You are on page 1of 6

5618 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 69, NO.

10, OCTOBER 2022

Detailed Study on the Role of Nature and


Distribution of Pinholes and Oxide Layer on the
Performance of Tunnel Oxide Passivated
Contact (TOPCon) Solar Cell
Sourav Sadhukhan, Shiladitya Acharya, Tamalika Panda, Nabin Chandra Mandal,
Sukanta Bose, Anupam Nandi, Gourab Das, Susanta Chakraborty, Member, IEEE,
Santanu Maity , Senior Member, IEEE, Partha Chaudhuri, Member, IEEE, and Hiranmay Saha

Abstract — Industrial silicon solar cells are now mostly I. I NTRODUCTION


based on aluminum back surface field (Al-BSF) or pas-
sivated emitter rear cell (PERC) technologies on p-type
crystalline silicon wafers. Recently tunnel oxide passivated
T HE tunnel oxide passivated contact (TOPCon) is an
emerging technology after passivated emitter rear cell
(PERC), which is expected to become important industrially
contact (TOPCon) solar cell on p-type Si wafers has
attracted attention due to its demonstrated higher efficiency in the coming decades [2]. A thorough understanding of this
than either Al-BSF or PERC type solar cell. Numerical structure is essential for further improvement of its compati-
analysis using 3-D Sentaurus Technology Computer Aided bility with the industrial production system. Again, the trend
Design (3-D-TCAD) software leads to the enhancement of of the market is based on p-type wafers [2], so p-type TOPCon
the efficiency of the p- and n-type TOPCon solar cells by
optimizing the size, nature, and number density of pinholes may become the main work horse for industry in coming
in the oxide layer; thickness of the oxide layer with and decades. In the TOPCon solar cell [see Fig. 1(a)], an ultrathin
without pinholes and B doping concentration in the hole silicon oxide (SiOx ) layer (∼1.5 nm) serves as the passivation
selective p+ poly-Si layer at the rear. Effects of both types layer as well as tunneling layer at the rear side, followed by
of pinholes, either completely through (physical contact) a highly doped polycrystalline silicon (p+ for hole selective
or partially through (localized thinner oxide), are studied
on cell performance. Simulation results show that pinholes and n+ for electron selective) layer, which again followed by
in tunnel oxide have an advantage in lowering of series SiNx deposition and the screen printed (Al/Ag or Ag) or plated
resistance and improvement of fill factor. To achieve opti- (required laser ablation) back contact, whereas the front side
mum performance, the size, nature, and number density of is kept intact as aluminum back surface field (Al-BSF) or
pinholes and thickness of the oxide layer should be opti- PERC solar cell [3], [4], [5], [6], [7], [8], [9], [10]. Main
mized. Considering both types of pinholes, the efficiency
achieved is 25.3% for p-TOPCon and 26% for n-TOPCon. advantage of the TOPCon solar cell is the lower recombina-
Also, the outputs of simulated p-TOPCon are compared tion current density J0 and higher efficiency than either the
with simulated p-PERC solar cell. The analysis shows that Al-BSF or PERC solar cell [1], [2], [3]. Experimental results
TOPCon solar cell on p-type wafer has significant ability to show that metal/p+ poly-Si/SiOx passivation contact on p-type
be adopted for industrial production. c-Si is poorer than metal/n+ poly-Si/SiOx passivation contact
Index Terms — Passivating contact, pinhole density, pin- on n-type c-Si substrate, and the efficiency of p-TOPCon
hole nature, pinhole size, poly-Si, tunnel oxide, tunnel oxide is far behind the efficiency of n-TOPCon solar cell in lab
passivated contact (TOPCon) solar cell, tunneling. scale [3], [4], [5], [11]. Highest efficiencies achieved are 26%
Manuscript received 4 May 2022; revised 25 July 2022; accepted for n-type TOPCon and 26.1% for poly-Si on oxide with
28 July 2022. Date of publication 17 August 2022; date of current version interdigitated back contact (POLO-IBC) structure [3] in lab
22 September 2022. This work was supported by the DST Solar PV Hub scale and values are higher than the efficiencies achieved for
Phase II at IIEST Shibpur under Project DST/ TMD/SERI/HUB/2(G). The
review of this article was arranged by Editor B. Hoex. (Corresponding Al-BSF or PERC solar cell.
authors: Santanu Maity; Susanta Chakraborty.) Ideally, transport of carriers through the ultrathin SiOx layer
Sourav Sadhukhan, Shiladitya Acharya, Tamalika Panda, Nabin was considered to be solely by quantum tunneling [9], [12].
Chandra Mandal, Sukanta Bose, Anupam Nandi, Gourab Das, Santanu
Maity, Partha Chaudhuri, and Hiranmay Saha are with the School of The SiOx layer is grown or deposited on the rear side of the Si
Advanced Materials, Green Energy and Sensor Systems, DST-IIEST wafer by wet chemical oxidation [12], ozone oxidation [13],
Solar PV Hub, Howrah 711103, India (e-mail: smaity.cegess@faculty. thermal oxidation [13], [14], plasma-assisted oxidation [14],
iiests.ac.in).
Susanta Chakraborty is with the Department of Computer Science [15], and atomic layer deposition (ALD) process [16]. In prac-
and Technology (CST), IIEST Shibpur, Howrah 711103, India (e-mail: tice, the ultrathin tunneling dielectric layer formed by the
susanta.chak@gmail.com). above methods contains pores through which poly-Si makes
Color versions of one or more figures in this article are available at
https://doi.org/10.1109/TED.2022.3196327. direct contact with Si substrate, which are defined as com-
Digital Object Identifier 10.1109/TED.2022.3196327 pletely through pinholes. The density and sizes of pinholes

