1876-6102 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer review by the scientific conference committee of SiliconPV 2015 under responsibility of PSE AG
doi: 10.1016/j.egypro.2015.07.103
Energy Procedia 77 ( 2015 ) 725 – 732
ScienceDirect
5th International Conference on Silicon Photovoltaics, SiliconPV 2015
The influence of ITO dopant density on J-V characteristics of
silicon heterojunction solar cells: experiments and simulations
Simon Kirnera, Manuel Hartigb, Luana Mazzarellaa, Lars Kortec, Tim Frijntsa, Harald
Scherg-Kurmesb, Sven Ringa, Bernd Stannowskia, Bernd Rechc, Rutger Schlatmanna
aPVcomB, Helmholtz-Zentrum Berlin für Materialien und Energie, Schwarzschildstr. 3, 12489 Berlin, Germany
b Technische Universität Berlin, Fakultät IV, HFT 5-2, Einsteinufer 25, 10587 Berlin, Germany
c Institut für Silizium Photovoltaik, Helmholtz-Zentrum Berlin für Materialien und Energie
Kekuléstraße 5, 12489 Berlin, Germany
Abstract
The TCO/a-Si:H(p) contact is a critical part of the silicon heterojunction solar cell. At this point, holes from the emitter have to
recombine loss free with electrons from the TCO. Since tunneling is believed to be the dominant transport mechanism, a high
dopant density in both adjacent layers is critical. In contrast to this, it has been reported that high TCO dopant density can reduce
field effect passivation induced by the a-Si:H(p) layer. Thus, in this publication, we systematically investigate the influence of a
thin (~10 nm) ITO contact layer with dopant densities ranging from Nd = 1019 - 1021 cm-3 placed between an ITO bulk layer of
70 nm with Nd= 2·1020 cm-3 and the a-Si:H(p) emitter on the J-V characteristics, with the aim to find an optimum Nd. We
accompanied our experiments by AFORS-HET simulations, considering trap-assisted tunneling and field dependent mobilities in
the a-Si:H(p) layer. As expected, two regimes are visible: For low Nd the devices are limited by inefficient tunneling, resulting in
S-shaped J-V characteristics. For high Nd a reduction of the field effect passivation becomes visible in the low injection range.
We can qualitatively reproduce these findings using device simulations.
© 2015 The Authors. Published by Elsevier Ltd.
Peer review by the scientific conference committee of SiliconPV 2015 under responsibility of PSE AG.
Keywords: Silicon hetero junction solar cells, HIT, TCO/a-Si:H(p) contact, ITO, tunneling, AFORS-HET,
1. Introduction
Silicon heterojunction (SHJ) solar cells consisting typically of a stack of transparent conducting oxide (TCO)/a-
Si:H(p/i)/c-Si(n) wafer/a-Si:H(i/n)/TCO/metal are well known for their very high conversion efficiency [1]. The key
advantage of this solar cell type is that there is no direct contact between the metal contacts and the absorber.
