You are on page 1of 8

Received: 14 July 2017 Revised: 18 September 2017 Accepted: 9 October 2017

DOI: 10.1002/pip.2960

EU PVSEC PAPER

Tunnel oxide passivating electron contacts as full‐area rear


emitter of high‐efficiency p‐type silicon solar cells
Armin Richter1 | Jan Benick1 | Ralph Müller1,2 | Frank Feldmann1,2 | Christian Reichel1 |

Martin Hermle1 | Stefan W. Glunz 1,2

1
Fraunhofer Institute for Solar Energy
Systems (ISE), Heidenhofstrasse 2, 79110 Abstract
Freiburg, Germany Tunnel oxide passivating contacts (TOPCon) consisting of an ultrathin tunnel oxide capped by a
2
Department of Sustainable Systems doped Si film exhibit excellent passivation and contact properties. The application of these con-
Engineering Albert Ludwig, University of tacts has so far resulted in efficiencies of up to 25.7% realized with an n‐type Si solar cell featur-
Freiburg, Georges‐Köhler‐Allee 103, 79110
ing a front‐side boron‐doped p+ emitter and n‐TOPCon as full‐area rear electron contact. In this
Freiburg, Germany
Correspondence
work, we study the same cell structure on p‐type Si. In this case, the p+ diffusion on the front acts
Armin Richter, Fraunhofer Institute for Solar as a front surface field (FSF) and the n‐TOPCon layer as a full‐area rear emitter. One benefit of
Energy Systems (ISE), Heidenhofstrasse 2, this rear‐junction cell design is that the whole base contributes to the hole transport towards
79110 Freiburg, Germany.
the local contacts on the front, which means that the lateral current transport within the FSF is
Email: armin.richter@ise.fraunhofer.de
less important than in the case of the front emitter of the n‐type cell. To study this, we addressed
the influence of the FSF lateral conductivity on the performance of these rear‐junction cells
Funding information
German Federal Ministry of Economic Affairs theoretically (based on a simulation study) as well as experimentally (with fabricated cells).
and Energy, Grant/Award Number: 0325827B Efficiencies up to 24.3% (independently confirmed) have been achieved with this structure
applying a FSF and up to 23.9% without the full‐area FSF. As such, these results demonstrate a
high device performance for these TOPCon rear emitter cells even without a lateral conductivity
in the FSF. This bears the potential to simplify the process chain quite substantially as no full‐area
boron diffusion is required.

KEY W ORDS

device simulation, front surface field, passivating contacts, silicon solar cells

1 | I N T RO D U CT I O N The standard solar cell structure in mass production features contacts


at the front and the rear, because of the less complex process
Carrier‐selective passivating contacts based on silicon thin films enable sequence compared with interdigitated back contact cells. The applica-
so far the highest energy conversion efficiencies for single junction tion of Si‐based passivating contacts on both surfaces of front and rear
crystalline silicon (c‐Si) solar cells.1-8 The best known example is the contacted solar cells results, however, in significant current losses due
heterojunction of hydrogenated amorphous silicon (a‐Si:H) and c‐Si, to parasitic absorption in the a‐Si:H or the poly‐Si layer and the cap-
which enables very high open‐circuit voltages, VOC, of up to ping transparent conductive oxide,16,17 although there are approaches
750 mV.9 Other Si thin film‐based passivating contacts such as to reduce these losses.18,19
10,11 2
SIPOS, tunnel oxide passivating contacts (TOPCon), or poly‐ In our approach, we combine the advantages of TOPCon with
Si12-14 make use of a high‐quality ultra‐thin intermediate SiOx layer those of a classical homojunction with its excellent electrical proper-
which is capped by a heavily doped nanocrystalline or polycrystalline ties and in particular optical transparency. We have realized this by
silicon layer. Most of these approaches have been demonstrated to using TOPCon as a full‐area electron contact on the rear surface of
result in highest conversion efficiencies in the range of 25% and an n‐type c‐Si solar cell with a boron‐doped front side emitter2,20,21
1,3-8
above, with an impressive record value of 26.7% realized with an (cell structure (A) in Figure 1). The excellent electrical and optical
interdigitated back contact solar cell and a‐Si:H based contacts.7,15 properties of the boron‐doped emitter are based on Al2O3 surface

Prog Photovolt Res Appl. 2017;1–8. wileyonlinelibrary.com/journal/pip Copyright © 2017 John Wiley & Sons, Ltd. 1
2 RICHTER ET AL.

