Light Trapping in Thin Film Silicon
Solar Cells on Plastic Substrates
Cover image: Microscope image of the grooves of ’Here comes the sun’ by the
Beatles, on vinyl.
Druk: Ipskamp Drukkers BV, Amsterdam
Light Trapping in Thin Film Silicon
Solar Cells on Plastic Substrates
Lichtopsluiting in dunnelaagsilicium
zonnecellen op plastic substraten
(met een samenvatting in het Nederlands)
Proefschrift
ter verkrijging van de graad van doctor aan de Universiteit Utrecht
op gezag van de rector magnificus, prof.dr. G.J. van der Zwaan,
ingevolge het besluit van het college voor promoties in het
openbaar te verdedigen op woensdag 16 januari 2013
des ochtends te 10.30 uur
door
Micha Minne de Jong
geboren op 6 maart 1981 te Laren
Promotor:
Co-promotor:
Prof.dr. R. E. I. Schropp
Dr. J. K. Rath
The work described in this thesis was financially supported by NL Agency
(Agentschap NL) of the Ministry of Economic Affairs, Agriculture and Innovation of The Netherlands: program EOS-LT (Energie Onderzoek Subsidie Lange Termijn).
Contents
1 Introduction
1.1 Renewable energy . . . . . . . . . .
1.2 Photovoltaic energy and solar cells
1.3 Silicon thin film solar cells . . . . .
1.4 Some basic solar cell physics . . . .
1.5 Low temperature flexible solar cells
1.6 Outline and objectives . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
9
9
10
12
14
16
18
2 Experimental techniques
2.1 Silicon depositions: Plasma-enhanced chemical vapour deposition
2.1.1 The ASTER deposition system . . . . . . . . . . . . . .
2.1.2 The IRIS plasma characterisation system . . . . . . . .
2.2 Materials characterization . . . . . . . . . . . . . . . . . . . . .
2.2.1 Reflection-transmission measurements . . . . . . . . . .
2.2.2 Constant-photocurrent method . . . . . . . . . . . . . .
2.2.3 Raman spectroscopy . . . . . . . . . . . . . . . . . . . .
2.3 Solar cell characterization . . . . . . . . . . . . . . . . . . . . .
2.3.1 The solar simulator . . . . . . . . . . . . . . . . . . . . .
2.3.2 Spectral response . . . . . . . . . . . . . . . . . . . . . .
21
21
22
24
25
25
27
27
29
29
30
3 The
3.1
3.2
3.3
31
31
33
34
34
role of temperature in plasma dust formation
Dusty plasmas: From α to γ’ . . . . . . . . . . . . . . . . . . .
The influence of temperature on dust formation . . . . . . . . .
Dust formation and OES . . . . . . . . . . . . . . . . . . . . .
3.3.1 Recording OES profiles . . . . . . . . . . . . . . . . . .
3.3.2 Dust formation as a function of power, hydrogen dilution, and temperature . . . . . . . . . . . . . . . . . . .
3.3.3 TEM images of dust . . . . . . . . . . . . . . . . . . . .
37
39
6
Contents
3.4
3.3.4
Mass
3.4.1
3.4.2
3.4.3
3.4.4
OES of pulsed Plasmas . . . . . . . . . .
spectrometry . . . . . . . . . . . . . . . . .
Clusters, the precursors of dust formation
Ion energies . . . . . . . . . . . . . . . .
Cluster formation and temperature . . .
Conclusions . . . . . . . . . . . . . . . .
4 Low temperature silicon layers
4.1 The role of substrate temperature in PECVD
4.2 Controlling the substrate temperature . . . .
4.2.1 Substrate stretch holder . . . . . . . .
4.2.2 Gas pressure . . . . . . . . . . . . . .
4.2.3 Plasma heating . . . . . . . . . . . .
4.3 Low temperature intrinsic layers . . . . . . .
4.3.1 a-Si:H intrinsic layers . . . . . . . . .
4.3.2 nc-Si:H intrinsic layers . . . . . . . . .
4.4 Low temperature doped layers . . . . . . . . .
4.5 Conclusions . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
5 Light trapping in amorphous silicon cells on
substrates
5.1 Light trapping techniques . . . . . . . . . . . .
5.1.1 Scattering . . . . . . . . . . . . . . . . .
5.1.2 Nanopyramid periodic structures . . . .
5.1.3 Geometric light trapping: micropyramid
tures . . . . . . . . . . . . . . . . . . . .
5.2 Low temperature solar cells on PC substrates .
5.2.1 Cells on PC: Experimental issues . . . .
5.2.2 Solar cell results . . . . . . . . . . . . .
5.3 Post-deposition treatments . . . . . . . . . . .
5.3.1 Shunt busting . . . . . . . . . . . . . . .
5.3.2 Post deposition annealing . . . . . . . .
5.3.3 Stability under light soaking . . . . . .
5.4 Conclusions . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
41
42
42
43
43
47
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
49
49
50
51
52
54
56
56
58
61
62
polycarbonate
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
periodic struc. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
6 Micromorph tandem cells on plastic substrates
6.1 Introduction . . . . . . . . . . . . . . . . . . . . .
6.2 nc-Si:H cells on glass substrates . . . . . . . . . .
6.3 Tandem cells on glass substrates . . . . . . . . .
6.3.1 Stability under light soaking . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
65
65
66
67
68
73
74
77
83
83
84
84
85
87
87
88
91
93
6.4
6.5
Contents
7
Tandem cells on plastic substrates . . . . . . . . . . . . . . . .
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
99
Bibliography
101
Summary
115
Samenvatting
119
List of publications
123
Nawoord
125
Curriculum Vitae
127
Chapter 1
Introduction
1.1
Renewable energy
The worlds thirst for energy expands rapidly. The International Energy Agency
(IEA) projects that the total energy consumption will grow 30% up to 2035
compared to 2010, leading to a 20% rise in carbon dioxide (CO2 ) emission [1].
90% of this increase in demand originates from emerging economies like China
and India, Brazil and the Middle East. These countries have a fast growing
middle class population that consumes more and more energy and have a rapid
economic growth. The energy consumption and the growth of these countries
are dominated by fossil fuels. Of the energy consumed in 2010 world wide, 81%
is generated from fossil fuel sources and this is estimated to be still 75% in
2035. If this scenario becomes reality, the IAE estimates a global temperature
rise in excess of 3.5°C. Therefore, new non-fossil based energy sources need to
be explored.
A very important candidate for alternative energy is solar energy. The
amount of radiative power the earth receives from the sun is multiple thousand
times the amount of energy consumed world wide. Therefore, if we are able to
harvest this energy in an economically viable manner, solar energy could fulfil
a large fraction of the worldwide energy demand.
Solar energy generation can generally be divided in two different technologies: Solar-thermal applications, in which solar radiation is used to heat up a
medium, which in turn is used to power a turbine. In photovoltaics (PV), the
radiation is directly converted into electrical energy.
If solar energy conversion is to become a large contributor to the energy
10
Chapter 1. Introduction
production, drastic cost reduction for solar cell modules is required. Solar
energy harvesting will only become an important means of energy production
if the price of solar energy can compete with conventional energy sources.
As technology progresses, the production costs have decreased over the last
few decades [2], while efficiencies have increased [3]. Also the upscaling of
production has been beneficial for cost reduction of solar panels. The price of
solar energy not only depends on the price of the equipment, but depends also
on the amount of sunlight available, which varies greatly at different locations
on the globe. If solar energy is to compete with energy from fossil sources, also
the price of other sources of energy plays a major role, which is also a location
dependent parameter. The point at which solar electricity can be harvested
at a price equal to that or lower than the price of electricity produced by
conventional energy plants, is called grid-parity. If this point is reached, solar
energy becomes a viable alternative to conventional energy sources without
subsidy support, although government policies may still play a dominant role
[4]. In some locations, solar energy harvesting has already reached grid parity
[2].
1.2
Photovoltaic energy and solar cells
In 1839, Alexandre-Edmond Becquerel discovered the photovoltaic effect [5].
He observed that light could induce a current when an interface of two liquids
was illuminated. Nowadays, we are familiar with a range of semiconductors
which we can use to convert photon energy into energetic charge carriers. If
we are able to extract these charge carriers from the semiconductor material,
we can generate a current that can be used to drive an external circuit. These
principles are the starting point of the development of solar cells.
When in 1954 Bell Laboratories reported on a p-n junction in silicon based
solar cell with an efficiency of around 6%, great interest was born for this
concept [6]. Initially, solar cells were far too expensive to be used for terrestrial
electricity generation, but became a standard source for space applications.
Ever since, reported efficiencies have continuously been increasing whereas the
production costs have decimated with production volume. In the last 10 years,
PV is one of the fastest growing industries with annual growth rates more than
40%. This increase is not only fueled by the progress in technology and lower
module prices, but also by increasing prices for fossil fuel based energy, and
to some extent by the awareness of the general public of the environment and
climate change and the government policies such as a feed-in-tariff for solar
energy [7].
1.2. Photovoltaic energy and solar cells
US$0.10/W
US$0.20/W
11
US$0.50/W
100
Thermodynamic
limit
80
60
US$1.00/W
III
40
20
Single bandgap
limit
I
II
US$3.50/W
0
0
100
200
Cost
300
400
500
(US$/m2)
Figure 1.1: Three generations of solar cells, showing their present and projected cost and efficiency. From [9].
The largest fraction of the PV industries is crystalline silicon (c-Si) based
technology, with a market share of around 90%. Thin-film based technologies
take up about 10% of the market share [7, 8].
In PV technology, three generations are distinguished. First generation
solar cells are c-Si based single junction solar cells, which in terms of efficiency,
will never cross the Shockley-Queisser (S-Q) limit of 30%. The S-Q limit
depends on the size of the band gap (Eg ) of the used material [10] and is 34%
for an optimum band gap of 1.4 eV. Crystalline silicon solar cells are based on
silicon wafers, which are sawed from single-crystal or multicrystalline silicon
ingots and have a typical thickness of a few hundred micrometers. Apart
from the efficiency limit, a major drawback of this technology is the relatively
high material usage. Second generation solar cell technology is based on thin
film technologies. The single junction type of 2nd generation solar cells have
again suffered from S-Q efficiency limits, however, a multijunction cell can in
principle lift the efficiency above the S-Q limit. Instead of bulk material as
the absorber material, these cells use thin films, which are deposited from the
gas phase. This has a number of advantages. (1) Because the thin films have
12
Chapter 1. Introduction
a thickness in the order of micrometers or even thinner, only a fraction of
the amount of material is needed. This makes them lighter, less fragile and
significantly cheaper to produce. (2) The gas phase deposition processes allow
for large glass substrates [11, 12] to be used as substrate materials, copying the
fabrication of displays. (3) Because other substrates than glass can be used,
thin film solar cells can be made flexible, thereby making it possible to fabricate
them in a to a roll-to-roll manufacturing process, which may drastically cut
production costs. Third generation solar cells can potentially break the S-Q
limit. Examples of third generation concepts are multi-junction cells [13], hotcarrier devices [14], spectral conversion techniques (up or down conversion)
[15, 16] or quantum dot-based devices [17]. A schematic representation of the
three generations of solar cells is given in figure 1.1.
1.3
Silicon thin film solar cells
The atoms in hydrogenated amorphous silicon (a-Si:H) do not form an ordered
matrix, unlike in c-Si, in which all atoms are fourfold coordinated. In a-Si:H
there is a certain degree of variation in the bond length and bond angle distribution, which has implications for the band gap and leads to the presence of
band tail states. Because not all atoms are bonded to four neighbouring atoms,
there will be non-bonding orbitals, the so-called dangling bonds. In a-Si:H material, most of these bonds will be passivated by bonding to incorporated H
atoms, although a number of dangling bonds will remain unpassivated, which
will act as midgap states or electronic defects in the silicon matrix. In doped
a-Si:H layers, due to a thermal equilibrium between dopants and defect creation, the defect density is even higher, leading to a higher recombination rate
for charge carriers. A conventional p-n junction, as used in c-Si solar cells
will therefore not perform adequately because most charge carriers are lost
in the silicon layers through trapping and recombination before they can be
extracted from the solar cell. Alternatively, by sandwiching an intrinsic a-Si:H
layer between doped layers, a device can be created in which the intrinsic layer
acts as the main light absorber. The p- and n-layers induce an electric field in
which the charge carriers drift towards the contacts.
Crystalline silicon is an indirect semiconductor with a band gap of 1.1 eV.
For every photon to be absorbed, momentum needs to be transferred to the
silicon lattice in the form of phonons to conserve momentum. Therefore the
absorption coefficient of c-Si is rather low and c-Si based cells are relatively
thick. Due to the structural disorder in a-Si:H, the material acts as a direct
band gap semiconductor with an Eg of around 1.8 eV, which results in a higher
1.3. Silicon thin film solar cells
13
absorption coefficient. Therefore, a-Si:H solar cells can be made much thinner
than their c-Si counterparts, with a typical i-layer thickness of smaller than
500 nm.
In 1965, Sterling showed the possibility to deposit a-Si:H in a radio frequency (RF) discharge [18]. Chittick showed in 1969 that a-Si:H can have a
broad range of photoconductivity, depending on the deposition temperature
[19]. His group also showed the possibility of fabricating n-type a-Si:H by substitutional doping by phosphorus. Ever since Spear discovered the possibility
of doping of a-Si:H both n-type and p-type by adding PH3 and B2 H6 to the gas
mixture in 1975 [20] and their following paper on the first a-Si:H p-n junction
in 1976 [21], numerous research groups and companies have been investigating
possible applications of this material [22]. It was Carlson and Wronski [23]
who made the first a-Si:H p-i-n solar cell, achieving a conversion efficiency of
2.4%. With the introduction of the first multi-chamber system by Kuwano
et al. [24], which separates reaction chambers for the deposition of n-doped,
p-doped and intrinsic layers to avoid dopant atom cross contamination, a new
solar cell record was set at 6.9% conversion efficiency. At present, the highest
reported stabilized conversion efficiency of a single junction a-Si:H solar cell
is 10.1% [3]. A major challenge for a-Si:H cells is the degradation under light
exposure, as first described by Staebler and Wronski in 1977 [25]. When the
cells are exposed to light, the midgap defect density (Nd ) increases, which
leads to degradation of the cell performance.
Another form of silicon that can be deposited from the gas phase is nanocrystalline silicon (nc-Si:H), which is also called microcrystalline silicon (µcSi). Nanocrystalline silicon differs from amorphous silicon by its structure; it
contains grains of crystalline silicon, embedded in the amorphous tissue. By
changing the atomic hydrogen concentration relative to silyl (SiH3 ) species
reaching the growth surface, the crystalline fraction can be controlled. Nanocrystalline silicon shows properties different from amorphous silicon. Due to
the silicon crystallites, nc-Si:H has a lower band gap of ∼1.1 eV and therefore
shows a higher absorption in the red part of the spectrum, and can have a
higher electron mobility if there are not too many grain boundary defects.
Furthermore, nc-Si:H solar cells are less sensitive to light induced degradation
(LID). The material was first reported by Vepřek in 1968 [26] and unintentionally oxygenated nc-Si:H was first used to make a thin film solar cells in
1992 [27], while in 1994 the first complete nc-Si:H cell was created by Meier et
al. from IMT [28]. It was also the researchers from IMT who introduced the
"micromorph" concept: a double junction solar cell, which comprises an a-Si:H
top cell and a nc-Si:H bottom cell to collect light from a broader spectrum
[13]. At present, the highest reported and certified nc-Si:H single junction cell
14
Chapter 1. Introduction
has an efficiency of 10.1% [29]. The highest reported stabilized (light soaked)
and certified efficiency for an a-Si:H/nc-Si:H tandem configuration is 12.2%
[30]. Because a multi-junction cell is basically a set of solar cells connected in
series, the total current is limited by the cell that generates the lowest current
density. Therefore, the current generated by the cells needs to be matched,
which imposes requirements on the thicknesses of the individual cells.
1.4
Some basic solar cell physics
Figure 1.2 shows a schematic representation of a superstrate type thin film
solar cell in the p-i-n configuration. It generally consist of a substrate, a transparent front contact (transparent conductive oxide or TCO), a p-doped layer,
an intrinsic (undoped) layer, an n-doped layer and a back reflector in succession. Metal deposited contacts are used for carrier extraction from the front
and back side of the cell. The heart of the cell is made out of a semiconductor
material, which can convert photon energy into excited charge carriers [5], if
the energy of the photon is higher than the band gap of the semiconductor.
In the case of a-Si:H, the material has an Eg of 1.8 eV at room temperature,
with band-tail states, caused by weak Si–Si bonds, and mid-gap states, caused
by silicon dangling bonds within the gap, caused by unbonded Si bonds in the
material. When a photon is absorbed, this can lead to the formation of an
electron-hole pair, which can move through the semiconductor material. The
charge carriers need to be moved to the external contacts by drift in an electric field to be able to recombine in an external circuit, because a transport
mechanism in the absorber layer based on diffusion alone would lead to a large
recombination loss. To accomplish this, an internal electric field is generated
by using doped layers. The absorber layer is made out of intrinsic silicon,
because it has a midgap defect density nearly two orders of magnitude lower,
compared to the doped layers.
Figure 1.3 shows the band diagram of a solar cell under short-circuit conditions (left) and under forward bias voltage conditions (right). Also shown are
the band gap, the quasi Fermi level (Ef ), and the electrons and holes. Applying a forward bias (Vb ) to the cell will reduce the electric field within the cell,
causing a higher recombination rate. When there is no current, the cell is in
open-circuit conditions. The applied voltage at this point is the open-circuit
voltage (Voc ).
1.4. Some basic solar cell physics
15
Incoming light
Substrate
TCO
p-layer
i-layer
Front contact
n-layer
Back reflector
Back contact
Electron potential
Figure 1.2: Schematic presentation of the layers in a p-i-n superstrate type
thin film solar cell.
-
Eg
Ef
Eg
-
+
+
p
eVb
i
n
p
i
n
Figure 1.3: Band diagram of a solar cell under short-circuit conditions (left)
and under forward bias conditions (right).
16
1.5
Chapter 1. Introduction
Low temperature flexible solar cells
For solar energy generation to become a feasible alternative to fossil fuelbased electricity generation or to other means of renewable energy generation,
the price of solar cells is extremely important. Thin film silicon solar cells
have the potential to be fabricated much cheaper than their crystalline silicon
counterparts. Not only the amount of material used is greatly reduced, also the
energy input to fabricate thin film solar cells is much smaller than for c-Si cells,
resulting in a lower so-called energy payback time [31]. If a flexible material
is chosen for the substrate, bendable cells can be made, which have a number
of advantages over rigid substrates, such as glass, without loss in efficiency.
First of all, plastics such as polyethylene terephthalate (PET), polyethylene
naphthalate (PEN) and polycarbonate (PC) are much cheaper than glass.
More importantly, flexible substrates enable a roll-to-roll process, in which cells
can be produced in a continuous process, rather than a batch-type process,
used for glass substrates. Especially when working with vacuum equipment,
roll-to-roll processing can yield large cost reductions [32]. Furthermore, flexible
cells can be shaped in many ways, making them integrable into buildings and
architecture. Using a lightweight and unbreakable substrate, such as plastics,
savings can be made in transport costs. It has to be noted that the highest
initial efficiency of 16.3% for thin film silicon solar cells has been obtained
on a flexible substrate (stainless steel foil), using a triple junction (a-Si:H/aSiGe:H/nc-Si:H) by Unisolar Ovonics [33].
The main challenge in depositing thin film solar cells directly on plastic
flexible substrates are the limitations imposed on deposition temperature. If
we want to use cheap plastics such as PET, PEN or PC, we are limited to deposition temperatures of 70-100°C, 150°C or 130°C, respectively [34]. In 1974,
Spear already observed changes in material quality of a-Si:H when changing
the substrate temperature in a plasma-enhanced chemical vapour deposition
(PECVD) [35]. Figure 1.4 shows the defect density of a-Si:H and nc-Si:H
layers as a function of substrate temperature. For both types of layers, a
minimum in dangling bond density is observed at a substrate temperature of
around 200-250°C [36], much higher than the temperatures permitted when using cheap plastics as a substrate. Depositing at lower substrate temperatures,
while keeping other deposition parameters constant, will produce layers with a
much higher defect density, resulting in lower solar cell efficiencies. Changing
the flow ratio of silane and hydrogen in the reaction chamber can be used to
compensate for these effects, but these measures will have direct consequences
for the deposition rate or can change the phase of the grown material [37].
Some of the changes in the material quality can be ascribed to the decrease
1.5. Low temperature flexible solar cells
17
Figure 1.4: Defect density in nc-Si:H and a-Si:H films as a function of substrate
temperature. From [36].
of energy transferred to the growing film by ions in the plasma [38]. Keeping
the film thickness low will reduce the charge carrier recombination losses, in
which case excellent light trapping schemes are needed for the solar cell.
Table 1.1 shows an overview of reported thin film silicon solar cells on flexible substrates. It is divided into two sections: cells in the p-i-n configuration
and cells in the n-i-p configuration. The table shows that low deposition temperatures generally lead to lower conversion efficiencies. Cells fabricated in a
transfer process are deposited on a temporary substrate that is able to withstand high temperatures and then transferred to a plastic substrate, enabling
high deposition temperatures. Most efforts at depositing low temperature cells
have focussed on n-i-p substrate type cells. Because n-i-p cells do not require
a transparent substrate, a wider range of substrate materials can be used.
To fabricate low-cost thin film solar cells, not only the material costs of
the TCO, absorber material and the substrate material are important, also
throughput is important. For micromorph tandem cells, the bottom nc-Si:H
cell has a typical thickness of 1 to 3 µm, which has an impact on deposition
18
Chapter 1. Introduction
Superstrate (p–i–n) cells :
Cell type
D/T
a-Si:H
D
a-Si:H
T
a-Si:H/nc-Si:H
T
Substrate (n–i–p) cells :
Cell type
D/T
a-Si:H/a-SiGe:H/nc-Si:H
D
a-Si:H/a-SiGe:H
D
a-Si:H
D
a-Si:H
D
nc-Si:H
D
a-Si:H
D
Substrate
PET
Polyester
Polyester
T (°C)
110
200
200
η (%)
4.9
7.7
9.1
Source
[39]
[40]
[41]
Substrate
SS
Kapton
PEN
E/TD
LCP
PET
T (°C)
n/a
n/a
150
140
180
100
η (%)
16.3
10.1
8.7
6.0
8.1
5.9
Source
[33]
[42]
[43]
[44]
[14]
[45]
E/TD: ethylene–tetracyclododecene copolymer, SS: stainless steel, LCP: liquid crystal polymer
Table 1.1: Present reported record initial efficiencies of silicon thin film solar
cells deposited on flexible substrates for p-i-n and n-i-p types of cells. D or T
denotes a direct deposition or a transfer-process.
times and therefore on manufacturing costs. Moreover, the top a-Si:H cell
has to be thin, because thick a-Si:H cells are more sensitive to light induced
degradation than very thin a-Si:H layers. For these reasons, thin absorber
layers should be used. However, to ensure good light absorption within the
active layers, light trapping schemes must be deployed. These schemes have
the purpose to enlarge the optical path of travelling photons in the cell, while
keeping the electrical paths short, enabling a good light absorption without
sacrificing too much on electrical performance. The traditional way in thin
film (superstrate) p-i-n solar cells is to create a rough interface between the
TCO and the p-layer of the cell, by texture etching the doped ZnO in an acidic
solution or depositing natively textured SnO2 [46, 47, 45]. Other schemes such
as rough silver, 1D or 2D [48] gratings or 3D nanopillar designs [49, 50] have
shown to improve the generated current in n-i-p cells.
1.6
Outline and objectives
The goal of this thesis is to investigate and find solutions for the difficulties that
are encountered when depositing thin film silicon solar cells on plastic flexible
substrates. On the level of deposition plasmas, we investigated the influence
1.6. Outline and objectives
19
of the deposition temperature on a number of plasma characteristics, with a
focus on the formation of dust, which is a temperature dependent process. The
influence of deposition temperature on the quality of silicon layers is studied.
It was investigated how the deleterious effects of low deposition temperature
on material quality can be compensated. Finally, solar cells were made at low
temperatures, both on glass substrates and on plastic substrates. Different
light trapping management techniques were tested for amorphous silicon solar
cells, nanocrystalline solar cells, and micromorph tandem solar cells.
Chapter two introduces the main experimental techniques that are used for
silicon and metal oxide depositions, optical and electrical material characterization and solar cell characterization techniques which are used for the study
for this thesis.
Chapter three is concerned with the changes that are induced in the plasmas when changing the deposition temperature. Using a newly developed
technique that utilizes the axial optical emission profile from the plasma we
were able to identify whether a plasma is dust-free or produces dust particles.
Furthermore, the influence of the substrate temperature is investigated on the
first stage of dust formation: cluster formation. Using mass spectroscopy, the
formation of polysilanes is studied as a function of substrate temperature.
The subject of chapter four is the study of the optical and electrical properties of amorphous and nanocrystalline silicon deposited at substrate temperatures below the optimum temperature and to develop plasma conditions at
which device-quality silicon layers can be deposited.
Chapter five treats the deposition and characterization of amorphous silicon thin film solar cells deposited at low temperature (130°C) on glass and the
deposition of these low temperature cells on plastic polycarbonate substrates.
First, a number of practical concerns are treated: the adhesion of layers to the
plastic substrates, the curving of plastic substrates due to thermal expansion
and the degassing of plastic substrates. A number of light trapping schemes
was studied: Scattering by a rough interface between the front contact (TCO)
layer and the p-layer, geometric light trapping by pyramids that are larger
then the effective wavelength of light in the material and finally light management by pyramidal structures comparable to the effective wavelength of light.
It was investigated how these texturization schemes enhance the light absorption, but also their influence on the electrical quality of the cells: Voc and fill
factor. Furthermore, a number of post-deposition treatments are investigated:
thermal annealing, shunt busting and light induced degradation.
The last chapter presents nanocrystalline solar cells and amorphous/ nanocrystalline silicon (micromorph) tandem solar cells deposited at low substrate
temperature (130°C) both on glass and on polycarbonate substrates. First
20
Chapter 1. Introduction
nanocrystalline solar cells are deposited on glass and characterized. a-Si:H/
nc-Si:H tandem cell are fabricated at low temperature on glass and on plastic
substrates.