Instead, the full area of the wafer surfaces is passivated by high quality a-Si:H(i) passivation layers allowing very
Available online at www.sciencedirect.com
© 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer review by the scientific conference committee of SiliconPV 2015 under responsibility of PSE AG
726 Simon Kirner et al. / Energy Procedia 77 ( 2015 ) 725 – 732
80 90 100
Nd
position in device (nm)
Wf,TCO
Xa-Si:H(p)
Eg
Eact
+
-
A decrease from N
d
= 10
21
to 10
19
cm
-3
corresponds
to a change in W
f, TCO
of +0.12 eV
(a) (b) (c)
80 90 100
-8
-6
-4
-2
0
Vacuum
energy
Conduction
band
Valence
band
Fermi
energy
E
act
E
g
X
a-Si:H(p)
electron energy (eV)
position in device (nm)
W
f,TCO
+
-
aͲSi:H(p),10nm
cͲSi(n),270μm)
Aggrid
aͲSi:H(n),10nm
ITOcontactlayer,10nm
TCO/Ag,80/200nm
aͲSi:H(i),8nm
aͲSi:H(i),8nm
ITObulklayer,70nm
low surface recombination velocities. This however makes contacting of the solar cell challenging. Especially the
emitter contact, consisting of a TCO, often indium tin oxide (ITO), deposited on the a-Si:H(p) emitter usually
induces a considerable contact resistance [2]. Several studies have addressed this issue theoretically by means of
device simulation. For example, Centurioni et al. have calculated using the device simulator AMPS that starting from
a TCO work function of
I
ITO = 5.1 eV, a change by -350 meV could reduce the energy conversion efficiency,
K
, of
the device by more than 40% due to the formation of an S-shaped current-voltage-(J-V-)characteristic. These
findings were ascribed to the formation of a rectifying contact at the ITO/a-Si:H(p) interface. A further reduction to
I
ITO = 4.3 eV lead to a complete blocking behavior of the cell. Similar results were calculated by Varache et al. [3]
and Zhao et al. [4] using AFORS-HET, by Rached et al. using Amorphous Semiconductor Device Modeling
Program [5] and Bivour et al. (AFORS-HET) [6]. All of these studies predict a high
K
only for
I
TCO > 4.8 eV. Then,
I
TCO approaches the value of
I
a-Si:H(p) ~ 5.2 eV, which can be calculated within the Anderson model to
I
a-Si:H(p) =
&
a-
Si:H(p) + Eg, a-Si:H(p) – Eact, a-Si:H(p), with
&
being the electron affinity, Eg the mobility gap and Eact the activation energy
(distance between Fermi-level and valence band), cf. Fig. 1c. If the Anderson model is valid, the band offset,
'I
,
becomes zero, when
I
TCO =
I
a-Si:H(p). In this case or when the work function difference is sufficiently low, transport
across the junction via thermionic emission is efficient enough, corresponding to a loss free contact. If
'I
is high,
tunneling is required for a loss free contact [7]. For an efficient tunnel contact, it is essential for both adjacent layers
to be highly doped. Under the assumption of a constant
&
TCO, this means a low
I
TCO is desirable for an efficient
tunnel contact. This aspect is only visible, if the TCO is calculated as a semi-conductor, as opposed to a metal as it
has been treated in the above mentioned references [2–6]. In Fig.1b and c, the difference between the two
approaches can be seen: If the TCO is treated as a metal, an increase in
I
TCO leads only to a decrease in band bending
on the a-Si:H(p) side, the negative effect on the TCO side is neglected. On the other hand, if the TCO is treated as a
semi-conductor, an increase in
I
TCO leads also to a band bending on the TCO side. This aspect has not been discussed
in much detail in literature so far.
Another aspect that has to be considered in this context is the interplay between TCO doping and absorber
passivation. It has recently been shown that the deposition of a highly doped TCO can reduce the effective life time
due to a reduction of the field effect passivation at the a-Si:H/c-Si interface, which is clearly visible in injection level
dependent measurements such as photo-conductance decay- or illumination dependent open-circuit voltage (Suns-
Voc) measurements [8,9]. This reduced field effect passivation can lead to a fill factor deterioration. This again
speaks for a low Nd, TCO (high
I
TCO) as being beneficial for the device performance.
As reported in literature, the work function of ITO,
I
ITO, can be manipulated via the deposition parameters over
the dopant concentration Nd
I
a-Si:H(p): e.g. Klein et al. show data between ~4.2 – 5.3 eV. A convenient way to
manipulate Nd is via the oxygen partial pressure during the deposition [10]. Further reported methods to manipulate
I
ITO are pre- and post-deposition treatments [11,12].
Fig. 1. (a) Device structure considered in this paper with the red box indicating the studied interface; (b) the respective band diagram zoomed-in
to the TCO/a-Si:H(p) contact in case the TCO is treated as a metal or (c) treated as a semi-conductor for different dopant densities, Nd and
corresponding work functions Wf of the TCO contact layer.