FIGURE 1 A, Schematic cross section of an n‐type Si solar cell with a boron‐doped front emitter and a tunnel oxide passivating contact (TOPCon)
at the rear. B and C, show the same cell design on a p‐type Si wafer with and without a full‐area front surface field (FSF), respectively [Colour figure
can be viewed at wileyonlinelibrary.com]

passivation.22-24 The full‐area TOPCon rear contact allows not only 2 | SIMULATION STUDY
high VOC values due to excellent passivation properties,5 but also
very high fill factors, FF, as there is no lateral current flow contribu-
2.1 | Simulation setup
tion for the electrons within the silicon bulk.20,25 Very recently, we
have achieved efficiencies up to 25.7% in combination with a very The device simulations of the TOPCon rear emitter solar cells were
5
high FF of 83.3%, which demonstrates the potential of this cell performed with Quokka v2,34 a solar cell simulation tool based on

structure and represents the highest efficiency reported so far for a the conductive boundary approach35 in which the surfaces are defined

both‐sides contacted Si solar cell.15 with recombination parameters J0 and sheet resistances, Rsheet. The

The application of exactly the same cell structure to a p‐type Si simulations were performed for a 2‐dimensional unit cell of the solar

wafer results in a rear junction solar cell (cell structure (B) in cell structure shown in Figure 1B, ie, with full‐area FSF. The main input
+
Figure 1). In this case, the p diffusion at the front acts as a high‐ parameters for these device simulations are listed in Table 1. They are

low junction (hole contact) also known as front surface field (FSF), all based on experimental measurements, eg, the J0 values were

while the TOPCon layer forms a full‐area pn‐junction (electron con- extracted from τeff measurements of symmetrically processed lifetime

tact), ie, emitter. It was already demonstrated that TOPCon forms a samples as described in Section 3.1 or the Rsheet of the FSF was mea-

high‐quality pn‐junction. 26,27


One benefit of this rear‐junction cell sured with the 4‐point probe technique. The optical model is based on

structure is that the full‐area metal contact to the TOPCon emitter a wavelength‐dependent front surface transmission, extracted from

omits lateral current transport losses within the emitter, which is the measured reflectance of the solar cells, and a parameterization of

quite a significant power loss for the front side emitter cell structure the pathlength enhancement Z as suggested in Ref. 36. The parameters
5
shown in Figure 1A. Another benefit is that the whole silicon bulk of this Z‐parameterization are given in Ref. 5. With respect to the bulk

contributes to the hole (majority carrier) transport towards the local


FSF contacts at the front, which means that the lateral current trans- TABLE 1 Main input parameters for the device simulations
port within the FSF is less important28,29 than in the case of the
Parameter Value
front‐side emitter of the n‐type Si cell. This bears the potential to
Wafer thickness W 250 μm
reduce the sheet resistance of the FSF or even omit the full‐area
++
Heavily doped p FSF below front contact:
FSF (cell structure (C) in Figure 1), which could reduce the complexity
Sheet resistance Rsheet,p++ 15 Ω/sq
of the fabrication process quite substantially, in particular if the high
++ Width 10 μm
temperature diffusion of the heavily doped p regions required
J0 of metallized area, J0e,p++,met 300 fA/cm2
directly below the front metal contacts can be replaced, for instance,
Contact resistivity ρC,p++ 0.4 mΩ cm2
by a laser‐based formation of the p++ region.30-33
Front metal grid:
In this work, we have studied the efficiency potential of these
Finger pitch 833 μm
p‐type Si solar cells with rear TOPCon emitter and diffused FSF, with
Finger width 11 μm
a special focus on the influence of the lateral conductivity of the FSF
Finger shading fraction 70%
on the device performance and to address the question if a full‐area
Grid resistance Rgrid 66 mΩ cm2
FSF is required. In order to investigate this theoretically, we have per-
Width of contact opening 8 μm
formed a device simulation study, addressing also the influence of the
Rear passivating contact (TOPCon):
Si bulk resistivity. In addition, we have fabricated test structures and
J0,TOPCon 7 fA/cm2
solar cells with and without the full‐area FSF in order to demonstrate
Contact resistivity ρC,TOPCon 50 mΩ cm2
the cell performance experimentally.
RICHTER ET AL. 3

properties of Si, eg, the charge carrier mobility, intrinsic carrier concen- interdependencies of all the parameters, as, for instance, reported by
tration, and Auger recombination, state‐of‐the‐art models have been Bivour et al for n‐type silicon solar cells with a silicon heterojunction
applied.37-40 The simulations were performed for standard testing con- at the rear and a diffused n+ FSF.41
ditions (STC), ie, for a temperature of 25°C. The simulated I‐V parameters are shown in Figure 2, on the left
for ρb = 1 Ω cm and on the right for ρb = 10 Ω cm. Note that J0,FSF
and Rsheet,FSF were varied independently, for which reason some com-
2.2 | Results and discussion binations of low J0,FSF and low Rsheet,FSF in the lower left corner of the
In order to study the influence of the lateral current transport in the diagrams might not be realized experimentally, as the lower limit for
full‐area FSF on the performance of the TOPCon rear junction cells, J0,FSF is given by the Auger recombination contribution of the FSF
the FSF sheet resistance Rsheet,FSF and the FSF passivation quality in diffusion. Note further that loss mechanisms such as a bulk lifetime
terms of the recombination parameter J0,FSF were varied indepen- limitation, perimeter recombination, or parallel shunt resistances have
dently. These variations were conducted for 2 different silicon bulk not been considered in these simulations. For this reason, the simu-
resistivities ρb of 1 Ω cm and 10 Ω cm, in order to identify lated I‐V parameters are slightly higher than experimentally expected.