A part of this work was done in collaboration with Wageningen University Glastuinbouw, who perform research on the use of micro- and nano textured glass and plastics for ultra-transparant greenhouse roofing to increase
crop production, and with Aquamarijn Micro Filtration BV, who provided the
micro-pyramid structured polycarbonate substrates.
Chapter 2
Experimental techniques
This chapter describes the experimental techniques that were used to deposit
different layers and solar cells used in this thesis and the techniques to characterize them.
2.1
Silicon depositions: Plasma-enhanced chemical vapour deposition
Plasma-Enhanced Chemical Vapour Deposition (PE-CVD) is a form of CVD
that can be used for thin film depositions. As opposed to most other forms of
CVD, PE-CVD can be operated at low temperatures. In the process, a source
gas is dissociated in an oscillating electric field between two parallel plates. In
our case, the substrate is mounted to the grounded electrode. Between the
plates, a source gas is introduced. For silicon thin film depositions, usually
SiH4 and H2 are used. In the electric field, electrons are accelerated and may
collide with a gas molecule. If this collision is sufficiently energetic, the impact
can cause ionization of the molecule, thereby creating an extra free electron,
which can in turn collide with a molecule. This avalanche of reactions will
result in a plasma containing (positive) ions and free electrons. Because the
electrons are much lighter than the ions, the electrons are faster and will be
collected at the electrodes, resulting in a positive plasma bulk.
Traditionally an excitation frequency of 13.56 MHz is used, due to legal
restrictions on the use of other radio frequency bands. Changing the excitation
frequency changes the plasma properties, such as the ion energies and bias
voltage in the plasma [51], which in turn can be beneficial or detrimental for
22
Chapter 2. Experimental techniques
the layer quality of the grown silicon layers.
2.1.1
The ASTER deposition system
The silicon layer depositions in this thesis are performed in the ASTER (Amorphous Semiconductor Thin-film Experimental Reactor) ultra-high vacuum multichamber deposition system [52], using Very High Frequency PE-CVD (VHF
PE-CVD). The chamber features 5 deposition chambers, mounted to a central
transport chamber: one for p-type silicon depositions, one for n-type depositions, two for intrinsic silicon depositions and one experimental reactor for
nanocrystal formation and deposition. A parking chamber is used for storing under vacuum and gradual cooling of samples. Samples can be up to
10 × 10 cm2 and are mounted on a titanium substrate holder. The holder is
inserted into the system through a loadlock chamber and can be transported between the separate chambers by a robot arm in the central chamber.
The deposition chambers are equipped with viewports to monitor the plasma,
either visually or using a spectrometer. In the intrinsic silicon chambers the
inter-electrode distance can be changed from 5 to 27 mm. These reactors are
equipped with showerhead-type powered electrodes for an even distribution of
the feedstock gasses into the plasma. All intrinsic layers are deposited at a
VHF excitation frequency of 60 Mhz, whereas the doped layers are grown at
50 MHz. The impedance of the plasma reactor can be matched to the 50 Ω of
the power input system through a set of adjustable capacitors, which form a
an L-type matching network. The amount of reflected power (measured with a
Rhode & Schwarz NAP power meter) can be minimized to less than 1% of the
input power. The area of the powered electrode is 170 cm2 in the chambers for
intrsinsic material deposition 150 cm2 in the chambers for doped depositions.
As source gasses for the depositions we use silane (SiH4 ) and molecular
hydrogen (H2 ) for the growth of intrinsic silicon layers. By changing the ratio
of the two source gasses, we can control the phase of the silicon to be either
amorphous, nanocrystalline or mixed-phase. For doped layers, dopant gasses
are added to the gas mixture. Trimethylboron (B(CH3 )3 or TMB) is added
for p-type doping and phosphine (PH3 ) is used for n-type doping.
Optical Emission Spectroscopy
During the deposition the process can be monitored in situ by optical emission
spectroscopy (OES). If a molecule A (or AB) is excited by electron impact
either by direct excitation
2.1. Silicon depositions: Plasma-enhanced chemical vapour deposition
23
4
S iH *
In te n s ity (a .u .)
3
2
H
β
H
α
S i*
1
0
2 0 0
3 0 0
4 0 0 5 0 0 6 0 0 7 0 0
W a v e le n g th (n m )
8 0 0
9 0 0
Figure 2.1: An example OES spectrum recorded from one of the ASTER
deposition chambers of a silane-hydrogen plasma. The different line emissions
are identified.
24
Chapter 2. Experimental techniques
A + e − → A∗ + e −
(2.1)
AB + e− → A∗ + B + e−
(2.2)
or by dissociative excitation
the excited molecule A∗ can relax to its ground state
A∗ → A + hν
(2.3)
emitting a photon. By recording the wavelength of the photons, we can identify
the species in the plasma. Species of interest are Si∗ (at 289 nm), SiH∗ (at
414 nm), Balmer alpha (Hα , 656 nm) and Balmer beta (Hβ , 490 nm). The
emission rates are associated with the dissociation rate of the different species. An example OES spectrum, recorded in ASTER from a silane/hydrogen
plasma, is presented in figure 2.1, showing the emitted lines. Furthermore,
information about the electron temperature can be extracted from the OES
information.
The light, emitted by the plasma, is monitored through a viewport and
an optical fibre and analysed and recorded using an Avantes MC2000 spectrometer. The peaks found in the emission spectrum are fitted to Gaussians after
subtracting a local background.
To reduce deposition on the viewport window, which would influence the
transmission of the window, it is shielded from the plasma using a valve when
no measurements are taken. An assembly of two horizontal slits is used to
obtain a vertical emission profile of the plasma. This technique is used for
the detection of dust in the plasma, as described in chapter 3. A schematic
representation of one of the ASTER deposition chambers and the OES setup
is given in figure 2.2.
2.1.2
The IRIS plasma characterisation system
The IRIS (Ions and Radicals in Silane plasmas) system is designed to examine
ions and molecules formed in the plasma. Therefore it is equipped with a
Hiden EQP300 mass spectrometer. The plasma chamber is a copy of one of the
deposition chambers from ASTER, but at the position of the substrate (at the
grounded electrode), an orifice is fitted that leads to a separate chamber, which
is pumped to an ultra-high vacuum. Behind the orifice, the mass spectrometer
is mounted. Ions and radicals that would normally reach the growing surface
at the substrate will now travel through the orifice, into the mass spectrometer.
2.2. Materials characterization
25
Optical fibre
Quartz
window
Heater/
substrate holder
Optical slits
VHF Plasma
Movable
platform
Optical port
Shutter
Showerhead
electrode
Gas inlet
Reactor wall
VHF
Figure 2.2: A schematic representation of one of the ASTER deposition chambers and the OES setup.
A set of electrostatic lenses leads the ions into the mass spectrometer to obtain
optimal yield. The mass spectrometer is not only mass-sensitive, but can also
distinguish between species of different energies. The chamber behind the
orifice is differentially pumped to make sure that the species do not collide
behind the orifice, thereby changing their energy and trajectory.
If neutral species are to be detected, they need to be ionized before they
enter the mass spectrometer. This is achieved by ionizing them by an electron
emission filament. The system is used for the detection of dust precursors, as
described in chapter 3 and described in more detail by E. Hamers [53]. The
system can also be fitted with an OES system, similar to the system attached
to the ASTER system.
2.2
2.2.1
Materials characterization
Reflection-transmission measurements
The R-T mini setup
The Reflection/Transmission setup (RT) from M. Theiss hard- and Software
[54] measures the specular reflection and transmission simultaneously on the
same spot on a sample. A halogen lamp, sample stage, connected with an
optical fibre to a spectrometer enable recording of the spectra from 380 to
1050 nm. The data can be analysed using a software package called ’SCOUT’
26
Chapter 2. Experimental techniques
Differentially
pumped chamber
To Hiden mass
spectrometer
Reactor wall
Heater
Orifice
VHF Plasma
Gas inlet
Reaction chamber
VHF electrode
VHF
Figure 2.3: A schematic representation of the IRIS reactor chamber and the
Hiden mass spectrometer attachment.
by W. Theiss [54], which can use several models for the calculation of the
wavelength-dependent absorption coefficient (α) and refractive index (n) for
different materials. These properties are then used to simulate transmission
and reflection spectra for layer stacks of different materials and different thicknesses. These can be fitted to the measured spectra. In the case of silicon, we
use the O’Leary, Johnson and Kim model (OJL) [55]. This model describes
mathematically the shape of the valence band and conduction band densities
of states of semiconductors, which can be used to calculate the absorption coefficients. The Kramers-Kronig relation is used to calculate the refractive index
(n) . By fitting these models to the acquired data we obtain the wavelength
dependent α and n of the measured material. From this information we can
distil the optical band gap of the material.
The UV-VIS spectrometer
Whereas the R-T mini setup only measures specular transmission and reflection on a sample, the Perkin Elmer Lambda 2S UV-VIS spectrometer setup is
equipped with an integrating sphere, which can also measure diffuse reflection
and transmission, which can, for instance, be important for light scattering
properties of substrates. In this way we can discriminate between specular,
diffuse and total reflection and transmission. Because the wavelength range of
2.2. Materials characterization
27
this apparatus is between 200 and 1100 nm, we can measure in the ultraviolet
(UV) range, in constrast to the R-T mini setup. Whereas the R-T mini setup
measures reflection and transmission on the same position on the sample simultaneously, this is not possible in the Perkin Elmer, because it needs to be
setup differently for transmission and reflection measurements.
2.2.2
Constant-photocurrent method
The sub-band gap absorption of silicon holds information on the slope of the
band edges of the band diagram and on the density of mid-gap defect states.
A method to measure the sub-band gap absorption is the use of the constantphotocurrent method (CPM). This technique is based on the photoconductivity of the sample when it is illuminated at a certain wavelength. The light is
supplied by a 250 W halogen lamp and guided through a filter wheel, containing 25 different interference filters in the red and infrared part of the spectrum.
For every filter, the lamp intensity is changed until the photocurrent matches
a predefined value, while (a relative measure for) the number of photons (Nph )
reaching the sample is recorded. Now the optical absorption coefficient α is
proportional to 1/Nph . The absolute absorption coefficients can be found by
calibrating the CPM absorption coefficient to to the absorption coefficients
found in RT measurements for λ< 1000 nm. For a-Si, from the absorption
coefficient at hν = 1.2 eV (α1.2 ) the defect density Nd can be calculated from
Nd = F α1.2 , where F is a calibration factor, found to be 1016 cm−2 [56]. The
band-edge absorption shows exponential behaviour, called the Urbach tail,
which can be expressed as
α(λ) = α0 e−E(λ)/E0
(2.4)
where α0 is a proportionality factor, E(λ) is the photon energy and E0 is the
so-called Urbach energy, which is obtained by fitting a logarithmic slope to
the absorption coefficient as a function of photon energy.
2.2.3
Raman spectroscopy
The crystalline volume fraction of a sample can be quantified by Raman spectroscopy measurements. In this measurement, the sample is locally illuminated by a strong laser. Besides the light that will be absorbed, transmitted
or specularly reflected, a portion of the light will be inelastically scattered. A
small fraction of the light shows a frequency shift (the Stokes shift) caused
by interactions with phonons in the material. This phenomenon is called
28
Chapter 2. Experimental techniques
Intensity HcountsL
6000
5000
4000
3000
2000
1000
0
250
300
350
400
450
-1
Raman shift Hcm L
500
550
Figure 2.4: An example of a Raman spectrum (dots) and the different fitted
Gaussians (solid lines) of a mixed phase thin film silicon layer deposited on a
PC substrate.
Raman scattering and gives information about the density of states of the
different phonons. The different measured modes in the case of silicon are
the transverse-optic mode (TO, 520 cm−1 ), associated with crystalline silicon
and the transverse-acoustic (TA, 100-200 cm−1 ), longitudinal-acoustic (LA,
300-360 cm−1 ), longitudinal-optic (LO, 380-450 cm−1 ) and the transverse optic (470-490 cm−1 ) modes associated with amorphous silicon [57, 58].
In our setup, a Spectra Physics Ar+ -ion laser (514 nm) illuminates the
sample at an angle of 30°. The light is polarized horizontally before hitting
the sample. After hitting the sample, backscattered light is focussed through
a set of lenses into a Spex triple monochromator and recorded by a nitrogen
cooled Roper Scientific CCD camera, after being polarized vertically. The
data is analysed by fitting a number of Gaussian peaks to the spectrum: 3
peaks fixed at 330, 440 and 480 cm−1 for the amorphous phase and two peaks
between 505 and 520 cm−1 for the crystalline phase. The area of these peaks
is then divided, resulting in the Raman crystalline fraction [59]:
Rc =
I510 + I520
I480 + I510 + I520
(2.5)
where Ix denotes the integrated intensity of the fitted peak at x cm−1 . Although the Rc gives a quantitative measure for the crystalline fraction, it does
not correspond exactly to a volume fraction. Nevertheless it is a useful quantity to compare the crystalline fraction of different samples. Figure 2.4 shows
2.3. Solar cell characterization
29
an example of a Raman spectrum, after background subtraction, measured
on a mixed phase a-Si/nc-Si silicon thin film deposited on a PC substrate at
low temperature. Also shown are the different Gaussians fit to the spectrum,
associated with the different phonon modes.
2.3
2.3.1
Solar cell characterization
The solar simulator
To test the conversion efficiency (η) of solar cells, we can perform currentvoltage (IV) measurements under standard illumination conditions. Internationally it is agreed that these measurements are performed under 100 mW/cm2 ,
AM1.5 illumination [60] at 25°C. In our setup the light is produced by a combination of a xenon and a halogen lamp, through a set of mirrors and lenses.
In the ideal case, the solar cell shows the behaviour of a diode, in parallel with
a current source Jph and a resistance Rp and in series with a resistance Rs ,
which can be described by
e(V − JRs )
V − JRs
) − 1) +
(2.6)
nd kT
Rp
where J0 is the reverse saturation current, e is the electron charge, nd is the
diode quality factor, k is Boltzmann’s constant and T is the temperature in
Kelvin. From the measurements under illumination we can obtain the short
circuit current density (Jsc ), which is the current when V = 0, the open circuit
voltage (Voc ), which is the voltage when J = 0 and the fill factor (FF), which
is defined as
J(V ) = −Jph+ J0 (exp(
Jmpp Vmpp
(2.7)
Jsc Voc
where Jmpp and Vmpp are the current and voltage at the point where the
product of the current and voltage peaks, the maximum power point. Then
the conversion efficiency is
FF =
η=
Pmpp
Jmpp Vmpp
F F Jsc Voc
=
=
Plight
Plight
Plight
(2.8)
in which Plight is the incident light power density, which in the case of AM1.5
light is 100 mW/cm2 . The slope of the curve at V = 0 is associated with Rp ,
whereas the slope at J = 0 is associated with Rs .
When IV characteristics are performed under dark conditions (Jph = 0) we
can extract the diode quality factor nd and the reverse saturation current J0 .
30
Chapter 2. Experimental techniques
2.3.2
Spectral response
The Spectral Response (SR) of a solar cell tells us the fraction of photons that
is converted to an electron that reaches an external circuit as a function of
photon wavelength, also called the quantum efficiency or external collection
efficiency. To measure the SR, modulated light of a xenon lamp is led through
a monochromator and led onto the sample, for which the generated current
is measured using a lock-in amplifier. The light incident on the sample is
calibrated using a photodiode with a known spectral response. The SR can be
measured from 350 up to 1100 nm and is equal to
SR(λ) =
Iph (λ)
nφ (λ) e
(2.9)
where Iph is the measured photocurrent, nφ is wavelength dependent amount
of photons directed at the sample per second and e is the electron charge.
From the wavelength dependent SR, the Jsc under AM1.5 illumination can
be calculated:
ˆ
Jsc = e SR(λ) φAM 1.5 (λ) dλ
(2.10)
in which φAM 1.5 (λ) is the wavelength dependent photon flux.
Commonly, an externally applied bias voltage is used during measurements.
A negative bias voltage is used to strengthen the internal electric field of the
solar cell to reduce the carrier recombination in the i-layer and thus measure
the maximum current generating capabilities of the cell. A positive bias voltage
can be used to investigate the cell’s performance under maximum power pointconditions. Bias light that resembles AM1.5 light (both in spectrum and intensity) can be applied to obtain SR under standardized operation conditions.
Because a tandem cell consists of two cells connected in series, the maximum measured current at a certain wavelength is limited by the cell that
generates the least current at that wavelength. Therefore, the individual cells
within a tandem solar cells can be measured by illuminating the cell with continuous (i.e. not chopped) bias light of different colours, thereby ’activating’
the cell that responds to that colour, such that we can measure its counterpart
by making sure that is does not limit the current from the cell that is being
tested with chopped monochromatic illumination. For this purpose, the bias
light can be led through a number of different filters: a red filter for measuring
the top cell and a blue filter for measuring the bottom cell in a tandem cell
consisting of an a-Si top cell and a nc-Si bottom cell.
Chapter 3
The role of temperature in
plasma dust formation
3.1
Dusty plasmas: From α to γ’
A challenging problem in low temperature depositions of thin-film silicon layers is the formation of dust particles in the plasma. When these particles are
incorporated in the silicon layers, they can introduce voids which will increase
the disorder in the amorphous network and will thus introduce electronic defects. Because the particles can be large compared to the film thickness, the
dust particles can also cause electrical shunts through the layers [61], although
recent research shows that controlled dust formation in the plasma can be
beneficial for device performance when the layers are deposited at high rate
[62].
The process of dust formation can be divided into three phases: starting
with a dust free plasma (the so-called α-regime), in the first phase, negatively
charged clusters may form through polymerization reactions in the plasma.
Figure 3.1 shows a schematic representation of the potential profile in the
bulk of the plasma. The bulk has a positive time-averaged potential, which
causes negatively charged clusters to be trapped inside the plasma bulk. If the
clusters collide with positively charged ions, they will become neutrals and will
leave the plasma by diffusion, unless they collide with electrons before they
leave the plasma, thereby collect negative charge and remain trapped. Because
the electron-capture cross section depends strongly on the size of the clusters
[63], only clusters that are large enough (> 2 nm) will be trapped inside the
32
Chapter 3. The role of temperature in plasma dust formation
Figure 3.1: Schematic presentation of a potential profile in a VHF plasma
reactor through a VHF cycle. The dashed and dotted lines show the plasma
potential at φ = 0.5π and φ =1.5π, respectively. The solid line shows the time
averaged potential. In this graph, the grounded electrode is situated on the
left whereas the powered electrode is on the right.
3.2. The influence of temperature on dust formation
33
plasma bulk. In the second phase, when these clusters reach a critical size and
concentration, the clusters will coagulate and dust will start to form [64, 65].
After coagulation has taken place, the particles have a typical size of a few tens
of nanometres and will quickly acquire negative charge and therefore repel each
other, preventing further aggregation. The last phase represents the growing of
the individual dust particles by the attachment of neutral particles or positive
ions. The dust particles are now either lost at the sides of the reactor, due to
a net gas flow towards the pumps or by diffusion, or are trapped in the reactor
until the plasma is switched off. When dust has formed that remains trapped
until the plasma is switched off, the plasma is in the so-called γ 0 -regime.
3.2
The influence of temperature on dust formation
The substrate temperature has a direct influence on the gas temperature inside
the reactor. It has been shown that in our system, in the temperature range
used, the average (including the inactive parts of the reactor) gas temperature
rises around 25° when the substrate temperature is increased 100° [66]. It
has been reported that choosing a higher substrate temperature can suppress
dust formation. Although the lower gas density at higher gas temperature
may play a role, the main mechanism is believed to be the dependence of the
polymerization rate of negative clusters on the gas temperature [63, 67]. When
the gas temperature decreases due to a decrease in the substrate temperature,
the critical cluster size and concentration for coagulation can be reached much
quicker. Therefore, especially at low substrate temperatures, it is important
to monitor the dust formation in the plasma during the deposition process and
to identify parameter windows for the deposition of dust free silicon. Laser
light scattering experiments, which measure the reflected light from dust in
the plasma, have shown to be a powerful tool to study the last phases of
the dust formation process [68, 69]. Recent investigations show nanoparticle
characterization using white light [70]. However, one of the drawbacks of these
optical techniques is that the deposited silicon films on the viewports absorb
a part of the light, resulting in a time-dependent signal [68]. A method that
uses no optical detection is the spectral analysis of the radio-frequency current
to monitor dust formation. By measuring the amplitude of the fundamental
and the higher harmonics of the current through the plasma the production
of nanometre sized particles can be detected [71, 72].
In this chapter we present a non invasive in-situ diagnostic tool for mon-
34
Chapter 3. The role of temperature in plasma dust formation
itoring dust formation, based on optical emission spectroscopy (OES). By recording the optical emission lines for several species in the plasma as a function
of the vertical position between the electrodes we construct the emission profile of the optical emission of the plasma. Using these profiles we are able
to identify the plasma regime. The advantage of this technique is that the
state of the plasma (dust free or dusty) is marked by the asymmetry of the
OES intensity profile and not by the absolute intensity value. Therefore this
technique is insensitive to the loss of transmittance of the viewport due to
silicon film deposition, which can be a disadvantage when using other optical
techniques. Because optical emission can also be used to predict the material
phase [73], this study shows that a single technique can be used as an insitu plasma diagnosis tool for monitoring of the amorphous to nanocrystalline
transition as well as the transition of the dust-free to the dusty regime without
a supplement technique. Therefore, it is a plasma monitoring tool to control
the complete silicon processing of “micromorph cell” manufacturing.
3.3
3.3.1
Dust formation and OES
Recording OES profiles
The presence of dust has an influence on the optical emission from a plasma.
After the coagulation phase, when particles in the plasma can have sizes of tens
of nanometres, their cross-section for capturing electrons is greatly increased.
This has a direct effect on the plasma properties. The electron density drops
an order of magnitude [63], while the electron temperature is greatly increased.
These properties are reflected in the optical emission from the plasma, as more
electrons have sufficient energy to excite the different plasma species. Because
the electron density in a plasma discharge can be significantly non-uniform
[74], space-resolved OES measurements can provide information on the local
electron temperature of the plasma. If we compare the optical emission from
a dust-producing plasma to a dust-free plasma, the local changes in emission
will tell us where the dust is located. Whereas for atomic species or small
clusters gravity can be neglected when compared to other forces like electrostatic forces or thermophoresis, larger dust particles will be influenced by
gravity and therefore be pulled towards the sheath at the bottom powered
electrode.
To record axial emission profiles from the plasma, we used an Avantes
MC2000 spectrometer connected to an optical fibre positioned behind an assembly of two horizontal slits of 1 mm wide placed at a distance of 80 mm
3.3. Dust formation and OES
O p tic a l E m is s io n (a .u .)
1 2
35
S iH *
H β3 x
1 0
H
8
α
2 x
6
4
2
G ro u n d e d
e le c tr o d e
0
0
P o w e re d
e le c tr o d e
5
1 0
1 5
2 0
2 5
V e r tic a l p o s itio n (m m )
3 0
Figure 3.2: Cross-sectional optical emission profile of a plasma in the α-regime.
Shown are the SiH∗ , Hα and Hβ lines. The position is measured from the upper
grounded electrode downwards.
from each other and 20 cm from the plasma centre. A quartz window is used
to ensure transmission in the ultra-violet. This system is mounted on a stage
that can be moved in the vertical direction. The position is measured from the
upper grounded electrode downwards. The spectral range of the spectrometer
is 200 nm to 900 nm. A schematic representation of the setup is given in figure
2.2. Using this system we recorded horizontal slices of the optical emission
from the plasma, with a spatial resolution of 1 mm. From the recorded spectra
we derived the relative intensity of the lines associated with different plasma
species, by subtracting a local background and fitting Gaussians to the peaks
found in the spectra. In this way we obtained the relative signal intensity of
Balmer-alpha (Hα ), Balmer-beta (Hβ ), excited SiH (SiH*) and excited Si (Si*)
as a function of vertical position.
Because it can take several minutes before the α to γ 0 transition takes place,
we waited for the emission to become stable before recording the spectra.
Figure 3.2 shows a typical optical emission profile for various emission lines
of a plasma in the α-regime in the ASTER deposition system. The plasma
parameters were P = 13 W (powered electrode area 170 cm2 ), R = 45 and
Ts = 200°C, where P is the applied power, R is the hydrogen flow dilution
36
Chapter 3. The role of temperature in plasma dust formation
O p tic a l E m is s io n (a .u .)
1 8
S iH *
H β3 x
1 6
1 4
H
1 2
1 0
8
6
4
2
0
α
2 x
0
G ro u n d e d
e le c tr o d e
P o w e re d
e le c tr o d e
5
1 0
1 5
2 0
2 5
V e r tic a l p o s itio n (m m )
3 0
Figure 3.3: Vertical optical emission profile of a plasma in the γ’-regime.
Shown are the SiH∗ , Hα and Hβ lines. The position is measured from the
upper grounded electrode downwards.
ΦH2 /ΦSiH4 and Ts is the substrate temperature. The inter-electrode distance
was 27 mm. This axial emission distribution is very typical for particle-free
plasmas: Maxima in emission in the plasma sheath near both electrodes and
a minimum in emission in the centre [74, 75]. The somewhat higher emission
near the bottom powered electrode can be ascribed to the asymmetric design
of our reactor where the powered (bottom) electrode has a smaller area than
the grounded (upper electrode+chamber wall) electrode. Figure 3.3 shows the
optical emission profile for a plasma using the same deposition parameters
but at an applied power of 16 W. This plasma is in the (dusty) γ’-regime,
which is confirmed (not shown) by a shift in the impedance towards a more
resistive plasma [76]. The profile changes from two rather symmetric peaks at
the plasma sheaths and low bulk emission to a large peak in emission at the
sheath near the lower (powered) electrode and a smaller peak at the upper
(grounded) electrode, along with higher bulk emission. Pulled by gravity,
the dust particles accumulate near the bottom electrode, where gravity is
counteracted by the force that the negatively charged particles experience from
the potential drop near the electrode. The increase in optical emission in this
lower region of the plasma in the γ’-regime can be ascribed to the presence of
3.3. Dust formation and OES
37
dust particles. Because the dust particles act as electron traps, the electron
density decreases and therefore the energy per electron increases, which in
turn enhances the emission. The emission intensity is proportional to the rate
constant of emission by the relation ISi =KSiH4 Ne .NSiH4 , where Ne and NSiH4
are the electron concentration and silane concentration respectively and KSiH4
is the rate constant that depends on the electron temperature. The increase of
electron temperature in the γ’ regime therefore increases the optical emission
intensity, especially at the powered electrode.