0018-9383 © 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Tsinghua University. Downloaded on June 23,2023 at 14:41:20 UTC from IEEE Xplore. Restrictions apply.
SADHUKHAN et al.: DETAILED STUDY ON THE ROLE OF NATURE AND DISTRIBUTION OF PINHOLES AND OXIDE LAYER 5619

TABLE I
P HYSICAL PARAMETERS FOR THE S IMULATION S TUDY

Fig. 1. (a) Schematic and (b) energy band diagram of p-TOPCon solar
cell.

in the oxide layer can be measured by conductive atomic


force microscopy (c-AFM) [17] and electron-beam-induced
current (EBIC) measurement [18]. Measured values of the
density of pinholes in the oxide layer have been found to be in
the range of 105 –1010 cm−2 , while their radii are in the range
of 2–40 nm [17], [18], [19]. All pinholes formed during the
growth of oxide layer or in the time of annealing may not be
completely through, i.e., polysilicon does not make physical
contact with silicon substrate through pinholes. A very thin
oxide in the order of a few ångströms may exist between
silicon and polysilicon, which is termed as partially through
pinholes [17].
In this work, we have considered both types of pinholes, i.e.,
completely through and partially through pinholes in the tunnel
oxide layer, and a comprehensive numerical analysis has been
done in detail on the role of the nature and distribution (i.e.,
completely through and partially through) of pinholes, their
sizes, their number densities for n- and p-type TOPCon solar
and at the surface is modeled by using Shockley–Read–Hall
cells, and the boron doping concentration in the p+ poly-
(SRH) recombination equations [25]. The tunnel current is
Si carrier selective layer for p-type TOPCon solar cell by
calculated using Tsu-Esaki model [26] with Wentzel–Kramer–
using a 3-D Sentaurus Technology Computer Aided Design
Brillouin (WKB) approximation [27]. Series resistance (Rs ) of
(3-D-TCAD) software.
the cell is extracted using multiple light method [28].
II. M ODEL OF THE S IMULATION S TUDY
III. S IMULATION O BSERVATION AND D ISCUSSION
The 3-D-TCAD is an advanced simulation tool, in which
optical and electrical simulations can be done simultaneously. A. Effect of the Thickness of Tunnel Layer (SiO2 ) for Pure
It can handle regions of a few nanometers to a few centimeters Tunneling of Majority Carriers
in a device, and results have been found to be accurate Fig. 1(b) shows the energy band diagram of Si, an ultrathin
enough to support the experimental data [20]. One can also SiO2 layer, and a highly doped p+ poly-Si. Band bending
reproduce the design of a complex device structure with the occurs at the interfaces due to the difference in electron affinity
help of 3-D-TCAD. In our study, we have used the p-type and doping concentration of the materials. Surface potential is
TOPCon structure, as shown in Fig. 1(a), with the individ- built at the Si/SiO2 interface due to the alignment of Fermi
ual layer and interface properties as briefed in Table I. For level [39], which repels the minority electrons away from the
optical simulation study, random upright pyramidal textured interface. On the other hand, majority holes in the valence
(maximum base width 6 μm and height 4 μm) surface is band of silicon accumulate at the interface and can tunnel
considered and advanced ray-tracing model [21] has been through the thin oxide layer.
used as optical solver. Optical generations are calculated Fig. 2(a) shows the variation of fill factor (FF) in percentage
by absorbed photon density and quantum yield model [21] and efficiency in percentage with tunnel oxide thickness in
considering complex refractive indices and thicknesses of all nanometer of a p-type TOPCon solar cell. From this figure,
materials [22], [23], [24] used in this simulation. Doping, it is clear that the FF of the cell decreases with increase
temperature, and field-dependent lifetime and mobility models in tunnel oxide thickness (dox ). For dox = 0.6 nm, the FF
are used. Defects and fixed charges are considered in poly-Si is 84.4 with series resistance (Rs ) 0.23 ·cm2 , and for
region and nitride/Si, oxide/Si, and oxide/poly-Si interfaces dox = 1.4 nm, FF becomes 67.4% with Rs 3.20 ·cm2 .
from experimental results reported in literatures (see Table I). It is well known that with increase in series resistance, the
Recombination of carriers through a defect level in the bulk FF decreases [39]. The increase of Rs is due to exponential

