Available online at www.sciencedirect.

com

Energy Procedia 33 (2013) 104 – 109

PV Asia Pacific Conference 2012

Numerical Modelling of Silicon p+ Emitters Passivated

by a PECVD AlOx/SiNx Stack
Fa-Jun Maa,b,*, Shubham Duttaguptaa,b, Marius Petersa,
Ganesh S. Samudraa,b, Armin G. Aberlea,b, Bram Hoexa
a
Solar Energy Research Institute of Singapore, National University of Singapore, 7 Engineering Drive 1, Block E3A,
Singapore 117574, Singapore
b
Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, Block E4,
Singapore 117576, Singapore

Abstract

Extraction of the exact surface recombination velocity at highly doped crystalline silicon (c-Si) surfaces is not
straightforward and typically involves advanced computer modelling. In this work, two theoretical methods (the
Kane-Swanson slope method and the general definition) for the extraction of the emitter saturation current density J0e
are compared in SENTAURUS TCAD. We find good agreement between the J0e values obtained by the two methods.
Experimental p+ emitter doping profiles on planar {100} samples are used to calibrate the process simulation,
followed by a calculation of the doping profiles for textured {111} samples featuring upright pyramids. The
experimentally measured J0e values of both textured and planar samples passivated by PECVD AlOx/SiNx stacks are
reproduced by adjusting the surface recombination velocity. The electron surface recombination velocity parameter
Sn0 at the p+ c-Si/AlOx interface is determined to be ~1 104 cm/s for all investigated p+ emitters on planar wafers,
comparable to values reported for lightly doped c-Si. A high density of fixed charge is found to reduce J0e by up to
90%, due to the suppression of Shockley-Read-Hall recombination at the surface.

© 2013
© 2013ThePublished by Elsevier
Authors. Published Ltd. Ltd.
by Elsevier Selection and/or
Open access peer-review
under CC BY-NC-ND under responsibility of Solar Energy
license.
Researchand
Selection Institute of Singapore
peer-review (SERIS) -ofNational
under responsibility University
Solar Energy ofInstitute
Research Singapore (NUS). The
of Singapore PV Asia
(SERIS) Pacific
– National
Conference
University 2012 was(NUS).
of Singapore jointly
Theorganised by cSERIS
PV Asia Pacifi and 2012
Conference the was
Asian Photovoltaic
jointly organised byIndustry Association
SERIS and the Asian
Photovoltaic
(APVIA). Industry Association (APVIA)

Keywords: Simulation; passivation; crystalline silicon; emitter saturation current density; AlOx/SiNx stacks

* Corresponding author. Tel.: +65 660 11032; fax: +65 6775 1943
E-mail address: Fajun.Ma@nus.edu.sg

1876-6102 © 2013 The Authors. Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of Solar Energy Research Institute of Singapore (SERIS) – National University of Singapore
(NUS). The PV Asia Pacific Conference 2012 was jointly organised by SERIS and the Asian Photovoltaic Industry Association (APVIA)
doi:10.1016/j.egypro.2013.05.046
 Fa-Jun Ma et al. / Energy Procedia 33 (2013) 104 – 109 105

1. Introduction

crystalline silicon (c-Si) solar cells are fabricated from p-type c-Si material.
Several advantages, such as higher lifetime and better impurity tolerance, are reported for n-type c-Si
material, which could potentially lead to higher efficiency solar cells at a lower cost [1]. However, one of
the main challenges of n-type c-Si solar cells is the effective passivation of p+ emitters by an industrially
feasible process. Dielectric films such as thermal SiO2 and a-SiNx:H are well suited for n+ emitters but
provide a less effective passivation on p+ emitters. This difference in passivation effectiveness is at least
partly attributed to an intrinsic positive fixed charge density in these dielectric films on c-Si [2].

In the recent years excellent passivation of p+ emitters was achieved by using negatively-charged
Al2O3 thin films grown by atomic layer deposition [3]. An Al2O3 thin film capped by PECVD SiNx was
found to provide better thermal stability during firing [4]. Recently, Duttagupta et al. have demonstrated
an excellent passivation of p+ emitters (with a sheet resistance in the range of 30 to 175 coated with
a PECVD AlOx/SiNx stack [5]. These stacks were deposited in an industrial PECVD reactor and
subsequently annealed in a standard industrial fast firing furnace. The demonstrated process is, therefore,
suitable for mass production.

In order to extract more information about the recombination at the highly doped c-Si surfaces, the
experimentally determined emitter saturation current densities J0e were analysed using advanced
computer modelling. First we compare two theoretical extraction methods for J0e in SENTAURUS TCAD
[6]. We then model symmetrical p+/n/p+ structures passivated on both sides by the PECVD AlOx/SiNx
stack, for both textured and planar surfaces, to reproduce the experimentally determined J0e of Ref. [5].