For large monodisperse injected particles in a non-reactive plasma, this
effect is limited mainly to the region close to the bottom electrode [74]. Because
in our reactor the dust is grown rather than injected, we expect a variety in
size and mass and therefore a broader axial distribution of the dust. Some of
the lighter dust particles will be located near or in the plasma bulk, thereby
also enhancing the optical emission from the bulk of the plasma.
Because we can assume a Maxwell-Boltzmann distribution for the electron
energy for a low-pressure plasma [77] and because the electron temperature is
well below the minimum electron energy to excite hydrogen for emission in the
Balmer series, the Hα and Hβ emissions can be ascribed to excitation of hydrogen by electrons in the high energy tail of the electron energy distribution.
Because of the differences in excitation energies (16.0 eV for Hα and 16.6 eV
for Hβ [78]), the ratio of the intensities Hβ /Hα can be used as a qualitative
measure for the electron temperature [79].
Figure 3.4 shows the spatially resolved Hβ /Hα emission ratio for a plasma
in the α-regime and for a plasma in the γ’-regime at the two above mentioned deposition conditions. We observe an increase in electron temperature
throughout the plasma reactor in the γ 0 -plasma, which is most pronounced
in the bulk of the plasma. This again reveals the presence of trapped dust
particles in the plasma.
3.3.2
Dust formation as a function of power, hydrogen
dilution, and temperature
Whether dust is produced in a plasma depends on the conditions under which
the plasma is maintained. It has been shown before that increasing power density input, decreasing hydrogen dilution, increasing gas pressure or lowering the
substrate temperature can change a dust free plasma into a dust producing
plasma [80, 81, 82]. Using our optical method we have investigated the influence of applied VHF power, hydrogen dilution, and substrate temperature on
the formation of dust in hydrogen diluted silane plasmas. We used substrate
temperatures Ts of 100°C, 150°C, and 200°C, which were reached through res-
38
Chapter 3. The role of temperature in plasma dust formation
0 .7 0
γ’ - r e g i m e
α- r e g i m e
0 .6 5
0 .5 5
H
b
/H
a
0 .6 0
0 .5 0
0 .4 5
0 .4 0
0
5
1 0
1 5
2 0
2 5
V e r tic a l p o s itio n (m m )
3 0
Figure 3.4: The intensity ratio Hβ /Hα as an indication of electron temperature
as a function of position between the electrodes in the α-regime (circles) and
in the γ’-regime (squares). The position is measured from the upper grounded
electrode downwards.
3.3. Dust formation and OES
39
istive heating of the substrate holder. Although some heating of the substrate
due to power dissipation from the plasma is expected, measurements have
shown the additional heating to be less than 3°C for a 20 W plasma at typical
deposition conditions. Because all powers used in these experiments are well
below 20 W, the plasma heating does not have a significant contribution to the
substrate temperature.
For all plasmas we used a process pressure of 1.1 mbar, an excitation frequency of 60 MHz, a total P ranging from 5 to 20 W and hydrogen dilution
R ranging from 20 to 60. The hydrogen flow was kept constant at 100 sccm
and the hydrogen dilution was changed by adjusting the silane flow. The distance between the horizontal powered lower electrode and the upper grounded
electrode was 27 mm.
If we fix the plasma parameters; pressure, hydrogen dilution, and substrate
temperature, we can control the plasma regime by changing the applied power.
If we start a plasma in the α-regime and increase the applied power, eventually
dust particles will start to form and the plasma will transit to the γ’-regime
resulting in the described change in the axial optical emission profile.
Using our optical technique of analysing the asymmetry of OES emission
distribution, we mapped the transition from the α-regime to the γ’-regime as
a function of hydrogen dilution, applied power, and substrate temperature.
Figure 3.5 shows the results from these investigations. Apart from applied
power and hydrogen dilution, we clearly see that the transition depends on
substrate temperature, going into the dusty regime at higher hydrogen dilution
or lower power at lower substrate temperatures. This implies that depositing at
low substrate temperature limits the parameter space for dust-free deposition.
Also shown is the transition from amorphous to nanocrystalline growth, which
mainly depends on hydrogen dilution. Together, the two transitions define a
parameter window in which we can grow dust-free amorphous silicon, which is
very limited at low substrate temperatures. The parameter window for dustfree amorphous silicon growth at a substrate temperature of 100° is indicated
in grey in the figure. Similar windows can be identified for depositions at
higher substrate temperatures. As the deposition rate is directly related to
the power dissipated to the plasma, limitations on applied power will limit the
maximum achievable deposition rate.
3.3.3
TEM images of dust
In both regimes we deposited amorphous silicon layers to investigate the presence of dust in the deposited layers. The layers were deposited on a 1 µm thick
ZnO:Al(0.5%) layer on a glass substrate. The ZnO layer was removed by chem-
40
Chapter 3. The role of temperature in plasma dust formation
F o rw a rd P o w e r (W )
2 0
a -S i
1 5
i
n c -S
1 0
0
5
200°C
150°C
100°C
100°C
Dust free
a-Si
2 0
3 0
4 0
5 0
6 0
Hydrogen Dilution (H2/SiH4)
7 0
Figure 3.5: Map of the transition from the α-regime to the γ’-regime as a
function of hydrogen dilution and applied power at substrate temperatures
of 200°C (squares), 150°C (circles) and 100°C (triangles). The amorphous
to nanocrystalline transition also depends on applied power and hydrogen
dilution. The parameter window for dust free amorphous silicon growth at
100°C is shown in grey.
Figure 3.6: TEM image of an a-Si layer deposited in the α-regime (a) and of
an a-Si layer deposited in the γ’-regime (b,c), as identified by analysis of the
optical emission profile. Image (c) shows a part of image (b) in more detail.
3.3. Dust formation and OES
41
ical etching in a 1.5% HCl solution for 1 hour, which leaves flakes of silicon
floating in the solution. This solution was filtered and afterwards washed with
ethanol. Finally, a transmission electron microscopy (TEM) grid was used to
scoop flakes from the ethanol and used as a sample in the TEM microscope.
Figure 3.6 shows TEM images of a layer deposited in the α-regime (a) and a
layer deposited in the γ’-regime (b and c). In the layer deposited in the γ’regime we observe particles with sizes ranging from several tens of nanometres
up to micrometers, whereas in the layers deposited in the α-regime we observe
no particles in the layers. The transition from the α-regime to the γ’-regime
was induced by a small increase in the applied power into the plasma, whereas
all the other plasma parameters were kept constant.
3.3.4
OES of pulsed Plasmas
A known method to suppress powder (dust) formation is the use of amplitude
modulated plasmas [83, 84, 85]. In contrast to continuous wave (CW) plasmas,
the VHF input signal is modulated by a square wave [86]. The behaviour
of modulated plasmas is quite different from CW driven plasmas, which is
manifested in a change in deposition rate and material properties and a change
in optical emission from the plasma. In this study, we investigate how the
optical emission from the plasma is influenced when a plasma that is in the
γ 0 -regime is influenced when the power input is pulsed. For this purpose
we chose the following plasma parameters: Gas flows of 35 sccm SiH4 and
175 sccm H2 , a time-average applied power 10 W, a VHF frequency of 60 MHz,
an electrode distance d of 27 mm, a pressure p of 0.6 mbar and a substrate
temperature of 130°C. The duty cycle was 50% in all cases. The pressure
was adjusted in such a way that, in a continuous wave plasma, the plasma
just crossed the boundary to the powder forming γ’-regime. For this plasma
we pulsed the power supply at different frequencies: 50 Hz, 500 Hz, 1 kHz,
10 kHz and 100 kHz. We recorded the axial emissions of these plasmas using
the method described in section 3.3.1. Figure 3.7 shows the axial profiles
under these conditions for the SiH* line emission. In the CW case, it shows a
typical emission profile from a γ’-plasma, showing an asymmetric profile and
a large contribution from the bulk. We do not observe a dip in the emission
from between the sheaths, which is due to the lower pressure compared to
the previous measurements. When the plasma is pulsed at 50 Hz, 500 Hz, or
1 kHz, the emission drops drastically over the whole profile, whereas the shape
changes into a typical profile from a plasma in the α-regime, with articulated
emissions from the plasma sheaths and a low bulk emission. Towards higher
modulation frequencies, the overall intensity increases and the bulk intensity
42
Chapter 3. The role of temperature in plasma dust formation
C W
P u
P u
P u
P u
P u
S iH * In te n s ity (a .u .)
2 5
2 0
ls e
ls e
ls e
ls e
ls e
d 5 0 H z
d 5 0 0 H z
d 1 k H z
d 1 0 k H z
d 1 0 0 k H z
1 5
1 0
0
5
0
5
1 0
1 5
2 0
2 5
V e r tic a l P o s itio n (m m )
3 0
Figure 3.7: Vertical optical emission profile of the SiH* line of continuous wave
and pulsed plasmas. The CW wave plasma was in the γ’-regime. Pulsing the
plasma changes the optical emission. Deposition on the window was minimized
by closing the shutter in between measurements.
also increases. Other studies have shown that layers grown in pulsed plasmas
show large amounts of small particles and that their size can be controlled
by changing the duty cycle of the modulation [87, 88, 89]. A longer duty
cycle will increase the size of the particles. If during the plasma off-period the
electric field within the plasma collapses, the particles can escape the plasma
zone if their typical diffusion time is shorter than the plasma off time. The
diffusion time depends on a number of plasma properties such as pressure
and temperature, and on the reactor geometry. Judging from figure 3.7, we
estimate the typical diffusion time to be between 0.05 and 0.5 ms in our reactor
under these specific plasma conditions.
3.4
3.4.1
Mass spectrometry
Clusters, the precursors of dust formation
It has been shown in earlier reports that in a silane plasma the energy distribution of ions that reach the substrate depends on the substrate temperature
3.4. Mass spectrometry
43
[38]. It is believed that the transfer of energy to the substrate by energetic
ions plays an important role in obtaining good quality materials [90]. This
hypothesis motivated us to study the substrate temperature dependence of
the ion energy distribution function (IEDF) of a silane/hydrogen plasma as a
function of substrate temperature. Using a Hiden EQP 300 energy-selective
mass resolved spectrometer we measured the IEDF of several species reaching
the growing surface, through a 30 µm sampling orifice fixed at the grounded
electrode, where in a deposition chamber the growing surface would be located.
For a fixed gas flow ratio ΦH2 /ΦH2 of 5, an applied VHF power of 10 W and
an excitation frequency of 50 MHz we measured the IEDFs of the Sin H+
2n+1
(n = 1. . . 5) and of H+
2 ions as a function of substrate temperature at different reaction chamber pressures from 0.05 to 0.25 mbar. This parameter space
corresponds to conditions we use for depositing silicon films for amorphous
silicon solar cells. At these process settings, the plasma is in the dust free
α -regime when the temperature is at the optimum deposition condition of
around 200 °C, but at low temperature and high pressure settings it is close
to the transition to the dusty γ’-regime. Because larger polysilane ions play
an important role in the initial phase of dust formation, the detection of these
ions can contribute to the understanding of the temperature dependence of
the regime change from the α to the γ’-regime [64, 82].
3.4.2
Ion energies
A summary of our results for the ion energies of SiH+
3 (mass = 31 amu) is
presented in figure 3.8 as a function of substrate temperature at different
pressures. In the pressure range studied, the ion energies for all measured
species show an increasing trend with increasing substrate temperature. This
trend is also observed for all the other measured ions in this study, including
H+
2 , which can be ascribed to a longer mean free path at a higher substrate
temperature, by complying to the ideal gas law: When the pressure is kept
constant, the gas density decreases when the gas temperature is increased.
3.4.3
Cluster formation and temperature
Figures 3.9 shows the count rate for SiH+
3 at the peak ion energy position as
a function of substrate temperature at different chamber pressures. Figure
+
3.10 shows the equivalent for Si4 H+
9 ions. The SiH3 ions’ count rate shows
an increasing trend with increasing temperature, whereas Si4 H+
9 shows an initial increase in maximum count rate with increasing temperature region to
reach a maximum at low substrate temperature and thereafter, a decrease in
44
Chapter 3. The role of temperature in plasma dust formation
2 5
P e a k Io n E n e r g y (e V )
S iH
+
3
2 0
1 5
0
p (m b a r)
0 .1 0
0 .2 0
0 .0 5
0 .1 5
0 .2 5
5 0
1 0 0
1 5 0
2 0 0
2 5 0
Substrate Temperature (°C)
Figure 3.8: The peak ion energy for SiH+
3 ions in a hydrogen diluted silane
plasma as a function of substrate temperature at various pressures. The ion
energy rises with increasing substrate temperature and decreases with increasing pressure.
3.4. Mass spectrometry
C o u n t r a te (C /s )
1 0 0 0 0 0
S iH
+
3
p (m b a r)
0 .1 0
0 .2 0
45
0 .0 5
0 .1 5
0 .2 5
8 0 0 0 0
6 0 0 0 0
4 0 0 0 0
2 0 0 0 0
0
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
Substrate Temperature (°C)
Figure 3.9: The count rate of SiH+
3 ions in a hydrogen diluted silane plasma
at the peak ion energy as a function of substrate temperature at different
pressures.
count rate at increasing temperature. For Si5 H+
11 ions (not shown) the count
rates decrease monotonically with increasing temperature for all pressures. We
speculate that the increase of the ion energies with substrate temperature corresponds to the larger mean free path at higher gas temperatures. Because
of this, the positive ions have a higher probability of travelling through the
plasma sheath near the grounded electrode without colliding; their energies
will be higher and so will be their count rate. Decreasing the pressure will
increase the mean free path of the clusters and will therefore have a similar
effect. Due to the increase in mean free path, the ion energy and the count
rate for low-mass ions increases, as observed for silyl ions. It is known that
a higher gas temperature reduces the polymerization rate of silyl ions into
larger polysilanes [64]. The increase in the number of positive ions through
the plasma sheath is counteracted by the decrease in polymerization rate, as
observed at higher temperatures. Figure 3.11 again shows count rates as a
function of pressure and temperature, but now for Si3 H+
7 ions. We observe
two opposing trends: At low pressures (0.05 and 0.10 mbar) the increase in
substrate temperature suppresses the formation of clusters, whereas at higher
pressures (0.15, 0.20, and 0.25 mbar), due to the larger number of collisions
46
Chapter 3. The role of temperature in plasma dust formation
1 5 0 0
C o u n t r a te (C /s )
S i4 H
1 0 0 0
+
9
p (m b a r)
0 .1 0
0 .2 0
0 .0 5
0 .1 5
0 .2 5
5 0 0
0
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
Substrate Temperature (°C)
Figure 3.10: The count rate of Si4 H+
9 ions in a hydrogen diluted silane plasma
at the peak energy as a function of substrate temperature at different pressures.
at elevated pressures, the trend towards higher substrate temperature is still
dominated by the increase in mean free path. It should be noted that we are
only able to measure positive ions. Because positively charged particles are
easily ejected from the plasma due to the plasma potential profile, the clusters
under investigation formed through two possible routes can be detected: if
the polymerization occurs fast, i.e. within the typical time needed for a positive particle to be ejected from the plasma, the cluster can spawn through
+
polymerization of positive particles by insertion of SiH2 into Sin Hm
ions,
although this reaction is believed to stop above n = 6 or 7 [67]. The other positive cluster generation scheme is through the positive charging of negatively
charged clusters, which can happen through the collision with a positively
charged ion (anion-cation neutralization [67]), making it a neutral particle,
+
followed by a discharge reaction (Sin H2n+2 + e− → Sin H2n+1
+ H + 2e− ),
ionizing the particle to a positively charged ion. Because the polymerization
of negative polysilanes occurs faster than the polymerization of positive chains
[91], the latter is the most probable route towards positively charged higher
silanes. Our results, showing a decrease in count rate for higher mass ions with
increasing temperature, confirms a negative influence of the gas temperature
3.4. Mass spectrometry
S i3 H
C o u n t r a te (C /s )
4 0 0 0
+
7
p (m b a r)
0 .1 0
0 .2 0
47
0 .0 5
0 .1 5
0 .2 5
3 0 0 0
2 0 0 0
1 0 0 0
0
5 0
1 0 0
1 5 0
2 0 0
2 5 0
Substrate Temperature (°C)
Figure 3.11: The count rate of Si3 H+
7 ions in a hydrogen diluted silane plasma
at the peak energy as a function of substrate temperature at different pressures.
The temperature dependence of the cluster growth changes with pressure.
on cluster formation. This confirms also the hypothesis that a higher substrate
temperature will prevent the plasma from going into the dusty γ’-regime by
suppressing the polymerization reactions of silyl into larger polysilanes.
3.4.4
Conclusions
We have presented a non-invasive in-situ technique to determine whether a
deposition plasma is in the dust free α-regime or in the dusty γ’-regime by
recording a spatially resolved optical emission profile perpendicular to the
electrodes. In the γ‘-regime we observe an increase in electron temperature in
the bulk, which indirectly confirms the presence of dust particles. We mapped
the transition from the α- to the γ’-regime as a function of hydrogen dilution, applied power and, substrate temperature. This method can generally
be used to determine the processing window in which dust free silicon films
can be deposited. Because optical emission can also be used to predict the
material phase, this study shows that a single technique can be used as an
in − situ plasma diagnosis tool for monitoring of the amorphous to nanocrystalline transition as well as the transition from the dust-free to the dusty
48
Chapter 3. The role of temperature in plasma dust formation
regime without a supplement technique. Therefore it is a plasma monitoring
tool to control the complete thin film silicon processing of ’micromorph’ cell
manufacturing.
Dust formation can be suppressed by using a pulsed power input. Changing
the modulation frequency and thus changing the plasma-off time determines
whether dust formation is suppressed.
We found a dependence of the initial growth of positively charged clusters
on substrate temperature, which explains the temperature dependence of the
α to γ’ transition.
Chapter 4
Low temperature silicon
layers
4.1
The role of substrate temperature in PECVD
The electronic quality of thin amorphous silicon (a-Si:H) and nanocrystalline
silicon (nc-Si:H) films is directly influenced by their deposition temperature.
For both materials a minimum in defect density (dangling bonds) is observed
for layers grown at a substrate temperature of around 200°C to 250°C [36].
Dangling bonds are unoccupied silicon bonds and these can act as recombination sites for charge carriers, but can be passivated by the attachment of a
hydrogen atom. Deposition at lower than standard substrate temperatures
will induce a higher defect density as well as a higher porosity and thus lower
refractive index [92, 93], and therefore, lead to lower solar cell efficiencies.
An a-Si:H deposition at standard temperature (~200°C to 250°) goes through
thermal equilibrium [94], which allows for minimisation of defects and optimum
hydrogen content for dangling bond passivation. In this thermal equilibrium,
there is a balance between strained intersilicon bonds (Si—H—Si) and Si—H
bonds: Si—H + Si— —Si ⇐⇒ Si—+ Si—H—Si [95]. Below standard temperature, which is associated with the glass transition temperature of a-Si:H,
the structure is ’frozen’, and the hydrogen cannot move around to passivate
defect sites. In this non-equilibrium regime, reactions in the growth zone of
the growing film are important. Passivation occurs when growth precursors
(mainly SiH3 [36]) move across the growing surface. Therefore the defect density is related to the diffusion length of the precursors, which is related to the
50
Chapter 4. Low temperature silicon layers
substrate temperature [96].
The surface mobility and therefore the material quality can be improved
by adding hydrogen to the feedstock gasses, so-called hydrogen dilution. By
doing this, even at a low substrate temperatures of 100°C or 75°C working
devices can be fabricated [97, 98]. Although the quality of low temperature
silicon layers can be improved by increasing the hydrogen dilution, doing so
will generally decrease the deposition rate .
It is generally accepted that energy transported to the growing surface
by ion bombardment in the plasma enhanced chemical vapour deposition
(PECVD) process can contribute to a higher surface mobility of these molecules. The kinetic energy flux carried by the energetic ions can be varied
by changing the plasma properties, such as the applied power density, the gas
pressure or the hydrogen dilution. It has also been shown that the ion energy flux towards the substrate decreases when the substrate temperature is
decreased, which in turn can be increased by adding more hydrogen. It has
been shown that, going from a deposition at 200°C to a deposition at 39°C,
the energy flux towards the growing surface can be restored to the original
level when the hydrogen dilution is increased [38].
Changing the substrate temperature also changes the optical properties of
the silicon. Figure 4.1 depicts the absorption coefficient and the refractive index as a function of wavelength of typical a-Si:H layers deposited at substrate
temperatures of 180°C and 130°C as measured by reflection-transmission measurements (R-T) and fitted to the OJL-model [55], showing a lower refractive
index over the whole spectrum and a lower absorption coefficient for the layer
deposited at low substrate temperature. These optical properties clearly indicate a higher bandgap for a-Si:H deposited at lower temperatures, which can
be attributed to a higher hydrogen content in the layers [92].
This chapter will cover our search for device quality intrinsic a-Si:H and
nc-Si:H layers, as well as p-type and n-type doped layers to be used for depositions directly onto polycarbonate (PC), which limits the substrate temperature to around 130-140°C, due to the glass transition temperature of PC
[34].
4.2
Controlling the substrate temperature
When depositing on plastic substrates, it is very important to accurately monitor and control the substrate temperature. A temperature higher than the
glass transition temperature (Tg ) of the substrate will result in deformation
or even melting of the substrate. Furthermore, in the range from 100°C to
4.2. Controlling the substrate temperature
)
a-Si (180°C)
a-Si (130°C)
ZnO:Al
-1
A b s . C o e ff. (c m
R e fr . In d e x
51
1 0
5
1 0
3
1 0
1
4
3
2
4 0 0
5 0 0
6 0 0
7 0 0
W a v e le n g th (n m )
8 0 0
Figure 4.1: Refractive index and absorption coefficient of a-Si:H deposited at
130°C and at 180°C and of sputter deposited ZnO:Al.
160°C, the silicon material quality is very sensitive to small variations in temperature [36] and therefore it is important to establish a controlled and stable
temperature.
4.2.1
Substrate stretch holder
When a flexible substrate is fixed to a rigid substrate holder and the thermal
expansion coefficient of the substrate material is higher than that of the substrate holder, the substrate will curve when it expands due to thermal expansion during deposition. This is the case for everyday plastics, which have
thermal expansion coefficients much higher than of titanium, which is the material our substrate holders are made of. This curving will cause a gap between
the substrate and the substrate holder, which will result in a suboptimal and
an inhomogeneous heat transfer from the substrate holder to the substrate,
resulting in a inhomogeneous and lower quality layer. Furthermore, because
introducing a gap between the substrate and substrate holder locally changes
the electrical properties of the plasma, a change in deposition rate occurs [99].
Therefore curving of the substrate should be avoided. To do this, we used
a specially designed substrate holder that can stretch the substrate when it
52
Chapter 4. Low temperature silicon layers
Figure 4.2: The substrate stretch holder, which is used to maintain a good
contact between the substrate holder and the substrate and to avoid curving
of the substrate. The maximum size of the substrate is 10 ×10 cm2 .
expands and thereby keeping it flat. Figure 4.2 shows a photograph of this
stretch holder, with in the inset a detailed image of one of the springs pulling
on the substrate. This configuration resembles the situation for a roll-to-roll
process, in which also the substrate is kept under tension during the deposition
[100]. Due to the design, there is always a small gap between the substrate and
the holder. The part of the stretch holder which holds the substrate is made
of stainless steel, whereas the normal substrate holders are completely made
of titanium. The problem of curving substrates is only relevant in the case of
batch-type processing, whereas in roll-to-roll processing we do not expect this
problem.
4.2.2
Gas pressure
To calibrate the actual temperature of the substrate to the accurately controlled heater temperature, we mounted both a glass substrate and a PC substrate to a regular and to the stretch substrate holder. To both substrates
we attached K-type (chromel-alumel) thermocouples in the centre and at 1 cm
from the edge of the substrate. To simulate deposition conditions we introduced 0.16 mbar or 5 mbar argon gas into the reactor and ignited a plasma
running at 5 or 20 Watts to mimic a-Si:H and nc-Si:H deposition conditions,
4.2. Controlling the substrate temperature
G la
G la
G la
P C
P C
5 /7
Substrate Temperature (°C)
2 0 0
1 7 5
s s ,
s s ,
s s ,
, 0 .1
, 5 m
r u le
v a c
0 .1
5 m
6 m
b a
53
u u m
6 m b a r A r
b a r A r
b a r A r
r A r
1 5 0
1 2 5
1 0 0
7 5
5 0
1 0 0
1 2 5
1 5 0
1 7 5
2 0 0
Heater Temperature (°C)
Figure 4.3: The substrate temperature measured on glass substrates in a regular substrate holder and on PC mounted on the stretch substrate holder,
as a function of set heater temperature at two different argon gas pressures,
corresponding to a-Si:H and nc-Si:H deposition pressure conditions.
respectively. For these configurations we measured the relation of the substrate
temperature to the heater temperature in the IRIS deposition chamber. As a
rule of thumb, we normally estimate the substrate temperature to be 5/7 of
the set heater temperature. The measurements were performed in a separate
deposition setup called IRIS, which uses an excitation frequency of 50 MHz,
whereas intrinsic layers in the ASTER system are deposited at 60 MHz and the
powered electrode is not a showerhead electrode, in contrast to the ASTER
setup. Otherwise the IRIS reactor chamber is a near-exact copy of one of the
ASTER deposition chambers.
Because there can only be a limited number of contact points between the
heater and the substrate holder and between the substrate holder and the substrate, most of the heat transfer (in high vacuum conditions) is of a radiative
nature and therefore the size of the gap between the heater and the substrate
holder and between the substrate holder and the substrate is important for
the heating of the substrate. Introducing gas into the reactor will induce
54
Chapter 4. Low temperature silicon layers
convective heat transfer, causing a higher substrate temperature at a fixed
heater temperature. At the end the substrate temperature will depend on the
properties of the substrate material, i.e. thermal conductivity and emissivity.
