You are on page 1of 8

Solar Energy Materials & Solar Cells 238 (2022) 111412

Contents lists available at ScienceDirect

Solar Energy Materials and Solar Cells


journal homepage: www.elsevier.com/locate/solmat

Intrinsic layer modification in silicon heterojunctions: Balancing transport


and surface passivation
Christoph Luderer , Dilara Kurt , Anamaria Moldovan , Martin Hermle , Martin Bivour *
Fraunhofer Institute for Solar Energy Systems, Heidenhofstrasse 2, 79110, Freiburg, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: In this work, the intrinsic hydrogenated amorphous silicon (a-Si:H) layer of silicon heterojunction (SHJ) solar
Amorphous silicon cells was modified to improve carrier transport while maintaining excellent passivation of the c-Si absorber
Transport surface. The microstructure of different multilayer intrinsic a-Si:H films was measured and its influence on
Surface passivation
contact resistance and passivation quality was investigated thoroughly. We show that a pronounced trade-off
Microstructure
Silicon heterojunction
between passivation and transport exists and that this trade-off is governed by the a-Si:H properties close to
Contact resistance the c-Si surface. The fact that the same trend was observed for hole and electron contact suggests that the
transport barrier formed by the interfacial a-Si:H layer is governed by a higher resistivity of the void-rich
interfacial layer or a less pronounced induced junction and not by asymmetric hole or electron barriers (band
offsets, tunnel efficiencies, …). Modified intrinsic layers have been tested on cell level, resulting in a series
resistance (Rs) reduction by about 0.3 Ωcm2 and an increase in fill factor (FF) by roughly 1.0 %abs. The power
conversion efficiency (η) was improved by about 0.3 %abs with respect to our baseline. Further, the beneficial
effect of a hydrogen plasma treatment (HPT) on passivation and transport of the hole contact was shown on
device level.

1. Introduction detrimental effects of the void-rich a-Si:H layer on Rs and the


short-circuit current density (Jsc) have been reported by Sai et al. [11]
The excellent surface passivation of crystalline silicon (c-Si) by a thin for the hole contact. They investigated the modification of the
hydrogenated amorphous silicon (a-Si:H) layer [1,2] enables world re­ i1-microstructure by varying deposition parameters like pressure and
cord efficiencies of SHJ solar cells [3]. A key to this outstanding surface power density and its influence on the passivation quality [11].
passivation is the prevention of epitaxial growth at the a-Si:H/c-Si Recently, Wu et al. reported on an increased valence band offset for
interface [4,5]. It was found that for single intrinsic layers, surface void-rich, high band gap a-Si:H(i) layers, being especially detrimental
passivation properties are best just before the transition from amor­ for the transport of holes [12]. Moreover, the influence of a hydrogen
phous to micro-crystalline Si [4,6], presumably because the a-Si:H plasma treatment (HPT) before, after or in between intrinsic a-Si:H
network is most relaxed in that case and comprises the lowest dangling deposition was studied by several groups [10,13–17] with respect to
bond density [7]. Therefore, careful control of the a-Si:H deposition improving the surface passivation.
parameters is necessary. Multilayer intrinsic a-Si films with proper In this work, we follow a similar approach as Sai et al. but extend the
interfacial layer properties have been tested to relax the requirements investigations by using test structures to directly compare passivation
for a single layer, i.e., the combination of efficient hydrogenation and and contact resistivity (ρc) of the different intrinsic layer stacks. We also
the avoidance of epitaxial growth [8–10]. address the role of the HPT after a-Si:H(i1) and prior to a-Si:H(i2)
A void-rich a-Si:H interfacial layer (i1) deposited in pure SiH4 plasma deposition to further modify passivation and transport. We performed
in combination with a dense a-Si:H layer (i2) deposited in highly H2 similar investigations for the hole and electron contact to check if an
diluted plasma has been shown to increase the passivation quality of SHJ intrinsic difference exists between both types of contacts. By doing so we
cells (compared to single intrinsic layers deposited in the transition zone provide a holistic picture of the trade-off between transport and
between amorphous and micro-crystalline Si) [9,10]. However, passivation at the SHJ and how this can be engineered towards more

* Corresponding author.
E-mail address: martin.bivour@ise.fraunhofer.de (M. Bivour).

https://doi.org/10.1016/j.solmat.2021.111412
Received 7 May 2021; Received in revised form 21 June 2021; Accepted 21 September 2021
Available online 14 January 2022
0927-0248/© 2021 Published by Elsevier B.V.
C. Luderer et al. Solar Energy Materials and Solar Cells 238 (2022) 111412