FIGURE 2 Simulated I‐V parameters for the TOPCon rear‐junction p‐type Si solar cells with FSF (structure (B) shown in Figure 1) as a function of
the FSF recombination parameter J0,FSF and the FSF sheet resistance, Rsheet,FSF. The results are shown on the left side for a silicon bulk resistivity of
1 Ω cm and on the right side for 10 Ω cm. The stars indicate the J0,FSF of the 300 Ω/sq FSF used in this study (see Table 2). Note that J0,FSF and
Rsheet,FSF were varied independently, for which reason some combinations of low J0,FSF and low Rsheet,FSF in the lower left corner of the diagrams
might not be realized experimentally, as those J0,FSF values can be lower than the Auger recombination contribution of the FSF [Colour figure can
be viewed at wileyonlinelibrary.com]
4 RICHTER ET AL.

For instance, perimeter recombination at small area high‐efficiency Si Altogether, the simulated efficiency, η, for cells made of 1 Ω cm Si
solar cells, as fabricated in this study, typically provokes an efficiency is rather insensitive to Rsheet,FSF. Although the cells made of 10 Ω cm Si
loss of ~0.3%abs.5,42 suffer from a decreasing FF with increasing Rsheet,FSF, the efficiency
The simulation results reveal that the influence of Rsheet,FSF and potential without FSF (Rsheet,FSF > 100 kΩ/sq) is almost as high as for
J0,FSF on the cell performance is quite different for 1 Ω cm and 10 Ω cm 1 Ω cm Si cells without FSF, as the lower FF is overcompensated by
Si bulk resistivity. For instance, the JSC is almost independent of a slightly higher VOC and in particular by a higher JSC. Thus, for cells
Rsheet,FSF and J0,FSF in the case of 10 Ω cm Si, while it strongly made of 10 Ω cm Si, a FSF seems also not to be required. Although
decreases with increasing J0,FSF in the case of 1 Ω cm Si especially for not studied in this work, it is worth mentioning that the higher conduc-
2
J0,FSF > 20 fA/cm . This different behavior, which has already been tivity of the 1 Ω cm Si is expected to be beneficial, if the pitch of the
reported for rear emitter solar cells,43 can be explained by the fact that front metal grid is increased.
for a given J0 value, the absolute surface recombination current is pro-
portional to the doping concentration under low injection conditions,44
as being present at short‐circuit conditions. The VOC of both bulk resis-
3 | S O L A R CE L L S
tivities is independent of Rsheet,FSF and increases very similar with
decreasing J0,FSF, as expected from the 1‐diode model.
With respect to the influence of the lateral conductivity of the
3.1 | Experimental
FSF, the distinct difference observed for the FF of both bulk resistiv- In order to study the influence of the FSF conductivity experimentally,
ities is most interesting. In the case of 1 Ω cm Si, the FF is rather TOPCon rear emitter p‐type Si solar cells with and without a FSF (cell
insensitive to an increase of Rsheet,FSF, which shows that the majority structures (B) and (C) in Figure 1, respectively) were fabricated analo-
carrier conductivity of 1 Ω cm Si is high enough to transport the gously to the front‐junction n‐type Si solar cells with rear TOPCon
majority carriers to the local FSF contacts without significant addi- electron contacts reported in Ref. 5. The cells with a designated area
tional series resistances losses. Thus, a FSF seems not be required of 2×2 cm2 were made of high‐quality (100)‐oriented p‐type float‐zone
for cells made of 1 Ω cm Si. On the other hand, the FF of 10 Ω (FZ) Si wafers with a resistivity of 10 Ω cm, a thickness of 250 μm, and
cm Si cells strongly decreases with increasing Rsheet,FSF, which shows a diameter of 100 mm.
that the cells made of 10 Ω cm Si are quite prone to the lateral con- The full‐area TOPCon emitter at the rear was formed by an ultra‐
ductivity in the FSF due to a too low Si bulk conductivity. At very thin wet‐chemically grown SiOx layer covered by a 15‐nm‐thick phos-
high Rsheet,FSF > 100 kΩ/sq, which can be considered to represent phorus‐doped SiC layer deposited with plasma‐enhanced chemical
cells without FSF, the FF of 10 Ω cm Si cells strongly increases with vapor deposition (PECVD), which was annealed afterwards in a tube
decreasing J0,FSF. This strong increase correlates with a significant furnace at 800°C (process details in Ref.2,27). In order to separate the
increase of the majority carrier concentration, as can be observed TOPCon emitter of the different cells located on each wafer, the
from the ratio of the maximum power point excess carrier density TOPCon layer was etched back in the perimeter region dry‐chemically
and the base doping concentration, Δnmpp/Ndop, shown in Figure 3. using a plasma etch process. This perimeter region was passivated
2
For Rsheet,FSF > 100 kΩ/sq and J0,FSF < 10 fA/cm , the majority car- afterwards with a stack of 10 nm Al2O3 deposited with plasma‐
rier concentration is increased by more than 50%, which is quite sub- assisted atomic layer deposition capped by 150‐nm PECVD SiOx. After
stantial as the charge carrier conductivity is proportional to the opening this Al2O3/SiOx stack at the active cell area wet‐chemically,
charge carrier concentration. Thus, for 10 Ω cm Si cells without the rear side metallization was formed by thermally evaporated Ag.
FSF represented here by Rsheet,FSF > 100 kΩ/sq, the FF increase with The front side was textured with random pyramids. The full‐area
decreasing J0,FSF can be explained by a significantly increasing major- FSF with a sheet resistance of Rsheet,FSF = 300 Ω/sq was diffused by
ity carrier concentration in excess of the bulk doping concentration, a BBr3 tube furnace process at 870°C followed by a drive‐in oxidation.
ie, the transition from low injection to high injection. To form front side contacts with low surface recombination losses, a