Simon Kirner et al. / Energy Procedia 77 ( 2015 ) 725 – 732 727
In this work, we use ITO with Nd ranging from ~1·1019 – 1·1021 cm-3 as a 10 nm thin contact layer placed in
between the a-Si:H(p) emitter and an ITO bulk layer (70 nm, Nd = 2·1020 cm-3) of SHJ solar cells. We observed that
solar cells with contact layers with Nd < 2·1019 cm-3 show a considerable decrease in FF. On the other hand, pseudo
FFs above 81% could still be measured with Nd ~ 1·1021 cm-3. The experimental results can be qualitatively
reproduced using the device simulator AFORS-HET when certain tunneling related models are considered. This
indicates that there is a trade-off between field effect passivation and a low resistivity tunnel contact when
considering the optimal Nd.
2. Experimental methods
ITO films have been deposited by means of DC magnetron sputtering in a Leybold Optics system designed for
substrate sizes of up to 30 × 30 cm² at a heater temperature of 220°C. The contact layer was deposited at 6.7·10-6 bar
and deposition power of 0.5 kW; the bulk layer was deposited with 3.6·10-6 bar and 1.0 kW. The total Ar- and mixed
gas flow (10% O2 in Ar) was 500 sccm and 250 sccm for the contact- and bulk-layer, respectively. The O2 partial
pressure in the contact layer was varied between 0.0 – 6.0%, which resulted in films with carrier densities of Nd
ranging from 1·1019 – 1·1021 cm-3 as obtained from Hall measurements on ITO/glass samples with ITO-thicknesses
ranging from 160 – 220 nm (results shown in Fig. 2). The bulk ITO layer was deposited with a constant O2 partial
pressure of 2.5% resulting in Nd = 2·1020 cm-3. Spectrophotometry measurements (not shown) revealed the expected
increase in free carrier absorption in the near infra-red region of the spectrum with increasing Nd.
Fig. 2. ITO dopant density Nd and resistivity U as obtained from Hall measurements as a function of O2 partial pressure. Film thickness was
between 160-220 nm.
The solar cells were fabricated on c-Si(n) wafers (FZ, ~3 :cm resistivity, <111> termination, 270 μm thickness).
Shiny etched wafer pieces were used to increase sensitivity to the contact resistivity. The layer stack is shown in Fig.
1a. Plasma enhanced chemical vapor deposition has been performed using an AKT-1600 tool to deposit the a-Si:H
i/n and i/p stacks. A lowly, trimethylboron-doped a-Si:H(p) layer has been used as emitter [13]. The back contact
consists of a ZnO:Al/Ag stack. The front TCO consists of the mentioned ITO contact- (10 nm) and bulk-layer
(70 nm) stack. A Ti/Ag grid with thicknesses of 10 and 1500 nm has been deposited on top of the front TCO.
Multiple cells per wafer were defined by means of photolithography with an area of 1 cm².
The devices were characterized by light J-V measurements using a sun simulator with dual-source illumination
under standard test conditions (25°C, class AAA spectrum) and illumination dependent open-circuit voltage (Suns-
Voc) measurements. Measurements were performed through an aperture of 1 cm². The reported
K
and current
density, J, values are calculated based on the aperture area (including the grid). Transmission-line-method-
(TLM-)measurements were performed to test the influence of Nd in the contact layer on the sheet resistance, Rsheet, of
the ITO layer stack on selected samples with dedicated test structures deposited on the same wafer as the solar cells.