Authorized licensed use limited to: Tsinghua University. Downloaded on June 23,2023 at 14:41:20 UTC from IEEE Xplore. Restrictions apply.
5620 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 69, NO. 10, OCTOBER 2022

Fig. 2. (a) Variation of FF in percentage and efficiency in percentage with


tunnel oxide thickness in nanometer. (b) Variation of series resistance
(Rs in Ω·cm2 ) and efficiency in percentage with p+ poly-Si doping
concentration (NA , poly-Si in cm−3 ).

decrease of tunneling probability with increase in barrier


(tunnel oxide) thickness [28]. From Fig. 2(a), it is also seen
that the efficiency rises to 24.6% for dox = 1.0 nm and
then decreases with further increase in dox . This is due to:
Fig. 3. Variation of (a) and (c) series resistance (Rs ) and FF (b) and
1) increase in open-circuit voltage (Voc ); 2) increase of short- (d) recombination current density (J0 ) and cell efficiency with pinhole
circuit current (Isc ) for (dox < 1.2 nm); and 3) very small radius (Rpin ) for tunnel oxide thickness (dox ) 1.4 nm and tunnel oxide
decrease of FF with increase in dox for dox ≤ 1.0 nm, and thickness for pinhole radius 4 nm.
beyond that, FF decreases rapidly. Isc should decrease with
increase in dox [27], but it increases with dox when tunnel C. Effect of Pinholes in the Oxide Layer
oxide is thin, i.e., dox < 1.2 nm. This is because of higher
optical generations on rear side due to the reflection of longer In Section III-A, it is seen that for oxide thickness 1.4 nm
wavelength lights caused by thicker oxide up to a certain with no pinhole, i.e., when tunneling is the only transport
thickness, where tunnel current is not affected too much (dox < mechanism, the FF of the cell is 67.4% and series resistance
1.2 nm). For dox > 1.2 nm, tunnel current decreases rapidly, (Rs ) is 3.2 ·cm2 , which are very poor for high-efficiency
which results in lower Jsc . solar cells, but the insertion of pinholes in the oxide layer
The FF becomes below 80% for oxide thickness 1.3 nm, decreases Rs and improves the FF to above 80%, which are
whereas Steinkemper et al. [7] shown for n-TOPCon, this very good for high-efficiency TOPCon solar cells. Higher den-
oxide thickness is beyond 1.7 nm. This is because the energy sity of pinholes further decreases Rs , but Voc also decreases.
offset between the conduction bands of Si and SiO2 is In Fig. 3(a), we can see Rs decreases with increase in pinhole
lower (∼3.4) than that (∼4.4) between the valance bands of density (Npin ) and pinhole radii (Rpin ) (2, 4, and 8 nm) in the
Si and SiO2 . That means with the same tunneling masses, tunnel oxide layer having thickness of 1.4 nm. Insertion of
holes (for p-TOPCon) are facing higher barrier than elec- pinholes in the oxide layer with Rpin = 2 nm and Npin =
trons (for n-TOPCon) for a particular tunnel oxide thick- 1 E4 cm−2 decreases Rs to 2.48 ·cm2 from 3.2 ·cm2 .
ness. Folchert et al. [9] have also shown that calculated Further increase in Npin results in rapid decrease of Rs and
maximum efficiency of n-type POLO junction is for broader saturates to 0.27 ·cm2 at very high pinhole density of 1 E10
oxide thickness (0.6–2 nm) than p-type POLO junction cm−2 . Similar nature is seen for other two series resistance
(0.9–1.2 nm). curves for Rpin = 4 and 8 nm. For Npin = 1 E4 cm2 , it is
seen that Rs decreases rapidly from 2.48  to 0.42 ·cm2 as
Rpin increases from 2 to 8 nm. At very high Npin of 1 E10
B. Effect of Doping Concentration in Polysilicon on cm−2 , Rs saturates to 0.23 and 0.22 ·cm2 for Rpin = 4 and
Specific Contact Resistance and Cell Output Parameters 8 nm, respectively. Due to decrease in Rs with increase in Npin
The variation of Rs and efficiency with the doping con- and Rpin , the FF of the cell increases accordingly [39] and
centration in the p+ poly-Si (N A,poly-Si in cm−3 ) is shown saturates at 84.4%. The decrement of Rs with increment of
in Fig. 2(b). Increase in N A,poly-Si reduces Rs of the device Npin as well as Rpin is due to the improvement of conductivity
due to the increase of the conductivity of poly-Si [40] and of the SiOx poly-Si junction by direct contact of poly-Si to
the reduction of metal/poly-Si specific contact resistance [41]. silicon substrate through pinholes.
On the other hand, at very high N A,poly-Si , the Auger recom- In Fig. 3(c), we can see that for a particular Rpin , Rs
bination increases resulting in poor carrier lifetime [36], [37] decreases with decrease in dox and increases with increase
and mobility [35] of the carriers. This is why initially with in dox . At Npin = 1 E4 cm−2 for dox = 1.4 nm, Rs is 0.7 ·cm2 ,
increase in doping concentration in poly-Si, the cell efficiency and it decreases to 0.32 ·cm2 at dox = 1 nm and increases to
increases and reaches to maximum and then starts to decrease. 1.28 ·cm2 at dox = 1.6 nm. Therefore, a large difference in
The maximum efficiency of 24.65% is achieved at doping Rs of 0.96 ·cm2 is seen with increase of dox from 1 to 1.6 nm
concentration 8 × 1019 cm−3 and oxide thickness of 1 nm. for Npin = 1 E4 cm−2 . This difference in Rs decreases rapidly
This doping concentration of 8 × 1019 cm−3 in poly-Si is [see Fig. 3(c)] with increase in Npin and becomes negligible for
used for further simulations. Npin > 5 E6 cm−2 . For small Npin (dox ≤ 1.4 nm), Rs increases

