Beruflich Dokumente
Kultur Dokumente
Donghwan Kima
a
Department of Material science and Engineering, Korea University, 145, Anam-ro, Seongbuk-gu, Seoul 02841, Korea
b
KU-KIST Green School Graduate School of Energy and Environment, Korea University, 145, Anam-ro, Seongbuk-gu, Seoul 02841, Korea
Keywords: Passivation quality of poly-Si contacts with different phosphorus doping concentration were investigated in this
Polysilicon study. Intrinsic poly-Si layers were deposited by LPCVD on a tunnel oxide surface, followed by n + poly-Si
Tunnel oxide doping and hydrogenation. For lightly doped poly-Si contacts with phosphorus concentration of 2.1 × 1019 cm 3 ,
Passivated contact higher temperatures and longer times increased iVOC achieving maximum value of 734 mV, as poly-Si grain size
N-type silicon
increases from 13 nm to 40 nm. However, for heavily doped poly-Si contacts with phosphorus concentration of
High efficiency
1.1 × 1020 cm 3 , iVOC decreased from 731 mV to 696 mV as annealing time increased from 10 to 60 min because
Solar cell
Auger recombination rate increased from 9.3 fA/cm2 to 21.6 fA/cm2 as phosphorus in-diffusion occurs. The
contact resistance of poly-Si contacts was also investigated to achieve a high fill factor. Finally, a poly-Si/SiOx/c-
Si passivated contact solar cell using a poly-Si contact on the back and boron diffused emitter on the front was
fabricated. As a result, high efficiency of 21.1% solar cell was achieved with VOC of 665 mV, JSC of 40.6 mA/cm2,
and fill factor of 78.3%.
1. Introduction efficiency of 21.2% [6–10]. About iVOC values, about 730 mV was ob-
tained using crystallized a-Si passivated contact [9,11] and LPCVD
Recombination losses at the interface between the metal contact and deposited poly-Si passivated contact [12]. The highest iVOC of 749 mV
doped-semiconductor region is a major efficiency-limiting factor in was reported by ISFH [13] for the latter with phosphorus implantation.
solar cells. To reduce losses from metal-semiconductor contact re- Although many groups involved poly-Si contact, the relation between
combination, the concept of passivated contact was introduced. A the poly-Si layer and passivation is still unclear because researchers
tunnel oxide passivated contact (TOPCon) solar cell [1] is composed of focused on the tunnel oxide quality. However, characteristics of the
a n + poly-Si/SiOx/c-Si contact on the rear side and boron doped poly-Si layer also affect on passivation quality of poly-Si contacts ac-
emitter on the front side, and has a reported highest efficiency of 25.8% cording to its annealing condition and doping concentration. Thus, we
[2,3]. One major method to form poly-Si/SiOx/c-Si passivated contacts focused our study, to better understand the effect of the poly-Si layer on
is to deposit an amorphous silicon layer (a-Si) and crystallized a-Si onto passivation quality, on the grain-size growth and phosphorus in-diffu-
poly-Si with subsequent annealing. Another way is to use the low- sion. High efficiency was achieved with a tunnel-oxide-passivated
pressure chemical vapor deposition method (LPCVD) for poly-Si de- contact solar cell with a homogeneous boron emitter on the front and
position, which provides better crystallinity. Using LPCVD-deposited poly-Si contact on the rear.
poly-Si/SiOx/c-Si passivated contact (hereafter, poly-Si contact), some For the experiments, the poly-Si contacts were annealed at different
groups reported poly-Si contact solar cells using POCl3 diffusion temperatures and times, and with different phosphorus doping con-
method and the highest efficiency was 21.5% [4,5]. Also, some groups centrations, in the poly-Si layer. Passivation quality was determined
reported poly-Si contact solar cells formed by ion implantation method using iVOC and J0 measured by the quasi-steady-state photoconductance
to form n + doping on i-poly-Si layer which achieved the highest (QSSPC) method [14]. The J0 values are determined under the high
⁎
Corresponding author.
E-mail addresses: jung1029@korea.ac.kr (H. Park), solar@korea.ac.kr (H.-S. Lee).
URL: http://solar.korea.ac.kr (D. Kim).
https://doi.org/10.1016/j.solmat.2018.09.013
Received 25 October 2017; Received in revised form 25 June 2018; Accepted 10 September 2018
0927-0248/ © 2018 Elsevier B.V. All rights reserved.