Figure 4.3 shows the temperature calibration results for a glass substrate in
a regular substrate holder and a PC substrate in the stretch holder. Measurements were performed under vacuum condition and in a 0.16 mbar and a
5 mbar argon environment with no plasma running. The same pressures are
used for a-Si:H or nc-Si:H depositions, respectively. Going from vacuum conditions to a low pressure condition such as 0.16 mbar of argon (although a
mixture of silane and hydrogen is used in silicon film depositions), the substrate temperature increases by a few degrees. A small difference in substrate
temperature is observed between the glass substrate in a regular substrate
holder and a PC substrate in the stretch holder, which could be attributed
to a small gap between the holder and substrate due to the holder design, or
due to the fact that the stretch part of the holder is made from stainless steel,
whereas the regular holder is entirely made of titanium. Titanium has a much
higher thermal conductivity coefficient (21.9 Wm−1 K−1 [101]) than stainless
steel (12-14 Wm−1 K−1 [102]).
4.2.3
Plasma heating
Running a plasma inside the reactor will raise the substrate temperature.
Coupling power into the plasma will induce heating of the substrate [103]. Energy from the plasma is transferred to the substrate through ion bombardment.
Figure 4.4 shows temperature calibration measurements of a glass substrate
and a PC substrate, both mounted in the stretch substrate holder, under two
different plasma conditions. Before the measurements we waited for the temperature to stabilize at a given pressure. The top graph shows the heating
of the substrate as a function of time under our most-used low temperature
a-Si:H deposition conditions, whereas the bottom graph shows the same for
low temperature deposition conditions for device quality nc-Si:H. The plasma
settings are listed in table 4.1. For both plasma regimes, the substrate heats
up during the deposition and the glass substrate temperature is always 2-3°C
higher than the PC substrate temperature. In both regimes, during the first
few minutes there is a rapid increase in substrate temperature, which gradually
changes into a linear temperature increase of around 1°C in 3.5 minutes. The
linear part of the heating is probably due to heating of the complete reactor
chamber. During a typical 300 nm a-Si:H i-layer deposition ( 25 minutes) the
heating due to the plasma is 7-8°C. During a 1000 nm nc-Si:H i-layer deposition
( 30 minutes) the plasma heating is estimated to be around 11°C.
Substrate Temperature (°C)
4.2. Controlling the substrate temperature
1 3 0
1 2 5
5 .0 W a tt
0 .1 6 m b a r
1 2 0
1 1 5
1 4 5
1 4 0
G la s s
P C
0
5
1 0
1 5
2 0
2 5
1 7 .5 W a tt
3 .0 m b a r
1 3 5
1 3 0
55
G la s s
P C
0
2
4
6
8
P la s m a -o n tim e (m in u te s )
1 0
Figure 4.4: The substrate temperature measured on a glass substrate and on a
PC substrate, both mounted on the stretch substrate holder, as a function of
plasma-on time. The plasma conditions are that of a standard low temperature
a-Si:H growing plasma and a nc-Si:H growing plasma.
Plasma
a-Si:H
nc-Si:H
p
P
ΦSiH4
ΦH 2
d
Th
mbar
W
sccm
sccm
mm
°C
0.16
3.0
5.0
17.5
35
5
175
100
27
10
180
170
p: pressure; P: applied plasma power; Φ: gas flow; d: interelectrode distance; Th : heater temperature
Table 4.1: Plasma properties as used for temperature calibrations in IRIS. The
calibration results are shown in figure 4.4.
Chapter 4. Low temperature silicon layers
6 x 1 0
5 x 1 0
4 x 1 0
1 7
3 x 1 0
1 7
2 x 1 0
1 7
1 0
1 7
7 0
1 7
1 7
6 5
6 0
D e fe c t D e n s ity
U rb a c h E n e rg y
5 5
U rb a c h E n e rg y (m e V )
D e fe c t D e n s ity (c m
-3
)
56
5 0
1 0 0
1 1 0
1 2 0
1 3 0
1 4 0
1 5 0
Substrate Temperature (°C)
Figure 4.5: Midgap defect density and Urbach energy as a function of substrate
temperature for a-Si:H layers, derived from CPM measurements.
4.3
4.3.1
Low temperature intrinsic layers
a-Si:H intrinsic layers
Temperature series
A series of a-Si:H layers was deposited on glass substrates with substrate temperatures ranging from 100°C to 145°C to investigate the influence of the
substrate temperature on the layer quality within our temperature range of
interest. The deposition parameters were: ΦSiH4 : 35 sccm, ΦH2 : 175 sccm,
p: 0.16 mbar, P: 5 W, d: 27 mm. Layer thicknesses were around 600 nm. The
temperature was varied by changing the controlled heater temperature and
applying the 5/7-rule. No large differences are found in band gap or light and
dark conductivities. However, we do find a clear trend in midgap defect density and Urbach energy (Eu ), both derived from constant-photocurrent method
measurements (CPM). The results from this study are shown in figure 4.5. Going from low to high temperature the defect density drops from over 5×1017
to 8×1016 cm−3 , whereas the Urbach tail energy decreases from 69 to 58 meV.
4.3. Low temperature intrinsic layers
8 x 1 0
1 7
6 x 1 0
1 7
7 5
0 .2 2
0 .2 0
R *
-3
D e fe c t D e n s ity (c m
0 .1 8
7 0
0 .1 6
0 .1 4
4 x 1 0
1 7
0 .1 2
1 0 0
1 2 5
1 5 0
1 7 5
2 0 0
2 2 5
6 5
H y d r o g e n F lo w ( s c c m )
2 x 1 0
6 0
1 7
D e fe c t D e n s ity
U rb a c h E n e rg y
1 0 0
5 5
U rb a c h E n e rg y (m e V )
1 8
)
1 0
57
1 2 5
1 5 0
1 7 5
2 0 0
H y d r o g e n F lo w (s c c m )
Figure 4.6: Defect density and Urbach energy as a function of hydrogen dilution, derived from CPM measurements for a-Si:H layers deposited at a substrate temperature of 130°C. The inset shows the microstructure factor, as
obtained from FTIR measurements.
Hydrogen dilution series
To achieve optimal layer quality for device performance, a hydrogen dilution
series was performed, using a silane flow of 35 sccm, and a varying hydrogen
gas flows between 105 and 200 sccm. The other deposition parameters were:
p: 0.16 mbar, P: 5 W, d: 27 mm, Ts : 130°C. The results for the midgap defect
density and Urbach energy, obtained from CPM measurements and the microstructure parameter R*, obtained from fourier transfer infrared (FTIR) data
are presented in figure 4.6. It shows an increasing layer quality trend toward
higher dilutions, expressed in a lower defect density, which changes almost an
order of magnitude while increasing the hydrogen to silane ratio from 3 to
around 6. Also the Urbach energy changes from more than 70 meV at the low
dilution end to 56 meV for the layers grown at the highest dilution. Because
also the deposition rate changes only a few percent, layers grown at high dilution would be most appropriate for device production. However, the highest
dilution ratio layers suffer from high compressive stress, resulting in peeling
of the layers from the substrate. It is known that a rise in compressive stress
58
Chapter 4. Low temperature silicon layers
occurs just before the transition from a-Si:H to nc-Si:H, which is attributed to
a higher hydrogen content in the layers, which partially is molecular hydrogen
trapped in microvoids that contributes to macroscopic stress [104].
4.3.2
nc-Si:H intrinsic layers
In the nc-Si:H growth regime, we performed a hydrogen dilution series and
a plasma-power series to find an optimal recipe for depositing intrinsic ncSi:H for solar cells, both in single-junction cells as well as for the bottom
cells of a-Si:H/nc-Si:H tandem cells. Because our aim is to deposit these
layers on polycarbonate substrates, the substrate temperature of these layers
is kept at 130°. The key physical parameter for these layers is the crystalline
fraction, which can be determined by Raman-spectroscopy. The procedure
for this is described in section 2.2.3. Silicon films grown near the transition
regime, consisting of a Raman crystalline ratio from 0.3 to 0.5 are referred to
as mixed-phase or transition-type silicon and are the preferred materials for
nc-Si:H cells and a-Si:H/nc-Si:H solar cells [105, 106]. Because these layers are
grown under plasma conditions for which the nucleation rate of crystallites is
low, small changes in the plasma or in the substrate material composition or
morphology can change the crystalline fraction, making it difficult for these
layers to reproduce. The use of a crystalline seed layer will solve this issue,
because nucleation has already taken place. Furthermore, if these layers are
grown in a p-i-n solar cell structure, they will be deposited on a nc-Si:H p-type
layer, which acts as a seed layer. Using such a p-layer as a seed layer would
render it useless for conductivity measurements of the layers. To mimic the
layer growth as if it was part of a solar cell structure, these layers were grown
on a thin (∼10 nm) nc-Si:H seed layer, which was deposited from a nc-Si:H
p-layer recipe, but without adding the trimethylborane to the gas mixture,
which is used to dope the material p-type. This is done so that we are still
able to do conductivity measurements. If we were to use a doped nc-Si:H seed
layer, the conductivity would be dominated by the conductivity of the p-layer.
Dilution series
Figure 4.7 shows the crystalline fraction and deposition rate of layers grown
at different hydrogen dilutions (while keeping the H2 flow fixed) around the
transition from a-Si:H to nc-Si:H growth. As expected, we observe a transition
from a-Si:H to nc-Si:H when the silane concentration in the gas feedstock mixture is increased. To evaluate whether the crystalline fraction is homogeneous
throughout the layer in the growth direction, the Raman spectra were meas-
4.3. Low temperature intrinsic layers
C r y s ta llin e R a tio
0 .8
C r y s ta llin e fr a c tio n
L a y e r S id e
G la s s S id e
D e p o s itio n R a te
0 .9
0 .8
0 .6
0 .7
0 .4
0 .6
0 .2
0 .5
0 .0
4
5
S ila n e F lo w
6
(s c c m )
D e p o s itio n R a te (n m /s )
1 .0
59
7
Figure 4.7: Crystalline fraction and deposition rate of Si layers deposited at
130°C on a nc-Si:H seed layer as a function of silane content of the plasma
feedstock mixture, measured by R-T and Raman-spectroscopy.
ured both from the top of the layer and through the Corning glass substrate.
This study reveals a comparable crystal fraction on both sides, indicating homogeneous growth in the growth direction. Although the measured crystal
fraction from the substrate side is influenced by the presence of the nc-Si:H
seed layer, the nc-Si:H seed layer only has a minor influence on the measured
crystalline fraction, because the penetration depth for Raman measurements is
around 100 nm for nc-Si:H, when using laser light with a wavelength of 514 nm
[107], whereas the seed layer is only 20 nm. The deposition rate decreases with
increasing hydrogen dilution. The hydrogen flow was 100 sccm, while the other
plasma parameters were p: 3.0 mbar, P: 20 W, d: 10 mm, Ts : 130°C. The layer
thickness was around 800 nm.
The desired Raman crystalline ratio is found at a silane flow of around
5 sccm. This layer is used for further optimalisation of the intrinsic nc-Si:H
layers for use in a-Si:H/nc-Si:H tandem structures. This is done by varying
the applied power into the plasma.
60
Chapter 4. Low temperature silicon layers
1 .0
D e p o s itio n r a te
C r y s ta llin e R a tio
0 .8
0 .6 0
0 .5 5
0 .6
0 .5 0
0 .4
C r y s ta llin e fr a c tio n
L a y e r S id e
G la s s S id e
0 .4 5
0 .4 0
0 .2
0 .3 5
0 .0
1 0
1 5
2 0
2 5
P o w e r (W )
D e p o s itio n R a te (n m /s )
0 .6 5
3 0
Figure 4.8: Crystalline fraction and deposition rate, measured by R-T and
Raman-spectroscopy, of Si layers deposited at 130°C on a nc-Si:H seed layer
as a function of applied power into the plasma.
Power series
The nc-Si:H layers developed from the hydrogen dilution series was used as
a basis for fine-tuning of the crystal fraction by changing the applied power
into the plasma. The power input was varied between 10 and 30 W in 5 W
increments. The gas flows were: silane: 5 sccm, hydrogen: 100 sccm. The
other plasma parameters were p: 3.0 mbar, d: 10 mm, Ts : 130°C. The layer
thickness was around 800 nm. The results for the Raman crystalline ratio,
measured both on the layer top surface and through the glass substrate as
well as the deposition rate are presented in figure 4.8. In this regime a transition from completely amorphous to a Raman-crystalline fraction of 80% takes
place. Between 15 W and 20 W, the crystal fraction changes from ∼25% to
∼50%, which is in the desired mixed-phase regime. For powers up to 20 W, the
deposition rate increases with increased power input but saturates at higher
applied powers, indicating silane depletion conditions. The layers deposited in
this power series were chosen as the i-layer material for nc-Si:H single junction
and a-Si:H/nc-Si:H tandem solar cells, as described in chapter 6.
4.4. Low temperature doped layers
Description
Ts
ΦSiH4
ΦH2
ΦCH4
ΦT M B
ΦP H3
P
p
°C
sccm
sccm
sccm
sccm
sccm
W
mbar
a-Si:H p
130
35
124
-
53
-
5
0.16
a-SiC:H p
200
30
-
40
18
-
5
0.15
a-Si:H n
130
30
23
-
-
7.5
5
0.5
a-Si:H n
200
30
23
-
7.5
5
0.5
nc-Si:H p
130
1
240
-
0.25
-
15
1.1
nc-Si:H n
130
1.2
180
-
-
0.27
15
1.1
61
Table 4.2: Gas flows, applied power and pressure used for plasma deposition
of doped layers at substrate temperatures of 130°C and 200°C. The measured
properties of these layers are summarized in table 4.3. TMB and PH3 are
diluted in H2 gas at a concentration of 2 at. %.
Nanocrystalline layers on PC
Because the nucleation of crystallites can be sensitive to the substrate material,
investigations have been performed on the crystal growth on PC substrates.
The results are presented in chapter 6.
4.4
Low temperature doped layers
Doped layers were deposited at a substrate temperature of 130°C by adapting
the substrate temperature for known optimised recipes at higher (standard)
temperatures or from optimized layers developed previously for solar cells at
100°C for deposition in the ASTER deposition system [100], and characterized. For this purpose p-type a-Si:H, n-type a-Si:H, p-type nc-Si:H and n-type
nc-Si:H were deposited and characterized. Table 4.2 shows the gas flows, applied power and reactor gas pressure of these depositions. All doped layers
were deposited at a plasma frequency of 50 MHz, an interelectrode distance of
27 mm and a substrate temperature of 130°C. Table 4.3 shows the measured
optical and electrical properties of the deposited layers, which were used for
low temperature silicon thin film depositions as used for single junction a-Si:H
and nc-Si:H cells and a-Si:H/nc-Si:H tandem solar cells.
The a-Si:H doped layers at 130°C show electrical properties very similar to
a-Si:H doped layers deposited at 200°C in our lab. Optically, the band gaps
of a-SiC:H at high temperature are higher due to the incorporation of carbon
in the network. For the nc-Si:H doped layers we have no information available
on the standard temperature (200°C) counterparts to compare with.
62
Chapter 4. Low temperature silicon layers
Description
Ts
rd
Eg (Tauc)
Eg (cubic)
E04
Ea
σd
Rc
°C
nm/s
eV
eV
eV
eV
(Ωcm)−1
%
-
a-Si:H p
130
0.244
2.04
1.69
2.03
0.41
2.3×10−6
a-SiC:H p
200
0.188
2.10
1.75
2.09
0.45
7.7×10−7
-
a-Si:H n
130
0.052
1.82
1.46
1.85
0.23
3.0 ×10−3
-
a-Si:H n
200
0.08
1.93
1.57
n/a
0.20
9.8×10−3
-
nc-Si:H p
130
0.052
2.09
1.75
1.95
0.05
7.8×10−1
57
nc-Si:H n
130
0.036
2.35
2.13
2.45
0.06
8.8×10−2
71
Ts : substrate temperature; rd : deposition rate; Eg : band gap; Ea : activation energy; σd : dark
conductivity; Rc : Raman crystalline ratio
Table 4.3: Optical and electrical properties of doped silicon thin films deposited
at a substrate temperatures of 130°C and 200°C. The deposition parameters
of these layers are given in table 4.2.
The insertion of a thin (5 to 10 nm) nc-Si:H p-layer between the ZnO:Al
TCO and the a-Si:H p-layer improves the contact between the (n-type) ZnO:Al
TCO and the p-type silicon layers [108], resulting in a higher Voc and FF. Because this layer is extremely thin, nucleation of crystallites needs to occur
within the first few nanometres of layer growth. To achieve this, we aim to develop thin, but highly crystalline nc-Si:H p-type doped layers. For this purpose
we performed a hydrogen dilution series, using hydrogen flows ranging from
150 to 270 sccm. The other plasma parameters were identical to the parameters listed in table 4.2 for the p-type nc-Si:H deposition. The Raman crystalline
ratio and deposition ratio for these layers are presented in figure 4.9. We find
an increasing Raman crystalline ratio with increasing hydrogen dilution. All
layers in this series show a similar activation energy of around 0.05 eV. The
dark conductivity at 300 K of the layers ranges from 3.1 ×10−1 (Ω cm)−1 at a
hydrogen flow of 150 sccm to 7.8 ×10−1 (Ω cm)−1 at 240 sccm H2 . The layer
deposited at a hydrogen flow of 270 sccm showed very inhomogeneous growth
and therefore we could not measure the dark conductivity. Based on these
properties we decided to use a hydrogen flow of 240 sccm H2 for the nc-Si:H
p-layer for solar cell growth.
4.5
Conclusions
To be able to deposit silicon thin films on plastic substrates without deforming them, the substrate temperature was calibrated for different substrates
and two different substrate holders. The deposition chamber gas pressure has
0 .7
0 .7
0 .6
0 .6
0 .5
0 .5
0 .4
0 .4
0 .3
0 .3
5
0
5
0
5
0
5
0
5
0
1 4 0
1 6 0
1 8 0 2 0 0 2 2 0 2 4 0 2 6 0
H y d r o g e n F lo w (s c c m )
0 .0
0 .0
0 .0
0 .0
0 .0
0 .0
0 .0
0 .0
0 .0
0 .0
2 8 0
63
7 5
7 0
6 5
6 0
5 5
5 0
4 5
4 0
3 5
3 0
D e p o s itio n R a te (n m /s )
C r y s ta llin e F r a c tio n
4.5. Conclusions
Figure 4.9: Raman crystalline ratio, measured by Raman spectroscopy and
deposition rate, measured by R-T spectroscopy, of p-type nc-Si:H layers deposited at 130°C on ZnO:Al coated glass substrates, as a function of hydrogen
flow.
64
Chapter 4. Low temperature silicon layers
a large influence on the substrate temperature. There is also a difference
between the substrate temperature of glass substrates in a regular substrate
holder compared to PC substrates in the substrate stretch holder. Running a
plasma will heat up the substrate 7°C to 8°C in a typical 300 nm a-Si:H i-layer
deposition. During a typical 1 µm nc-Si:H i-layer deposition, the additional
heating of the substrate is estimated to be 11°C.
We optimized both intrinsic layers and doped layers for use in solar cells
deposited at a substrate temperature of 130°. In the amorphous regime, ilayers show a monotonous increase in layer quality with increasing hydrogen
dilution, but at high hydrogen dilutions stress becomes a problem, and should
therefore be avoided. In the nc-Si:H regime we can control the crystal fraction
of layers both by the hydrogen dilution as well as the applied power into the
plasma. In this way we can control the crystallinity of the layers to a desired
value.
For doped layers we adapted known recipes at higher and lower substrate
temperatures to our desired substrate temperature and found layer characteristics suitable for solar cell depositions.
Chapter 5
Light trapping in
amorphous silicon cells on
polycarbonate substrates
5.1
Light trapping techniques
In order to develop highly efficient thin film silicon solar cells in general, and
on plastics in particular, light management is a key feature to optimize the
absorption of light over the complete solar spectrum to obtain a high current
density. Because thick silicon absorber layers are detrimental for the electronic properties of the solar cells due to collapse of the electric field in the
intrinsic layer bulk that leads to increased recombination, light management
techniques are needed that boost the absorption such that an adequate light
induced current can be generated while using thin absorber layers. For thin
film silicon solar cells, several methods have been investigated to enhance the
light absorption in the active layers of solar cells. The use of textured front
TCOs to scatter the incoming light in superstrate type of cells and a rough
ZnO-silver interface at the back reflector (BR) for substrate type of cells has
been extensively studied [109, 110]. Other techniques use 2D or 3D gratings
on back reflectors in substrate type of cells [48, 111] or gratings implemented in the substrate surface to induce diffraction of the light into the active
areas of the cells in superstrate configured cells [112]. Also research is done
on nanopillar [49] structured 3D back reflectors in thin film silicon cells in a
66
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
p-i-n configuration. Since cost is a very important factor in the success of
a certain type of cell, the light management techniques do not only need to
be successful in light absorption enhancement, but should also be suitable for
large scale processing. Furthermore, using thinner absorber layers will potentially have a number of positive consequences: lower material usage, lower
processing times, and therefore higher throughput, improved stability against
light induced degradation and a higher open-circuit voltage [25, 113, 28].
In this research, we present results obtained for three different light trapping techniques, which were used for thin film amorphous silicon solar cells
deposited at a substrate temperature below 130°C: (1) Scattering, by using a
texture-etched TCO front contact, obtained by etching the ZnO:Al, coated on
glass in an HCl solution in water; (2) Using regular textures comparable in size
to the effective wavelength of visible light; (3) We will introduce the concept
of geometric light trapping, which is based on refraction and reflection of light
on structures larger than the wavelength of visible light.
To test our light trapping techniques, we have developed amorphous silicon
intrinsic layers at deposition temperatures of 130°C with acceptable electronic
quality for use in solar cells. We also developed n-type and p-type doped
amorphous silicon layers at these low deposition temperatures. Using ZnO:Al
as a front transparent conductive oxide and as a BR, we fabricated a-Si solar
cells on our different structured polycarbonate (PC) substrates. On these
cells we measured current density-voltage (J-V) characteristics and spectral
response (SR). As a reference we deposited a-Si cells under the same conditions
on flat glass/ZnO and on Asahi U-type TCO glass.
5.1.1
Scattering
A traditional way to enhance light absorption in a superstrate cell configuration, is to use a randomly textured TCO, like texture-etched aluminium doped
zinc oxide (ZnO:Al) or natively textured materials such as fluorinated tin oxide
(SnO2 :F) as used for the commercially available Asahi TCO glass. A rough
interface between the TCO and the silicon layers causes the light to scatter
into the optically active silicon layers, causing a longer light path through the
silicon and thereby enhancing the absorption without increasing the thickness
of the silicon layers. If the scattering angle is large enough, the light that is
reflected from the back of the cell will be trapped inside the cell through total
internal reflection [114, 45]. Because also electrically inactive layers like the
TCO or p-layer have a relatively high absorption in the blue part of the solar
spectrum, a major drawback of using textured interfaces is that the scattering
is most pronounced in the short wavelength region, which also causes substan-
5.1. Light trapping techniques
67
1 μm
Figure 5.1: AFM image of a 1 µm ZnO:Al(0.5%) layer, texture etched in a
1.5% HCl solution for 10 seconds, used as a front TCO in p-i-n solar cells.
tial absorption in these layers. As the absorption coefficient of the intrinsic
layer material in the red part of the spectrum is low, a large light path increase
is needed to capture a substantial part of the red light.
5.1.2
Nanopyramid periodic structures
Several groups have investigated the use of 1D and 2D periodic structures
as light trapping schemes. These gratings can either be integrated into the
front contact in superstrate (p-i-n) type of cells or used in the back contact
in substrate (n-i-p) type of cells. Although the far-field effects of these periodic structures are well understood, understanding the near-field interaction
of the periodic structures and the incoming light is crucial for optimization of
the structures. Different theoretical investigations have been done for several
different types of nanostructures. The main parameter in these investigations
concerns the period (or pitch) of the nanostructures. Numerical simulations
by Haase et al. [115] and Dewan et al. [116] for a nc-Si cell in a p-i-n configuration on pyramid-structures predict an optimal pitch of 850 and 700 nm,
respectively. Campa et al. [117] and Gomard et al. [118] theoretically invest-
68
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
500 nm
500 nm
Figure 5.2: AFM images of the convex nanopyramid structured PC (left) and
concave nanopyramid structured PC (right).
igated the use of rectangular gratings and photonic crystals for a-Si cells in an
n-i-p configuration and found optimum pitches of 300 and 400 nm respectively.
Ferry et al. [48, 119] and Zhu et al. [120] demonstrate absorption enhancement effects of nanocones and nanodomes in n-i-p a-Si solar cells with very
thin absorber layers and find optimum periods of 500 and 450 nm. Eisele et
al. show reduced reflectance from 1D gratings in a-Si cell-like layer stacks for
gratings with 389 and 798 nm gratings in a p-i-n configuration [121].
In this chapter, we present a-Si cells deposited on nanopyramid structured
surfaces in a p-i-n configuration. As a substrate we used PC, embossed with
400 nm pyramids on a square base. We experimented both with inverted pyramids (facing inwards, called type I from now) and normal pyramids (facing
outwards, type II). AFM images of the structures used are displayed in figure
5.2, showing convex pyramids on the left and concave pyramids on the right.
The pyramid base for both structures is 400 nm, which is expected to give
maximum current enhancement for a-Si thin film solar cells, as studied in the
literature.
5.1.3
Geometric light trapping: micropyramid periodic
structures
We present a light trapping scheme based on reflection and refraction, which
does not rely on scattering on nanostructured interfaces. We fabricated such
structures on PC substrates, by embossing them with 8 µm base pyramids
5.1. Light trapping techniques
69
Figure 5.3: SEM images of convex pyramids (left) and concave pyramids
(right) embossed on polycarbonate. These substrates are used for geometric light trapping.
with positive pyramid angles, using a hot embossing technique [122]. Because the feature size is much larger than the effective wavelength of light, no
scattering is expected. For a perpendicular light ray, when it hits the cell, a
reflected light ray will have a high probability to hit another plane of one of
the pyramids, which results in a reduction of the reflection from the cell and
will therefore have a higher light absorption. Furthermore, the slanted nature
of the pyramids will induce a light path increase, resulting in more absorbed
light. When the angles of the light ray on various interfaces within the cell are
large enough, the light may be trapped within the active layers of the solar cell
through total internal reflection. A schematic representation of this scheme
is shown in figure 5.4. Using a 3D ray tracing program, explained in the following section, we modelled the absorption enhancement of p-i-n silicon layer
stacks on such pyramidal structures, while varying the angle of the pyramid
planes with respect to the substrate surface and calculated at which angles
maximum absorption enhancement takes place. We have done this for both
direct and diffuse incoming light. For both types of illumination a significant
light absorption enhancement is expected.