FIGURE 3 Ratio of the excess carrier density at maximum power point and the base doping concentration Δnmpp/Ndop for the simulations shown
in Figure 2. Note that for 1 Ω cm p‐type Si (left), the contour plot is completely red because Δnmpp/Ndop ranges only from 0.01 to 0.05, indicating
low level injection over the whole J0,FSF and Rsheet,FSF range [Colour figure can be viewed at wileyonlinelibrary.com]
RICHTER ET AL. 5

heavily doped p++ FSF is located underneath the front side metal con- TABLE 2 Surface passivation properties and sample structures of the
tacts. This p++ FSF was also diffused via a BBr3 tube furnace process 3 different symmetrical lifetime samples
and a subsequent drive‐in diffusion. The front surface, either with or TOPCon Emitter With FSF Without FSF
without full‐area FSF, was passivated with 10 nm of Al2O3 deposited iVOC (mV) 719 721 736
with plasma‐assisted atomic layer deposition. The front side was addi- iFF (%) 84.8 83.3 85.2
tionally coated with a double‐layer antireflection coating consisting of J0 (fA/cm2) 7.1 7.5 3.1
50‐nm PECVD SiNx and 90‐nm thermally evaporated MgF2. Prior to
the MgF2 deposition, contact openings were etched wet‐chemically
through the front side Al2O3/SiNx layer stack. The front side metal
contacts were formed with a stack of e‐beam evaporated Ti/Pd/Ag
(833‐μm pitch). All structuring processes required for the fabrication
of these solar cells are based on photolithography processes (eg, to
structure thermally‐grown SiO2 mask layers) and the lift‐off technique also rather good passivation properties. The rather low iFF of this sam-
(eg, for the front side metal contacts). ple is at least to some extent associated with sample edge recombina-
In order to determine the surface passivation quality of the front tion as was reported for such lifetime samples with pn‐junction if a
surface with and without the 300 Ω/sq FSF in terms of J0, symmet- unpassivated edge is present,52 which is the case for this sample. For
rically processed lifetime test samples were fabricated without the the sample without FSF, an iVOC level of 736 mV and an iFF level of
heavily doped p++ regions. Similar lifetime test samples were also fab- >85% were achieved, which clearly outperforms the passivation qual-
ricated for the phosphorous‐doped TOPCon emitter. These lifetime ity of the 300 Ω/sq FSF sample. This demonstrates that an excellent
samples were made of 10 Ω cm p‐type FZ Si wafers as used for cells, surface passivation can be achieved with Al2O3 even on this random‐
except for the sample with FSF, which was made of a (100)‐oriented pyramids textured p‐type Si surfaces without FSF.
n‐type FZ Si wafer with a resistivity of 10 Ω cm and a thickness of The extracted J0 values of these samples can be used to estimate
200 μm. The surface morphology of the lifetime test samples was the efficiency potential of the TOPCon rear emitter solar cell with and
identical to the respective surface of the solar cells, ie, planar for without FSF based on the simulation results shown in Figure 2. The
the TOPCon emitter and random‐pyramids textured for the samples stars mark the efficiency potential of the 300 Ω/sq FSF. From the sim-
with and without FSF. J0 values were extracted from injection‐ ulations, it can be observed that the predicted efficiency potential of
dependent effective lifetime, τeff, data according to the high‐injection cells with FSF is slightly higher for 10 Ω cm Si compared with 1 Ω
method proposed by Kane and Swanson45 applying the correction cm Si. The efficiency potential of cells without FSF can be estimated
routines described by Kimmerle et al. 46
τeff was measured with the from the very high Rsheet,FSF > 100 kΩ/sq region. For the J0 value of
photo‐conductance decay technique using the Sinton lifetime tester ~3 fA/cm2, as extracted for the lifetime test sample without FSF, the
WCT120 in quasi‐steady‐state mode (QSSPC) 47
and the generalized predicted efficiency potential is slightly higher for 1 Ω cm Si. Thus,
analysis proposed by Nagel et al.48 The optical factor fopt required 10 Ω cm Si has the highest efficiency potentials for cells with FSF
for the QSSPC analysis was determined for each sample by adapting and 1 Ω cm Si for cells without FSF, although the difference in partic-
the QSSPC measurements to transient photo‐conductance decay mea- ular for cells without FSF is very small (less than 0.2%abs). Considering
surements in an excess carrier density range around 3×1015 cm−3. these small differences in the predicted efficiency potential, we chose
For properly adapted fopt, a good agreement of both measurements 10 Ω cm Si wafers for the first batch of TOPCon rear emitter cells fab-
was observed for all samples. Note that this J0 evaluation is also well ricated in this study.
suited for the sample without FSF, which is an Al2O3‐passivated p‐type
Si wafer without a diffused junction where the negative fix charges of
Al2O349,50 form an accumulation layer at the surface. This follows from
3.3 | Solar cell results
the surface recombination parameter analysis reported by McIntosh The measured I‐V parameters of the fabricated TOPCon rear emitter
and Black.51 cells with and without FSF are summarized in Table 3 for the best cells
per variation together with averaged data. The pseudo fill factor, PFF,
and total series resistance of the cells under illumination, Rs,light, are
3.2 | Surface passivation properties shown as well. The PFF was extracted from illumination intensity vs
Table 2 gives an overview of the surface passivation properties of the VOC (Suns‐VOC) measurements53,54 using the Suns‐VOC setup from
fabricated lifetime test samples and the respective sample structures. Sinton Instruments. Rs,light was determined from the voltage difference
The surface passivation properties are quantified with the implied of the Suns‐VOC and the I‐V data at maximum power point.55,56
open‐circuit voltage, iVOC, the implied fill factor, iFF, and the J0 values, Efficiency values in the range of 24% have been achieved for cells
as extracted from the injection‐dependent τeff data. The TOPCon emit- with and without FSF, which demonstrates experimentally a very high
ter lifetime sample shows very good passivation properties indicated device performance for these TOPCon rear emitter cells even without
by a high iVOC of around 720 mV and an iFF close to 85%, which clearly additional lateral conductivity of the FSF. With an independently con-
demonstrates that this TOPCon emitter is well suited for the high‐effi- firmed peak efficiency of 24.3%, the cells with FSF show a slightly
ciency p‐type Si solar cells considered in this study. The lifetime test higher performance than the cells without FSF, for which a peak effi-
sample with the 300 Ω/sq FSF shows a high iVOC of 721 mV, and thus, ciency of 23.9% was achieved. A comparison of the I‐V parameters
6 RICHTER ET AL.