efficient transport. Using a void-rich i-layer and the hydrogen plasma treatment (i1+HPT)
instead, increased the iVoc significantly at the cost of higher ρc. The stack
2. Experimental details of i1+HPT+i2, in which both i1 and i2 were 3 nm thick, showed the best
passivation but also the highest ρc. These results clearly show that a
Solar cells and lifetime test structures were fabricated on random void-rich i-layer (i1) is necessary for good passivation at the hole and
pyramid textured 1 Ωcm, 180 μm thick n-type FZ silicon wafers (Fig. 1). electron contact but it is also mainly responsible for resistive losses.
The monofacial solar cells were fabricated in the rear-emitter design and
metallized at the front and rear side via screen printing of an Ag paste 3.2. a-Si:H(i1) modification
cured at 200◦ C in a belt furnace. As TCO, indium tin oxide (ITO) was
used, deposited via DC magnetron sputtering with a target composition In the previous section, it was shown that i1 is necessary to obtain
of In2O3:SnO2 (90:10 weight-%). Contact resistance test structures for excellent passivation of the c-Si absorber, but ρc is increased thereby.
the hole (electron) contacts where fabricated on textured p-type (n-type) The goal for further optimization is therefore to modify the i1 micro­
FZ silicon wafers as described elsewhere [18]. For the doped a-Si:H structure towards slightly more dense properties to improve transport
layers, the plasma-enhanced chemical vapor deposition (PECVD) pa­ while maintaining a high passivation quality. By varying the deposition
rameters were kept constant throughout this work and an effective conditions the a-Si:H(i) microstructure can be altered.
thickness on textured surface of ≈12 nm and ≈9 nm on test structures Fig. 3a shows the absorption band associated with Si–H stretching
and solar cells was used, respectively. For the intrinsic a-Si:H multi­ modes obtained from infrared spectroscopy of different a-Si:H(i) layers
layers, the deposition parameters like power density, pressure and H2 with a total thickness of 9 nm on a planar c-Si wafer. A figure of merit for
dilution (ratio R of H2 to total gas flow) were varied. The intrinsic a-Si:H the a-Si:H microstructure is the so-called microstructure factor R*,
stack was formed by the deposition of two a-Si:H layers and additionally which is determined from the Gaussian deconvolution of the Si–H
an intermediate hydrogen plasma treatment (HPT) was used to further stretching modes into a low (LSM) and a high stretching mode (HSM),
modify the properties of the first layer and of the a-Si:H/c-Si interface located at ≈ 2000 cm− 1 and ≈2100 cm− 1, respectively [20–22]:
(i1+HPT+i2). The a-Si:H(i) layer properties and microstructure was IHSM
studied via spectroscopic ellipsometry and Fourier-transformed infrared R* = (1)
ILSM + IHSM
spectroscopy (FTIR). For FTIR, 9 nm thick a-Si:H(i) layers on a planar
c-Si substrate were investigated using an attenuated total reflection A clear difference in the microstructure between i1, i2 and the stack
(ATR) setup to increase the absorption signal of the thin film. can be seen. In the dense i-layer i2 the LSM is dominant, corresponding
to isolated monohydrides (SiH) [23], resulting in a low R* of 0.15. In
3. Results contrast to that, in the void-rich i-layer i1 the HSM plays a major role,
corresponding to hydrogen bonded to a void surface [21] as well as
3.1. Influence of different a-Si:H(i) layers and layer stacks dihydrides (SiH2) [23], displayed by a high microstructure factor (R* =
0.88). Consequently, in the stack of i1 and i2 a combination of both
Fig. 2 shows ρc of SHJ electron (a) and hole (b) contacts comprising microstructures was measured (R* = 0.34).
the intrinsic a-Si:H stack or only the individual layer, all with a similar As one approach to lower the void density and to improve the
total thickness of ≈6 nm. Different colors represent different cumulative structural quality of i1 (thereby reducing the i1 R* value), the H2 dilu­
post-deposition thermal treatments of 10 min each. Average implied tion ratio R in the plasma was increased (Fig. 3b). Whereas the LSM
open-circuit voltages (iVoc) measured by the equally processed lifetime intensity was basically unchanged, the HSM intensity decreased mark­
samples at 1 sun illumination are indicated for each group as a measure edly (red). This can be explained by an increased diffusion energy on the
of the passivation quality. The passivation quality was found to be in­ c-Si surface by a higher rate of exothermic hydrogen exchange reactions
dependent of the annealing temperature up to 220 ◦ C. Basically, a very due to the increased density of atomic hydrogen in the plasma [24].
similar trend was observed for hole and electron contacts but with lower The optical band gap Eg, determined by spectroscopic ellipsometry,
overall ρc and recombination for the electron contact. Without intrinsic decreased with increased H2 dilution ratio (not shown), also indicating a
a-Si:H layer, the passivation quality was poor and ρc limited by the TCO/ densification of the intrinsic layer. In Ref. [25] the positive correlation
a-Si:H(n/p) interface [19]. Adding a dense intrinsic a-Si:H layer (i2) between the microstructure factor R* obtained from infrared spectros­
between c-Si and doped a-Si:H did not increase ρc. The iVoc was copy and the optical band gap Eg extracted from spectroscopic ellips­
improved to some extent, but still poor, especially for the hole contact. ometry measurements was shown. Therefore, usually a higher Eg and
higher R* indicate a film with more voids, less order whereas a lower Eg
and lower R* indicate a denser, more ordered film.
Further, a plasma power reduction during layer growth (green)
similarly decreased R* and Eg, presumably due to a lower void fraction
as a result of a lower deposition rate.
R* was also lower for the i1 after an HPT of 35 s (blue) compared to
the untreated reference i1 (black). In this case, however, a different
trend in Eg was observed as Eg increased with increasing HPT duration
(not shown). The total intensity of the Si–H stretching band and espe­
cially the LSM intensity was increased after HPT, which is a sign for
more hydrogen in the a-Si:H layer [23].
The a-Si:H(i) thickness was unchanged up to 35 s of HPT and
decreased for longer HPT durations (not shown). This confirms the
etching effect of such an HPT reported in Refs. [14,16,26] with an in­
cubation time of ≈35 s for our process conditions during which
hydrogen diffusion into the film dominates over etching [14,26].
Fig. 1. Sketches of the utilized test structures and solar cells for a-Si:H(i) As a next step, transport (ρc) and passivation (τ) with modified
variation at the hole contact. For a-Si:H(i) variation at the electron contact intrinsic a-Si:H layers at the hole contact were tested on co-fabricated
resistance test structures with n-type doping was used for c-Si and a-Si:H at the test structures. In Fig. 4, ρc after 180 ◦ C annealing of various hole con­
front and rear side. tact stacks comprising different a-Si:H(i) layers is plotted against the

2
C. Luderer et al. Solar Energy Materials and Solar Cells 238 (2022) 111412

Fig. 2. ρc and iVoc of SHJ electron (a) and hole (b) contacts comprising different undoped a-Si:H layer (stacks) as indicated by the sketch above the graphs. Colors
refer to different annealing temperature. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3. (a) Infrared spectroscopy signal of different a-Si:H(i). (b) Infrared spectroscopy signal as indicator of the a-Si:H(i1) microstructure for several process pa­
rameters. The intensity was normalized on the a-Si:H(i) layer thickness.