Authorized licensed use limited to: Tsinghua University. Downloaded on June 23,2023 at 14:41:20 UTC from IEEE Xplore. Restrictions apply.
SADHUKHAN et al.: DETAILED STUDY ON THE ROLE OF NATURE AND DISTRIBUTION OF PINHOLES AND OXIDE LAYER 5621

rapidly with dox , because for small Npin , the transport is


mainly due to tunneling of holes, and from Section III-A, it is
known that when tunneling is the only transport mechanism,
then Rs increases with increase in dox . Now, for large Npin
(>5 E6 cm−2 ), the transport is mainly through pinholes, that
is why the effect of dox on Rs becomes negligible.
Simulation results show that although with increase in Npin ,
Rs decreases, but it also increases the recombination current
( J0 ) [see Fig. 3(b) and (d)]. This is due to higher recombi-
nation of minority carriers in pinholes, because with increase Fig. 4. Variation of hole current ratio with pinhole density (Npin ) for
in Npin , total volume of pinholes increases and pinholes have (a) different pinhole radii (Rpin in nanometer) and dox = 1.4 nm and
higher defects than oxide region (see Table I). Now, for Npin > (b) different oxide thicknesses (dox in nanometer) and Rpin = 4 nm.
5 E6 cm−2 , J0 increases rapidly with Rpin and becomes
190.8 fA/cm−2 at Rpin = 8 nm and Npin = 1 E10 cm-2 [see
Fig. 3(b)]. So, at Npin = 1 E10 cm−2 , the difference in J0
becomes 129.6 fA/cm−2 with increase in Rpin from 2 to 8 nm.
Also, J0 increases with dox , and at Npin = 1 E10 cm−2 and
Rpin = 4 nm, J0 becomes 99.6 fA/cm−2 for dox = 1.6 nm
[see Fig. 3(d)]. At Npin = 1 E10 cm−2 with increase in dox
from 1 and 1.6 nm, the difference in J0 becomes 17 fA/cm−2 ,
which is much smaller than that for Rpin . This is because with
increase in Rpin , the increment in total volume of pinhole is
much higher than that with increase in dox . It has been seen
that surface SRH recombination rate at Si/SiO2 interface is
3.2 E15 cm−2 and that at Si/poly-Si interface is 1.1 E17 cm−2 .
The higher surface recombination rate at Si/poly-Si interface
is due to higher interface defects density than Si/SiO2 interface
(see Table I). The surface SRH recombination can be further
reduced by using Al2 O3 as a tunnel layer. The positive fixed
charges at p-Si/SiO2 interface (+1 × 1011 cm−2 ) induce
the minority electrons at the interface causing higher surface
Fig. 5. Variation of (a) series resistance (Rs in Ω·cm2 ) and FF (in
SRH recombination. On the other hand, negative fixed charges percentage), (b) recombination current density (J0 in fA·cm−2 ), and
(−4 E12) present at p-Si/Al2 O3 interface [32], which repeals (c) efficiency (%) with pinhole density (cm−2 ) for different percentages
minority electrons from the interface. of density of completely through pinholes (Nth ) and partially through
pinholes (Nnon-th ) for a particular pinhole density (Npin ) in the oxide
In Fig. 3(b), it is seen that the efficiency peak is shifting layer, pinhole radius of 4 nm, and oxide thickness of 1.2 nm. (d) J–V
toward lower Npin side with increase in Rpin . This means characteristics of p-TOPCon solar cell for pinhole density of 5 E6 cm−2
that highest efficiency can be achieved with smaller Npin for (Nth = 60% and Nnon-th = 40%), radius of 4 nm, oxide thickness of 1.2 nm,
and bulk lifetime (τBulk ) of 1 ms.
higher Rpin . Here, highest efficiency achieved is 24.7% at
Rpin = 4 nm, Npin = 5 E6 cm−2 , and dox = 1.4 nm. On the
achieved at Npin = 5 E6 cm−2 . This means pinhole dominated
other hand, efficiency peak is shifting toward higher Npin
transport is achieved at lower Npin for thicker oxide, but for
side with increase in dox [see Fig. 3(d)]. This means higher
better performance, higher density of pinholes is required.
number of pinholes are required to achieve highest efficiency
for thicker tunnel oxide. Here, highest efficiency achieved is
24.8% at dox = 1.2 nm, Npin = 1 E6 cm−2 , and Rpin = 4 nm. D. Effect of Completely Through and Partially Through
Fig. 4(a) shows the variation of hole current ratio (tunnel- Pinholes in the Oxide Layer
ing/pinhole transport) with Npin for different Rpin values(dox Simulation has been done by considering both types of pin-
is 1.4 nm). It is seen that the current ratio becomes 0.1 at holes, and results are compared with the results of completely
Npin = 5 E5 cm−2 , 0.09 at 1 E5 cm−2 , and 0.07 at 5 E4 cm−2 . through pinholes. In case of partially through pinholes, tunnel
These values of current ratio decrease rapidly with further oxide of thickness 0.2 nm is considered between polysilicon
increase in Npin becoming more pinhole dominated transport and silicon. We have started with dox = 1.2 nm, Rpin =
and the values increases with decrease in Npin shows tunnel 4 nm, and no partially through pinholes (Nnon-th = 0) and
dominated transport. Again, with increase in pinhole radius, then observed the variation of Rs with Npin with increase
pinhole dominated transport is achieved at lower Npin . It is in Nnon-th in Npin [see Fig. 5(a)]. Now, it is seen that with
also seen that with thinner oxide, pinhole dominated trans- insertion of Nnon-th along with completely through pinholes
port is achieved with higher Npin [see Fig. 4(b)]. Even for (Nth ) in the oxide layer, the Rs curve is shifting upward. That
dox = 1.4 nm, carrier transport becomes pinhole dominated means for a particular Npin , the Rs value becomes higher as
at Npin = 1 E5 cm−2 . For dox = 1.4 nm, the FF is 83.4% the percentage of partially through pinholes increases in the
at Npin = 1 E5 cm−2 , whereas for dox = 1 nm, this FF is oxide layer. Results show that at Npin = 5 E4 cm−2 , Rs is