0.00 0.02 0.04 0.06
10
19
10
20
10
21
N
d
(cm
-3
)
O
2
partial pressure
10
-3
10
-2
10
-1
U(:cm)
728 Simon Kirner et al. / Energy Procedia 77 ( 2015 ) 725 – 732
3. AFORS-HET model and tunneling models
A general description of the used simulation tool AFORS-HET can be found in Ref. [14]. The calculated layer
stack consists of TCO/a-Si:H(p/i)/c-Si(n)/a-Si:H(i/n). For all of the layers except the TCO, AFORS-HET text book
structures were used*. The data for the TCO and the a-Si:H(p) layer are given in the appendix. The surface
recombination velocities at both contacts has been set to 1·107 cm/s to impose no additional loss mechanism.
To simulate tunneling at the TCO/a-Si:H(p) interface, two additional models have been used: the trap assisted
tunneling (TAT) model based on the theory of Hurkx et al. [15] as well as the field-dependence of the mobilities
(FDM) suggested by Willemen et al. [16,17]. TAT allows holes to tunnel into states in the band gap and thus the
effective tunneling mass, m*, has to be considered as a fraction of the electron mass, m*. FDM is based on the
finding that the mobility of a-Si:H increases exponentially under very high fields and a field constant F0 has to be
considered [18]. Both models depend heavily on the magnitude of the electric field, thus, have a very high
sensitivity on the dopant densities in the adjacent layers. For m* and Fo values from literature were taken [16,17]. In
Fig. 3, the influence of the models on the J-V characteristics is shown. As can be seen, by considering both models,
the blocking J-V characteristic of the device, induced by the n/p/n structure is avoided by making the TCO/a-Si:H(p)
junction ohmic. A weakness of our approach is the fact that we assume Maxwell-Boltzmann- as opposed to Fermi-
Dirac-statistics to calculate electron- and hole-densities in the TCO based on the density of states. This could
theoretically lead to unrealistically high numbers. However, by carefully initializing the numerical model, we can
confirm realistic electron densities close to Nd in the whole TCO (cf. Fig. 1c). Implementation of Fermi-Dirac
statistics into AFORS-HET is ongoing. We expect a small difference in quasi Fermi level position, which would lead
to minor differences when calculating an ideal Nd, but would not affect the general trend discussed in this paper.
Fig. 3. Simulated J-V characteristics with and without the tunneling models TAT and FDM applied in the a-Si:H(p) emitter.
4. Results & discussion
4.1. Experimental results
The solar cell characteristics short-circuit current density, Jsc, fill factor, FF, open-circuit voltage, Voc and power
conversion efficiency,
K
, are shown in Fig. 4 as a function of Nd in the ITO contact layer. Also given are the
differential resistances, Rsc, calculated from the inverse of the light J-V slope near short circuit conditions and Roc
calculated analogously near open circuit conditions. In the graph showing the FF, also the pseudo-FF data as
obtained from Suns-Voc measurements is shown for comparison. The following observations can be made: Jsc shows
*to be found online: http://www.helmholtz-berlin.de/forschung/oe/ee/si-pv/projekte/asicsi/afors-het/index_en.html
-0.6 -0.3 0.0 0.3 0.6 0.9
-30
0
30
60
90
X
TCO
= 4.5 eV
J (mA/cm²)
V (V)
no TAT,
no FDM
TAT m*=0.012,
no FDM
TAT m*=0.012,
F
0
= 2*10
5
V/cm
N
d
= 1*10
19
cm
-3
Simon Kirner et al. / Energy Procedia 77 ( 2015 ) 725 – 732 729
no systematic trend, some scattering is presumably related to a process error leading to increased under-evaporation
during grid deposition for some samples. Increase in parasitic absorption due to increased free carrier absorption for
very high Nd is believed to play only a minor role given the low contact layer thickness. The FF is overall on a
relatively low level. We explain the large discrepancy to the pseudo FF by the low doping of the a-Si:H(p) layer and
the resulting losses at the TCO/a-Si:H(p) contact. With respect to the influence of Nd, one can observe an increase
for contact layers up to >2·1019 cm-3. Above this value, the trend is ambiguous due to the small number of data
points but a small decrease in FF and pseudo FF is apparent. When looking at Roc, Rsc and the shape of the curve at
high forward bias shown in Fig. 5a, one can observe that the reduction in FF is due to a more pronounced S-shape of
the J-V curve. This, as we show below in more detail, can be explained by increasingly inefficient tunneling though
the increasing space charge region at the TCO/a-Si:H(p) interface (Schottky-diode as opposed to ohmic behavior of
the TCO/a-Si:H(p) junction) [19]. The small observable decrease in FF for very high Nd is accompanied by a
pronounced reduction in effective carrier lifetime for low minority carrier densities, as obtained from SunsVoc
measurements and shown in Fig. 5b. This can be explained by the reduction of field effect passivation, which
usually affects more the FF than the Voc due to the shape of the lifetime versus minority carrier density curves [8].