minority carrier lifetime (τ). Each data point represents the median
value of at least ten measurements. The symbol size was chosen pro­
portional to Eg of the particular intrinsic layer. The reference (closed
black square) comprised a high power i1 (3 nm) with no H2 dilution, an
HPT of 7 s and a highly H2 diluted i2 (3 nm) and yielded τ ≈ 4.3 ms and
ρc ≈ 214 mΩcm2.
As indicated by the arrow, generally an increase in τ comes at the cost
of higher ρc. However, there were i-layers providing similar passivation
as the baseline reference (closed black square) while ρc was reduced.
Reducing the total thickness of our reference a-Si:H stack from ≈6
(closed black square) to ≈ 4.5 nm (open black square) reduced both τ
(from 4.3 to 1.3 ms) and ρc (from 214 to 91 mΩcm2) significantly,
highlighting the strong trade-off between high passivation quality and
efficient transport linked to the intrinsic layer’s thickness. For compar­
ison, results of groups "only i2" and "only i1+HPT" from Fig. 2b are
Fig. 4. ρc and τ at an injection level of 1.0 × 1015 cm− 3 for i-layer stacks with
shown as dark red respectively yellow closed circles. The mediocre
different microstructure and i1 thickness fractions. The black closed square
passivation level of "only i1" (1.2 ms) was further reduced to 1.1 ms by
represents the reference process with an i1 thickness ratio of 0.55 and a i1 R
value of 0%. Closed symbols refer to ≈6 nm and open symbols to ≈4.5 nm total reducing the total a-Si:H(i) thickness from ≈6 to ≈ 4.5 nm, while ρc was
effective i-layer stack thickness on textured surface. Semi-closed symbols reduced significantly from 185 to 110 mΩcm2 (closed vs. open yellow
represent i1 deposition with lower power. Different shades of red and yellow circle).
represent the i1 thickness fraction variation and different shades of blue In another attempt, starting from "only i1" the i1 thickness fraction
represent the i1 H2 dilution variation. (For interpretation of the references to (i1 thickness to total a-Si:H(i) thickness) was gradually reduced to zero
color in this figure legend, the reader is referred to the Web version of ("only i2") while a total a-Si:H thickness of 6 nm was maintained (see
this article.) legend, closed yellow to red circles). The optimum i1 fraction regarding
τ remained at 0.55, as for our baseline reference. However, a slightly
reduced i1 thickness fraction (0.44) led to similarly high τ and a

3
C. Luderer et al. Solar Energy Materials and Solar Cells 238 (2022) 111412

reduction of ρc from 214 to 150 mΩcm2. comparable to the reference. Jsc (not shown) was very similar for all
Reducing the power during i1 deposition, but maintaining similar groups. This resulted in an increased η by about 0.3 %abs compared to
thickness, reduced ρc by about ≈50 mΩcm2 while τ degraded only our baseline reference with very similar maximum η of 22.0% for all
slightly ("i/p 2") compared to the reference. Using the lower power i1 at three modifications. Please note that the cell performance is limited by a
reduced i1 thickness fraction led to further ρc reduction down to 140 low Jsc due to a simplistic grid design with a busbar shading of 4% of the
mΩcm2 and similar τ ("i/p 3"). Increasing the H2 dilution of i1 while total cell area.
trying to keep the thickness constant (closed blue symbols) led to a In another cell batch, the total thickness of the i-layer stack at the
reduction in ρc by about ≈80 mΩcm2 down to 130 mΩcm2 at R = 30% hole contact was varied by changing the i1 thickness, while the HPT and
while τ was essentially unaffected ("i/p 1"). Increasing R further reduced the i2 thickness was constant (Fig. 6). Again, the lower power i1 was
τ significantly for R > 30% while ρc gradually approached the lower used, as in “i/p 2” and “i/p 3”. A strong correlation of the i1 thickness and
limit of ρc ≈ 90 mΩcm2 for samples without intrinsic layer (not shown). Rs was observed. A linear fit was performed, yielding a slope of ≈340
The positive influence of an HPT after i1 deposition on transport and mΩcm2/nm. Since exact determination of the i1 thickness is difficult and
passivation can be seen by comparing the reference with a group that did considerable scatter is present in the Rs data, this is only intended as a
not receive the treatment (black square vs. green diamond). rough estimate to highlight the significant impact of the i1 thickness on
To summarize, compared to the baseline reference (Ref., black Rs.
square) a slight reduction in the i1 thickness fraction from 0.55 to 0.44
(orange closed circle), as well as a slightly denser i1 due to deposition
with lower power ("i/p 2") or using some H2 dilution ("i/p 1") reduced ρc
markedly, while τ was still on a high level. Another interesting candidate
was the combination of a slightly denser i1 by using lower power and
adapting the i1 thickness fraction to 0.44 ("i/p 3"). The i1 microstructure
can be further modified with a post-deposition HPT, which can be
beneficial for both transport and passivation.

3.3. Solar cells with modified a-Si:H(i) at the hole contact

The most promising modified a-Si:H(i)/a-Si:H(p) hole contact stacks


taken from Fig. 4 were tested on cell level. Namely, "i/p 1" with a 30% H2
diluted plasma during i1 deposition (dark blue pentagon), "i/p 2" with a
reduced i1 deposition power (black semi-closed square) and "i/p 3" with
lower i1 power as in "i/p 2" and reduced i1 thickness ratio (orange semi-
closed circle). Solar cell parameters of the three modified hole contacts
are shown in Fig. 5 in comparison to the baseline reference. The bene­
ficial transport properties seen on test structures are reflected in the
significant reduction in Rs and increase in FF. The difference in Voc is
small, which shows that the passivation quality is at similar level with a
Fig. 6. Rs as a function of the i1 thickness in the i-layer stack at the hole
Voc loss of 1–2 mV for the i/p 3 process. The pseudo fill factor (pFF), also
contact. The HPT and i2 thickness were kept constant.
reflecting the passivation quality, remained on a high level for all groups

Fig. 5. SHJ solar cell parameters of three modified i/p stacks compared to our reference process. Group i/p 1 comprised a H2 diluted i1. For groups i/p 2 and i/p 3 the
i1 deposition power was lower and for group i/p 3 the i1 thickness fraction was also lower compared to the reference. Jsc (not shown) was constant for all groups.