70
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
Incoming Light
Angle
PC
1000 nm ZnO:Al
Pyramid
angle
275 nm a-Si
Figure 5.4: A schematic 2D representation of the model used for ray tracing
calculations. For the simulations we used a ZnO:Al thickness of 1000 nm and
an a-Si thickness of 275 nm. The pyramid angle is defined as the angle between
the substrate surface and one of the pyramid faces.
Simulations
Performing optical modelling on features in the (hundreds of) nanometre range,
such as textured interfaces, used for scattering, requires complex and time consuming calculations, such as finite difference time domain calculations [123].
Because our pyramids have dimensions much larger than the wavelength of
light we can use ray optics to calculate the absorption of light in the different layers. Using a model that combines reflection, refraction, and absorption
[124], we calculated the absorption enhancement for several periodically textured substrates at all wavelengths of interest. For the wavelength dependent
absorption properties and refractive indices of the silicon and aluminium doped
zinc oxide TCO layers we used data obtained by measuring the optical properties of these thin films deposited for this study and fitting the optical data
using the OJL model [55]. The absorption coefficient is lower for the a-Si deposited at 130°C compared to the a-Si:H deposited at 180°C that is used for
5.1. Light trapping techniques
71
optimal layer properties, due to the higher band gap of the material deposited
at low temperature. In the ray tracing calculations we modelled the solar cells
as follows: (1) A halfspace of air with a refractive index of 1 (2) as a substrate
we used PC with a fixed refractive index of 1.5 and no absorption. (3) As a
TCO we modelled 1000 nm of ZnO:Al with refractive index and absorption
as shown in figure 4.1. For the amorphous silicon we used the measured optical properties of the low temperature (130°C) sample as shown in figure 4.1
and a layer thickness (perpendicular to the substrate surface) of 275 nm. Subsequently, we changed the shape of the substrate into square based pyramids
with varying pyramid angles. Taking the pyramid angle as the angle between
the substrate surface and the pyramid wall, we varied this angle from -60° to
60° in steps of 5°, where a negative pyramid angle denotes an inverted pyramid
facing inwards into the substrate. We simulated light hitting the cell in two
ways: (1) direct, perpendicular to the substrate surface and (2) diffuse, using
a Lambertian angle distribution. We used light with wavelengths from 400 nm
to 800 nm in steps of 10 nm. For every wavelength the following was calculated: (1) The total fraction of light lost through reflection. (2) The fraction
of light absorbed by the ZnO:Al. (3) The fraction of light absorbed by the
a-Si and (4) the fraction of light lost by transmission through the complete
cell. For every wavelength we took the average result of 1000 rays hitting the
cell on a randomly chosen position on the cell.
Figure 5.5 shows the calculated reflection and absorption by the ZnO:Al
layer, the absorption of the a-Si:H layer of a flat cell and of a cell with pyramid
angles of 40°, illuminated with direct light. According to the simulations, the
pyramidal structure suppresses the reflection over the whole spectral range,
which results in an enhanced absorption in the a-Si, which in turn will results
in a higher generated current. There is also a small increase in absorption
in the ZnO:Al of the front electrode. If we now integrate the absorbed light
over the AM1.5 spectrum we can calculate the expected generated current for
different pyramid angles and different types of light (direct or diffuse). The
calculated current density results for direct light and diffuse light, for pyramid
angles from -60° to 60° are shown in figure 5.6. For direct light, the simulated
current density changes from around 9 mA/cm2 for a flat surface to a maximum
of 13.5 mA/cm2 for pyramid angles of around 45° degrees for both positive and
negative pyramid angles. For diffuse light, the current density increases from
around 8.5 mA/cm2 to 12 mA/cm2 for pyramid angles of 30° and higher, for
both normal and inverted pyramids. The increase in absorption is mainly
caused by a lower reflection over the whole spectrum, resulting in a relative
increase in absorption over the whole spectrum and a relatively high absorption
enhancement in the red part (λ > 600 nm) of the spectrum. According to our
72
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
R e fle c tio n , A b s o r p tio n
1 .0
0 .8
a-Si:H absorption
Flat Substrate
40° Micropyramids
0 .6
0 .4
ZnO:Al absorption
Reflection
0 .2
0 .0
4 0 0
5 0 0
6 0 0
7 0 0
W a v e le n g th (n m )
8 0 0
Figure 5.5: Simulated total reflection, ZnO:Al absorption and a-Si absorption
in an a-Si:H cell on micropyramid structured substrates with a pyramid angle
of 40° and on a flat substrate for direct incoming light.
simulations, for direct light, a current density enhancement of 45% can be
achieved at a negative or positive pyramid angle of 45°, compared to a flat
cell. For diffuse light the weighted absorption enhancement is 40% at pyramid
angles larger than 40°.
Substrates
Experimentally, the pyramids are regularly arranged and have a pyramid angle
of 54.7°. Between the pyramids is a 2 µm wide flat surface area, which accounts
for 36% of the sample surface. The production of the mold and the embossing
of the PC substrates was carried out by Aquamarijn microfiltration B.V. The
mold for embossing was produced from a silicon wafer (100) by applying a
positive photoresist and patterning this with a square pattern with a line
width of 2 µm using UV-lithography. After development of the photoresists
the wafer was anisotropically etched in a KOH solution at 80°C to create a
pattern of pyramid structures on the surface. The photoresist was stripped
away and a 200 nm layer of copper was evaporated on the etched wafer and
subsequently 500 µm nickel was electroplated on the wafer (Technic Elevate Ni
5.2. Low temperature solar cells on PC substrates
73
C u r r e n t D e n s ity (m A /c m
2
)
1 4
1 3
1 2
1 1
1 0
D ir e c t L ig h t
D iffu s e L ig h t
8
9
-6 0
-4 0
-2 0
0
2 0
4 0
Micro-V angle (°)
6 0
Figure 5.6: Simulated current density of a-Si cells on micropyramid structured substrates for direct light (solid) and diffuse light (dashed) for different
pyramid angles.
5910). The silicon wafer was removed from the nickel by dissolution in KOH.
The obtained mold was used for hot embossing of the micropyramids in PC.
The hot embossing was done by clamping the mold on a 1 mm thick piece of
PC between two aluminium plates. The clamps were tightened and placed in
a vacuum oven. After evacuation the temperature was gradually increased to
200°C and kept stable for 1 hour. After that the oven was cooled down to
100°C and the substrate was removed from the mold. An SEM image of the
structures is shown in figure 5.3. One type has pyramids facing outward, out
of the substrate and the other type of substrate with pyramids facing inwards.
5.2
Low temperature solar cells on PC substrates
To be able to deposit thin film cells on plastics, the process temperature
should never exceed the glass transition temperature of the substrate used,
for all the processing steps. For polycarbonate, this transition temperature is
around 145°C [34], which is lower than the experimentally found optimal sub-
74
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
strate temperature of 200°C for a-Si deposition [36]. The consequences of low
substrate temperature for the electronic quality and optical properties of the
a-Si:H material have been discussed in the previous chapter. In this section
we will discuss the deposition of front TCOs on flexible substrates and their
adhesion to these substrates. Because we use plastics as a substrate, we will
also check whether there is degassing from the material that could contaminate the growing layer or the vacuum system. Because the thermal expansion
coefficient of PC is rather high, compared to the substrate holder material, we
employed a specially designed substrate holder that allows stretching of the
substrate while it expands, as discussed in chapter 4.
5.2.1
Cells on PC: Experimental issues
TCO adhesion to plastic substrates
Figure 5.7 shows optical microscope images of ZnO:Al deposited directly onto
flat polycarbonate substrates. The ZnO:Al shows cracks. The layers grown at
100 W RF power in the sputter deposition (right) show less cracks than the
ones grown at higher powers (300 W, left). The sheet resistance of the latter
layers is much higher than is suitable for use as a front TCO in solar cells.
Earlier studies of ZnO:Al grown on flexible plastic substrates do not report
adhesion problems [125, 126]. The adhesion difficulties for ZnO on flat PC
substrates may be attributed to the difference in thermal expansion of ZnO
(4×10−6 K−1 [127]) and PC (65×10−6 K−1 [128]). Because the substrate heats
up during the deposition, it expands. After the deposition, when the substrate
contracts as it cools down, the ZnO layer cracks under compressive stress. The
microscope images in figure 5.7 show debris that agree with this hypothesis
of compressive stress. Depositions on our micro- and nanopyramid structured
PC substrates do not show adhesion problems, because the textures act as
stress relievers. Because glass has a thermal expansion coefficient comparable
to that of ZnO (8×10−6 K−1 ), these problems do not occur when depositing
ZnO on glass substrates.
Plastic substrate degassing
Before a deposition on a substrate can commence, we need to be sure that there
is no degassing from the substrate while heating it up before the deposition or
during the deposition itself. Volatile elements (such as carbon and moisture)
from the substrate could contaminate the layer, resulting in material properties different than anticipated. Furthermore, contamination of the reactor
5.2. Low temperature solar cells on PC substrates
75
Figure 5.7: Microscope pictures of cracks formed in ZnO:Al layers deposited
on flat PC substrates at different powers. The layers grown at 100 W (right)
show less cracks than the layers grown at 300 W (left).
chamber could have deleterious effects on subsequent depositions. Therefore,
we checked whether our PC substrates release gasses when introduced in the
vacuum chamber. We did this by monitoring the background pressure over
time after introduction of the substrate into the reaction chamber, while closing the valve to the pumps. Even when there is no substrate present, the
background pressure will increase over time when the pumps are disconnected
from the chamber. Now, if a substrate releases gasses, the increase in pressure
will occur faster. Because every newly introduced substrate (or any object,
such as a substrate holder) from outside the vacuum carries moisture with it,
the background pressure is expected to rise after introduction into the vacuum,
but will quickly drop over time. Figure 5.8 shows the change in background
pressure after introduction of a PC substrate into chamber 4 of the ASTER
system. The heater temperature was set to 180°C, which corresponds to a slow
heating of the substrate up to 130°C. Every few minutes we disconnected the
pumps from the chamber and monitored the increase of pressure over time,
which was higher when a PC sample was inserted than without a sample for
the first 2 hours, but after that both situations show similar behaviour. Therefore we conclude that there is no continuing degassing from the substrates and
that they are therefore safe to use in our deposition system.
76
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
Closed pump Valve
Heater set to to 130°C
P re s s u re (m b a r)
1 E -3
1 E -4
No Sample
10 minutes
30 minutes
55 minutes
155 minutes
Background pressure
1 E -5
(in minutes)
1 E -6
1 0
1 0 0
1 0 0 0
T im e (s )
Figure 5.8: Pressure increase over time when the pump valve is closed after
different times of heating of the substrate in vacuum. For comparison the
time dependent pressure increase of an empty chamber is also shown. The
background pressure over time (with substrate and substrate holder mounted,
in minutes) with the pump valve open shows a usual pressure decrease.
5.2. Low temperature solar cells on PC substrates
5.2.2
77
Solar cell results
To test the light trapping abilities of our different structured substrates, we
deposited single junction a-Si solar cells in a p-i-n configuration on the different
substrates. As a front TCO we used ZnO:Al, deposited in our RF magnetron
sputtering system, SALSA, from a ZnO:0.5%Al2 O3 target at room temperature. As a BR we used ZnO:Al from a ZnO:2% Al2 O3 target. The silicon
layers were deposited in our ultra high vacuum multi-chamber system ASTER
by VHF-PECVD at a frequency of 60 MHz, using a showerhead electrode.
The p- and n-layers were deposited at 50 MHz in separate chambers. The
substrate temperature during deposition was set at 130°C and is expected to
increase 7°C to 8°C during the deposition. The metal contacts were deposited
by thermal evaporation of silver and aluminium. The cell structure was as
follows: The substrate, a 1000 nm ZnO:Al front TCO, a double p-structure
of nanocrystalline and amorphous silicon of 15 nm (in total), a 275 nm thick
intrinsic a-Si layer, a 30 nm thick a-Si n-layer, a 100 nm ZnO:Al BR and a
silver/aluminium back contact. After deposition, the cells were annealed in a
nitrogen environment for 1 hour at 125°C. The area of the cells was 0.16 cm2 .
As a reference, cells were deposited on 2 other types of substrates: (1) Asahi
U-type natively textured fluorinated tin oxide (SnO2 :F) TCO glass and (2)
flat glass (Corning Eagle 2000) substrates with an untreated flat ZnO:Al TCO
layer, under the same low temperature deposition conditions. For reference
purposes, a cell deposited on a flat PC surface would be the best candidate,
but the stress of the ZnO:Al layers prevented flat PC to be used as a substrate,
due to cracking of the ZnO:Al layers, as discussed above.
Figure 5.9 shows the J-V measurements of the cells on flat glass, Asahi-U
TCO glass, micropyramid structured PC and on nanopyramid structured PC
substrates. Table 5.1 shows the electrical properties for all types of cells, as
obtained by J-V measurements under dark conditions and under AM1.5 illumination after annealing at 125°C for 1 hour in a nitrogen environment. The
cell on micropyramid structured PC shows an increase in Jsc of 22%, compared to the cell on a flat glass substrate. The cell on concave nanopyramids
(type I) shows a short-circuit current density increase of 22%, whereas the
convex nanopyramids (type II) enhance the current density by 28%. The current generated by the cells on convex nanopyramid structured PC substrates
is slightly higher than that of the reference cells on Asahi U-type TCO. Figure 5.10 shows the spectral response data for three types of cells on PC and
the cells on flat glass and on Asahi U-type TCO glass. All cells on structured or textured substrates show a significantly higher quantum efficiency at
wavelengths higher than 400 nm, when compared to the cell on a flat substrate,
78
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
5
fla t
m ic
n a n
n a n
A s a
s s
y ra m
y ra m
y ra m
U -ty p
id
id I
id II
e T C O
-5
J
(
m A /c m
2
)
0
g la
ro p
o p
o p
h i
-1 0
-1 5
-0 .5
0 .0
0 .5
V o lta g e (V )
1 .0
Figure 5.9: Current density - Voltage characteristics for cells deposited on
different embossed PC substrates at 130°C. As a reference J-V characteristics
for cells on flat glass and on Asahi textured TCO glass are also shown.
Substrate type
J0
Rs
Rp
η
(mA/cm2 )
(Ωcm2 )
(Ωcm2 )
(%)
1.49
5.6×10−10
7.5
1166
5.6
59
1.90
4.7×10−8
9.7
801
6.4
0.88
63
1.69
4.7×10−9
9.3
964
6.8
12.98
0.89
64
1.98
1.3×10−7
9.0
1116
7.4
12.81
0.93
64
1.49
2.0×10−10
9.1
1250
7.6
Jsc
Voc
FF
(mA/cm2 )
(V)
(%)
Flat glass
10.14
0.93
64
Micropyramid
12.35
0.87
Nanopyramid I
12.33
Nanopyramid II
Asahi U-type
n
Jsc : short-circuit current density; Voc : open-circuit voltage; FF: fill factor; n: diode quality
factor; J0 : reverse saturation current; Rs : series resistance; Rp : parallel resistance; η: conversion
efficiency. Type I: concave pyramids; type II: convex pyramids
Table 5.1: Initial electrical properties of cells deposited at 130°C on flat ZnO:Al
coated glass substrates, micro- and nanopyramid structured PC substrates and
Asahi U-type textured glass as measured by J-V measurements under AM1.5
illumination and under dark conditions.
5.2. Low temperature solar cells on PC substrates
1 .0
fla t
m ic
n a n
n a n
A s a
0 .8
E C E
0 .6
g la
ro p
o p
o p
h i
s s
y ra m
y ra m
y ra m
U -ty p
79
id
id I
id II
e T C O
0 .4
0 .2
0 .0
4 0 0
5 0 0
6 0 0
7 0 0
W a v e le n g th (n m )
8 0 0
Figure 5.10: External collection efficiency measurements of cells deposited on
different embossed PC substrates at 130°C. As a reference the measurements
for cells on flat glass and on Asahi textured TCO glass are also shown.
which can be ascribed to anti-reflective properties and better response in the
red part of the spectrum. If we compare the spectral response of the cells
on nanopyramids to that of the cell on Asahi TCO, we observe a comparable
response at wavelengths above 600 nm, a small decrease in quantum efficiency
for wavelengths between 500 and 600 nm, whereas the response between 400
and 500 nm is higher. This results in a total short-circuit current density generation that is slightly higher for the cell on convex pyramids and a bit smaller
for the cell on concave pyramids, compared to the cell on Asahi TCO. The
cell on micropyramid structured PC, when compared to the cell on Asahi Utype TCO-glass, shows a total current density that is about 0.5 mA/cm2 lower.
This is mainly caused by the difference in response above 500 nm.
In the ultra-violet part of the spectrum, below 400 nm, absorption of light
by the substrate and the TCO dominate spectral response behaviour. The cell
on Asahi TCO, which is made of SnO2 :F, shows transparency down to 300 nm,
whereas ZnO:Al, due to its lower band gap for the samples in this study, cuts
off the light below 350 nm. PC is not transparent for light of wavelengths
below 380 nm. This is illustrated in figure 5.11, which shows the transmission
of glass and PC with and without the ZnO:Al front contact and of Asahi U-
80
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
1 .0
T r a n s m is s io n
0 .8
0 .6
P o ly c a
Z n O :A
C o r n in
Z n O :A
A s a h i-
0 .4
0 .2
0 .0
3 0 0
4 0 0
rb o n a te
l o n P C
g g la s s
l o n g la s s
U T C O
5 0 0
6 0 0
7 0 0
W a v e le n g th (n m )
8 0 0
Figure 5.11: Total transmission of glass and PC substrates with and without
ZnO:Al and the total transmission of Asahi U-type TCO glass.
type TCO glass. These differences in ultra-violet transmission account for the
difference in spectral response in this wavelength region.
The cells on flat glass and on Asahi TCO show a high Voc of 0.93 V. This
is caused by the high optical band gap of the intrinsic a-Si:H deposited at low
temperature (1.9 eV), which is attributed to a higher hydrogen content in the
film. The cell on structured PC shows a lower Voc of 0.87 V, which can be
attributed to a higher dark current (J0 ). The cell on micropyramid structured
PC shows a conversion efficiency of 6.4%, which is around 1% absolute lower
than the cell on textured Asahi TCO, which has a slightly higher Voc and a
higher fill factor of 64%, compared to 59% for the cell on micropyramid structured PC. The cells on nanopyramid structured concave and convex pyramids
show an efficiency of 6.8% and 7.4%, respectively, which is slightly lower than
that of the cell on Asahi TCO, due to a lower Voc .
Our simulation results (figure 5.6) had shown that a current enhancement
for micropyramid structured substrates is possible up to 45%. Experimentally,
for a pyramid angle of 54°, which is close to the calculated optimum pyramid
angle, a current enhancement of 22% has been observed, compared to cells on
a flat glass substrate. The micropyramid structured substrates that we used
have a flat surface area of around 30%, due to the fabrication method, which
explains a lower current enhancement than expected for a substrate that is
completely filled with pyramids. A better coverage, using the same embossing
5.2. Low temperature solar cells on PC substrates
130°C a-Si cell on micropyramid PC
130°C a-Si cell on nanopyramid PC I
130°C a-Si cell on nanopyramid PC II
130°C a-Si cell on flat glass
1 .0
0 .8
R e fle c ta n c e
81
0 .6
0 .4
0 .2
0 .0
4 0 0
5 0 0
6 0 0
7 0 0
W a v e le n g th (n m )
8 0 0
Figure 5.12: Total reflection of a-Si cells deposited on flat glass substrates,
on micropyramid structured PC substrates and on both types of nanopyramid
structured PC. All cells were deposited at a substrate temperature of 130°C.
technique, could be achieved by using larger-sized pyramids. According to
the simulations, the increase in absorbed light is mainly caused by a decrease
in reflection of light from the top surface of the cells. Figure 5.12 shows the
measured total reflection from the cell on micropyramid structured PC, both
types of nanopyramid structured PC and the cell on flat glass. The cell on
the micropyramid structured substrate shows a lower reflection over the whole
measured spectral range. This results in a generated current density in the
cell on structured PC which is comparable to the cell on Asahi TCO, but the
cell on structured PC suffers from a lower Voc and a lower FF than the cell
on Asahi. The cell on PC has a higher diode quality factor (n) value of 1.90
compared to the cell on Asahi (1.49) and has a saturation current density
which is 2 orders of magnitude higher than the latter.
Although the silicon layers of the cells are made under identical deposition conditions, the qualities of the diodes differ. There are a few possible
explanations for this. Firstly, defects could be created in the silicon due to
the thermal expansion of the substrate, the coefficient of thermal expansion of
PC, being much higher than that of amorphous silicon. This could result in
external stress in the silicon layers and thereby induce defects in these layers.
Secondly, controlling the substrate temperature is very important for obtaining good quality a-Si films. Although good quality films can be deposited even
82
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
Defective region
Figure 5.13: Cross-sectional TEM images of a-Si cells deposited on micropyramid structured PC substrates. On the left a crack through the complete
cell is visible. On the right, defective regions are present in the shoulder of a
pyramid.
at room temperature, tuning of the hydrogen dilution is necessary to secure a
device quality film [129]. When depositing on plastic substrates, intrinsic stress
of the layer on the substrate will cause curving of the substrate, resulting in a
decrease in heat transfer from the heater to the substrate, which could result
in a lower substrate temperature. Although our depositions were done in a
specially designed substrate holder, which can compensate for the expansion
by moving one end of the holder outwards by a pulling spring, when a layer
is deposited which shows compressive stress, the bending of the substrate will
cause a gap between the substrate holder and the substrate. Thirdly, studies
have shown that the substrate morphology can have an influence on the defect
formation during the deposition of a-Si [40]. Defects are formed within the
concave regions of the substrate. Similar defective regions were also found in
nanocrystalline silicon thin films [130, 131]. Figure 5.13 shows cross-sectional
transmission electron microscopy (XTEM) images of the cell on micropyramid
structured PC, on which we can identify defective regions: On the left, we
observe a complete crack through the silicon layers. On the right side, nanocracks (elongated voids) can be identified in the silicon layers near a valley in
the TCO-silicon interface.
5.3. Post-deposition treatments
83
A fte r s h u n t b u s tin g
In itia l
5
-5
-1 0
J
(
m A /c m
2
)
0
-1 5
-0 .2
0 .0
0 .2
0 .4
0 .6
0 .8
1 .0
V o lta g e (V )
Figure 5.14: Shunt busting a cell can change a short-circuited (shunted) cell
(dashed) to a working cell (solid).
5.3
5.3.1
Post-deposition treatments
Shunt busting
In thin film solar cells, short-circuit paths, also known as shunts, can destroy
the diode behaviour of the cell. Especially when cells have very thin silicon
layers or are on very rough surfaces, shunting of cells can drastically bring
down the yield of solar cells. Earlier studies reveal that these shunt paths are
metastable, i.e. shunt paths can be created and removed by applying forward
and reverse bias voltages respectively [132]. The most likely way a shunt path
can form is through the incursion of Al from the ZnO:Al BR into pinholes or
macroscopic defects, formed due to dust on the surface during the deposition
process. During shunt-busting, the Al diffuses out of the a-Si [133]. Figure
5.14 shows the J-V characteristics of a standard a-Si thin film solar cell before
and after our shunt-busting procedure, which consists of applying a linearly
increasing reverse bias voltage from 0 V to -5 V in 6 seconds. Care is taken
so that the PC substrate is not damaged due to overheating during shunt
busting. Although this procedure can recover some of the shunted cells, it
does not work for all short-circuited cells. For some cells, the PC substrate
heats up too much before it can recover, resulting in a damaged substrate and
a destroyed cell.
84
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
1000
5
10
)
2
1E-3
J (mA/cm
)
2
J (mA/cm
0
0.1
125°C
135°C
140°C
1E-5
1E-7
1E-9
125°C
135°C
140°C
-0.5
0.0
0.5
Voltage (V)
1.0
1.5
-5
-10
-15
-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0
Voltage (V)
Figure 5.15: Dark (left) and light J-V characteristics of a cell on convex nanopyramid structured PC after annealing at different temperatures. Annealing
up to 135°C improves the quality of the solar cell, whereas after annealing at
140°C the cell performance collapses. The slight shift of the minimum from
zero voltage for the curves under dark conditions is an experimental artefact.
5.3.2
Post deposition annealing
Figure 5.15 shows the J-V-characteristics of cells deposited on convex nanopyramid structured PC, after annealing at different temperatures. Directly after
the deposition, annealing was done in a nitrogen environment for 4 hour at
125°C. Subsequently, the cell was annealed at 135°C for 4 hours. The third and
last annealing step was done at 140°C for 90 minutes. Annealing at 125°C and
135°C increases the performance of the cell, which is reflected in the fill factor.
After the annealing at 140°C, the fill factor collapses and the cell performance
drops drastically. Although this is below the glass transition temperature of
the PC material of 145°C [34], the cell is still adversely affected. This may
be due to structural changes of the PC below the glass transition temperature
[134].
5.3.3
Stability under light soaking
Degradation of a-Si solar cells under light soaking conditions reduces the cell
performance over time [25]. To test the light induced degradation of a certain
cell, a standard test is undertaken in which the cell is soaked under AM1.5like illumination conditions for 1000 hours, while keeping the temperature
constant at 50°C under open-circuit conditions. As the metal-halide lamp
5.4. Conclusions
5
85
Annealed 125°C
Soaked 0.7 AM1.5 1000 h
Re-annealed 125°C
-5
J
(
m A /c m
2
)
0
-1 0
-1 5
-0 .4
-0 .2
0 .0
0 .2
0 .4
0 .6
0 .8
1 .0
V o lta g e (V )
Figure 5.16: J-V characteristics of an a-Si cell on nanopyramid structured PC
before and after light soaking under AM1.5-like illumination conditions for
1000 hours. A re-annealing step improves the Voc and FF.
degrades gradually while in use, the intensity (in this case) was roughly 70%
the intensity of the AM1.5 spectrum. Normally, we would monitor the cell
during the complete soaking time, but as the cells on PC are easily damaged
during the measurements, we chose to only measure the cell performance before
and after light soaking. Figure 5.16 shows these measurements. After light
soaking, the Voc drops from 0.88 V to 0.85 V and the fill factor drops from
0.66 to 0.56. After a re-annealing step at 125°C for 1 hour, the Voc increases
to 0.88 V and the fill factor increases up to 0.61.