Authorized licensed use limited to: Tsinghua University. Downloaded on June 23,2023 at 14:41:20 UTC from IEEE Xplore. Restrictions apply.
5622 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 69, NO. 10, OCTOBER 2022

TABLE II
C OMPARISON OF O UTPUT PARAMETERS B ETWEEN P -PERC AND
P -TOPC ON S OLAR C ELLS FOR τBULK = 1 MS

0.38 ·cm2 for Nnon-th = 0 and Nth = 100%, 0.42 ·cm2 for
Nnon-th = 40% and Nth = 60%, and 0.51 ·cm2 for Nnon-th =
70% and Nth = 30%. That means the difference in Rs is
0.13 ·cm2 between no partially through pinhole and 70%
partially through pinholes. This difference in Rs decreases with
increasing pinhole density in the oxide layer. The increase in
Rs is due to the fact that for a particular pinhole density when
a percentage of completely through pinholes is replaced by
partially through pinholes, this reduces number of direct paths Fig. 6. Hole current density near pinhole region. (a) 3-D view.
for hole transport to tunneling of holes, which provides higher (d) Cross-sectional view for completely through pinhole. (c) 3-D view.
series resistance. Therefore, partially through pinholes in the (e) Cross-sectional view for partially through pinhole with pinhole density
of 5 E6 cm−2 (Nth = 60% and Nnon-th = 40%), pinhole radius of 4 nm,
oxide layer show disadvantage in case of series resistance of and oxide thickness of 1.2 nm. (b) Hole in the tunnel oxide at pinhole
the cell. region for completely through pinhole.
In Fig. 5(b), it is seen that J0 curve is shifting downward
as Nnon-th increases in Npin . That means for a particular Npin , (Nth = 60% and Nnon-th = 40%), Rpin = 4 nm, dox = 1.6 nm,
J0 decreases with increase in density of partially through and τBulk = 5 ms, the efficiency achieved is 26% with J0 =
pinholes (Nnon-th ). At Npin = 1 E7 cm−2 , J0 is 47.7 fA·cm−2 28.6 fA·cm−2 . Morisset et al. [42] achieved an implied Voc
for Nnon-th = 0 and Nth = 100%, 44.8 fA·cm−2 for Nnon-th = of 714 mV for oxide thickness of 1.3–1.4 nm and boron
40% and Nth = 60%, and 42.4 fA·cm−2 for Nnon-th = 70% and dopped poly-Si layer of thickness 15 nm deposited to avoid
Nth = 30%. This difference in J0 increases with increase of blistering effect. Mack in 2021 achieved an efficiency of 21.2%
Npin and decreases with decrease in Npin and becomes negli- with Voc of 671 mV, Jsc of 39.6 mA/cm2 , FF of 80%, and
gible for Npin < 5 E6 cm−2 . Increasing Nnon-th in Npin reduces Rs of 0.6 ·cm2 for oxide thickness of 1.2–1.4 nm. They
the total poly-Si/Si contact area, and as surface recombination also showed poly-Si/metal contact resistivity of 4–5 m·cm2
rate at SiO2 /Si interface is lower than that at poly-Si/Si and metal recombination current density of 63 fA/cm2 for
interface, it results in less recombination of minority electrons, poly-Si thickness of 240 nm and doping concentration of
and hence, J0 decreases. Therefore, increasing Nnon-th in Npin 6.3 E19 cm−3 [29].