The latter is therefore, as expected, relatively unaffected by the change in Nd and thus
K
follows mainly the trend of
the FF.
To confirm that the decrease in FF is predominantly related to the contact resistance, the sheet resistance of the
ITO contact and bulk layer stack has been measured on the TLM structures on the samples with Nd = 1.9·1019 cm-3
(average FF = 68%) and Nd = 2.3·1019 cm-3 (average FF = 71%). It increased by only <5 :ƶ. Given the geometry of
the metal grid, this would lead to an expected increase in series resistance of < 0.02 :cmϡ, which can hardly
explain the drop in FF. This and the pronounced S-shaped J-V characteristics are strong indications that insufficient
tunneling at the TCO/ a-Si:H(p) interface is the dominating mechanism reducing the FF.
Fig. 4. Solar cell characteristics as obtained from light J-V- and Suns-Voc-measurements as a function of the dopant density, Nd in the ITO contact
layer. Lines are guides to the eye indicating the influence of Nd on the two mechanisms: Tunneling and field effect passivation.
31
32
33
34
35
65
70
75
80
0.66
0.68
0.70
0.72
0.74
14
15
16
17
10
19
10
20
10
21
0
10
20
30
10
19
10
20
10
21
4
6
8
J
sc
(mA/cm²)
sun simulator
SunsVoc (pseudo-FF)
FF (%)
V
OC
(V)
K (%)
R
SC
(k:cm²)
N
d
(cm
-3
) in ITO contact layer
R
OC
(:cm²)
730 Simon Kirner et al. / Energy Procedia 77 ( 2015 ) 725 – 732
Fig. 5. (a) Experimentally obtained light J-V characteristics under high forward bias and (b) effective carrier densities as a function of minority
carrier density for selected cells with different Nd. The color code is the same for both graphs.
4.2. Simulation results
Variations of the dopant density, Nd, in the TCO of the above described model structure have been performed and
for two different electron affinities
&
TCO
and the light J-V characteristics have been calculated. The results are
shown in Fig. 6. Simulations show qualitatively similar trends as the experiments: J-V curves of structures with low
Nd become S-shaped due to insufficient tunneling at the TCO/a-Si:H(p) interface. There is only a small reduction of
Voc visible, which could be related to a reduction in field-effect passivation for high Nd. Comparing the two
&
TCO, it
can be seen that the dependence on Nd is less pronounced for high
&
TCO. This can be explained by the fact that the
band bending is mostly affected by
&
TCO and its change affects
'I
directly, whereas the change in work function
from Nd = 1·1019 to 1·1021 cm-3 is only ~120 meV.
Fig. 6. Simulated light J-V characteristics at high forward bias for electron affinitys of (a)
&
TCO= 4.5 eV and (b)
&
TCO= 4.0 eV and varying Nd.