2. Simulation setup

State-of-the-art models such as Fermi-Dirac statistics, bandgap narrowing [7], carrier mobility [8], and
effective intrinsic carrier density [9] were used in our simulations. As Auger recombination has a large
contribution to J0e, significant differences were found for J0e extracted with different Auger
parameterisation sets [10]. We have implemented the Auger parameterisation by Kerr and Cuevas [11] in
the simulator to get an accurate estimation of Auger recombination for different injection levels.

Experimentally, J0e is typically extracted by [12]. In simulations,

however, the J0e value is typically calculated according to its general definition [13]. From the general
definition, it is not so difficult to compute the normalised electron recombination current at the edge of
the space charge region for planar surfaces. However, it is rather difficult to determine J0e for textured
surfaces, as integration is required. In the present study the J0e values are thus extracted in the following
two-step procedure for both textured and planar surfaces:

a. By ramping up the illumination intensity of a long-wavelength light source (1150 nm), the
injection level ( ) dependent effective carrier lifetimes eff are simulated, which in principle is
the same as the quasi-steady-state photoconductance (QSSPC) technique [14].
b. The J0e value is determined from the best linear fit of the inverse Auger-corrected eff vs. curve
at = 1×1016 cm-3 ± 30 %.

Table 1 summarises the simulation parameters used in our simulations.

106 Fa-Jun Ma et al. / Energy Procedia 33 (2013) 104 – 109

Table 1. Simulation parameters.

Parameters [unit] Value Parameters [unit] Value

n-type substrate doping [cm-3] 5.68×1014 Fixed charge density Qf in the -3×1012
Substrate thickness [μm] 150 AlOx/SiNx stack [q/cm2]
Substrate lifetime ( n0 = p0) [ms] 3 Illumination wavelength [nm] 1150

3. Results and discussion

To estimate their relative difference, J0e values were extracted with both methods from p+/n/p+ and
n /p/n+ planar wafers. The p+ and n+ emitter profiles were assumed to have a Gaussian distribution with a
+

peak position right at the surface and a constant junction depth of 1.0 μm. The surface doping
concentration varied from 2×1018 to 1×1021 cm-3. Additionally, surface recombination and fixed charges
were assumed to be absent, to ensure that only Auger recombination contributes to J0e even with a low
surface doping level. The simulated relative J0e difference between the two methods is within 7 % for n+
emitters and within 2 % for p+ emitters, as shown in Fig. 1. J0e values extracted via the general definition
are slightly larger for highly doped n+ emitters, which may
be attributed to the shift of the depletion region. J0e values extracted from slightly lower injection levels
(i.e. 5×1015 cm-3) show a difference of less than 1 %. Therefore, the two methods are comparable in terms
of J0e extraction.

10 20
10
Planar + 30 Ohm/sq
n emitter
Relative J0e difference [%]

8 No fixed charge + 55 Ohm/sq

p emitter 80 Ohm/sq
No surface recombination 19 30 Ohm/sq
10
6 55 Ohm/sq
p doping [cm ]
-3

80 Ohm/sq

4 18
10

2
17
10
+

-2 18 19 20 21
16
10
10 10 10 10 0.0 0.5 1.0 1.5 2.0
-3
Surface doping [cm ] Depth [ m]

Fig. 1. Simulated relative J0e difference between two extraction Fig. 2. Simulated p+ emitter doping profiles (lines) agree well
methods, the general definition [13] and the Kane & Swanson with experimental p+ emitter doping profiles (symbols) on
method [12], from p+/n/p+ and n+/p/n+ planar samples. planar surfaces.

From secondary ion mass spectrometry (SIMS) or electrochemical capacitance-voltage (ECV)

measurements, the p+ emitter doping profiles in planar silicon samples were determined. These
experimental profiles were used to calibrate the process simulation in SENTAURUS TCAD. High
temperature steps (> 800 °C) including temperature ramp-down steps were modelled with one-
dimensional process simulation. The thickness of the boron silicate glass, doping level of the boron
silicate glass, the Arrhenius pre-factor and the activation energy of the boron segregation coefficient at
the silicon/oxide interface were tuned to fit the experimental p+ emitter doping profiles. As shown in Fig.
2, simulated profiles show good agreements with experimental p+ emitter doping profiles for planar
samples. The Arrhenius pre-factor and activation energy in this study were determined to be around
 Fa-Jun Ma et al. / Energy Procedia 33 (2013) 104 – 109 107

10000 and 1.45 eV, respectively. With the four determined parameters, two-dimensional process
simulations were carried out to get the p+ emitter doping profiles on textured {111} surfaces using the
same thermal steps as for planar {100} surfaces.