TABLE 3 I‐V parameters as well as PFF and Rs,light of the best cells per variation together with averaged values. All cells were made of 10 Ω cm FZ
Si wafers, have a designated area of 2 × 2 cm2, and were measured under STC
Front side VOC, mV JSC, mA/cm2 FF, % η, % PFF, % Rs.light, Ω cm2

with FSF Best 714.2 42.4 80.8 24.3a 82.4 0.38


Ave. of 4 cells 713.7 ± 0.4 42.2 ± 0.2 79.8 ± 0.8 24.0 ± 0.3 82.1 ± 0.4 0.38 ± 0.03
w/o FSF Best 708.4 42.4 79.7 23.9 83.1 0.65
Ave. of 3 cells 711.5 ± 2.9 42.3 ± 0.1 78.9 ± 0.7 23.8 ± 0.2 82.6 ± 0.5 0.65 ± 0.02
a
Independently confirmed by Fraunhofer ISE CalLab.

reveals a very similar JSC of both cell types, which is supported by the outside the active cell area. Due to a non‐perfect masking layer
almost identical internal quantum efficiency shown in Figure 4. VOC (defects in a sacrificial SiOx layer), the TOPCon emitter in the active
values in the range of 710 mV and above were obtained for both var- cell area was harmed during this process which led to local shunts
iations, which demonstrates that a high quality surface passivation is and increased recombination currents. It is important to emphasize
provided by the phosphorus‐doped TOPCon emitter at the rear and that this issue is only related to our small‐area laboratory‐type cell
the Al2O3 passivated front surface. A slightly higher VOC level was process sequence and would not appear at full‐size solar cells, as
observed for the cells with FSF, which was not expected from the sim- no etch‐back is required there. We are quite confident to improve
ulation results shown in Figure 2, where a ~5 mV higher VOC was pre- the cell performance of our small‐area cells by the application of an
dicted for the cells without FSF. The lower VOC of the cells without improved etch‐back process, as currently being implemented in a
FSF is—at least to some extent—due to processing issues (contact new cell batch.
opening wider than p++ diffusion as detected in microscope images),
as the cells without FSF are more prone to this issue due to a higher
J0,FSF increase. 4 | CO NC L U SI O NS A N D O U TL OO K
The most significant influence on the device performances origi-
nates from the FF. The FF of the cells without FSF is ~1%abs lower In this work, we have investigated the performance of p‐type silicon
than the FF of the cells with FSF, although the PFF of the cells with- solar cells with a TOPCon rear emitter theoretically by performing a
out FSF is on average even 0.5%abs higher. The lower FF of the cells device simulation study as well as experimentally with fabricated solar
without FSF can be attributed to a ~70% higher series resistance, as cells. The focus of the investigation was on the influence of the lateral
can be observed from the Rs,light data given in Table 3. Actually, such conductivity of a FSF. The simulation study revealed that for a silicon
a ~1%abs lower FF for cells without FSF was expected from the sim- bulk resistivity of 1 Ω cm, the cell performance is quite insensitive to
ulation results, but on a significantly higher FF level around 83.5%. the sheet resistance of the FSF, indicating that a FSF is not required.
Thus, both types of cells suffer from an overall too low FF level of The simulation results showed further that cells made of 10 Ω cm Si
~80%, which can be attributed to an overall too low PFF level of only are more sensitive to the sheet resistance of the FSF, which is, how-
~82.5%. Consequently, the FF is not limited by series resistances, as ever, almost completely compensated by a higher VOC and JSC level.
supported by the reasonable Rs,light level of ~0.4 Ω cm2 for the solar Experimentally, TOPCon rear emitter cells were fabricated with and
cells with FSF, which is the same Rs,light level as observed also for the without FSF using 10 Ω cm Si wafers. Efficiency values up to 23.9%
>25% efficient front emitter n‐type Si cells reported in Ref. 5. This and 24.3% were obtained for cells without and with FSF, respectively.
overall low FF level originates—at least to some extent—from the The slightly lower performance of the cells without FSF is related to a
etch back process, which is necessary to remove the TOPCon emitter moderate series resistance increase, as predicted by the device
simulations.
100 As such, these results demonstrate a high device performance for
90 the TOPCon rear emitter cells even without a lateral conductivity in
80 the FSF. This bears the potential to simplify the process chain quite
R IQE substantially as no full‐area boron diffusion is required. There is, how-
70
IQE and R (%)

with FSF
60 ever, still a highly‐doped p++ region required underneath the front grid
without FSF
50 for contact passivation. Although a diffused local p++ region was used

40
in this study, alternative processes for the formation of local p++

30
regions have been reported, in particular based on laser doping.30-33
Thus, the high device performance obtained in this study for the cells
20
without FSF can be considered as a first step towards a high‐efficiency
10
diffusion‐free rear‐junction p‐type Si solar cell, and further work needs
0
400 600 800 1000 1200 to address whether such alternative processes can replace the high‐
wavelength (nm) quality, diffused p++ regions underneath the front grid. As rear emitter

FIGURE 4 Internal quantum efficiency IQE and reflectance R solar cells are more prone to defect recombination in the silicon bulk,57
measured on cells either with or without the FSF [Colour figure can future work needs also to address the requirements regarding the
be viewed at wileyonlinelibrary.com] electronic quality of the p‐type Si bulk material.
RICHTER ET AL. 7

ACKNOWLEDGEMEN TS 18. van Cleef MWM, Rath JK, Rubinelli FA, van der Werf CHM, Schropp
REI, van der Weg WF. Performance of heterojunction p+ microcrystal-
The authors would like to thank S. Seitz, A. Leimenstoll, F. Schätzle, N. line silicon n crystalline silicon solar cells. J Appl Phys.
Brändlin, K. Zimmermann, A. Seiler, and H. Steidl for their contribu- 1997;82(12):6089‐6095.
tions during solar cell processing and E. Schäffer and F. Martin for 19. Bivour M, Temmler J, Steinkemper H, Hermle M. Molybdenum and
performing measurements. This work was partially supported by the tungsten oxide: high work function wide band gap contact materials
for hole selective contacts of silicon solar cells. Sol. Energy Mater. Sol.
German Federal Ministry of Economic Affairs and Energy under grant
Cells. 2015;142:34‐41.
No. 0325827B “26+”.
20. Glunz SW, Feldmann F, Richter A, Bivour M, Reichel C, Steinkemper H,
Benick J, Hermle M. The irresistible charm of a simple current flow pat-
ORCID tern – 25% with a solar cell featuring a full‐area back contact, In:
Proceedings of 31st European Photovoltaic Specialists Conference
Armin Richter http://orcid.org/0000-0002-7232-8385
and Exhibition, Hamburg, Germany, 2015, pp. 259–263.