4
C. Luderer et al. Solar Energy Materials and Solar Cells 238 (2022) 111412

Motivated by the beneficial effect of the HPT after i1 deposition on ρc and close to the c-Si surface necessary for efficient hole extraction from
and on τ seen in Fig. 4, an HPT duration time variation was conducted on the absorber [29]. With respect to the detrimental role of the i-layer for
cell level. The HPT duration time was varied between 0 and 35 s. As can transport, it was shown for p-type and n-type doped a-Si:H [30,31] and a
be seen in Fig. 7, the HPT is clearly beneficial for the surface passivation. high work function metal oxide [32] that the a-Si:H(i) layer can reduce
This is reflected in an increased Voc and pFF of groups with HPT the induced band bending in the c-Si absorber which leads to reduction
compared to the group without HPT. No clear trend for the passivation of accumulated holes and thus to a reduction of hole conductivity in the
quality in dependence on the duration time was observed. The Voc is very vicinity of the a-Si/c-Si interface.
similar for 35 and 7 s HPT, whereas pFF values for the 7 s process are Thus, one possible explanation for the higher ρc for structures con­
slightly higher compared to 18 respectively 35 s. taining i1 could be that the structural differences between i1 and i2
Regarding transport, Rs after 7 s HPT is similar to the untreated cause a greater potential drop across i1 than across i2, and thus less c-Si
group, but a significant Rs reduction was obtained for 18 and 35 s HPT. band bending. In an attempt to probe this effect, the surface photo­
The beneficial effect of the HPT on transport can be also seen in the voltage method [33] was used. However, no significant difference in the
increased FF with increasing HPT duration, reaching a maximum value induced bend bending for the hole contacts comprising an i-layer in
of 83.5% for the 35 s process. Overall, this trend transferred to η with a Fig. 2b could be detected, i.e., either only i2, only i1 (with and without
maximum of 22.6% after 35 s of HPT. Longer HPT durations up to 90 s HPT) or the i1+HPT+i2 stack (not shown). However, this does not mean
did not further increase η (not shown). An HPT duration variation was that this effect must be excluded, as it is possible that the resolution of
also performed at the electron contact on cell level (not shown), but the measurement method is not sufficient.
besides an initial improved passivation after 7 s of HPT, a longer HPT did Besides insufficient induced band bending, structural properties of a-
neither affect passivation nor transport. Note that ρc of the electron Si:H(i) and the a-Si:H/c-Si interface might influence transport. In Fig. 8,
contact is generally much lower and that the i-layer stack at the electron a sketch of the band structure of the a-Si:H(p)/a-Si:H(i)/c-Si hetero­
contact and had a lower total thickness and a lower i1 thickness fraction junction (under illumination at maximum-power point) is shown.
compared to the reference i-layer stack used for the hole contact. Possible hole transport paths from the c-Si into the a-Si:H(p) contact
layer are indicated, as partly adapted from Ref. [34]:
4. Discussion
1) emission of charge carriers across the hetero-barrier
4.1. Induced band bending and possible transport paths 2) trap-assisted tunneling (TAT) across the hetero-barrier
3) tunnel hopping in valence band tail states
For efficient passivation at the hole contact, recombination of pho­ 4) tunneling and subsequent emission.
togenerated minority electrons and majority holes at and close to the a-
Si:H/c-Si interface must be avoided. For efficient transport, a sufficiently In addition to a large hole concentration in the contact region,
high hole conductivity in the contact region along the hole transport thermionic emission across the band offsets requires a small barrier
path, from the induced c-Si junction towards the TCO electrode, needs to height and trap-assisted tunneling transport requires sufficient valence
be ensured [27,28]. band tail states and defect states at the right energetic position. In the
Regarding transport, a sufficiently high band bending at the surface following, we try to correlate the experimental trends with the possible
of the c-Si absorber and in the undoped a-Si:H(i), induced by the doped influencing factors mentioned above.
a-Si:H layer, is a necessary condition for a well-working contact, i.e., low
ρc. It provides the high hole concentration and thus hole conductivity at

Fig. 7. SHJ solar cell parameters in dependence of the HPT duration at the hole contact. Jsc (not shown) was constant for all groups.