The fixed charge density Qf in the PECVD AlOx/SiNx stack was experimentally determined to be
about -3×1012 cm-2 by contactless capacitance-voltage (C-V) measurements on undiffused samples [15]. ,
which is consistent with results reported in the literature for such stacks [16]. Negative fixed charges
attract holes and repel electrons, which has a significant impact on Shockley-Read-Hall (SRH)
recombination at the surface. They only alter the carrier concentrations in a narrow surface region and
thus they are expected to have little impact on the total Auger recombination losses in the emitters. By
setting the electron surface recombination velocity parameter Sn0 to 0 cm/s, only Auger recombination
contributes to J0e. From Fig. 3 it can be seen that no appreciable difference can be observed in the J0e
value determined for samples with fixed surface charges or without fixed surface charges. However, the
presence of fixed charges will likely cause significant difference when Sn0 is non-zero, e.g. 1×103 cm/s.
Among the four investigated p+ emitters, the impact of the negative fixed charge on the J0e seems to be
most significant for the emitter. This is not unexpected, as this emitter has the lowest surface
concentration and is expected to be most sensitive to fixed surface charges.

The effect of the minority charge carrier (in this case electron) capture cross-section and interface trap
distribution are lumped together into the parameter Sn0, whereby a lower Sn0 indicates a better interface
quality. By matching experimental and simulated J0e values, Sn0 can be determined for each emitter. With
the choice of different fixed charge density, the extracted Sn0 from planar wafers, however, varies
significantly as shown in Fig. 4. Assuming a lower fixed charge density results in a lower Sn0 and thus
better interface quality. For the measured fixed charge density (-3×1012 cm-2), Sn0 was determined to be
~1×104 cm/s for all p+ emitters. Using the electron capture cross-section [17] and the interface trap
distribution [18] reported in the literature for the c-Si/Al2O3 interface, Sn0 was also determined to be
~1×104 cm/s for lightly doped planar c-Si surfaces [19]. This result indicates the p+ c-Si/AlOx interface in
this study is as good as that on lightly doped c-Si, possibly due to relatively low surface doping. There are
also reports on much lower Sn0 (~1×102 cm/s) for p+ emitters passivated by PECVD AlOx [20]. However,
those reports probably have not considered the impact of fixed charge as shown in Fig. 4.
2
5
J0e 62.9 46.9 27.6 18.7 [fA/cm ]
200 10
2
Planar
2
Sn0 = 0 cm/s, Qf = 0 q/cm Qf [q/cm ]
12 2 12
150 Sn0 = 0 cm/s, Qf = -3x10 q/cm -4x10
12
3
Sn0 = 1x10 cm/s, Qf = 0 q/cm
2 4
10 -3x10
J0e [fA/cm ]

12
2

Sn0 [cm/s]

3 12 2
Sn0 = 1x10 cm/s, Qf = -3x10 q/cm -2x10
100
12
3
-1x10
10
50
No Qf
0 2
10
0 50 100 150 200 0 50 100 150 200
Sheet resistance [ /sq] Sheet resistance [ /sq]
Fig. 3. Simulated J0e values as a function of the emitter sheet Fig. 4. Sn0 values as a function of the sheet resistance of the p+
resistance for various values of Sn0 and Qf. emitter, for various values of Qf. The experimental J0e values are
shown at the top of the graph [5].
108 Fa-Jun Ma et al. / Energy Procedia 33 (2013) 104 – 109

4. Conclusion

In this work, using modelling with SENTAURUS TCAD, two widely used J0e extraction methods
( slope method and the general definition) were found to give comparable results.
The observed discrepancies for heavily doped n+ emitters were found to be related to numerical errors in
SENTAURUS TCAD. p+ emitters passivated by a PECVD AlOx/SiNx stack on pyramid-textured and
planar silicon surfaces were modelled to reproduce the measured J0e. p+ emitter doping profiles on planar
surfaces were used to calibrate the process simulation in order to obtain doping profiles on textured
surfaces. A high fixed charge density in the AlOx/SiNx stack was found to have little impact on the total
Auger recombination in the emitter, but significant impact on the SRH recombination rate at the surface.
J0e was reduced by up to 90% due to suppressed SRH recombination at the surface. Sn0 was determined to
be ~1×104 cm/s for all p+ emitters on planar surfaces, which is as good as that on lightly doped c-Si. The
very low Sn0 values (~1×102 cm/s) reported in the literature may well be underestimated, as the impact of
fixed charge was not properly taken into account.

Acknowledgements

SERIS is sponsored by the National University of Singapore (NUS)

Research Foundation (NRF) through the Singapore Economic Development Board (EDB). This work was
sponsored by NRF grant NRF2009EWT-CERP001-056.