shows advantage in case of J0 for p-TOPCon solar cell. Fig. 6 shows that highest hole current density for completely
The peak of efficiency curve shifts toward higher Npin side through pinhole is 8.5 E6 A·cm−2 and that for partially through
as percentage of Nnon-th increases in Npin for a particular dox pinhole is 2.4 E6 A·cm−2 . The highest efficiency is achieved
and Rpin [see Fig. 5(c)]. The highest efficiency achieved is when both types of pinholes are considered because partially
24.9% for Npin = 5 E6 cm−2 (Nth = 60% and Nnon-th = 40%), through pinholes are not affecting the hole current too much,
Rpin = 4 nm, dox = 1.2 nm, N A,poly-Si = 8 × 1019 cm−3 , and but it is reducing J0 even if Rs is increased a little bit.
τBulk = 1 ms with Voc = 713.7 mV, Jsc = 41.2 mA·cm−2 ,
FF = 84.6%, Rs = 0.42 ·cm2 , and overall recombination IV. C ONCLUSION
current density (J0 ) = 42.5 fA·cm−2 . Fig. 5(d) shows the The p-type TOPCon structure has been analyzed with SiO2
corresponding J –V curve. as the tunneling layer for majority carriers (holes). For oxide
Also, simulated p-type PERC solar cell considers thickness <1.4 nm, hole transport takes place by tunneling,
Al2 O3 /Si3 N4 as rear passivating dielectric stack and partial rear and to achieve FF >80%, oxide thickness should be ≤1.2 nm.
contact, and the results are compared with p-type TOPCon High boron doping concentration in poly-Si reduces Rs but
solar cell (see Table II). For p-PERC, the highest efficiency increases recombination. Therefore, to achieve best perfor-
achieved is 22.8% at 8% rear contact, which is 2.1% lower mance, boron doping concentration should be optimized. Only
than simulated p-TOPCon solar cell. From Table II, we can see tunneling of carriers results in high series resistance, whereas
that for p-TOPCon, Voc has improved and total recombination insertion of pinholes in the oxide layer reduces the series
current density on rear side (J0,rear ) has reduced significantly resistance and improves the FF, but high density of pinholes
than p-PERC. Effective surface recombination velocity on results in very high recombination current, which reduces the
rear side (Seff,rear ) is also reduced significantly in simulated open-circuit voltage. For higher diameter pinholes, the best
p-TOPCon solar cell than p-PERC solar cell. performance of the cell can be achieved at lower density of
This efficiency of p-TOPCon is further improved to 25.3% pinholes. On the other hand, for thicker oxide with fixed
for τBulk = 5 ms. In case of n-TOPCon for Npin = 5 E6 cm−2 pinhole diameter, the best performance can be achieved at

Authorized licensed use limited to: Tsinghua University. Downloaded on June 23,2023 at 14:41:20 UTC from IEEE Xplore. Restrictions apply.
SADHUKHAN et al.: DETAILED STUDY ON THE ROLE OF NATURE AND DISTRIBUTION OF PINHOLES AND OXIDE LAYER 5623