5. Conclusions and outlook
Solar cells with ITO contact layers with various dopant densities, Nd, have been investigated. The systematic
formation of S-shaped J-V characteristics (increased Roc) is observed for ITO contact layers with decreasing doping
densities (and presumably increased work function) for Nd < 2·1019 cm-3. On the other hand, the negative influence
of reduced field passivation on the device performance for very high charge carrier densities is relatively low. These
1014 1015 1016
0.6
0.9
1.2
1.5
Weff (ms)
minority carrier density (cm-3)
Nd (cm-3)
1.110
21
1.910
20
2.310
19
2.010
19
1.710
19
9.910
18
0.4 0.5 0.6 0.7 0.8
-30
-20
-10
0
10
20
J (mA/cm²)
V (V)
0.50.60.70.8
-40
-20
0
20
J (mA/cm²)
V (V)
N
d
(cm
-3
)
2.5*10
19
3.5*10
19
5*10
19
8*10
19
10*10
19
(a) (b)
X
TCO=4.5eV
X
TCO=4.0eV
(a) (b)
Simon Kirner et al. / Energy Procedia 77 ( 2015 ) 725 – 732 731
trends can qualitatively be reproduced using the device simulator AFORS-HET when tunneling models are
considered, which suggests that inefficient tunneling can lead to the S-shaped J-V characteristics.
Under the assumption that the work function increases with reducing Nd (an assumed constant electron affinity),
the presented data indicate that there are two competing mechanisms: (1) a good field effect passivation demands a
high work function (low Nd), whereas (2) an efficient tunnel recombination contact demands a low work function
(high Nd). This aspect should be considered in the search for an “ideal” TCO contact layer for HIT solar cells.
The best manufactured solar cell had a contact layer with Nd = 2.3·1019 cm-3. The data indicates that a slight
increase in Nd could further improve the contact resistivity. For more meaningful and predictive simulations in the
future, a calibration of the model and determination of the many unknown parameters by in-depth characterization
and analyses is essential.
Acknowledgements
The authors thank K. Bhatti, T. Hänel, T. Henschel, J. Kerstin and M. Wittig for technical support and Dr. S.
Schmidt for fruitful discussions. This work was supported by the co-financing of the Thüringer Aufbaubank and the
Europäischen Sozialfonds in the framework of the project OptiSolar, partially by the European Commission through
the FP7-ENERGY project “HERCULES” (Grant No. 608498) and by the Federal Ministry of Education and
Research (BMBF) and the state government of Berlin (SENBWF) in the framework of the program
“Spitzenforschung und Innovation in den Neuen Ländern” (grant no. 03IS2151)
Appendix A. important model input parameters
ParameterThickness
(nm)
Electron
affinity,
X(eV)
Mobility
gap(eV)
Dopantdensity,Nd
(cmͲ3),activation
energy(meV)
Defect
density
(cmͲ3)
Electron(hole)
mobility
(cm²/Vs)
F0
(V/cm)
m*/
melectron
TCO804.5,4.84variable–40(20)––
aͲSi:H(p)841.721020,2004ͼ102020(5)2ͼ1050.012
References
[1] Taguchi M, Yano A, Tohoda S, Matsuyama K, Nakamura Y, Nishiwaki T, Fujita K, Maruyama E, 24.7 % Record Efficiency HIT Solar Cell
on Thin Silicon Wafer. IEEE J Photovolt; 2014 4:96.
[2] Lee S-Y, Choi H, Li H, Ji K, Nam S, Choi J, Ahn S-W, Lee H-M, Park B. Analysis of a-Si:H/TCO contact resistance for the Si
Heterojunction Back-Contact Solar Cell. Sol En Mat and Sol Cells; 2014 120:412.
[3] Varache R, Kleider JP, Gueunier-Farret ME, Korte L. Silicon Heterojunction Solar Cells: Optimization of Emitter and Contact Properties
from Analytical Calculation and Numerical Simulation. Mat Sci and Eng; 2013 9:593.
[4] Zhao L, Zhou CL, Li HL, Diao HW, Wang WJ. Role of the Work Function of Transparent Conductive Oxide on the Performance of
Amorphous/Crystalline Silicon Heterojunction Solar Cells Studied by Computer Simulation. Phys Stat Sol A; 2008 205: 1215.