References

[1] Cotter JE, Guo JH, Cousins PJ, Abbott MD, Chen FW, Fisher KC. p-type versus n-type silicon wafers: prospects for high-
efficiency commercial silicon solar cells. IEEE Transactions on Electron Devices 2006;53:1893-1901..
[2] Altermatt PP, Plagwitz H, Bock R, Schmidt J, Brendel R, Kerr MJ, Cuevas A. The surface recombination velocity at boron-
doped emitters: comparison between various passivation techniques. Proc. 21st European Photovoltaic Solar Energy
Conference, Dresden, Germany; 2006, p. 647.
[3] Hoex B, Schmidt J, Bock R, Altermatt PP, van de Sanden MCM, Kessels WMM. Excellent passivation of highly doped p-
type Si surfaces by the negative-charge-dielectric Al2O3. Appl. Phys. Lett. 2007;91:112107.
[4] Schmidt J, Veith B, Brendel R. Effective surface passivation of crystalline silicon using ultrathin Al 2O3 films and Al2O3/SiNx
stacks. Physica Status Solidi-Rapid Research Letters 2009;3:287-9.
[5] Duttagupta S, Lin F, Shetty KD, Aberle AG, Hoex B. Excellent boron emitter passivation for high-efficiency Si wafer solar
cells using AlO x/SiNx dielectric stacks deposited in an industrial inline plasma reactor. Progress in Photovoltaics: Research
and Applications, doi: 10.1002/pip.1259, 2012.
[6] SENTAURUS Manual, F-2011.09 ed. Synopsys Inc., Mountain View, CA, 2011.
[7] Schenk A. Finite-temperature full random-phase approximation model of band gap narrowing for silicon device simulation.
Journal of Appl.Phys. 1998;84:3684-95.
[8] Klaassen DBM. A unified mobility model for device simulation I. Model-equations and concentration-dependence. Solid-
State Electronics 1992;35:953-9.
[9] Altermatt PP, Schenk A, Geelhaar F, Heiser G. Reassessment of the intrinsic carrier density in crystalline silicon in view of
band-gap narrowing. Journal of Appl. Phys. 2003;93:1598-1604.
[10] Mäckel M, Varner K. On the determination of the emitter saturation current density from lifetime measurements of silicon
devices. Progress in Photovoltaics: Research and Applications, doi: 10.1002/pip.2167, 2012.
[11] Kerr MJ, Cuevas A. Recombination at the interface between silicon and stoichiometric plasma silicon nitride. Semiconductor
Science and Technology 2002;17:166-72.
[12] Kane DE, Swanson RM. Measurement of the emitter saturation current by a contactless photoconductivity decay method.
Proc. 18th IEEE Photovoltaic Specialists Conference, Las Vegas, USA; 1985, pp. 578-583.
[13] Schumacher JO, Altermatt PP, Heiser G, Aberle AG. Application of an improved band-gap narrowing model to the numerical
simulation of recombination properties of phosphorus-doped silicon emitters.Solar Energy Materials and Solar Cells 2001;
65:95-103.
 Fa-Jun Ma et al. / Energy Procedia 33 (2013) 104 – 109 109

[14] Sinton RA, Cuevas A. Contactless determination of current-voltage characteristics and minority-carrier lifetimes in
semiconductors from quasi-steady-state photoconductance data. Appl. Phys. Lett. 1996;69:2510-12.
[15] Schroder DK. Contactless surface charge semiconductor characterization. Materials Science and Engineering B 2002;91-
92:196-210.
[16] Laades A, Sperlich HP, Sturzebecher U, Angermann H, Tofflinger JA, John W, Blech M, Bahr M, Lawerenz A. Interface
issues of all-PECVD synthesized AlO x/SiNx passivation stacks for silicon solar cells. Proc. 27th European Photovoltaic Solar
Energy Conference, Frankfurt, Germany; 2012, pp. 888-895.
[17] Werner F, A. Cosceev A, Schmidt J. Interface recombination parameters of atomic-layer-deposited Al2O3 on crystalline
silicon. Journal of Appl. Phys. 2012;111:073710.
[18] Werner F, Veith B, Zielke D, Kuhnemund L, Tegenkamp C, Seibt M, Brendel R, Schmidt J. Electronic and chemical
properties of the c-Si/Al2O3 interface. Journal of Appl. Phys. 2011;109:113701.
[19] Ma FJ, Samudra GG, Peters M, Aberle AG, Werner F, Schmidt J, Hoex B. Advanced modeling of the effective minority
carrier lifetime of passivated crystalline silicon wafers. Journal of Appl. Phys. 2012;112:054508.
[20] Saint-Cast P, Richter A, Billot E, Hofmann M, Benick J, Rentsch J, Preu R, Glunz SW. Very low surface recombination
velocity of boron doped emitter passivated with plasma-enhanced chemical-vapor-deposited AlOx layers. Thin Solid Films
2012;522:336-9.