higher density of pinholes. For large Npin (>5 E6 cm−2 ), [18] F. Feldmann et al., “A study on the charge carrier transport of passivating
the effect of oxide thickness on device series resistance contacts,” IEEE J. Photovolt., vol. 8, no. 6, pp. 1503–1509, Nov. 2018.
[19] A. S. Kale et al., “Understanding the charge transport mechanisms
becomes negligible. The efficiency of p-TOPCon can be through ultrathin SiOx layers in passivated contacts for high-efficiency
further improved by considering a tunnel layer having negative silicon solar cells,” Appl. Phys. Lett., vol. 114, no. 8, Feb. 2019,
fixed charge and tunneling masses equal to or less than that of Art. no. 083902.
[20] Y. C. Wu and Y. R. Jhan, “Introduction of synopsys sentaurus TCAD
SiO2 layer. Overall, Voc and efficiency of simulated p-TOPCon simulation,” in Proc. 3D TCAD Simulation CMOS Nanoeletronic
solar cell have been improved significantly over simulated Devices. Singapore: Springer, 2018, pp. 1–17.
p-PERC, which shows better rear-side passivation by tunneling [21] Sentaurus TCAD User Manual, Synopsys, Mountain View, CA, USA,
2020, pp. 645–677.
in p-TOPCon than partial rear contact in p-PERC solar cell. [22] Renewable Resource Data Center. (2004). National Renewable
By this simulation, we have tried to establish optimization of Energy Laboratory. [Online]. Available: http://rredc.nrel.gov/solar/
partially through and completely through pinholes in the oxide spectra/am1.5/
[23] PV Light House Refractive Index Library. Accessed: Dec. 7, 2020.
layer, which improves the efficiency of TOPCon cell and also [Online]. Available: https://www.pvlighthouse.com.au/refractive-index-
allows use of thicker tunneling oxide layer leading to better library
robustness of fabrication of TOPCon solar cells. [24] Refractive Index Info. Accessed: Dec. 8, 2020. [Online]. Available:
https://refractiveindex.info/
[25] M. Z. Rahman, “Modeling minority carrier’s recombination lifetime of
R EFERENCES p-Si solar cell,” Int. J. Renew. Energy Res., vol. 2, no. 1, pp. 117–122,
[1] M. A. Green et al., “Solar cell efficiency tables (version 58),” Prog. 2012.
Photovolt., Res. Appl., vol. 29, no. 7, pp. 657–667, Jun. 2021. [26] R. Tsu and L. Esaki, “Tunneling in a finite superlattice,” Appl. Phys.
[2] (2021). International Technology Roadmap for Photovoltaic. [Online]. Lett., vol. 22, no. 11, pp. 562–564, Jun. 1973.
Available: https://itrpv.vdma.org/ [27] D. J. Griffiths, Introduction to Quantum Mechanics. London, U.K.:
[3] S. W. Glunz et al., “Silicon-based passivating contacts: The TOPCon Pearson, 2005.
route,” Prog. Photovoltaics: Res. Appl., pp. 1–19, Dec. 2021. [28] K. C. Fong, K. R. McIntosh, and A. W. Blakers, “Accurate series
[4] D. Yan et al., “Silicon solar cells with passivating contacts: Classification resistance measurement of solar cells,” Prog. Photovolt., Res. Appl.,
and performance,” Prog. Photovolt., Res. Appl., pp. 1–17, May 2022. vol. 21, no. 4, pp. 490–499, Jun. 2013.
[5] A. Richter et al., “Design rules for high-efficiency both-sides-contacted [29] S. Mack, D. Herrmann, M. Lenes, M. Renes, and A. Wolf, “Progress
silicon solar cells with balanced charge carrier transport and recombi- in p-type tunnel oxide-passivated contact solar cells with screen-printed
nation losses,” Nature Energy, vol. 6, no. 4, pp. 429–438, Apr. 2021. contacts,” Sol. RRL, vol. 5, Mar. 2021, Art. no. 2100152.
[6] B. Grübel et al., “Progress of plated metallization for industrial bifacial [30] Z. Liu et al., “Numerical and experimental exploration towards a 26%
TOPCon silicon solar cells,” Prog. Photovoltaics: Res. Appl., vol. 30, efficiency rear-junction n-type silicon solar cell with front local-area
no. 6, pp. 615–621, Jun. 2022. and rear full-area polysilicon passivated contacts,” Sol. Energy, vol. 221,
[7] H. Steinkemper, F. Feldmann, M. Bivour, and M. Hermle, “Numerical pp. 1–9, Jun. 2021.
simulation of carrier-selective electron contacts featuring tunnel oxides,” [31] A. Schenk and G. Heiser, “Modeling and simulation of tunneling
IEEE J. Photovolt., vol. 5, no. 5, pp. 1348–1356, Sep. 2015. through ultra-thin gate dielectrics,” J. Appl. Phys., vol. 81, no. 12,