5.4
Conclusions
We studied the light trapping in a-Si cells deposited on PC substrates using
three different substrate structures: Asahi U-type, nanopyramids and pyramids much larger than the wavelength of light. We have achieved an initial solar
cell efficiency of 7.4% for a cell directly deposited on nanopyramid structured
PC after annealing for 1 hour at 125°C. Compared to cells on Asahi U-type,
these cells show a slightly higher current density, but suffer from a lower Voc .
Cells on micropyramid structured PC had a maximum initial efficiency of 6.4%
after annealing, which is lower due to a lower current and a lower fill factor.
86
Chapter 5. Light trapping in amorphous silicon cells on polycarbonate
substrates
XTEM studies show that the cells deposited on PC substrates have more defects than the cells grown on glass substrates, which could be caused by a large
difference in thermal expansion of PC and a-Si, or by the sharp features of the
structured PC samples. This is reflected in a higher reverse saturation current
for cells deposited on PC.
Post-annealing improves the efficiency of a-Si cells deposited at low temperature. Some short-circuited cells can be recovered by applying a shunt-busting
procedure, in which a reverse bias is applied for several seconds.
Chapter 6
Micromorph tandem cells
on plastic substrates
6.1
Introduction
Nanocrystalline silicon (nc-Si:H) differs from amorphous silicon (a-Si:H) in a
number of ways. A very important property for solar cell applications is its
lower band gap (1.1 eV) compared to a-Si:H (1.8 eV), enabling it to absorb light
with wavelengths up to ∼1100 nm. A stacked structure of an a-Si:H cell and
a nc-Si:H cell is called the micromorph concept, which was first introduced by
IMT (now EPFL) [13]. This concept allows the blue and green part of the
spectrum to be absorbed by the a-Si:H, the top cell, whereas the remaining
light is passed on to the nc-Si:H cell, the bottom cell, which will absorb mostly
red light. In this way, the mismatch between the energy of the absorbed
photons and the band gap is minimized, and therefore thermalization losses
are reduced, compared to a nc-Si:H cell. Compared to a single junction a-Si:H
cell, light of a broader spectrum is absorbed.
The absorption coefficient of nc-Si:H is low, compared to amorphous material with a similar band gap, such as a-(Si)Ge, due to a predominantly indirect
nature of the band-gap. Therefore rather thick absorber layers are used for
solar cells, with typical thicknesses of 1 to 3 µm. Because in a micromorph
tandem cell the individual cells are connected in series, the total cell current is
equal to the lowest current generated by one of the two cells. Therefore good
current matching between the sub-cells in a tandem cell is crucial for good
electrical performance.
88
Chapter 6. Micromorph tandem cells on plastic substrates
In this chapter, we report on our studies towards depositing nc-Si:H and aSi:H/nc-Si:H tandem cells at a substrate temperature of 130°C, and our results
on direct depositions of cells onto polycarbonate (PC) substrates. For this we
needed to adapt the deposition processes to obtain device quality materials
at low temperature. The material studies on the layers that are used in this
chapter are described in chapter 4. For the a-Si:H top cells, the cells treated
in chapter 5 are used as a basis.
As described in chapter 4, to obtain device quality material at low temperatures, silicon layers have to be deposited at a higher hydrogen to silane
gas flow ratio (hydrogen dilution) than that is required at higher substrate
temperatures [37]. As a result, deposition rates decrease, resulting in longer
deposition times. This time factor becomes even more severe when we consider
multi-junction cells, because of the thick hydrogenated nanocrystalline silicon
bottom cell needed to achieve adequate current matching between the top and
the bottom cell. We tackle this problem by reducing the total thickness of the
cell to around 1000 nm. This concept has been applied to cells at high temperatures [135], sometimes making use of intermediate reflecting layers between
the top and bottom cells [136].
6.2
nc-Si:H cells on glass substrates
Based on the nc-Si:H intrinsic layer series as a function of applied plasma
power input, as described in chapter 4, we produced solar cells on glass in a
p-i-n configuration. As a front transparant conducting oxide (TCO) we used
aluminium doped zinc oxide (ZnO:Al) which was sputter-deposited, followed
by texture etching in a hydrochloric acid solution. As a back contact we
used evaporated silver, after sputter-depositing a ZnO:Al back reflector. The
thickness of the intrinsic layers of the cells was aimed at 700 nm, based on
deposition rate. Before characterization, the cells were annealed in a nitrogen
environment for 1 hour at 125°C. The size of the cells was 0.16 cm2 .
Figure 6.1 shows the current density-voltage (J-V) characteristics of ncSi:H cells, as a function of applied plasma power around the transition from
a-Si:H to nc-Si:H. Going from 12.5 W to 22.5 W, the material (as shown in
figure 4.8) changes from mostly amorphous to almost fully crystalline. Figure
6.2 shows the short-circuit density (Jsc ), open-circuit voltage (Voc ), fill factor
(FF) and the resulting conversion efficiency (η) before and after annealing for
1 hour at 125°C. For the cells made from the layers in this series this results in
an increasing current density up to 20 W applied plasma power, which can be
attributed to a lower band gap for the material made at at higher power, owing
6.2. nc-Si:H cells on glass substrates
89
C u r r e n t D e n s ity (m A /c m
2
)
5
1 2 .5 W
1 5 W
1 7 .5 W
2 0 W
2 2 .5 W
0
-5
-1 0
-1 5
-2 0
-0 .4
-0 .2
0 .0
0 .2
0 .4
0 .6
0 .8
V o lta g e (V )
Figure 6.1: J-V curves for nc-Si:H cells deposited at 130°C on glass substrates
as a function of applied VHF power. The structure of the cell is glass/20 nm
p-nc-Si:H/700 nm i-nc-Si:H/30 nm n-a-Si:H/100 nm ZnO:Al/Ag/Al.
to a larger crystalline fraction. At 20 W power, the i-layer is fully crystalline
and the band gap for the material made at higher powers does not decrease
any more. At the same time, the Voc decreases from 0.62 V to 0.49 V for the
annealed cells. The FF shows a maximum at 17.5 W applied power, where the
crystalline fraction is roughly 40%. The resulting efficiencies show the same
trend, peaking at 17.5 W. At this point, annealing raises the Voc by about
60 meV, while the FF is lifted by about 8% absolute. The highest achieved
conversion efficiency is 6.9%.
As mentioned in the introduction, deposition rate becomes an issue at low
deposition temperature. We have been successful in depositing nc-Si:H at
a reasonably high deposition rate (0.51 nm/s) using high pressure (3 mbar)
and high power and a showerhead cathode for the gas distribution in the
plasma zone for the deposition. This is an adaptation of the deposition process
which delivered 10% efficiency nc-Si:H cells at standard deposition temperature
(180°C) [137].
Chapter 6. Micromorph tandem cells on plastic substrates
0 .6 4
2 0
0 .5 6
1 8
1 6
Before Annealing
Annealed 125°C 1 hour
1 4
1 2
1 0
1 5
2 0
2 5
1 0
0 .4 8
1 5
2 0
0 .4 0
2 5
O p e n C ir c u it V o lta g e (V )
2 2
7
0 .6 5
6
E ffic ie n c y (% )
F ill F a c to r (% )
C u r r e n t D e n s ity (m A /c m
2
)
90
0 .6 0
5
0 .5 5
4
1 0
1 5
2 0
P o w e r (W )
2 5
1 0
1 5
2 0
2 5
P o w e r (W )
Figure 6.2: Electrical properties of nc-Si:H cells deposited at 130°C on glass
substrates as a function of applied VHF power, before and after an annealing
step at 125°C of 1 hour.
6.3. Tandem cells on glass substrates
91
1 .0
7 0 0 n m
9 0 0 n m
0 .8
T a n d e m
B o tto m
B o tto m
C e ll
C e ll
T o ta ls
E C E
0 .6
0 .4
T o p C e lls
B o tto m
C e lls
0 .2
0 .0
4 0 0
5 0 0
6 0 0 7 0 0 8 0 0 9 0 0
W a v e le n g th (n m )
1 0 0 0
Figure 6.3: Spectral response characteristics for tandem cells deposited on glass
substrates at 130°C, showing cells with two different bottom cell thicknesses:
700 nm (solid) and 900 nm (dashed). The spectral response for the top cell,
bottom cell and summed up response are shown individually.
6.3
Tandem cells on glass substrates
The single junction a-Si:H cells as described in chapter 4 together with the ncSi:H cells from the previous section were combined to produce a-Si:H/nc-Si:H
tandem cells at 130°C. A double p-layer (nc-Si:H/a-Si:H) was used to make
proper contact with the texture etched ZnO:Al front TCO. The complete solar
cell structure was as follows: Superstrate/ZnO:Al TCO/p nc-Si:H/p a-Si:H/i
a-Si:H/n a-Si:H/p nc-Si:H/i nc-Si:H/n a Si:H/ZnO:Al/Ag/Al.
We used very thin layers as the photo-active layers: 275 nm of a-Si:H for
the top cell, combined with a nc-Si:H bottom cell with an i-layer of 700 nm.
Figure 6.3 (solid) shows the external collection efficiency (ECE) data for the
resulting tandem solar cell structure, showing the spectral response for the
top a-Si:H cell and the bottom nc-Si:H cell separately and the total (sum)
ECE of the structure. The calculated current densities for the cells showed
a mismatch between the top and bottom cells: 8.8 mA/cm2 for the top cell
and 7.6 mA/cm2 for the bottom cell. Based on these measurements we redeposited the solar cell structure, but now using a thicker bottom cell of 900 nm
92
Chapter 6. Micromorph tandem cells on plastic substrates
5
J (m A /c m
2
)
0
-5
-1 0
a -S
n c ta n
ta n
-1 5
-2 0
-0 .5
0 .0
0 .5
V o lta g e (V )
1 .0
i, 2
S i,
d e m
d e m
7 5
7 0
,
,
n m
0 n m
B C 7 0 0 n m
B C 9 0 0 n m
1 .5
Figure 6.4: J-V characteristics for cells deposited on glass substrates at 130°C
under AM1.5 light conditions, showing cells with two different bottom cell
thicknesses: 700 nm (dashed) and 900 nm (solid). Also the J-V characteristics
of the single junction cells on which the tandem was based are shown.
for better current matching. The results are also in figure 6.3 (dashed). Calculated from the spectral response measurements, the bottom cell current
increased to 9.1 mA/cm2 , whereas the top cell current also increased slightly,
to 9.2 mA/cm2 . The increase in top cell current is probably due to a slight
change in deposition conditions.
Figure 6.4 shows the J-V curves under AM1.5 illumination of the a-Si:H cell
(275 nm) and the nc-Si:H cell (700 nm) on which the deposition recipe for the
tandem was based, and the resulting tandem cell. Also shown is the tandem
cell with an i-layer thickness of 900 nm for the bottom cell. The J-V data for
the a-Si:H cell shows a rather high Voc of 0.90 V, which is attributed to the
high band gap of the a-Si:H layer deposited at 130°C. The characteristics of the
curves, especially near Voc , confirms that the tunnel recombination junction
with the low temperature doped layers is working well, showing no S-character.
The micromorph tandem cell has a Voc of 1.40 V, a Jsc of 10.5 mA/cm2 and a
FF of 65%, resulting in an initial efficiency of 9.5%.
Figure 6.5 shows a bright field cross-sectional transmission electron microscopy (XTEM) image of the complete cell structure showing the different
6.3. Tandem cells on glass substrates
93
layers. Analysis of the XTEM images shows a top cell thickness of 285 ± 10
nm and a bottom cell thickness of 960 ± 60 nm, including p- and n-layers. The
deposition rates of the a-Si:H and the nc-Si:H i-layer are 0.22 nm/s and 0.51
nm/s, respectively, which results in a total deposition time for the i-layers of
only 51 minutes. The deposition time for all Si layers of the tandem cell is just
over 80 minutes. The p- and n-layers were not optimized for deposition speed.
Apart from the greatly reduced deposition time, reducing the thickness of the
bottom cell has a number of advantages. First of all, it results in a decrease
in material usage. Secondly, the thinner layers mitigate the deleterious effect
of a relatively high defect density (resulting from deposition at lower than optimum temperature) on FF and Voc . Thirdly, thinner layers induce less stress
on the substrate, which is a very important property when using plastics as a
substrate.
Cost reduction (together with flexibility) is the main advantage of using
plastics as a substrate material. Apart from the substrate, cost reductions
can be achieved by speeding up the manufacturing process (througput), i.e.
reducing deposition times. This can be achieved by either increasing the deposition rate of the layers or by decreasing the thickness of the layers. In
this study we have shown that excellent spectral splitting can be achieved in
very thin micromorph tandem solar cells deposited at 130°. Because the low
temperature a-Si:H has a high band gap of 1.9 eV, corresponding to light with
a wavelength of 653 nm, the top cell will transmit more (red) light towards
the bottom cell than an a-Si:H cell deposited at higher temperature, while
the Voc of the completed cell will rise. This results in the possibility to use a
thinner bottom cell, both because there is more light available for the bottom
cell and because of the lower current generated by the top cell. The resulting
lower Jsc of the complete cell is partly compensated by the higher Voc . In our
configuration, we achieved excellent current matching using a 275 nm thick
a-Si:H top cell and a 900 nm thick nc-Si:H bottom cell.
6.3.1
Stability under light soaking
Degradation of solar cells under light conditions reduces the cell performance
over time [25]. This is especially true for a-Si:H solar cells. To test the light
induced degradation of a low temperature tandem cell, a standard test is undertaken in which the cell is illuminated by an AM1.5-like spectrum, in our
light soaking set up, which is described in Chapter 2. The power density
approaches 100 mW/cm2 . The temperature of the samples is controlled and
kept at a constant 50°C. During the light soaking, J-V measurements are performed at exponentially increasing time intervals. Between the measurements,
94
Chapter 6. Micromorph tandem cells on plastic substrates
Al/Ag
ZnO:Al
nc-Si:H bottom cell
TRJ
a-Si:H top cell
ZnO:Al
Figure 6.5: XTEM image of the a-Si:H/nc-Si:H tandem solar cell, showing the
ZnO:Al TCO, a-Si:H top cell and nc-Si:H bottom cell. A clear boundary is
visible where the tunnel recombination junction (TRJ) is between the a-Si:H
top cell and the nc-Si:H bottom cell. The cells have total thicknesses of 285 nm
for the top cell and 960 nm for the bottom cell.
1 .0 0
0 .9 8
V
95
o c
F F
0 .9 6
N o r m a liz e d V
o c
, F ill F a c to r
6.4. Tandem cells on plastic substrates
0 .9 4
0 .9 2
0 .9 0
0 .0 1 0 .1
1
1 0
1 0 0 1 0 0 0
L ig h t S o a k in g T im e (H o u r s )
Figure 6.6: The normalized degradation of the FF and Voc of an a-Si:H/ncSi:H tandem cell deposited at 130°C.
the cells are kept under open-circuit conditions. The light intensity during the
light soaking was monitored using a crystalline silicon reference diode.
Figure 6.6 shows these measurements for a-Si:H/nc-Si:H tandem cells deposited at a substrate temperature of 130°C on glass. We can see that the Voc
decreases roughly 7.5% and the FF decreases 9% over time. The main part
of the degradation of the fill factor occurs within 10 to 100 hours, whereas
of the Voc degradation occurs within the first hours of light soaking. The
fast degradation of the Voc implies that a part of the degradation has already
taken place when the J-V characteristics of the cells were measured in our
solar simulator. The different degradation times indicate different degradation mechanisms. The degradation of the fill factor is related to the formation
of dangling bonds, which act as recombination centres, which cause a decrease
in current, especially under forward bias-conditions.
6.4
Tandem cells on plastic substrates
For the deposition of an a-Si:H/nc-Si:H tandem cells on polycarbonate we
copied the recipe from the previous section and performed a deposition run
Chapter 6. Micromorph tandem cells on plastic substrates
-1
-1
-1
-
-
-
7 .5
5 .0
2 .5
0 .0
2 .5
5 .0
7 .5
0 .0
2 .5
5 .0
-0 .5
1 .0
ΦS
iH 4
/ ΦH
2
= 4 .5 /1 0 0
0 .8
N o B ia s L ig h t
B o tto m c e ll
T o p c e ll
0 .6
E C E
J (m A /c m
2
)
96
ΦS
ΦS
0 .0
0 .5
1 .0
V o lta g e (V )
/ ΦH =
iH
iH
5 /1 0 0
2
4
4
/ ΦH = 4 . 5 / 1 0 0
0 .4
0 .2
2
1 .5
0 .0
4 0 0
6 0 0
8 0 0
W a v e le n g th (n m )
1 0 0 0
Figure 6.7: (left) J-V curves under AM1.5 light conditions of tandem cells
deposited on micro-structured PC substrates, using two different hydrogen
dilutions for the bottom cell i-layer. The inset shows Raman spectroscopy
data on two bottom cell i-layers at ΦSiH4 /ΦH2 = 5/100 on glass and on PC.
(right) Spectral response data for the tandem cell deposited at the higher
hydrogen dilution, showing the response of the top cell, the bottom cell and
the response under dark conditions.
using different types of structured PC, described in chapter 5, as substrates in
the stretch substrate holder. Figure 6.7 (left, solid) shows the J-V measurements under AM1.5 light conditions on the resulting cell on micro-structured
PC. The high Voc (1.58 V) and the very low current density (3.7 mA/cm2 )
point towards an a-Si:H bottom cell, whereas for the same deposition run on a
glass substrate, the bottom cell showed nc-Si:H behaviour. The a-Si:H growth
on PC substrates was confirmed by Raman spectroscopy measurements, shown
in the inset, which indeed show only a very low Raman crystalline ratio (0.07),
whereas the cell deposited on a glass substrate shows a Raman crystalline ratio
of 0.47. For this purpose we deposited Si:H layers using the same recipe as the
bottom cell i-layer on top of a ZnO:Al/p nc-Si:H structure on micro-structured
PC.
Consequently we deposited a tandem cell structure on different types of
structured PC, but now changing the gas flows from ΦSiH4 /ΦH2 = 5/100 to
ΦSiH4 /ΦH2 = 4.5/100. Figure 6.7 (left, dashed) shows the J-V curves of the
resulting cell on micro-structured PC, showing a higher Jsc of 7.9 mA/cm2 and
a Voc of 1.25 V, indicating that now the bottom cell is indeed nanocrystalline.
6.4. Tandem cells on plastic substrates
G la s s , d
5
P C , d
P C , d
B C
B C
B C
97
= 9 0 0 n m
= 9 0 0 n m
= 1 3 5 0 n m
-5
J
(
m A /c m
2
)
0
-1 0
-0 .5
0 .0
0 .5
1 .0
1 .5
V o lta g e (V )
Figure 6.8: J-V measurements of tandem cells deposited on micro-structured
PC, using two different bottom cell thicknesses. As a comparison, the data for
a low temperature cell on glass (900 nm bottom cell) in also shown. The inset
shows the corresponding dark J-V data.
Figure 6.7 (right) shows the spectral response curves for the same cell. The
generated current in the bottom cell is much lower (6.9 mA/cm2 ) than the
current from the top cell (8.2 mA/cm2 ) and therefore the current in the cell
is limited by the current generated in the bottom cell. For the cells deposited
on both types of nanopyramid structured PC (as described in chapter 5), the
current generated in the bottom cells is lower, between 5.5 and 6 mA/cm2 ,
because the nanostructures are designed for light trapping in a-Si:H cells. For
light trapping in the red and infra-red part of the spectrum, larger sized pyramids are needed [112]. Coming back to our working a-Si/nc-Si tandem cell on
micro-structured PC, the black line in figure 6.7 shows the spectral response
under dark conditions. The measured spectral response in the blue light region
under dark conditions, which does not follow the spectral response of the bottom cell in this region, indicates that the bottom cell can conduct current even
when it is not illuminated. When this happens, we say that the cell is leaking
[138], which is probably due to low-quality nc-Si:H which can have shunt paths
and/or a high midgap defect density in the layer, which has adverse effects on
the Voc and fill factor of the cell [139].
98
Chapter 6. Micromorph tandem cells on plastic substrates
1 .0
0 .8
T a n d e m
o n g la s s , d
T a n d e m
o n P C , d
T a n d e m
o n P C , d
B C
B C
B C
= 9 0 0 n m
= 9 0 0 n m
= 1 3 5 0 n m
C e ll T o ta ls
E C E
0 .6
0 .4
T o p C e lls
B o tto m
C e lls
0 .2
0 .0
4 0 0
5 0 0
6 0 0 7 0 0 8 0 0 9 0 0
W a v e le n g th (n m )
1 0 0 0
Figure 6.9: Spectral response measurements on tandem cells deposited on PC,
using two different bottom cell thicknesses. As a comparison, the data for the
best low temperature tandem cell on glass in also shown.
To obtain a better matching of the currents generated by the top and
bottom cell we deposited new tandem cells, but now using a bottom cell i-layer
thickness of 1350 nm, as opposed to the 900 nm used in the previous runs. It is
also believed that a thicker bottom cell will reduce the leaking of the bottom
cell. Figure 6.8 shows the J-V characteristics under AM1.5 illumination for
three a-Si:H/nc-Si:H tandem cells: two cells deposited on micro-structured
PC substrates, utilizing two different bottom cell thicknesses, and the tandem
deposited on glass substrates (900 nm bottom cell), as described in the previous
section. Figure 6.9 shows the corresponding spectral responses. Although the
response of the bottom cell increases (8.3 mA/cm2 ) when a thicker bottom
cell is used, the current of the complete cell is now limited by the top cell
(7.2 mA/cm2 ). The lower current of the top cell is caused by a lower spectral
response in the 350-500 nm region. This is probably due to a thicker p-layer
or a less transparent TCO layer. Unfortunately, unforeseen circumstances
prevented us from redepositing this run.
When we compare the cells deposited on PC to the cell on glass, we observe
a lower current generated in the bottom cell. The low voltages for tandem cells
on PC can be attributed to the leakage in the bottom cells. As opposed to
6.5. Conclusions
99
dBC
η
Jsc
Voc
FF
Rs
Rp
JT C
JBC
nm
(%)
(mA/cm2 )
(V)
(%)
(Ωcm2 )
(Ωcm2 )
(mA/cm2 )
(mA/cm2 )
Glass/TE ZnO:Al
700
9.0
9.7
1.37
68
14.9
2800
8.6
7.6
Glass/TE ZnO:Al
900
9.5
10.5
1.40
65
15.9
2008
9.2
9.1
Micro-struct. PC
900
6.1
7.9
1.25
61
20.4
2172
8.2
6.9
Micro-struct. PC
1350
5.3
6.5
1.26
65
23.0
3426
7.2
8.3
Nano-struct. PC I
900
5.6
6.2
1.27
72
21.0
7025
8.0
5.5
Nano-struct. PC I
1350
5.6
8.0
1.19
59
20.7
1449
n/a
8.3
Substrate type
Nano-struct. PC II
1350
5.2
6.4
1.27
65
41.0
6925
7.9
5.8
dBC : bottom cell thickness; η: conversion efficiency; Jsc : current density; Voc : open-circuit
voltage; FF: fill factor; Rs : Series resistance; Rp : parallel resistance; JT C : top cell current
density; JBC : bottom cell current density
Table 6.1: Electrical properties of a-Si:H/nc-Si-H tandem cells deposited on
glass and on micro- and nano-structured PC at a substrate temperature of
130°C. Values are obtained from J-V measurements under AM1.5 illumination
and from spectral response measurements. All top cell i-layers are 275 nm
thick.
the a-Si:H cells described in chapter 4, the micro-structured substrates do not
exhibit light trapping comparable to texture-etched ZnO:Al in nc-Si:H cells.
The cells on micro-structured PC show a decrease in Voc , compared to the cells
on glass. When we look at the J-V curves measured under dark conditions
(figure 6.8, inset), we see that the cell on glass has a lower diode quality
factor n of 3.5, compared to the cells on PC (∼3.9) and a reverse saturation
current of 9.1 × 10−10 , which is about one order of magnitude lower than the
cells deposited on PC. This indicates a lower material quality of the materials
deposited on PC. A similar observation was made for a-Si:H single junction
solar cells deposited on PC substrates.
An overview of the solar cell properties of all tandem cells described in this
chapter is given in table 6.1.
6.5
Conclusions
In this chapter we presented nc-Si:H and a-Si:H/nc-Si:H tandem solar cells
deposited on glass and on PC substrates. We optimized the nc-Si:H layer
quality by tuning the hydrogen dilution when depositing the layer and were
able to accurately control the Raman crystallinity ratio by changing the applied power into the plasma. These layers were used to fabricate nc-Si:H cells
100
Chapter 6. Micromorph tandem cells on plastic substrates
on glass substrates, which gave us good cell performance, using thin (700 nm)
i-layers. Combined in an a-Si:H/nc-Si:H tandem cell structure, we were able
to deposit a tandem solar cell at a substrate temperature of 130°C with an
(initial) conversion efficiency of 9.5%. Because the used silicon layers in this
cell are rather thin, deposition times can be kept to a minimum, which results
in a total deposition time for all silicon layers of 80 minutes, and less than an
hour for the i-layers only.
When this recipe was transferred to PC substrates, the crystal nucleation
behaved differently, resulting in a-Si:H/a-Si:H tandem cells, which suffer from
a very low current, due to bad current matching. Changing the hydrogen
dilution solved this problem. The light trapping in structured PC substrates
is less pronounced in the bottom cell, compared to texture etched ZnO:Al,
which is used for light trapping in our tandem cells on glass. Therefore we
deposited tandem cells with a thicker bottom cell to obtain better current
matching.
Bibliography
[1] International Energy Agency. World energy outlook 2011. 2011.