5
C. Luderer et al. Solar Energy Materials and Solar Cells 238 (2022) 111412

stack. Reasons for this are the dominance of interfaces over bulk effects
and that the properties of the films, which are only a few nanometers
thin, are determined by the initial growth regime. Thus, an alternative
description is that the thin, void-rich a-Si:H(i1) layer might have an
effectively smaller contact area to the c-Si due to a lower effective sur­
face coverage of the c-Si absorber [36]. This might contribute to the
increased ρc for structures including i1.
Moreover, a reduced i1 thickness increases the probability of direct
tunneling transport through i1. This is supported by the observation that
ρc with a i1+HPT+i2 stack of a total thickness of 4.5 nm (i1 thickness ≈
2 nm) dropped to ≈ 90 mΩcm2 (Fig. 4), a value also reached with no i-
layer present. Thus, although an i1 and the corresponding a-Si:H/c-Si
interface was present, the i-layer contribution to ρc was negligible,
which might indicate very efficient transport through i1 via tunneling
once the i1 thickness is below a certain critical thickness.
The above-mentioned effects modify both hole and electron trans­
Fig. 8. Sketched band diagram of the a-Si:H/c-Si junction at maximum-power port, explaining the same qualitative trend for the hole and electron
point at the hole contact. Recombination at the a-Si:H/c-Si interface or at defect contact in Fig. 2. However, the increase in ρc due to the insertion of i1 is
states close to the interface must be avoided for high-quality surface passiv­ about three times higher for the hole contact (≈+150 mΩcm2 compared
ation. Possible transport paths 1)-4) are illustrated and discussed in the text. to ≈+50 mΩcm2 for the electron contact). The electron contact is not as
The resistance is governed by the a-Si:H/c-Si interface and the i1 and i2 "bulk" heavily reliant on tunneling as the hole contact, since thermionic
properties, like the energetic position and density of defects and valence band emission across the smaller conduction band offset is more efficient.
tail states, as indicated by the series connection at the top. Moreover, the doping efficiency in a-Si:H(n) is higher [37] and as a
result of the asymmetric band-line up, the induced junction is less pro­
4.2. i1 vs. i2: Passivation and transport nounced for the hole contact compared to the electron contact [30]. A
further role for less efficient majority carrier transport for the hole
Using a single layer of dense a-Si:H(i2) without a-Si:H(p) layer contact might be the lower mobility of holes compared to electrons [38].
deposited on top, a good passivation of the c-Si surface with iVoc values With respect to the i-layer stack, the passivation quality is further
above 730 mV (not shown) can be obtained. However, after a-Si:H(p) improved with the deposition conditions of the dense i2 (Fig. 2) on top of
layer deposition, the passivation quality was impeded substantially and the i1 and the c-Si interface. Hydrogen can penetrate the void-rich i1 and
the iVoc decreased to ~640 mV, as depicted in Fig. 2b. Doped layer saturate defects at the a-Si:H/c-Si interface and thereby further reduce
deposition was found detrimental by several other groups (e.g., Refs. the interface defect density [10]. Beyond this, the role of the i2 in the
[10,35]). De Wolf and Kondo explained that with a lowered Si–H bond i-layer stack might also be to function as a “spacer” to screen the a-Si:H
rupture energy in a-Si:H(i) due to the shift of the Fermi level in the (p) defects from the minority carriers at the well passivated a-Si:H
presence of a doped layer [35]. Liu et al. argued that partial crystalli­ (i1)/c-Si interface [39].
zation occurred at the a-Si:H/c-Si interface by hydrogen bombardment
during a-Si:H(p) deposition [10]. In that case, dangling bonds at the c-Si 4.3. Influence of lower power and H2 dilution
surface are no longer passivated efficiently, resulting in a high defect
density at the interface [12]. The high power density enabling a high i1 deposition rate ( > 1.0
With respect to transport, the defect state density in i2 will be lower nm/s compared to ≈ 0.1 nm/s for i2) is one key factor to achieve a void-
than in p-doped a-Si:H, but sufficient states at the right energetic posi­ rich a-Si:H(i) layer and to prevent epitaxial growth [11]. Hence,
tion might still be available to explain the negligible difference in ρc reducing the power during i1 deposition ("i/p 2", "i/p 3" in Figs. 4 and 5)
between groups "w/o i-layer" and "only i2" in Fig. 2. leads to a denser a-Si:H(i), as indicated by the lower R* (and lower Eg) in
Applying a void-rich a-Si:H(i1) ensures an abrupt a-Si:H(i)/c-Si Fig. 3b. Similarly, H2 dilution results in a denser a Si:H(i) layer, as
interface with a low defect density (with and without HPT) [10]. On the mentioned above ("i/p 1"). Since denser layers more readily lead to
one hand, this low-defective interface enables a lower minority carrier epitaxial growth, this can also lead to increased defect density at the
recombination rate and therefore higher iVoc compared to "only i2". On a-Si:H(i)/c-Si heterojunction. This is accompanied by a higher majority
the other hand, ρc was increased with a void-rich a-Si:H(i) at the c-Si and minority carrier recombination at and close to the interface. This
surface. One explanation for that involves path 1 in Fig. 8. Due to the has a detrimental effect on surface passivation, as observed in the lower
larger Eg for void-rich i1 (Eg ≈ 1.9 eV) than for dense i2 (Eg ≈ 1.7 eV), the minority carrier lifetime on test structures and in the slightly lower Voc
barrier height for thermionic emission would be increased in the case of on cell level compared to the reference.
i1. The structural properties of the modified i1 layers are still very
A reduction of the i1 thickness also reduced ρc and Rs ("i/p 3", Figs. 4 different from those of i2, but have been altered in this direction (e.g.,
and 5). Thus, besides increased band offsets at the a-Si:H/c-Si interface, lower R*). Following the argument above, this could imply bandgap
a-Si:H “bulk” properties, like the energetic position and density of gap states with more favorable energetic positions for tunneling transport in
defect states and valence band tail states, might contribute to the higher the modified i1 than in the reference i1.
ρc of i1 than of i2. This would mean that the energetic position of these
states, which enable transport paths 2, 3, and 4 in Fig. 8, are more 4.4. Influence of the HPT
favorable in i2 than in i1.
A lower “bulk” conductivity of void-rich a-Si:H compared to denser, The effect of an HPT on the passivation quality was already discussed
lower band-gap a-Si:H was reported in Ref. [25]. It must be mentioned, by other groups: The HPT introduces hydrogen which can diffuse easily
however, that these results were obtained from measuring the lateral through the void-rich a-Si:H(i1) and saturate dangling bonds at the
conductivity of “thick” a-Si:H(i) films deposited on glass and this might interface, thereby improve the passivation [10,17]. The HPT can also
not be relevant for the vertical transport through the contact stack of the induce a-Si:H network relaxation by replacing strained Si–Si bonds with
ultra-thin layers discussed here. In general, the properties of such “thick” Si–H bonds [40]. This can even cause an epitaxial layer at the a-Si:H/c-Si
layers might have limited relevance for the multi-layer heterojunction interface for a transition-zone a-Si:H film [10]. In the case of heavily