H. Park et al. Solar Energy Materials and Solar Cells 189 (2019) 21–26
2. Experimental method
22
H. Park et al. Solar Energy Materials and Solar Cells 189 (2019) 21–26
correlated with annealing temperature, and that the high-temperature size of the poly-Si was estimated as 20 nm for the LD poly-Si contacts
annealing process is very important for achieving high-quality passi- sample annealed at 950 °C for 10 min. Also, the grain size for UD, LD
vation of the poly-Si contacts solar cell. and HD contact samples increased from 13 nm to 22 nm as annealing
Second, we changed the annealing time at a fixed temperature of temperature increased from 800° to 950°C. This result matches well
950 °C, because the highest iVOC was achieved at 950 °C, as shown in with Harbeke's result [17], which indicates that the grain size of the as-
Fig. 2 (a). As shown in Fig. 2 (b), with increase in the annealing time dep polysilicon layer deposited at 600 °C was 11 nm and it grew after
from 10 to 60 min, iVOC increased from 721 mV to 734 mV for the LD annealing at 900–1000 °C. The final grain size of polysilicon annealed
poly-Si contacts. However, the iVOC of the HD poly-Si contacts sample at 1000 °C was 18 nm from Harbeke's result [17], thus indicating a si-
decreased as the annealing time increased. This result indicates that milar grain size and growth effect in our experiments. And also, the
higher temperature and longer annealing time improves passivation result of increased grain size indicates the increased iVOC of the UD poly-
quality after low-concentration poly-Si doping. To analyze the cause of Si contact was caused by the effect of the grain size growth.
this difference in behavior, we focused on the grain size of the poly-Si Increasing the annealing time induced poly-Si grain growth. The
layer and phosphorus in-diffusion, and the effect of these factors on grain sizes of the poly-Si contacts annealed for 60 min at 950 °C were
passivation quality. twice as large as the poly-Si contacts sample annealed for 10 min
The average grain size of the poly-Si layer was measured using the (20–40 nm). Increasing the grain size of the poly-Si contacts sample
GI-XRD method and Scherrer's equation [15] resulted in an increase in iVOC of the LD poly-Si contacts. This result
indicates that the grain size of the poly-Si layer induced a high inter-
0.9
t= face-passivation quality in the poly-Si contacts between poly-Si and
Bcos (1)
SiOx while P in-diffusion decreases the iVOC of the HD poly-Si contact.
where t is the average grain size, is the wavelength of x-ray, B is full The grain size effect on passivation at the poly-Si/SiOx interface was
width half maximum and is the radians. As a result, the average grain analyzed by two physical reason. First, the energy band changed at the
Fig. 3. (a) 3D schematic of the energy band diagram for the poly-Si/SiOx/c-Si poly-Si contacts structure and 2D schematic of the energy band diagram at the interface
between poly-Si/SiOx/c-Si grains and their boundaries [18] and (b) measured J0 as a function of average grain size for LD poly-Si contacts and simulation result.
23
H. Park et al. Solar Energy Materials and Solar Cells 189 (2019) 21–26
poly-Si grain boundaries with SiOx, compared with those within the poly-Si contacts but slightly increased for LD poly-Si contacts. From the
grains, as shown in Fig. 3(a). At the grain region, the schematic energy simulation result and lifetime curve, the phosphorus in-diffusion is
band diagram is the same as the energy band of an n-type crystalline partially responsible for the Auger recombination and lowered iVOC . The
silicon/SiOx interface. However, at poly-Si grain boundary regions, the difference between the measured J0 and the calculated J0 Auger can be
energy band changes because electrons are depleted to the defect levels induced from the increase in the surface recombination caused by the
that exist at the grain boundaries. This causes the Fermi level of the
grain boundary to decrease and align with the intrinsic Fermi level.
Because of the different energy band structure, a barrier height for
electrons is formed and holes, which are minority carriers, can tunnel
through the tunnel oxide layer. Moreover, holes tunneling through
oxide move to the poly-Si grain boundaries and recombine with elec-
trons captured at the interface defect of the grain boundary. Thereby,
the poly-Si grain size affects the passivation quality at the poly-Si/SiOx
interface.
To analyze the relationship between grain size and the recombina-
tion rate of the poly-Si contacts solar cell, we made a simple assumption
that each grain is rectangular-shaped with square of length d and depth
t, resulting in a grain volume of d 2· t . From the assumption, interface
defect produced from the grain boundary is characterized by defect
concentration at the interface of SiOx/Si dangling bonds per unit length
(Ndis ) and grain size (d) and equation of interface defect becomes
Ndis
Dit =
d (2)
qni2
Jo = Seff
ND (3)
24
H. Park et al. Solar Energy Materials and Solar Cells 189 (2019) 21–26
poly-Si annealed under the same conditions on the back. On the front of
the solar cell, a homogeneous boron-diffused emitter was applied and
Al2O3/SiNx stack layer was deposited on the front surface as passivation
layers. For metallization, the evaporation method was used for front
and back metal contacts.