[2] K. Branker, M. J. M. Pathak, and J. M. Pearce. A review of solar photovoltaic levelized cost of electricity. Renewable and Sustainable Energy
Reviews, 15(9):4470–4482, 2011.
[3] M. A. Green, K. Emery, Y. Hishikawa, and W. Warta. Solar cell efficiency tables (version 37). Progress in Photovoltaics: Research and
Applications, 19(1):84–92, 2011.
[4] P. D. Lund. Boosting new renewable technologies towards grid parity
- economic and policy aspects. Renewable Energy, 36(11):2776–2784,
2011.
[5] A.E. Becquerel. Comptes Rendus de l’Academie des Sciences, 9:145,
1839.
[6] J. Perlin. From Space to Earth: The Story of Solar Electricity. Harvard
University Press, 2002.
[7] A. Jäger-Waldau. Photovoltaic Technology, Chapter 8: Overview of the
Global PV industry. Elsevier, 2012.
[8] International Energy Agency. Trends in photovoltaic application. 2010.
[9] Gavin Conibeer. Third-generation photovoltaics.
10(11):42 – 50, 2007.
Materials Today,
[10] W. Shockley and H.J. Queisser. Detailed balance limit of efficiency of
p-n junction solar cells. Journal of Applied Physics, 31(3):510–519, 1961.
102
Bibliography
[11] O. Gabriel, S. Kirner, C. Leendertz, M. Gerhardt, A. Heidelberg,
H. Bloeß, R. Schlatmann, and B. Rech. Large area PECVD of a-Si:H/aSi:H tandem solar cells. Physica Status Solidi (C): Current Topics in
Solid State Physics, 8(10):2982–2985, 2011.
[12] R. E. I. Schropp. Large-area thin-film silicon: Synergy between displays and solar cells. Japanese Journal of Applied Physics, 51(3 PART
2):03CA07, 2012.
[13] J. Meier, S. Dubail, R. Fluckiger, D. Fischer, H. Keppner, and A. Shah.
Intrinsic microcrystalline silicon - a promising new thin film solar cell
material. In Conference Record of the 24th. IEEE Photovoltaic Specialists
Conference, volume 1, pages 409 – 412, 1994.
[14] T. Takeda, M. Kondo, and A. Matsuda. Thin film silicon solar cells
on liquid crystal polymer substrate. In Proceedings of the 3rd World
Conference on Photovoltaic Energy Conversion, volume B, pages 1580–
1583, 2003.
[15] J. de Wild, A. Meijerink, J. K. Rath, W. G. J. H. M. Van Sark, and
R. E. I. Schropp. Upconverter solar cells: Materials and applications.
Energy and Environmental Science, 4(12):4835–4848, 2011.
[16] E. Klampaftis, D. Ross, K. R. McIntosh, and B. S. Richards. Enhancing the performance of solar cells via luminescent down-shifting of the
incident spectrum: A review. Solar Energy Materials and Solar Cells,
93(8):1182–1194, 2009.
[17] A.J Nozik. Quantum dot solar cells. Physica E: Low-dimensional Systems and Nanostructures, 14(14):115 – 120, 2002.
[18] H. F. Sterling and R. C. G. Swann. Chemical vapour deposition promoted by r.f. discharge. Solid State Electronics, 8(8):653–654, 1965.
[19] R.C. Chittick, J.H. Alexander, and H.F. Sterling. The preparation and
properties of amorphous silicon. Journal of the Electrochemical Society,
116(1):77–81, 1969.
[20] W.E. Spear and P.G. Le Comber. Substitutional doping of amorphous
silicon. Solid State Communications, 17:1193–1196, 1975.
[21] W. E. Spear, P. G. Le Comber, S. Kinmond, and M. H. Brodsky. Amorphous silicon p-n junction. Applied Physics Letters, 28(2):105–107, 1976.
Bibliography
103
[22] A. Madan. Amorphous silicon - From doping to multi-billion dollar
applications. Journal of Non-Crystalline Solids, 352:881–886, 2006.
[23] D. E. Carlson and C. R. Wronski. Amorphous silicon solar cell. Applied
Physics Letters, 28(11):671–673, 1976.
[24] Y. Kuwano, M. Ohnishi, S. Tsuda, Y. Nakashima, and N. Nakamura.
Preparation and properties of amorphous silicon produced by a consecutive, separated chamber method. Japanese Journal of Applied Physics,
21(3):413–417, 1982.
[25] D. L. Staebler and C. R. Wronski. Reversible conductivity changes in
discharge-produced amorphous Si. Applied Physics Letters, 31(4):292–
294, 1977.
[26] S. Vepřek and V. Mareček. The preparation of thin layers of Ge and Si by
chemical hydrogen plasma transport. Solid State Electronics, 11(7):683–
684, 1968.
[27] M. Faraji, S. Gokhale, S. M. Choudhari, M. G. Takwale, and S. V.
Ghaisas. High mobility hydrogenated and oxygenated microcrystalline
silicon as a photosensitive material in photovoltaic applications. Applied
Physics Letters, 60(26):3289–3291, 1992.
[28] J. Meier, R. Fluckiger, H. Keppner, and A. Shah. Complete microcrystalline p-i-n solar cell - Crystalline or amorphous cell behavior? Applied
Physics Letters, 65:860–862, 1994.
[29] K. Yamamoto, M. Toshimi, T. Suzuki, Y. Tawada, T. Okamoto, and
A. Nakajima. Thin film poly-Si solar cell on glass substrate fabricated at
low temperature. In Material Research Society Symposium Proceedings,
volume 507, pages 131–138, 1998.
[30] A. Terakawa, M. Hishida, S. Yata, W. Shinohara, A. Kitahara,
H. Yoneda, Y. Aya, I. Yoshida, M. Iseki, and M. Tanaka. SANYO‘s
R&D on thin-film silicon solar cells. In Proceedings of the 26t h European
Photovoltaic Solar Energy Conference and Exhibition, pages 2362–2365,
2011.
[31] K. L. Chopra, P. D. Paulson, and V. Dutta. Thin-film solar cells: An
overview. Progress in Photovoltaics: Research and Applications, 12(23):69–92, 2004.
104
Bibliography
[32] M. Izu and T. Ellison. Roll-to-roll manufacturing of amorphous silicon
alloy solar cells with in situ cell performance diagnostics. Solar Energy
Materials and Solar Cells, 78(1-4):613–626, 2003.
[33] B. Yan, G. Yue, L. Sivec, J. Yang, S. Guha, and C-S. Jiang. Innovative
dual function nc-SiOx :H layer leading to a > 16% efficient multi-junction
thin-film silicon solar cell. Applied Physics Letters, 99(11):113512, 2011.
[34] S. Fakirov, F. J. Balta Calleja, and M. Krumova. On the relationship
between microhardness and glass transition temperature of some amorphous polymers. Journal of Polymer Science Part B: Polymer Physics,
37(13):1413–1419, 1999.
[35] W.E. Spear, R.J. Roveland, and A. Al-Sharbaty. The temperature
dependence of photoconductivity in a-Si. Journal of Non-Crystalline
Solids, 15:410–422, 1974.
[36] A. Matsuda. Thin-film silicon - Growth process and solar cell application.
Japanese Journal of Applied Physics, 43:7909, 2004.
[37] J. K. Rath, M. M. de Jong, A. Verkerk, M. Brinza, and R. E. I. Schropp.
Gas phase conditions for obtaining device quality amorphous silicon at
low temperature and high deposition rate. Materials Research Society
Symposium Proceedings, 1153:463–468, 2009.
[38] A. D. Verkerk, M. M. de Jong, J. K. Rath, M. Brinza, R. E. I. Schropp,
W. J. Goedheer, V. V. Krzhizhanovskaya, Y. E. Gorbachev, K. E. Orlov,
E. M. Khilkevitch, and A. S. Smirnov. Compensation of decreased
ion energy by increased hydrogen dilution in plasma deposition of thin
film silicon solar cells at low substrate temperatures. Materials Science
and Engineering B: Solid-State Materials for Advanced Technology, 159160(C):53–56, 2009.
[39] Y. Ishikawa and M.B. Schubert. Flexible protocrystalline silicon solar
cells with amorphous buffer layer. Japanese Journal of Applied Physics,
45:6812–6822, 2006.
[40] J. K. Rath, Y. Liu, A. Borreman, E. A. G. Hamers, R. Schlatmann, G. J.
Jongerden, and R. E. I. Schropp. Thin film silicon modules on plastic
superstrates. Journal of Non-Crystalline Solids, 354(19-25):2381–2385,
2008.
Bibliography
105
[41] E.A.G. Hamers, E. Buiskool, E. Beverloo, D. Semerel, A. Heessels,
J. Donkerbroek, B. Kelder, D. Klomp, B.J. Kingma, and G.J. Jongerden. Transfer of 9% a-Si:H/µc-Si:H tandem technology to roll-to-roll
production for large area flexible PV laminates. In 26th European Photovoltaic Solar Energy Conference Proceedings, Hamburg, pages 2379 –
2383, 2011.
[42] Y. Ichikawa, T. Yoshida, T. Hama, H. Sakai, and K. Harashima. Production technology for amorphous silicon-based flexible solar cells. Solar
Energy Materials and Solar Cells, 66(1-4):107–115, 2001.
[43] T. Söderström, F.-L. Haug, V. Terrazzoni-Daudrix, X. Niquille,
M. Python, and C. Ballif. N/I buffer layer for substrate microcrystalline
thin film silicon solar cell. Journal of Applied Physics, 104(10):104505,
2008.
[44] H. Mase, M. Kondo, and A. Matsuda. Microcrystalline silicon solar cells
fabricated on polymer substrate. Solar Energy Materials and Solar Cells,
74(1-4):547–552, 2002.
[45] J. K. Rath, M. Brinza, Y. Liu, A. Borreman, and R. E. I. Schropp.
Fabrication of thin film silicon solar cells on plastic substrate by very high
frequency PECVD. Solar Energy Materials and Solar Cells, 94(9):1534–
1541, 2010.
[46] W. Beyer, J. Hüpkes, and H. Stiebig. Transparent conducting oxide films
for thin film silicon photovoltaics. Thin Solid Films, 516(2-4):147–154,
2007.
[47] M. Berginski, J. Hüpkes, M. Schulte, G. Schöpe, H. Stiebig, B. Rech,
and M. Wuttig. The effect of front ZnO:Al surface texture and optical
transparency on efficient light trapping in silicon thin-film solar cells.
Journal of Applied Physics, 101(7):074903, 2007.
[48] V. E. Ferry, M. A. Verschuuren, H. B. T. Li, R. E. I. Schropp, H. A.
Atwater, and A. Polman. Improved red-response in thin film a-Si:H
solar cells with soft-imprinted plasmonic back reflectors. Applied Physics
Letters, 95(18):183503, 2009.
[49] Y. Kuang, K. H. M. van der Werf, Z. S. Houweling, and R. E. I. Schropp.
Nanorod solar cell with an ultrathin a-Si:H absorber layer. Applied
Physics Letters, 98(11):1131111, 2011.
106
Bibliography
[50] Y. Kuang, K. H. M. van der Werf, Z. S. Houweling, M. Di Vece, and
R. E. I. Schropp. Fabrication and characterization of nanorod solar
cells with an ultrathin a-Si:H absorber layer. Journal of Non-Crystalline
Solids, 358:2209–2213, 2012.
[51] R. E. I. Schropp and M. Zeman. Amorphous and Microcrystalline Silicon Solar Cells - Modelling, Material and Device Technology. Kluwer,
Boston/Dordrecht/London, 1998.
[52] J.K. Rath, A. Verkerk, R.H. Franken, C. van Bommel, C.H.M. van der
Werf, A. Gordijn, and R.E.I. Schropp. Implementation of highly stable
microcrystalline silicon by vhf pecvd at high deposition rate in micromorph tandem cells. In Proc. IEEE 4th World Conference on Photovoltaic energy conversion, page 1473, 2006.
[53] E. A. G. Hamers. PhD thesis, Utrecht University, 1998.
[54] W. Theiss Hard- and Software. http://www.wtheiss.com.
[55] S. K. O’Leary, S. R. Johnson, and P. K. Lim. The relationship between
the distribution of electronic states and the optical absorption spectrum
of an amorphous semiconductor: An empirical analysis. Journal of Applied Physics, 82(7):3334–3340, 1997.
[56] N. Wyrsch, F. Finger, T. J. McMahon, and M. Vanecek. How to reach
more precise interpretation of subgap absorption spectra in terms of
deep defect density in a-Si:H. Journal of Non-Crystalline Solids, 137138(PART 1):347–350, 1991.
[57] J. E. Smith Jr., M. H. Brodsky, B. L. Crowder, M. I. Nathan, and
A. Pinczuk. Raman spectra of amorphous Si and related tetrahedrally
bonded semiconductors. Physical Review Letters, 26(11):642–646, 1971.
[58] A. Zwick and R. Carles. Multiple-order raman scattering in crystalline
and amorphous silicon. Physical Review B, 48(9):6024–6032, 1993.
[59] E. Bustarret, M. A. Hachicha, and M. Brunel. Experimental determination of the nanocrystalline volume fraction in silicon thin films from
Raman spectroscopy. Applied Physics Letters, 52(20):1675–1677, 1988.
[60] R. Hulstrom, R. Bird, and C. Riordan. Spectral solar irradiance data
sets for selected terrestrial conditions. Solar Cells, 15(4):365–391, 1985.
Bibliography
107
[61] G. S. Selwyn, J. Singh, and R. S. Bennett. In situ laser diagnostic studies of plasma-generated particulate contamination. Journal of Vacuum
Science & Technology A: Vacuum, Surfaces and Films, 7(4):2758–2765,
1989.
[62] G. Parascandolo, R. Bartlome, G. Bugnon, T. Söderström, B. Strahm,
A. Feltrin, and C. Ballif. Impact of secondary gas-phase reactions on
microcrystalline silicon solar cells deposited at high rate. Applied Physics
Letters, 96(23):233508, 2010.
[63] A. A. Fridman, L. Boufendi, T. Hbid, B. V. Potapkin, and A. Bouchoule.
Dusty plasma formation: Physics and critical phenomena. theoretical
approach. Journal of Applied Physics, 79(3):1303–1314, 1996.
[64] A.A. Howling, C. Courteille, J.-L. Dorier, L. Sansonnens, and Ch. Hollenstein. From molecules to particles in silane plasmas. Pure and Applied
Chemistrys, 68(5):1017–1022, 1996.
[65] R. Martins, V. Suva, I. Ferreira, A. Domingues, and E. Fortunato. Role
of the gas temperature and power to gas flow ratio on powder formation
and properties of films grown by the PECVD technique. Materials Science and Engineering B: Solid-State Materials for Advanced Technology,
69:272–277, 2000.
[66] A.D. Verkerk. Plasma deposition of thin film silicon at low substrate
temperature and at high growth rate. PhD thesis, Utrecht University,
2009.
[67] J. Perrin, C. Bohm, R. Etemadi, and A. Lloret. Possible routes for
cluster growth and particle formation in rf silane discharges. Plasma
Sources Science and Technology, 3(3):252–261, 1994.
[68] A. Bouchoule and L. Boufendi. Particulate formation and dusty plasma
behaviour in argon-silane RF discharge. Plasma Sources Science and
Technology, 2(3):204–213, 1993.
[69] C. Courteille, Ch. Hollenstein, J. L. Dorier, P. Gay, W. Schwarzenbach,
A. A. Howling, E. Bertran, G. Viera, R. Martins, and A. Macarico. Particle agglomeration study in silane plasmas: In situ study by polarizationsensitive laser light scattering. Journal of Applied Physics, 80(4):2069–
2078, 1996.
108
Bibliography
[70] S. Mitic, M. Y. Pustylnik, G. E. Morfill, and E. Kovacevic. In situ
characterization of nanoparticles during growth by means of white light
scattering. Optics Letters, 36(18):3699–3701, 2011.
[71] L. Boufendi, J. Gaudin, S. Huet, G. Viera, and M. Dudemaine. Detection
of particles of less than 5 nm in diameter formed in an argon-silane
capacitively coupled radio-frequency discharge. Applied Physics Letters,
79(26):4301–4303, 2001.
[72] Z. Shen and U. Kortshagen. Experimental study of the influence of
nanoparticle generation on the electrical characteristics of argon-silane
capacitive radio-frequency plasmas. Journal of Vacuum Science and
Technology, Part A: Vacuum, Surfaces and Films, 20(1):153–159, 2002.
[73] A Gordijn. Microcrystalline silicon for thin-film solar cells. PhD thesis,
Utrecht University, 2005.
[74] S. Mitic, M. Y. Pustylnik, and G. E. Morfill. Spectroscopic evaluation
of the effect of the microparticles on radio frequency argon plasma. New
Journal of Physics, 11:083020, 2009.
[75] Ph. Belenguer and J. P. Boeuf. Transition between different regimes of
RF glow discharges. Physical Review A, 41(8):4447–4459, 1990.
[76] H. Águas, L. Raniero, L. Pereira, E. Fortunato, and R. Martins. Effect
of the discharge frequency and impedance on the structural properties
of polymorphous silicon. Thin Solid Films, 451-452:264–268, 2004.
[77] M. A. Lieberman and A. J. Lichtenberg. Principles of Plasma Discharges
and Materials Processing. Wiley, 1994.
[78] J. Perrin and S. P. M. Schmitt. Emission cross sections from fragments
produced by electron impact on silane. Chemical Physics, 67(2):167–176,
1982.
[79] M. Takai, T. Nishimoto, M. Kondo, and A. Matsuda. Effect of highersilane formation on electron temperature in a silane glow-discharge
plasma. Applied Physics Letters, 77(18):2828–2830, 2000.
[80] A. Bouchoule, A. Plain, L. Boufendi, J. P. Blondeau, and C. Laure. Particle generation and behavior in a silane-argon low-pressure discharge
under continuous or pulsed radio-frequency excitation. Journal of Applied Physics, 70(4):1991–2000, 1991.
Bibliography
109
[81] K. B. Chai, W. Choe, C. R. Seon, and C. W. Chung. Role of hydrogen in
evolution of plasma parameters and dust growth in capacitively coupled
dusty plasmas. Applied Physics Letters, 97(20):201503, 2010.
[82] M. N. van den Donker, E. A. G. Hamers, and G. M. W. Kroesen. Measurements and semi-empirical model describing the onset of powder formation as a function of process parameters in an RF silane-hydrogen
discharge. Journal of Physics D: Applied Physics, 38(14):2382–2389,
2005.
[83] C. Anandan, C. Mukherjee, T. Seth, P. N. Dixit, and R. Bhattacharyya.
Effect of pulse parameters on the deposition rate of hydrogenated amorphous silicon in a modified pulsed plasma discharge. Applied Physics
Letters, 66:85–87, 1995.
[84] A. C. W. Biebericher, J. Bezemer, W. F. van der Weg, and W. J. Goedheer. Deposition rate in modulated radio-frequency silane plasmas. Applied Physics Letters, 76(15):2002–2004, 2000.
[85] C. Mukherjee, C. Anandan, T. Seth, P. N. Dixit, and R. Bhattacharyya.
Optoelectronic properties of hydrogenated amorphous silicon films grown
using a modified pulsed plasma discharge. Applied Physics Letters,
68:194–196, 1995.
[86] A. C. W. Biebericher. Deposition Techniques of Hydrogenated Amorphous Silicon using Modulated Radio-Frequency Plasmas. PhD thesis,
Utrecht University, 2002.
[87] M. Shiratani, S. Maeda, K. Koga, and Y. Watanabe. Effects of gas temperature gradient, pulse discharge modulation, and hydrogen dilution
on particle growth in silane rf discharges. Japanese Journal of Applied
Physics, Part 1: Regular Papers and Short Notes and Review Papers,
39(1):287–293, 2000.
[88] P. Chaudhuri, A. Bhaduri, A. Bandyopadhyay, S. Vignoli, P. P. Ray, and
C. Longeaud. High diffusion length silicon germanium alloy thin films
deposited by pulsed RF PECVD method. Journal of Non-Crystalline
Solids, 354(19-25):2105–2108, 2008.
[89] K. Kamataki, H. Miyata, K. Koga, G. Uchida, N. Itagaki, and M. Shiratani. Impacts of amplitude modulation of RF discharge voltage on the
growth of nanoparticles in reactive plasmas. Applied Physics Express,
4(10):105001, 2011.
110
Bibliography
[90] E. A. G. Hamers, W. G. J. H. M. Van Sark, J. Bezemer, H. Meiling,
and W. F. van der Weg. Structural properties of a-Si:H related to ion
energy distributions in VHF silane deposition plasmas. Journal of NonCrystalline Solids, 226(3):205–216, 1998.
[91] J. Perrin, A. Lloret, G. De Rosny, and J. P. M. Schmitt. Positive and
negative ions in silane and disilane multipole discharges. International
Journal of Mass Spectrometry and Ion Processes, 57(3):249–281, 1984.
[92] W. Beyer. Hydrogen effusion: a probe for surface desorption and diffusion. Physica B: Physics of Condensed Matter, 170(1-4):105–114, 1991.
[93] A. Fontcuberta I Morral, P. Roca I Cabarrocas, and C. Clerc. Structure and hydrogen content of polymorphous silicon thin films studied by
spectroscopic ellipsometry and nuclear measurements. Physical Review
B - Condensed Matter and Materials Physics, 69(12):1253071–12530710,
2004.
[94] R. A. Street, J. Kakalios, C. C. Tsai, and T. M. Hayes. Thermalequilibrium processes in amorphous silicon.
Physical Review B,
35(3):1316–1333, 1987.
[95] G. Ganguly and A. Matsuda. Defect formation during growth of hydrogenated amorphous silicon. Physical Review B, 47:3661–3670, 1993.
[96] P. C. P. Bronsveld, J. K. Rath, R. E. I. Schropp, A. Mates, B. Fejfar,
Rezek, and J. Kočka. Internal structure of mixed phase hydrogenated
silicon thin films made at 39◦ C. Applied Physics Letters, 89(5), 2006.
[97] M. Brinza, J. K. Rath, and R. E. I. Schropp. Thin film silicon n-i-p solar
cells deposited by VHF PECVD at 100◦ C substrate temperature. Solar
Energy Materials and Solar Cells, 93(6-7):680–683, 2009.
[98] C. Koch, M. Ito, and M. Schubert. Low-temperature deposition of
amorphous silicon solar cells. Solar Energy Materials and Solar Cells,
68(2):227–236, 2001.
[99] W. G. J. H. M. Van Sark, H. Meiling, J. Bezemer, M. B. Von Der Linden,
R. E. I. Schropp, and W. F. van der Weg. Inhomogeneities in PECVD
deposited a-Si:H films induced by a spacing between substrate and substrate holder. Solar Energy Materials and Solar Cells, 45(1):57–63, 1997.
[100] P. Bronsveld. Development of Hydrogenated Silicon Solar Cells and Materials < 100◦ C. PhD thesis, Utrecht University, 2012.
Bibliography
111
[101] W.M. Haynes and D.R. Lide, editors. CRC Handbook of Chemistry and
Physics, 92th edition, Internet Version 2012. CRC Press, 2012.
[102] J. Blumm, A. Lindemann, B. Niedrig, and R. Campbell. Measurement of
selected thermophysical properties of the npl certified reference material
stainless steel 310. International Journal of Thermophysics, 28(2):674–
682, 2007.
[103] M. N. van den Donker, R. Schmitz, W. Appenzeller, B. Rech, W. M. M.
Kessels, and M. C. M. van de Sanden. The role of plasma induced
substrate heating during high rate deposition of microcrystalline silicon
solar cells. Thin Solid Films, 511-512:562–566, 2006.
[104] U. Kroll, J. Meier, A. Shah, S. Mikhailov, and J. Weber. Hydrogen
in amorphous and microcrystalline silicon films prepared by hydrogen
dilution. Journal of Applied Physics, 80(9):4971–4975, 1996.
[105] O. Vetterl, F. Finger, R. Carius, P. Hapke, L. Houben, O. Kluth, A. Lambertz, A. Mück, B. Rech, and H. Wagner. Intrinsic microcrystalline silicon: A new material for photovoltaics. Solar Energy Materials and Solar
Cells, 62(1):97–108, 2000.
[106] T. Roschek, T. Repmann, J. Müller, B. Rech, and H. Wagner. Comprehensive study of microcrystalline silicon solar cells deposited at high rate
using 13.56 MHz plasma-enhanced chemical vapor deposition. Journal of
Vacuum Science and Technology, Part A: Vacuum, Surfaces and Films,
20(2):492–498, 2002.
[107] C. Droz, E. Vallat-Sauvain, J. Bailat, L. Feitknecht, J. Meier, and
A. Shah. Relationship between raman crystallinity and open-circuit
voltage in microcrystalline silicon solar cells. Solar Energy Materials
and Solar Cells, 81(1):61–71, 2004.
[108] J. C. Lee, V. Dutta, J. Yoo, J. Yi, J. Song, and K. H. Yoon. Superstrate
p-i-n a-Si:H solar cells on textured zno:al front transparent conduction
oxide. Superlattices and Microstructures, 42(1-6):369–374, 2007.
[109] T. Söderström, F. L. Haug, X. Niquille, and C. Ballif. Tcos for NIP
thin film silicon solar cells. Progress in Photovoltaics: Research and
Applications, 17(3):165–176, 2009.
[110] J. K. Rath, Y. Liu, M. M. de Jong, J. de Wild, J. A. Schuttauf,
M. Brinza, and R. E. I. Schropp. Transparent conducting oxide layers
112
Bibliography
for thin film silicon solar cells. Thin Solid Films, 518(24 SUPPL.):e129–
e135, 2010.
[111] B. Curtin, R. Biswas, and V. Dalal. Photonic crystal based back reflectors for light management and enhanced absorption in amorphous silicon
solar cells. Applied Physics Letters, 95(23), 2009.
[112] R. Dewan, M. Marinkovic, R. Noriega, S. Phadke, A. Salleo, and
D. Knipp. Light trapping in thin-film silicon solar cells with submicron
surface texture. Optics Express, 17(25):23058–23065, 2009.
[113] A. V. Shah, H. Schade, M. Vanecek, J. Meier, E. Vallat-Sauvain,
N. Wyrsch, U. Kroll, C. Droz, and J. Bailat. Thin-film silicon solar
cell technology. Progress in Photovoltaics: Research and Applications,
12(2-3):113–142, 2004.