6
C. Luderer et al. Solar Energy Materials and Solar Cells 238 (2022) 111412

disordered, void-rich a-Si:H films, however, network relaxations re­ with a possibly lower effective contact area of such void-rich films, and
quires more energy [10]. Therefore, an epitaxial layer is avoided for the the alteration of the induced band bending, affecting the hole density in
void-rich i1 even after HPT. A thinner i1 could further facilitate the contact region (a-Si:H(i), c-Si surface). Carefully modifying the void-
hydrogen diffusion to the interface [10] and might therefore be bene­ rich a-Si:H(i) toward more order and higher density as well as adapting
ficial for passivation. However, a too thin i1 might not protect the a-Si: its thickness, reduced Rs significantly and resulted in improved η despite
H/c-Si interface sufficiently from the plasma, resulting in damage slightly decreased passivation quality. Further, a prolonged HPT after a-
detrimental for passivation [13–15]. Schulze et al. reported on the band Si:H(i1) deposition at the hole contact was found to decrease Rs while
gap widening effect of hydrogen [41]. They also discussed that the also being beneficial for passivation. This was ascribed to the simulta­
retreat of the valence band is responsible for the increased Eg. As a result, neous reduction of void density enabling more efficient transport and
the valence band offset increased systematically with higher hydrogen the saturation of interface states during HPT for a high passivation
content (higher Eg), whereas the conduction band offset remained quality.
basically unchanged [41]. The introduced hydrogen might therefore be The results shown allow further fine-tuning of the hole and electron
responsible for the band gap widening after HPT observed here and also contact in terms of the trade-off between passivation and transport, but
in Ref. [16]. The increased valence band offset was used by Wu et al. to also show that there is still a need for further improved fundamental
explain impeded hole transport after hydrogen modification of the understanding of thin-film heterojunction stacks.
intrinsic layer [12]. However, they mentioned that this effect was
overestimated in their simulation since they did not include tunneling CRediT authorship contribution statement
transport. Nevertheless, here a significant Rs reduction by applying the
HPT at the hole contact was observed. The beneficial structural changes Christoph Luderer: Conceptualization, Methodology, Investigation,
regarding transport induced by the HPT clearly over-compensated the Writing – original draft, Visualization. Dilara Kurt: Investigation,
potentially increased valance band offset, i.e., transport barrier. In Visualization. Anamaria Moldovan: Writing – review & editing. Mar­
addition, since the hole transport is heavily reliant on tunneling anyway tin Hermle: Writing – review & editing, Supervision. Martin Bivour:
[28,42], the possibly slightly further increased barrier height due to a Conceptualization, Writing – review & editing, Supervision.
valence band retreat after HPT might not be relevant. Possible would be
an increased valence band tail state density due to changes in the a-Si:H
Declaration of competing interest
network introduced by the HPT. This could facilitate hole transport
(paths 2 and 3 in Fig. 8), if the broadened valence band tail (increased
The authors declare that they have no known competing financial
Urbach energy) over-compensates the retreat of the valence band (by an
interests or personal relationships that could have appeared to influence
increased Si–H bond density) [41]. However, that case would result in a
the work reported in this paper.
decreased Eg measured via spectroscopic ellipsometry after HPT, which
was not observed. Mews et al. also reported that an HPT did not increase
Acknowledgements
the Urbach energy of their a-Si:H(i) layers [17].
The HPT might reduce mid-gap defects in the i1 and at the a-Si:H/c-
The authors would like to thank A. Leimenstoll, F. Schätzle, K.
Si interface and by that reduce surface recombination and increase the
Zimmermann for sample preparation, Z. Newcomb-Hall for sample
hole concentration at the c-Si surface, which would increase the prob­
preparation and dark I–V measurements, F. Martin for measuring the
ability for (trap-assisted) tunneling transport to the a-Si:H(p). However,
solar cells and C. Messmer for fruitful discussions. This work was funded
this would not explain the overall strong trade-off between surface
by the German Federal Ministry for Economic Affairs and Energy under
passivation and transport observed in Fig. 4.
contract numbers 03EE1031A (PaSoDoble) and 0324293E (Dynasto).
R* was decreased after HPT, a sign for a slightly more ordered film
with a lower void density [16,21]. This correlates with a slightly
decreased broadening term observed via fitting the spectroscopic References
ellipsometry data with a Tauc-Lorentz model [43] (not shown). For the
[1] S. de Wolf, A. Descoeudres, Z.C. Holman, C. Ballif, High-efficiency silicon
cells presented in Fig. 7, the HPT therefore apparently i) provided a heterojunction solar cells: a review, green, 2, 7-24, https://doi.org/10.1515/green-
low-defective interface by supplying hydrogen for the saturation of 2011-0018, 2012.
dangling bonds and ii) reduced voids and improved the transport across [2] S. Dauwe, J. Schmidt, R. Hezel, Very low surface recombination velocities on p-
and n-type silicon wafers passivated with hydrogenated amorphous silicon films,
the i1, similarly to the other i1 modifications. The etching effect of the in: Conference Record of the Twenty-Ninth IEEE Photovoltaic Specialists
HPT and accompanied i1 thickness reduction only plays a minor role for Conference, New Orleans, LA, USA, IEEE, Piscataway, NJ, 2002, pp. 1246–1249.
the cell results in Fig. 7, since the i1 thickness was almost constant up to [3] K. Yoshikawa, W. Yoshida, T. Irie, H. Kawasaki, K. Konishi, H. Ishibashi, T. Asatani,
D. Adachi, M. Kanematsu, H. Uzu, K. Yamamoto, Exceeding conversion efficiency
35 s (incubation regime) and only started to decrease for longer of 26% by heterojunction interdigitated back contact solar cell with thin film Si
durations. technology, Sol. Energy Mater. Sol. Cell. 173 (2017) 37–42, https://doi.org/
Basically, more structural characterization like the measurement of 10.1016/j.solmat.2017.06.024.
[4] H. Fujiwara, M. Kondo, Impact of epitaxial growth at the heterointerface of a-Si:H/
Urbach energies with very high spatial resolution is necessary to find a c-Si solar cells, Appl. Phys. Lett. 90 (2007) 13503, https://doi.org/10.1063/
definitive explanation for the experimental results discussed here. 1.2426900.
Detailed numerical device simulations would further help to deepen the [5] T.H. Wang, E. Iwaniczko, M.R. Page, D.H. Levi, Y. Yan, H.M. Branz, Q. Wang, Effect
of emitter deposition temperature on surface passivation in hot-wire chemical
understanding.
vapor deposited silicon heterojunction solar cells, Thin Solid Films 501 (2006)
284–287, https://doi.org/10.1016/j.tsf.2005.07.196.
5. Conclusion [6] A. Descoeudres, L. Barraud, R. Bartlome, G. Choong, S. De Wolf, F. Zicarelli,
C. Ballif, The silane depletion fraction as an indicator for the amorphous/
crystalline silicon interface passivation quality, Appl. Phys. Lett. 97 (2010)
In this article, we showed that the void-rich a-Si:H(i1) used to pre­ 183505, https://doi.org/10.1063/1.3511737.
vent epitaxial growth is the main source of i-layer induced transport [7] S. De Wolf, M. Kondo, Abruptness of a-Si:H/c-Si interface revealed by carrier
lifetime measurements, Appl. Phys. Lett. 90 (2007) 42111, https://doi.org/
losses. Higher resistance was observed for both the electron and the hole
10.1063/1.2432297.
contact, showing that it is not dominated by asymmetric effects like [8] H. Fujiwara, T. Kaneko, M. Kondo, Application of hydrogenated amorphous silicon
different band offsets for electrons or holes. Instead, several other oxide layers to c-Si heterojunction solar cells, Appl. Phys. Lett. 91 (2007) 133508,
possible influencing factors have been discussed. Among those, less https://doi.org/10.1063/1.2790815.
[9] Y. Zhang, C. Yu, M. Yang, L.-R. Zhang, Y.-C. He, J.-Y. Zhang, X.-X. Xu, Y.-Z. Zhang,
favorable energetic position of defect states was suggested to be the X.-M. Song, H. Yan, Significant improvement of passivation performance by two-
reason for impeded transport using void-rich a-Si:H(i), in combination step preparation of amorphous silicon passivation layers in silicon heterojunction