RE FE R ENC E S 21. Hermle M, Feldmann F, Eisenlohr J, Benick J, Richter A, Lee B, Stradins


P, Rohatgi A, Glunz SW. Approaching efficiencies above 25% with both
1. Adachi D, Hernández JL, Yamamoto K. Impact of carrier recombination sides‐contacted silicon solar cells, in: Proceedings of 42nd IEEE Photo-
on fill factor for large area heterojunction crystalline silicon solar cell voltaic Specialists Conference, New Orleans, LA, 2015.
with 25.1% efficiency. Appl Phys Lett. 2015;107(23):233506.
22. Benick J, Hoex B, van de Sanden MCM, Kessels WMM, Schultz O,
2. Feldmann F, Bivour M, Reichel C, Hermle M, Glunz SW. Passivated rear
Glunz SW. High efficiency n‐type Si solar cells on Al2O3‐passivated
contacts for high‐efficiency n‐type Si solar cells providing high interface
boron emitters. Appl Phys Lett. 2008;92(25):253504.
passivation quality and excellent transport characteristics. Sol. Energy
Mater. Sol. Cells. 2014;120:270‐274. 23. Hoex B, Schmidt J, Bock R, Altermatt PP, van de Sanden MCM, Kessels
WMM. Excellent passivation of highly doped p‐type Si surfaces by the
3. Masuko K, Shigematsu M, Hashiguchi T, et al. Achievement of more
negative‐charge‐dielectric Al2O3. Appl Phys Lett. 2007;91(11):112107.
than 25% conversion efficiency with crystalline silicon heterojunction
solar cell. IEEE J. Photovoltaics. 2014;4(6):1433‐1435. 24. Richter A, Benick J, Hermle M. Boron emitter passivation with Al2O3
4. Nakamura J, Asano N, Hieda T, Okamoto C, Katayama H, Nakamura K. and Al2O3/SiNx stacks using ALD Al2O3. IEEE J. Photovoltaics.
Development of heterojunction back contact Si solar cells. IEEE Journal 2013;3(1):236‐245.
of Photovoltaics. 2014;4(6):1491‐1495. 25. Steinkemper H, Hermle M, Glunz SW. Comprehensive simulation study
5. Richter A, Benick J, Feldmann F, Fell A, Hermle M, Glunz SW. n‐Type Si of industrially relevant silicon solar cell architectures for an optimal
solar cells with passivating electron contact: identifying sources for effi- material parameter choice. Prog. Photovolt: Res. Appl. 2016;
ciency limitations by wafer thickness and resistivity variation. Sol. 24(10):1319‐1331.
Energy Mater. Sol. Cells. 2017;173:96‐105. 26. Feldmann F, Simon M, Bivour M, Reichel C, Hermle M, Glunz SW.
6. Smith DD, Cousins P, Westerberg S, de Jesus‐Tabajonda R, Aniero G, Carrier‐selective contacts for Si solar cells. Appl Phys Lett.
Shen Y‐C. Toward the practical limits of silicon solar cells. IEEE J. Photo- 2014;104(18):181105.
voltaics. 2014;4(6):1465‐1469.
27. Feldmann F, Simon M, Bivour M, Reichel C, Hermle M, Glunz SW. Effi-
7. Yoshikawa K, Kawasaki H, Yoshida W, et al. Silicon heterojunction solar cient carrier‐selective p‐ and n‐contacts for Si solar cells. Sol. Energy
cell with interdigitated back contacts for a photoconversion efficiency Mater. Sol. Cells. 2014;131:100‐104.
over 26%. Nat Energy. 2017;2:17032.