[8] J. Melskens, B. W. H. van de Loo, B. Macco, L. E. Black, S. Smit, pp. 7900–7908, Jun. 1997.
and W. M. M. Kessels, “Passivating contacts for crystalline silicon solar [32] V. Sharma, Study of Charges Present in Silicon Nitride Thin Films
cells: From concepts and materials to prospects,” IEEE J. Photovolt., and Their Effect on Silicon Solar Cell Efficiencies. Tempe, AZ, USA:
vol. 8, no. 2, pp. 373–388, Mar. 2018. Arizona State Univ., 2013.
[9] N. Folchert, R. Peibst, and R. Brendel, “Modeling recombination and [33] W. D. Eades and M. R. Swanson, “Calculation of surface generation
contact resistance of poly-Si junctions,” Prog. Photovolt., Res. Appl., and recombination velocities at the Si-SiO2 interface,” J. Appl. Phys.,
vol. 28, no. 12, pp. 1289–1307, Dec. 2020, doi: 10.1002/pip.3327. vol. 58, no. 11, pp. 4267–4276, 1985.
[10] N. Balaji et al., “Pathways for efficiency improvements of industrial [34] J. Y. W. Seto, “The electrical properties of polycrystalline silicon films,”
PERC silicon solar cells,” Sol. Energy, vol. 214, pp. 101–109, Jan. 2021. J. Appl. Phys., vol. 46, no. 12, pp. 5247–5254, Dec. 1975.
[11] A. Harter et al., “Influence of intrinsic silicon layer and intermediate [35] D. B. M. Klaassen, “A unified mobility model for device simulation—I.
silicon oxide layer on the performance of inline PECVD deposited Model equations and concentration dependence,” Solid-State Electron.,
boron-doped TOPCon,” IEEE J. Photovolt., vol. 11, no. 4, pp. 936–943, vol. 35, no. 7, pp. 953–959, Jul. 1992.
Jul. 2021. [36] D. B. M. Klaassen, “A unified mobility model for device simulation—II.
[12] F. Feldmann, M. Bivour, C. Reichel, H. Steinkemper, M. Hermle, and Temperature dependence of carrier mobility and lifetime,” Solid-State
S. W. Glunz, “Tunnel oxide passivated contacts as an alternative to Electron., vol. 35, no. 7, pp. 961–967, Jul. 1992.
partial rear contacts,” Sol. Energy Mater. Sol. Cells, vol. 131, pp. 46–50, [37] P. P. Altermatt, “Models for numerical device simulations of crystalline
Dec. 2014. silicon solar cells—A review,” J. Comput. Electron., vol. 10, no. 3,
[13] J.-I. Polzin et al., “Temperature-induced stoichiometric changes in pp. 314–330, Sep. 2011.
thermally grown interfacial oxide in tunnel-oxide passivating contacts,” [38] Renewable Resource Data Center. (2004). National Renewable
Sol. Energy Mater. Sol. Cells, vol. 218, Dec. 2020, Art. no. 110713. Energy Laboratory. [Online]. Available: http://rredc.nrel.gov/solar/
[14] N. C. Mandal et al., “Study of the properties of SiOx layers prepared spectra/am1.5/
by different techniques for rear side passivation in TOPCon solar cells,” [39] S. M. Sze and K. K. Ng, Physics of Semiconductor Devices, 3rd ed.
Mater. Sci. Semicond. Process., vol. 119, Nov. 2020, Art. no. 105163. Hoboken, NJ, USA: Wiley, 2007.
[15] T. N. Truong et al., “Morphology, microstructure, and doping behaviour: [40] M. S. Raman, T. Kifle, E. Bhattacharya, and K. N. Bhat, “Physical
A comparison between different deposition methods for poly-Si/SiOx model for the resistivity and temperature coefficient of resistivity in
passivating contacts,” Prog. Photovolt., Res. Appl., vol. 29, no. 7, heavily doped polysilicon,” IEEE Trans. Electron Devices, vol. 53, no. 8,
pp. 857–868, Mar. 2021. pp. 1885–1892, Aug. 2006.
[16] M. Lozac’h, S. Nunomura, and K. Matsubara, “Double-sided TOPCon [41] A. Y. C. Yu, “Electron tunneling and contact resistance of metal-silicon
solar cells on textured wafer with ALD SiOx layer,” Sol. Energy Mater. contact barriers,” Solid-State Electron., vol. 13, no. 2, pp. 239–247,
Sol. Cells, vol. 207, Apr. 2020, Art. no. 110357. Feb. 1970.
[17] R. Peibst et al., “Working principle of carrier selective poly-Si/c-Si [42] A. Morisset et al., “Conductivity and surface passivation properties of
junctions: Is tunnelling the whole story?” Sol. Energy Mater. Sol. Cells, boron-doped poly-silicon passivated contacts for c-Si solar cells,” Phys.
vol. 158, pp. 60–67, Dec. 2016. Status Solidi A, vol. 216, no. 10, May 2019, Art. no. 1800603.

Authorized licensed use limited to: Tsinghua University. Downloaded on June 23,2023 at 14:41:20 UTC from IEEE Xplore. Restrictions apply.

You might also like