7
C. Luderer et al. Solar Energy Materials and Solar Cells 238 (2022) 111412

solar cells, Chin. Phys. Lett. 34 (2017) 38101, https://doi.org/10.1088/0256- [26] A. Fontcuberta i Morral, P. Roca i Cabarrocas, Etching and hydrogen diffusion
307X/34/3/038101. mechanisms during a hydrogen plasma treatment of silicon thin films, J. Non-
[10] W. Liu, L. Zhang, R. Chen, F. Meng, W. Guo, J. Bao, Z. Liu, Underdense a-Si:H film Cryst. Solids 299-302 (2002) 196–200, https://doi.org/10.1016/S0022-3093(01)
capped by a dense film as the passivation layer of a silicon heterojunction solar 01001-8.
cell, J. Appl. Phys. 120 (2016) 175301, https://doi.org/10.1063/1.4966941. [27] M. Hermle, F. Feldmann, M. Bivour, J.C. Goldschmidt, S.W. Glunz, Passivating
[11] H. Sai, P.-W. Chen, H.-J. Hsu, T. Matsui, S. Nunomura, K. Matsubara, Impact of contacts and tandem concepts: approaches for the highest silicon-based solar cell
intrinsic amorphous silicon bilayers in silicon heterojunction solar cells, J. Appl. efficiencies, Appl. Phys. Rev. 7 (2020) 21305, https://doi.org/10.1063/
Phys. 124 (2018) 103102, https://doi.org/10.1063/1.5045155. 1.5139202.
[12] Z. Wu, L. Zhang, W. Liu, R. Chen, Z. Li, F. Meng, Z. Liu, Role of hydrogen in [28] C. Messmer, M. Bivour, J. Schön, M. Hermle, Requirements for efficient hole
modifying a-Si:H/c-Si interface passivation and band alignment for rear-emitter extraction in transition metal oxide-based silicon heterojunction solar cells,
silicon heterojunction solar cells, J. Mater. Sci. Mater. Electron. 31 (2020) J. Appl. Phys. 124 (2018) 85702, https://doi.org/10.1063/1.5045250.
9468–9474, https://doi.org/10.1007/s10854-020-03486-5. [29] M. Bivour, C. Messmer, L. Neusel, F. Zähringer, J. Schön, S.W. Glunz, M. Hermle,
[13] S.N. Granata, T. Bearda, F. Dross, I. Gordon, J. Poortmans, R. Mertens, Effect of an Principles of carrier-selective contacts based on induced junctions, in: Proc. 33rd
in-situ H$_2$ plasma pretreatment on the minority carrier lifetime of a-Si:H(i) Eur. Photovolt. Sol. Energy Conf, . Exhib., Amsterdam, The Netherlands, 2017,
passivated crystalline silicon, Energy Procedia 27 (2012) 412–418, https://doi. pp. 348–352.
org/10.1016/j.egypro.2012.07.086. [30] C. Leendertz, Effizienzlimitierende Rekombinationsprozesse in amorph/
[14] J. Geissbühler, S. De Wolf, B. Demaurex, J.P. Seif, D.T.L. Alexander, L. Barraud, kristallinen und polykristallinen Siliziumsolarzellen. Dissertation, Technische
C. Ballif, Amorphous/crystalline silicon interface defects induced by hydrogen Universität Berlin, Berlin, 2012.
plasma treatments, Appl. Phys. Lett. 102 (2013) 231604, https://doi.org/10.1063/ [31] C. Leendertz, N. Mingirulli, T.F. Schulze, J.P. Kleider, B. Rech, L. Korte, Discerning
1.4811253. passivation mechanisms at a-Si:H/c-Si interfaces by means of photoconductance
[15] J. Schüttauf, C. van der Werf, W. van Sark, J.K. Rath, R. Schropp, Comparison of measurements, Appl. Phys. Lett. 98 (2011) 202108, https://doi.org/10.1063/
surface passivation of crystalline silicon by a-Si:H with and without atomic 1.3590254.
hydrogen treatment using hot-wire chemical vapor deposition, Thin Solid Films [32] L. Neusel, M. Bivour, M. Hermle, Selectivity issues of MoO x based hole contacts,
519 (2011) 4476–4478, https://doi.org/10.1016/j.tsf.2011.01.319. Energy Procedia 124 (2017) 425–434, https://doi.org/10.1016/j.
[16] A. Descoeudres, L. Barraud, S. De Wolf, B. Strahm, D. Lachenal, C. Guérin, Z. egypro.2017.09.268.
C. Holman, F. Zicarelli, B. Demaurex, J. Seif, J. Holovsky, C. Ballif, Improved [33] K. Heilig, Determination of surface properties by means of large signal
amorphous/crystalline silicon interface passivation by hydrogen plasma treatment, photovoltage pulses and the influence of trapping, Surf. Sci. 44 (1974) 421–437,
Appl. Phys. Lett. 99 (2011) 123506, https://doi.org/10.1063/1.3641899. https://doi.org/10.1016/0039-6028(74)90128-9.
[17] M. Mews, T.F. Schulze, N. Mingirulli, L. Korte, Hydrogen plasma treatments for [34] T.F. Schulze, L. Korte, E. Conrad, M. Schmidt, B. Rech, Electrical transport