28. Schultz‐Wittmann O, Turner A, Eggleston B, de Ceuster D, Suwito D.,
8. Haase F, Kiefer F, Schäfer S, et al. Interdigitated back contact solar cells van Kerschaver E, Baker‐Finch S, Prajapati V. High volume manufactur-
with polycrystalline silicon on oxide passivating contacts for both polar- ing of high efficiency crystalline silicon solar cells with shielded metal
ities. Jpn J Appl Phys. 2017;56(8S2): 08MB15. contacts, In: Proceedings of 32st European Photovoltaic Specialists
9. Taguchi M, Yano A, Tohoda S, et al. 24.7% record efficiency HIT solar Conference and Exhibition, München, Germany, 2016, pp. 456–459.
cell on thin silicon wafer. IEEE J. Photovoltaics. 2014;4(1):96‐99.
29. Rüdiger M, Schmiga C, Rauer M, Hermle M, Glunz SW. Efficiency
10. Kwark Y, Sinton RA, Swanson RM. Low J0 contact structures using potential of n‐type silicon solar cells with aluminum‐doped rear p+
sipos and polysilicon films, In: Proceedings of 20th IEEE Photovoltaic emitter. IEEE Trans Electron Devices. 2012;59(5):1295‐1303.
Specialists Conference, Las Vegas, USA, 1988, pp. 787–792.
30. Fernandez‐Robledo S, Kluska S, Greulich J, Nekarda J. Selective boron
11. Yablonovitch E, Gmitter T, Swanson RM, Kwark YH. A 720 mV open emitters using laser‐induced forward transfer versus laser doping from
circuit voltage SiOx:c‐Si:SiOx double heterostructure solar cell. Appl borosilicate glass. IEEE J. Photovoltaics. 2017; in press
Phys Lett. 1985;47(11):1211‐1213.
31. Poulain G, Blanc D, Focsa A, et al. Selective laser doping from boron sil-
12. Gan JY, Swanson RM. Polysilicon emitters for silicon concentrator solar icate glass. Energy Procedia. 2012;27:455‐459.
cells, In: Proceedings of 21st IEEE Photovoltaic Specialists Conference,
Kissimmee, USA, 1990, pp. 245–250. 32. Hameiri Z, Mai L, Puzzer T, Wenham SR. Influence of laser power on
the properties of laser doped solar cells. Sol. Energy Mater Sol. Cells.
13. Rohatgi A, Rounsaville B, Ok Y‐W, et al. Fabrication and modeling of
2011;95(4):1085‐1094.
high‐efficiency front junction N‐type silicon solar cells with tunnel oxide
passivating back contact. IEEE J. Photovoltaics. 2017;7:1236‐1243. 33. Kluska S, Granek F. High‐efficiency silicon solar cells with boron local
back surface fields formed by laser chemical processing. IEEE Electron
14. Römer U, Peibst R, Ohrdes T, et al. Recombination behavior and contact
Device Lett. 2011;32(9):1257‐1259.
resistance of n+ and p+ poly‐crystalline Si/mono‐crystalline Si junctions.
Sol. Energy Mater. Sol. Cells. 2014;131:85‐91. 34. Fell A. A free and fast three‐dimensional/two‐dimensional solar cell
15. Green MA, Hishikawa Y, Warta W, et al. Solar cell efficiency tables (ver- simulator featuring conductive boundary and quasi‐neutrality approxi-
sion 50). Prog. Photovolt: Res. Appl. 2017;25(7):668‐676. mations. IEEE Trans. Electron Devices. 2013;60(2):733‐738.