[5] Rached D, Mostefaoui R. Influence of the Front Contact Barrier Height on the Indium Tin Oxide/Hydrogenated p-Doped Amorphous Silicon
Heterojunction Solar Cells. Thin Sol Films; 2008 516:5087.
[6] Bivour M, Schröder S, Hermle M. Numerical Analysis of Electrical TCO/a-Si:H(p) Contact Properties for Silicon Heterojunction Solar Cells.
En Proc; 2013 38:658.
[7] Kanevce A, Metzger WK. The Role of Amorphous Silicon and Tunneling in Heterojunction with Intrinsic Thin Layer (HIT) Solar Cells. J
Appl Phys; 2009 105: 094507.
[8] Rößler R, Leendertz C, Korte L, Mingirulli N, Rech B. Impact of the Transparent Conductive Oxide Work Function on Injection-Dependent
a-Si:H/c-Si Band Bending and Solar Cell Parameters. J Appl Phys; 2013 113: 144513.
[9] Macco B, Deligiannis D, Smit S, van Swaaij RACMM, Zeman M, Kessels WMM. Influence of Transparent Conductive Oxides on
Passivation of a-Si:H/c-Si Heterojunctions as Studied by Atomic Layer Deposited Al-doped ZnO. Semicond Sci and Tech, 2014 29:122001.
732 Simon Kirner et al. / Energy Procedia 77 ( 2015 ) 725 – 732
[10]Klein A, Körber C, Wachau A, Säuberlich F, Gassenbauer Y, Harvey SP, Proffit DE, Mason TO. Transparent Conducting Oxides for
Photovoltaics: Manipulation of Fermi level, Work Function and Energy Band Alignment. Materials; 2010 3: 4892.
[11]Sugiyama K, Ishii H, Ouchi Y, Seki K. Dependence of ITO Work Function on Surface Cleaning Method as Studied by Ultraviolet and X-ray
Photoemission Spectroscopies. J Appl Phys; 2000 87:295.
[12]Kim JS, Lägel B, Moons E, Johansson N, Baikie ID, Salaneck WR, Friend RH, Cacialli F. Kelvin Probe and Ultraviolet Photoemission
Measurements of Indium Tin Oxide Work Function: a Comparison. Synth Metals; 2000 111: 311.
[13]Mazzarella L, Kirner S, Mews M, Conrad E, Korte L, Stannowski B, Rech B, Schlatmann R. Comparison of TMB and B2H6 as Precursors for
Emitter Doping in High Efficiency Silicon Hetero Junction Solar Cells. En Proc; 2014 60:123.
[14]Leendertz C, Stangl R, Modelling an a-Si:H/c-Si Solar Cell with AFORS-HET. In: van Sark WGJHM , Korte L, Roca F, editors, Physics and
Technology of Amorphous-Crystalline Heterostructure Silicon Solar Cells, Berlin: Springer; 2012. p. 459–482.
[15]Hurkx GAM, Klaassen DBM, Knuvers MPG. A new Recombination Model for Device Simulation Including Tunneling. IEEE Trans Elec
Dev; 1992 39:331.
[16]Willemen JA, Zeman M, Metselaar JW. Computer Modelling of Amorphous Silicon Tandem Cells. In: Proc. of 1st WCPC-I, Hawaii; 1994 p.
599–602.
[17]Zeman M, Willemen JA, Vosteen LLA, Tao G, Metselaar JW. Computer Modelling of Current Matching in a-Si:H/a-Si:H Tandem Solar
Cells on Textured TCO Substrates. Sol En Mat and Sol Cells; 1997 46:81.
[18]Juska G, Kocka J, Arlauskas K, Jukonis G. Electron Drift Mobility in a-Si:H under Extremely High Electric Field. Sol State Comm; 1990
75:531.
[19]Bivour M, Reichel C, Hermle M, Glunz SW. Improving the a-Si:H(p) Rear Emitter Contact of n-Type Silicon Solar Cells. Sol En Mat and
Sol Cells; 2012 106:11.