passivation of amorphous-crystalline silicon-heterojunctions on surfaces promoting mechanisms in a-Si:H/c-Si heterojunction solar cells, J. Appl. Phys. 107 (2010)
epitaxy, Appl. Phys. Lett. 102 (2013) 122106, https://doi.org/10.1063/ 23711, https://doi.org/10.1063/1.3267316.
1.4798292. [35] S. De Wolf, M. Kondo, Boron-doped a-Si:H/c-Si interface passivation: degradation
[18] C. Luderer, C. Messmer, M. Hermle, M. Bivour, Transport losses at the TCO/a-Si:H/ mechanism, Appl. Phys. Lett. 91 (2007) 112109, https://doi.org/10.1063/
c-Si heterojunction: influence of different layers and annealing, IEEE J. 1.2783972.
Photovoltaics 10 (2020) 952–958, https://doi.org/10.1109/ [36] T.F. Schulze, H.N. Beushausen, C. Leendertz, A. Dobrich, B. Rech, L. Korte,
JPHOTOV.2020.2983989. Interplay of amorphous silicon disorder and hydrogen content with interface
[19] C. Luderer, L. Tutsch, C. Messmer, M. Hermle, M. Bivour, Influence of TCO and a- defects in amorphous/crystalline silicon heterojunctions, Appl. Phys. Lett. 96
Si:H doping on SHJ contact resistivity, IEEE J. Photovoltaics 11 (2021) 329–336, (2010) 252102, https://doi.org/10.1063/1.3455900.
https://doi.org/10.1109/JPHOTOV.2021.3051206. [37] W.E. Spear, P.G. Le Comber, Substitutional doping of amorphous silicon, Solid
[20] M.H. Brodsky, M. Cardona, J.J. Cuomo, Infrared and Raman spectra of the silicon- State Commun. 17 (1975) 1193–1196, https://doi.org/10.1016/0038-1098(75)
hydrogen bonds in amorphous silicon prepared by glow discharge and sputtering, 90284-7.
Phys. Rev. B 16 (1977) 3556–3571, https://doi.org/10.1103/PhysRevB.16.3556. [38] R.A. Street, Trapping parameters of dangling bonds in hydrogenated amorphous
[21] A.H.M. Smets, W.M.M. Kessels, M.C.M. van de Sanden, Vacancies and voids in silicon, Appl. Phys. Lett. 41 (1982) 1060–1062, https://doi.org/10.1063/1.93400.
hydrogenated amorphous silicon, Appl. Phys. Lett. 82 (2003) 1547–1549, https:// [39] M. Bivour, Silicon Heterojunction Solar Cells: Analysis and Basic Understanding,
doi.org/10.1063/1.1559657. Fraunhofer Verlag, Stuttgart, 2015.
[22] J. Mouro, A. Gualdino, V. Chu, J.P. Conde, Microstructure factor and mechanical [40] S. Sriraman, S. Agarwal, E.S. Aydil, D. Maroudas, Mechanism of hydrogen-induced
and electronic properties of hydrogenated amorphous and nanocrystalline silicon crystallization of amorphous silicon, Nature 418 (2002) 62–65, https://doi.org/
thin-films for microelectromechanical systems applications, J. Appl. Phys. 114 10.1038/nature00866.
(2013) 184905, https://doi.org/10.1063/1.4829020. [41] T.F. Schulze, L. Korte, F. Ruske, B. Rech, Band lineup in amorphous/crystalline
[23] A.A. Langford, M.L. Fleet, B.P. Nelson, W.A. Lanford, N. Maley, Infrared absorption silicon heterojunctions and the impact of hydrogen microstructure and topological
strength and hydrogen content of hydrogenated amorphous silicon, Phys. Rev. B 45 disorder, Phys. Rev. B 83 (2011) 554, https://doi.org/10.1103/
(1992) 13367–13377, https://doi.org/10.1103/PhysRevB.45.13367. PhysRevB.83.165314.
[24] A. Matsuda, Thin-film silicon –growth process and solar cell application, Jpn. J. [42] P. Procel, H. Xu, A. Saez, C. Ruiz-Tobon, L. Mazzarella, Y. Zhao, C. Han, G. Yang,
Appl. Phys. 43 (2004) 7909–7920, https://doi.org/10.1143/JJAP.43.7909. M. Zeman, O. Isabella, The role of heterointerfaces and subgap energy states on
[25] W. Liu, L. Zhang, X. Yang, J. Shi, L. Yan, L. Xu, Z. Wu, R. Chen, J. Peng, J. Kang, transport mechanisms in silicon heterojunction solar cells, Prog. Photovoltaics Res.
K. Wang, F. Meng, S. De Wolf, Z. Liu, Damp-heat-stable, high-efficiency, industrial- Appl. 28 (2020) 935–945, https://doi.org/10.1002/pip.3300.
size silicon heterojunction solar cells, Joule 4 (2020) 913–927, https://doi.org/ [43] G.E. Jellison, F.A. Modine, Parameterization of the optical functions of amorphous
10.1016/j.joule.2020.03.003. materials in the interband region, Appl. Phys. Lett. 69 (1996) 371–373, https://
doi.org/10.1063/1.118064.

You might also like