[114] J. Müller, B. Rech, J. Springer, and M. Vanecek. TCO and light trapping
in silicon thin film solar cells. Solar Energy, 77(6):917–930, 2004.
[115] C. Haase and H. Stiebig. Thin-film silicon solar cells with efficient periodic light trapping texture. Applied Physics Letters, 91:061116, 2007.
[116] R. Dewan, I. Vasilev, V. Jovanov, and D. Knipp. Optical enhancement
and losses of pyramid textured thin-film silicon solar cells. Journal of
Applied Physics, 110(1):013101, 2011.
[117] A. Čampa, O. Isabella, R. Van Erven, P. Peeters, H. Borg, J. Krč,
M. Topič, and M. Zeman. Optimal design of periodic surface texture
for thin-film a-Si:h solar cells. Progress in Photovoltaics: Research and
Applications, 18(3):160–167, 2010.
[118] G. Gomard, E. Drouard, X. Letartre, X. Meng, A. Kaminski, A. Fave,
M. Lemiti, E. Garcia-Caurel, and C. Seassal. Two-dimensional photonic
crystal for absorption enhancement in hydrogenated amorphous silicon
thin film solar cells. Journal of Applied Physics, 108(12):123102, 2010.
[119] V. E. Ferry, A. Polman, and H. A. Atwater. Modeling light trapping in
nanostructured solar cells. ACS Nano, 5(12):10055–10064, 2011.
[120] J. Zhu, C. M. Hsu, Z. Yu, S. Fan, and Y. Cui. Nanodome solar cells with
efficient light management and self-cleaning. Nano Letters, 10(6):1979–
1984, 2010.
Bibliography
113
[121] C. Eisele, C. E. Nebel, and M. Stutzmann. Periodic light coupler gratings in amorphous thin film solar cells. Journal of Applied Physics,
89(12):7722–7726, 2001.
[122] M. Heckele and W. K. Schomburg. Review on micro molding of thermoplastic polymers. Journal of Micromechanics and Microengineering,
14(3):R1–R14, 2004.
[123] C. Rockstuhl, S. Fahr, F. Lederer, K. Bittkau, T. Beckers, and R. Carius.
Local versus global absorption in thin-film solar cells with randomly
textured surfaces. Applied Physics Letters, 93(6):061105, 2008.
[124] J. E. Cotter. Raysim 6.0 - a free geometrical ray tracing program for
silicon solar cells. In Proceedings of the 31st PVSC, Lake Buena Vista,
USA, volume 1066, pages 1165–1168, 2005.
[125] Y. S. Jung, H. W. Choi, K. H. Kim, S. J. Park, and H. H. Yoon. Properties of AZO thin films for solar cells deposited on polycarbonate substrates. Journal of the Korean Physical Society, 55:1945–1949, 2009.
[126] J. Lee, D. Lee, D. Lim, and K. Yang. Structural, electrical and optical
properties of ZnO:Al films deposited on flexible organic substrates for
solar cell applications. Thin Solid Films, 515:6094–6098, 2007.
[127] A. Witvrouw and F. Spaepen. Viscosity and elastic constants of amorphous Si and Ge. Journal of Applied Physics, 74(12):7154–7161, 1993.
[128] P. Zoller. Study of the pressure-volume-temperature relationships of
four related amorphous polymers: Polycarbonate, polyarylate, phenoxy
and polysulfone. Journal of polymer science.Part A-2, Polymer physics,
20(8):1453–1464, 1982.
[129] P. Alpuim, V. Chu, and J. P. Conde. Amorphous and microcrystalline
silicon films grown at low temperatures by radio-frequency and hot-wire
chemical vapor deposition. Journal of Applied Physics, 86(7):3812–3821,
1999.
[130] H. B. T. Li, R. H. Franken, J. K. Rath, and R. E. I. Schropp. Structural
defects caused by a rough substrate and their influence on the performance of hydrogenated nano-crystalline silicon n-i-p solar cells. Solar
Energy Materials and Solar Cells, 93(3):338–349, 2009.
114
Bibliography
[131] E. Vallat-Sauvain, J. Bailat, J. Meier, X. Niquille, U. Kroll, and A. Shah.
Influence of the substrate’s surface morphology and chemical nature on
the nucleation and growth of microcrystalline silicon. Thin Solid Films,
485(1-2):77–81, 2005.
[132] T. J. McMahon and M. S. Bennett. Metastable shunt paths in a-Si solar
cells. Solar Energy Materials and Solar Cells, 41-42:465–473, 1996.
[133] S. Dongaonkar, J. D. Servaites, G. M. Ford, S. Loser, J. Moore, R. M.
Gelfand, H. Mohseni, H. W. Hillhouse, R. Agrawal, M. A. Ratner,
T. J. Marks, M. S. Lundstrom, and M. A. Alam. Universality of nonohmic shunt leakage in thin-film solar cells. Journal of Applied Physics,
108(12):124509, 2010.
[134] C. Bauwens-Crowet and J.-C. Bauwens. Annealing of polycarbonate
below the glass transition: quantitative interpretation of the effect on
yield stress and differential scanning calorimetry measurements. Polymer, 23(11):1599–1604, 1982.
[135] S. Schicho, D. Hrunski, R. Van Aubel, and A. Gordijn. High potential
of thin (< µm) a-Si:H/µc-Si:H tandem solar cells. Progress in Photovoltaics: Research and Applications, 18(2):83–89, 2010.
[136] F. Meillaud, A. Feltrin, M. Despeisse, F-J. Haug, D. Domin, M. Python,
T. Söderström, P. Cuony, M. Boccard, S. Nicolay, and C. Ballif. Realization of high efficiency micromorph tandem silicon solar cells on glass
and plastic substrates: Issues and potential. Solar Energy Materials and
Solar Cells, 95(1):127–130, 2011.
[137] A. Gordijn, J. K. Rath, and R. E. I. Schropp. High-efficiency µc-si
solar cells made by very high-frequency plasma-enhanced chemical vapor deposition. Progress in Photovoltaics: Research and Applications,
14(4):305–311, 2006.
[138] F. A. Rubinelli, R. L. Stolk, A. Sturiale, J. K. Rath, and R. E. I. Schropp.
Sensitivity of the dark spectral response of thin film silicon based tandem
solar cells on the defective regions in the intrinsic layers. Journal of NonCrystalline Solids, 352(9-20 SPEC. ISS.):1876–1879, 2006.
[139] J. Löffler, A. Gordijn, R. L. Stolk, H. Li, J. K. Rath, and R. E. I. Schropp.
Amorphous and ’micromorph’ silicon tandem cells with high open-circuit
voltage. Solar Energy Materials and Solar Cells, 87(1-4):251–259, 2005.
Summary
In the search for new and renewable energy sources, solar energy can fulfil
a large part of the growing demand. The biggest threshold for large-scale
solar energy harvesting is the price of the solar panels, of which at present
the workhorse is the crystalline silicon solar cell, made from silicon wafers. A
method to decrease solar panel prices is the use of thin-film solar cells, which
require only a fraction of the raw material compared to crystalline silicon cells.
For further cost reductions, continuous fabrication using low-cost substrates
can be a solution. In this thesis, we investigate the possibilities of depositing
thin film film solar cells directly onto cheap plastic substrates. Apart from the
low cost, flexible solar cell can be used on speciality products, such as clothing
or security papers.
Thin silicon films are deposited from the gas phase, in our case using (very
high frequency) plasma-enhanced chemical vapour deposition. In this process,
feedstock gasses (silane and hydrogen) are decomposed in a reaction chamber
using a radio frequency discharge. Decomposition of these gasses produces
radicals, which can reach the substrate, where a thin film grows. In these types
of processes, the temperature of the growing surface has a large influence on
the quality of the grown films.
Because plastic substrates limit the maximum tolerable substrate temperature, new methods have to be developed to produce device-grade silicon layers. But lowering the substrate temperature does not only alter the behaviour
of depositing species on the growth surface of the films grown, it also changes
the behaviour of the plasma inside the reaction chamber. Apart from growing
a film on the substrate, silyl radicals can also polymerize in the plasma bulk.
If the resulting particles are negatively charged, they are trapped inside the
plasma and continue to grow. At a certain size and density, they will coagulate and form dust particles in the plasma, which can cause a serious threat to
device operation if they are captured in the thin films. We have shown that using an all-optical technique, we can identify whether dust particles are present
116
Summary
in the plasma or not. For this, we study the axial optical emission from the
plasma (caused by relaxation of excited state of species in the plasma). Due
to the presence of the particles in the plasma that capture electrons, the electron temperature increases, inducing a higher optical emission. Because the
(large) dust particles encounter gravity, they are pulled towards the bottom
of the reactor and therefore the plasma shows an asymmetric optical emission
profile.
To understand why the formation of dust is temperature-sensitive, we monitored the formation of polysilanes as a function of temperature using a massspectrometer attached to a plasma reactor chamber. Counting the different
polysilane radicals at different temperatures in a dust-free regime, but in the
dust forming incubation phase, we found the polymerization rate to be influenced by the substrate temperature, which can explain the temperature
dependence of dust formation.
As a substrate material for solar cells, we chose polycarbonate (PC), because of its excellent transparency and its relatively high glass transition temperature of 130-140°C. At 130° we searched for deposition recipes that yield
good quality silicon layers. For this purpose we first investigated how we can
accurately control the substrate temperature. Diluting the feedstock silane gas
with hydrogen has a large influence on the material properties. In the case of
amorphous silicon (a-Si), increasing the hydrogen dilution generally improves
the quality of the silicon until we reach the nanocrystalline silicon (nc-Si) regime. Just before this regime, the a-Si layers show high intrinsic stress, which
might result in detachment of the silicon layer from the substrate. In the nc-Si
regime, together with changing the power input into the plasma, the hydrogen
dilution can be used to control the volume fraction of crystallites within the
silicon layer.
Using these silicon layers, including doped silicon layers at low temperature,
a-Si thin film solar cells were fabricated with an intrinsic layer thickness of
275 nm, both on glass and PC substrates. Because low temperature silicon is
generally not as good as its high temperature counterpart, recombination of
photogenerated charge carriers can be a problem, resulting in a lower Voc and
fill factor. These problems can be mitigated when thinner silicon layers are
used and therefore an adequate light trapping technique needs to be employed.
For a-Si cells, we have simulated and experimentally tested three light trapping
techniques, using embossed structures in PC substrates and random structures
on glass, using features of different sizes: regular pyramid structures larger
than the wavelength of light (micropyramids), pyramid structures comparable
to the wavelength of light (nanopyramids) and random nano-textures as used
in commercial fabrication (Asahi U-type TCO glass). Both micropyramid
Summary
117
and nanopyramid substrates enhance the light absorption within the cells.
Using nanopyramid substrates we could achieve current densities in cells on
PC comparable to current densities achieved on Asahi U-type TCO glass.
Using these techniques we could achieve initial conversion efficiencies for a-Si
cells on PC of 6.4% on micropyramid substrates and 7.4% on nanopyramid
substrates, compared to 7.6% for cells deposited under identical conditions on
Asahi U-type TCO glass. For nc-Si cells on texture etched aluminium doped
zinc oxide (ZnO:Al) on glass, we could achieve an initial conversion efficiency
of 6.9% using a very thin absorber layer of 750 nm.
Combining low temperature a-Si and nc-Si cells we fabricated tandem solar
cells in the ’micromorph’ concept at 130°C. By optimizing the thicknesses of
the different silicon layers and controlling the crystalline fraction of the bottom
(nc-Si) cell, we could achieve an initial conversion efficiency of 9.5% on texture
etched ZnO:Al coated glass. When this recipe was transferred to structured
PC substrates, the crystal nucleation behaved differently, resulting in a-Si/aSi tandem cells, which suffer from a very low current, due to bad current
matching. Changing the hydrogen dilution could solve this problem. The
light trapping in structured PC substrates is less pronounced in the bottom
cell, compared to texture etched ZnO:Al, which is used for light trapping in
our tandem cells on glass.
Samenvatting
Op zoek naar nieuwe betaalbare alternatieve energiebronnen is zonne-energie
een veelbelovende. De aarde ontvangt meer dan 1000 keer zoveel energie dan
er verbruikt wordt. De grootste belemmering voor grootschalige oogst van
zonne-energie is de prijs van zonnepanelen, waarvan het meestgebruikte type
de zogenaamde kristallijne zonnecel is, gemaakt van in plakken gezaagde brokken extreem zuiver en daardoor duur silicium. Een alternatief voor deze techniek is het gebruik van dunne films, waarvan de productie slechts een fractie van het materiaal van zijn kristallijne tegenhanger nodig heeft. Verdere
kostenbesparingen zouden kunnen worden geboekt door goedkope substraten
te gebruiken, het liefst in een continu proces. Dit proefschrift beschrijft mijn
onderzoek naar technieken om dunne-film silicium zonnecellen direct te produceren op goedkope flexibele plastic substraten. Los van de kostenbesparingen,
zouden flexibele zonnecellen niche-markten kunnen bedienen, zoals zonnecellen
op kleding of waardepapieren.
Dunne silicium films worden gefabriceerd in vacuümreactors, waar een
plasma de procesgassen silaan (SiH4 ) en waterstof (H2 ) ontleedt tot radicale moleculen, die wanneer ze op het substraat neerslaan, daar langzaam
een dunne laag vormen. Deze radicalen kunnen, buiten een laag vormen, ook
aan elkaar plakken, zodat in het plasma materieklontjes of ’stof’ ontstaat. Dit
proces van stofvorming blijkt erg temperatuurgevoelig te zijn. Omdat ons uiteindelijk doel het fabriceren van zonnecellen op plastic substraten is, kunnen
we niet met hoge temperaturen werken. Omdat we hebben gekozen voor polycarbonaat (PC) (omdat het hoogtransparant is en verwerkt kan worden bij
relatief hoge temperaturen) mag de temperatuur niet hoger worden dan 130°C
om de substraten niet te beschadigen. De eerste stap bij het ontstaan van stof
is het groeien van silaanclusters. Als deze clusters negatief geladen zijn, worden
de deeltjes ’gevangen’ in de depositiereactor, waar ze verder kunnen groeien.
Bij een bepaalde grootte en dichtheid van de clusters klonteren ze samen en
vormen stofdeeltjes, die de werking van de gedeponeerde zonnecellen kunnen
120
Samenvatting
saboteren. We laten zien dat we kunnen aantonen of er stofdeeltjes aanwezig
zijn in het plasma, op basis van een optische techniek. Hiervoor bestuderen
we de optische emissie (door relaxatie van aangeslagen atomen en moleculen)
van het plasma als functie van de positie in het plasma. Stofdeeltjes vangen
elektronen in, waardoor de elektrontemperatuur lokaal stijgt, resulterend in
een hogere optische emissie. Doordat de stofdeeltjes zich naar de onderkant
van de reactor begeven door de zwaartekracht, verraden zij hun aanwezigheid
door een asymmetrisch uitgezonden emissieprofiel.
Met een massaspectrometer bestudeerden we de grootte en en aanwezigheid van clusters in het plasma, als functie van de temperatuur, om zo inzicht
te krijgen in de temperatuurafhankelijkheid van stofvorming. Er bleek een
temperatuurafhankelijkheid te bestaan van de groei van de clusters, wat kan
verklaren waarom plasma’s bij lage temperatuur eerder stofproducerend worden.
Bij een substraattemperatuur van 130°C (de maximaal toelaatbare temperatuur voor het gebruik van PC) zochten we naar depositiemethodes voor
siliciumlagen van goede kwaliteit. Hiervoor was het belangrijk de substraattemperatuur nauwkeurig te kunnen beheersen. In het algemeen geldt voor
de productie van amorf silicium (a-Si), dat het verhogen van de waterstofverdunning van het brongas silaan, de kwaliteit van het materiaal positief
beïnvloedt, totdat het materiaal nanokristallijn wordt. Net voor deze overgang vertoont het materiaal hoge interne spanning, wat tot gevolg kan hebben
dat de siliciumlagen losspringen van het substraat. In het nc-Si gebied kan de
kristalfractie van het materiaal beïnvloed worden door het veranderen van de
waterstofverdunning en door het veranderen van het toegevoerde vermogen.
Met deze siliciumlagen, samen met gedoteerde siliciumlagen gedeponeerd
bij lage temperatuur, fabriceerden we a-Si zonnecellen met een intrinsieke laag
van 275 nm, zowel op glassubstraten als op PC substraten. Omdat over het
algemeen siliciumlagen gedeponeerd bij lage temperatuur niet zo goed van
kwaliteit zijn als lagen gedeponeerd bij optimale (hogere) temperaturen, kan
recombinatie van door licht gegeneerde ladingsdragers de werking van zonnecellen verslechteren, door een lagere open-klemspanning en vulfactor. Deze
problemen kunnen verminderd worden door een dunnere laag intrinsiek silicium te gebruiken, maar hierdoor wordt een goede lichtopsluitingstechniek
onmisbaar. Voor a-Si hebben we verschillende lichtopsluitingstechnieken gesimuleerd en experimenteel getest, door gebruik te maken van geperste structuren in PC en willekeurige piramidestructuren op glas: Regelmatige piramides,
veel groter dan de effectieve golflengte van zichtbaar licht op PC (micropiramides), regelmatige piramidestructuren vergelijkbaar met de effectieve golflengte
van licht op PC (nanopiramides) en piramidestructuren op nanoschaal, zo-
Samenvatting
121
als gebruikelijk is voor commerciële zonnecelproductie (Asahi U-type). Beide
structuren op PC zorgen voor een verhoogde lichtabsorptie van de cellen. Met
micropiramide substraten behaalden we een initiële conversie-efficiëntie van
6.4% en 7.4% op nanopiramide substraten, beide op PC. Cellen gedeponeerd
onder dezelfde omstandigheden op Asahi U-type hadden een initiële efficiëntie van 7.6%. Nanokristallijn silicium cellen met een ruwgeëtste aluminium
gedoteerde zinkoxidelaag (ZnO:Al) op glassubstraten vertoonden een initiële
efficiëntie van 6.9%, met een intrinsieke absorptielaag van slechts 750 nm.
Nadat we de dikte van de intrinsieke laag van de nc-Si cel hadden geoptimaliseerd, vertoonde een combinatie van een a-Si cel en een nc-Si zonnecel
in een tandemstructuur (het zogenaamde micromorph concept), gedeponeerd
bij 130°, een initiële efficiëntie van 9.5%, op ruwgeëtste ZnO:Al. Wanneer
dezelfde fabricagemethode werd gebruikt om cellen aan te groeien op gestructureerd PC, bleek dat de kristalgroei van de nc-Si cel zich anders gedroeg, wat
resulteerde in a-Si/a-Si tandemstructuren. Door de waterstofverdunning van
de de nc-Si intrinsieke laag te veranderen konden we toch een micromorphe cel
maken. De lichtopsluiting in de nc-Si deelcel van de tandemcel op gestructureerd PC werkte minder goed dan die van de tandemcel op ruwgeëtste ZnO:Al.
List of Publications
Publications within the scope of this thesis
M.M. de Jong, J. de Koning, J.K. Rath and R.E.I. Schropp, Identification
of various plasma regimes in very high frequency PECVD of amorphous and
nanocrystalline silicon near the phase transition, P roceedings of the 25th
EU P V SEC Conf erence, V alencia, Spain, 3149-3151, 2010.
M.M. de Jong, A. Mohan, J.K. Rath, and R.E.I. Schropp, Temperature dependence of the ion energy distribution in a hydrogen diluted silane VHF plasma,
AIP Conf erence P roceedings, 1397, 411-412, 2011.
M.M. de Jong, J.K. Rath, R.E.I. Schropp, P.J. Sonneveld, G.L.A.M. Swinkels,
H.J. Holterman, J. Baggerman and C.J.M. van Rijn, Geometric light confinement in a-Si thin film solar cells on micro-structured substrates, P roceedings
of the 26th EU P V SEC Conf erence, Hamburg, Germany, 370-372, 2011.
M.M. de Jong, J. De Koning, J.K. Rath and R.E.I. Schropp. An optical
analysis tool for avoiding dust formation in very-high frequency hydrogen diluted silane plasmas at low substrate temperatures, P hysics of P lasmas, 19,
020703, 2012.
M.M. de Jong, J.K. Rath, R.E.I. Schropp, P.J. Sonneveld, G.L.A.M. Swinkels,
H.J. Holterman, J. Baggerman, C.J.M. van Rijn and E.A.G. Hamers, A novel
structured plastic substrate for light confinement in thin film silicon solar cells
by a geometric optical effect, Journal of N on − Crystalline Solids, 358(17),
2308-2312, 2012.
M.M. de Jong, J.K. Rath and R.E.I. Schropp, Very thin micromorph tandem solar cells deposited at low substrate temperature, M aterials Research
124
List of Publications
Society Symposium P roceedings, 1426, 45-49, 2012.
M.M. de Jong, J. Baggerman, C.J.M. van Rijn, P.J. Sonneveld, G.L.A.M.
Swinkels, H.J. Holterman, J.K.Rath and R.E.I. Schropp, Scattering, diffraction and geometric light trapping in thin film amorphous silicon solar cells
on plastic substrates, M aterials Research Society Symposium P roceedings,
1426, 155-160, 2012.
M.M. de Jong, P.J. Sonneveld, J. Baggerman, C.J.M. van Rijn, J.K. Rath and
R.E.I. Schropp, Utilization of geometric light trapping in thin film silicon solar
cells: Simulations and experiments, P rogress in P hotovoltaics, published
online, DOI: 10.1002/pip.2299
Publications outside the scope of this thesis
J.K. Rath, M.M. de Jong, A. Verkerk, M. Brinza and R.E.I. Schropp, Gas
phase conditions for obtaining device quality amorphous silicon at low temperature and high deposition rate, M aterials Research Society Symposium
P roceedings, 1153, 463-468, 2009.
A.D. Verkerk, M M. de Jong, J.K. Rath, M. Brinza, R.E.I. Schropp, W.J.
Goedheer, V.V. Krzhizhanovskaya, Y.E. Gorbachev, K.E. Orlov, E.M. Khilkevitch and A.S. Smirnov, Compensation of decreased ion energy by increased
hydrogen dilution in plasma deposition of thin film silicon solar cells at low
substrate temperatures, Materials Science and Engineering B : Solid − State
M aterials f or Advanced T echnology, 159-160(C), 53-56, 2009.
J.K. Rath, Y. Liu, M.M. de Jong, J. De Wild, J.A. Schuttauf, M. Brinza and
R.E.I. Schropp, Transparent conducting oxide layers for thin film silicon solar
cells, T hin Solid F ilms, 518(24SUPPL.), e129-e135, 2010.
Nawoord
Nou, het is af. Juhluh!
Het schrijven van een proefschrift is veel werk. Gelukkig heb ik het niet
alleen hoeven doen. Daarom wil ik graag een aantal mensen bedanken.
Allereerst dank ik mijn begeleiders. Jatin, dank voor de dagelijkse begeleiding, voor je rijke kennis als het gaat om halfgeleiders en voor je positieve
kijk op de wereld, maar vooral op wetenschappelijke resultaten. Ik vond het
prettig dat je altijd beschikbaar was voor advies. Ruud ben ik vooral dankbaar dat hij me de gelegenheid gaf om aan dit project te beginnen en voor
het nauwkeurig lezen en verbeteren van alle teksten, plaatjes en praatjes die
ik heb geproduceerd de afgelopen jaren.
Onderzoek met een sterk technische ondertoon kan niet zonder technici.
Ik ben dan ook mijn dank verschuldigd aan Bart, Martin, Karine, Caspar,
Gerard, Ruurd en Roberto voor het geduldig maken van talloze laagjes (zeker
bij voorstellen als ’Misschien moeten we deze serie nog een keer doen’), oplossen
van vacuümproblemen, op peil houden van de characterisatietools en hun kijk
op praktische problemen.
Piet (en later Jim en Theo), Gert-Jan en Henk Jan uit Wageningen dank ik
voor de samenwerking. De inkijkjes in de wondere wereld van de kastuinbouw
vond ik intrigerend. Cees en Jacob dank ik voor de verschillende substraten.
Zonder had ik een groot deel van mijn onderzoek niet uit kunnen voeren.
Het werk van mijn studenten die ik begeleid heb: Jaap, Rob, Robin, Yalda
en in zekere zin kleine Casper hebben allemaal een plekje in het proefschrift
gevonden, waarvoor dank.
Uiteindelijk ben ik lang aanwezig geweest in de groep ’Physics of Devices’.
Dat is geen toeval. Ik heb het altijd een leuke groep mensen gevonden. Als
dunnefilmgroentje werd ik aan de arm genomen door Arjan, Monica en Hongbo. Met ’generatiegenoten’ Yanchao, Jan-Willem, Jessica, Sylvester, RuudB
en Diederick heb ik veel plezier gehad, op het werk, maar ook op conferenties
126
Nawoord
en daarbuiten. Verder hoop ik dat Kuang, Henriette, Xin, Zachar, Kees, Pim,
Akshatha, Caterina, Oumkelthoum, Lourens, Marcel en Wilfried net zoveel
plezier zonder me hebben. De lol was er voor mij even af toen duidelijk werd
dat we als groep naar Eindhoven moesten verdwijnen, maar ik hoop dat jullie
daar ook je draai vinden. Zoals Riny laatst zei: ’Het was gewoon een heel gezellig zootje en dat mis ik.’ Riny, dank voor je gezelligheid en levenswijsheid.
Ik denk dat het voor mij ook zo gaat zijn.
Curriculum Vitae
The author was born on March 6, 1981 in Laren (NH), the Netherlands. He
obtained his secondary school degree in 1999 from ’Gymnasium Celeanum’ in
Zwolle. From 2000 to 2008 he studied Physics at Utrecht University, from
which he graduated in 2008 with a master degree titled ’Nanomaterials: Chemistry and Physics’. For his master research he carried out research on the ion
energy distributions in silane and hydrogen deposition plasmas in the group
’Physics of Devices’ at Utrecht University. In the same group, he started his
PhD research under the supervision of Prof. dr. R.E.I. Schropp and Dr. J.K.
Rath on the topic of thin film silicon solar cells deposited at low deposition
temperatures and direct deposition of these cells on plastic substrates. Several
light trapping techniques were investigated. The results are presented in this
thesis.