Light I-V measurements were made of the solar cell, and the results
are shown in Fig. 6. The solar cell showed 21.1% efficiency with a VOC
of 665 mV, JSC of 40.6 mA/cm2, and fill factor of 78.3% with 2 × 2 cm2
cell size. The difference between implied VOC and cell VOC is expected to
be further reduced by the optimization of the emitter, front passivation,
and metallization process. The limitation of VOC at the cell level is ex-
pected to 725 mV based of the QUOKKA simulation program [28].
4. Summary
Fig. 6. (a) Schematic illustration of solar cell fabricated using HD poly-Si contact annealed at 950 °C for 30 min structure and (b) corresponding light I-V curve.
SiOx interface defect density and the oxide break-up, since the latter can results showed that the passivation quality of the poly-Si contacts im-
be enhanced when the doping concentration at poly-Si increases [26]. proved with annealing at higher temperature (950 °C) and longer time
Further, the increase in the dopant concentration in the poly-Si layer (60 min). This improvement is due to a grain growth effect on passi-
also affect the barrier height between the poly-Si/SiOx/c-Si layers, with vation quality because the grain size of poly-Si affects the recombina-
an increased barrier height from 0.217 eV to 0.259 eV by increasing the tion rate. However, high doping concentrations of poly-Si result in in-
dopant concentration from 2.1 × 1019 cm 3 (LD) to 1.1 × 1020 cm 3 (HD). creasing phosphorus in-diffusion, with increase the annealing time,
This may affect the interface passivation quality, but a deeper and more which is different from the case of low-concentration doped poly-Si
detailed study is required to better understand physical relationship contacts. This result was learned after simulation of the recombination
between the barrier height and passivation quality. current density from Auger recombination. Moreover, the lifetime
curve under the high-injection condition also decreased, meaning that
3.3. Fabrication of solar cell using poly-Si contact the Auger recombination rate increased. The conclusion is that high
temperature and long-time annealing are required to achieve large
To fabricate a poly-Si/SiOx/c-Si passivated contact solar cell using grain size and high iVOC in cases where critical degradation from
the poly-Si contacts, contact resistances were measured using Ag eva- phosphorus in-diffusion does not occur. Based on these experimental
porated TLM patterns with reactive ion etching to remove poly-Si re- results, a solar cell was fabricated, and achieved an efficiency of 21.1%.
gion without being covered by Ag. The contact resistance of poly-Si/
SiOx/c-Si was analyzed using a 2D TLM method [27].
The results indicate that the contact resistance of the LD poly-Si Acknowledgement
contacts decreased as annealing time increased to 394.27 m cm2 , as
shown in Fig. 5. Better contact characteristics were achieved with lower This work was supported by the New & Renewable Energy Core
contact resistance for HD poly-Si contacts, achieving contact resistance Technology Program of the Korea Institute of Energy Technology
below 46.31 m cm2 . The lowest contact resistance was 21.85 m cm2 , Evaluation and Planning (KETEP), which was granted financial re-
achieved for HD poly-Si contacts annealed at 950 °C for 60 min. sources from the Ministry of Trade, Industry, and Energy, Republic of
Given the optimal contact resistance and high iVOC of 731.5 mV for Korea (Nos. 20163030014020 and 20154030200760). This work was
the HD poly-Si contacts annealed at 950 °C for 30 min, we fabricated a also supported by the KU-KIST Graduate School Project, Republic of
solar cell applying the poly-Si contact structure with heavily doped Korea.
25
H. Park et al. Solar Energy Materials and Solar Cells 189 (2019) 21–26
To drive the equation between the grain size of the poly-Si layer and the interface recombination rate, we made the simple assumption that each
grain is rectangular-shaped with a square of length d and depth t, resulting in a grain volume of d 2· t . The total area of the sample was set as A. The
grain boundary region at the interface is the same as the product of the grain boundary length of one grain and number of grains in the total area.
Since each grain is surrounded by four edges (length of 4d ) and the edges are shared with four adjacent grains, the grain boundary length of one grain
has a length of 2d . The number of grains are equal to the total volume divided by the grain volume, which is At / d 2t . Then, the grain boundary length
becomes A/ d and by dividing by the total area, the grain boundary length per unit area (LGB ) becomes;
At 2
LGB = 2d × ÷A=
d 2t d (A.1)
From the above equation, the grain boundary length at the interface of SiO x/ poly-Si is inversely proportional to the grain size. Since we assumed
that the grain boundary act as interface recombination centers, we set the constant NGB as the defect density per grain boundary length (NGB ). As a
result, the equation for the interface defect produced from the grain boundary (Dit ) becomes,
NGB
Dit =
d (A.2)
26