635, 2013.
[13] O. Kunz, Z. Ouyang, J. Wong, and A. G. Aberle, “Advances in evaporated
solid-phase-crystallized poly-Si thin-film solar cells on glass (EVA),” Adv.
OptoElectron., vol. 2008, pp. 1–10, 2008.
[14] A. Kumar, H. Hidayat, C. Ke, S. Chakraborty, G. K. Dalapati, P. I. Widen-
borg, C. C. Tan, S. Dolmanan, and A. G. Aberle,” Impact of the n+
emitter layer on the structural and electrical properties of P-type polycrys-
talline silicon thin-film solar cells,” J. Appl. Phys., vol. 114, pp. 134505-
1–134505-7, 2013.
[15] J. Bovatsek, A. Tamhankar, R. Patel, N. M. Bulgakova, and J. Bonse, “
Effects of pulse duration on the ns-laser pulse induced removal of thin film
materials used in photovoltaics,” Proc. SPIE, vol. 7201, pp. 1–13, 2009.
[16] A. G. Aberle and P. I. Widenborg, “Crystalline silicon thin-film so-
lar cells via high-temperature and intermediate-temperature approaches,”
in Handbook of Photovoltaic Science and Engineering, London, U.K.:
Wiley, 2011, pp. 452–486.
[17] Z. Ouyang, O. Kunz, A. B. Sproul, and S. Varlamov, “Influence of the
absorber doping for p-type polycrystalline silicon thin-film solar cells on
glass prepared by electron beam evaporation and solid-phase crystalliza-
tion,” J. Appl. Phys., vol. 109, pp. 054510-1–054510-7, 2011.
[18] M. I. Keevers, A. Turner, U. Schubert, P. A. Basore, and M. A. Green,
“Remarkably effective hydrogenation of crystalline silicon on glass mod-
ules,” in Proc. 20th Eur. Photovoltaic Sol. Energy Conf., Barcelona, Spain,
2005, pp. 1305–1308.
[19] Y. Qiu, O. Kunz, A. Fejfar, M. Ledinsk´
y, B. T. Chan, I. Gordon, D. V.
Gestel, S. Venkatachalm, and R. Egan, “On the effects of hydrogenation
of thin film polycrystalline silicon: A key factor to improve heterojunction
solar cells,” Sol. Energy Mat. Sol. Cells, vol. 122, pp. 31–39, 2014.
[20] C.-F Yeh, T.-J Chen, C. Liu, J. T. Gudmundsson and M. A. Lieberman,
“Hydrogenation of polysilicon thin-film transistor in a planar inductive
H2/Ar discharge,” IEEE Electron. Device Lett., vol. 20, no. 5, pp. 223–
225, May 1999.
[21] B. Rau, T. Weber, B. Gorka, P. Dogan, F. Fenske, K. Lee, S. Gall, and
B. Rech, “Development of a rapid thermal annealing process for polycrys-
talline silicon thin-film solar cells on glass,” Mater. Sci. Eng., B, vol. 159,
pp. 329–332, 2009.
[22] P. I. Widenborg and A. G. Aberle, “Hydrogen-induced dopant neutral-
isation in p-type AIC poly-Si seed layers functioning as buried emit-
ters in ALICE thin-film solar cells on glass,” J. Cryst. Growth, vol. 306,
pp. 177–186, 2007.
[23] S. Honda, T. Mates, M. Ledinsk´
y, A. Fejfar, J. Koˇ
cka, T. Yamazaki,
Y. Uraoka, T. Fuyuki, H. Boldyryeva, A. Mackov´
a, and V. Peˇ
rina, “Defects
generation by hydrogen passivation of polycrystalline silicon thin films,”
Sol. Energy, vol. 80, pp. 653–657, 2006.
[24] J. Wong, J. Huang, S. Varlamov, M. A. Green and M. Keevers, “The roles
of shallow and deep levels in the recombination behaviorof polycrystalline
silicon on glass solar cells,” Prog. Photovoltaics: Res. Appl., vol. 20, pp.
915–922, 2012.
[25] P. A. Basore, “Numerical modeling of textured silicon solar cells us-
ing PC-1D,” IEEE Trans. Electron Devices, vol. 37, no. 2, pp. 337–343,
Feb. 1990.
Authors’ photographs and biographies not available at the time of publication.