16. Feldmann F, Reichel C, Müller R, Hermle M. The application of poly‐Si/ 35. Brendel R. Modeling solar cells with the dopant‐diffused layers treated
SiOx contacts as passivated top/rear contacts in Si solar cells. Sol. as conductive boundaries. Prog Photovolt: Res Appl. 2012;20(1):31‐43.
Energy Mater Sol. Cells. 2017;159:265‐271. 36. Fell A, McIntosh KR, Fong KC. Simplified device simulation of silicon
17. Wolf SD, Descoeudres A, Holman ZC, Ballif C. High‐efficiency silicon solar cells using a lumped parameter optical model. IEEE J. Photovol-
heterojunction solar cells: a review, green 2 (1) (2012). taics. 2016;6(3):611‐616.
8 RICHTER ET AL.

37. Altermatt PP, Schenk A, Geelhaar F, Heiser G. Reassessment of the 48. Nagel H, Berge C, Aberle AG. Generalized analysis of quasi‐steady‐
intrinsic carrier density in crystalline silicon in view of band‐gap state and quasi‐transient measurements of carrier lifetimes in semicon-
narrowing. J Appl Phys. 2003;93(3):1598. ductors. J Appl Phys. 1999;86(11):6218‐6221.
38. Klaassen DBM. A unified mobility model for device simulation: II. Tem- 49. Schuldis D, Richter A, Benick J, Saint‐Cast P, Hermle M, Glunz SW.
perature dependence of carrier mobility and lifetime. Solid State Properties of the c‐Si/Al2O3 interface of ultrathin atomic layer depos-
Electron. 1992;35(7):961‐967. ited Al2O3 layers capped by SiNx for c‐Si surface passivation. Appl
Phys Lett. 2014;105(23):231601.
39. Richter A, Glunz SW, Werner F, Schmidt J, Cuevas A. Improved quanti-
tative description of Auger recombination in crystalline silicon. Phys Rev 50. Hoex B, Heil SBS, Langereis E, van de Sanden MCM, Kessels WMM.
B. 2012;86(16):165202. Ultralow surface recombination of c‐Si substrates passivated by
plasma‐assisted atomic layer deposited Al2O3. Appl Phys Lett.
40. Schenk A. Finite‐temperature full random‐phase approximation model 2006;89(4):42112.
of band gap narrowing for silicon device simulation. J Appl Phys.
51. McIntosh KR, Black LE. On effective surface recombination parame-
1998;84(7):3684‐3695.
ters. J Appl Phys. 2014;116(1):14503.
41. Bivour M, Rüdiger M, Reichel C, Ritzau K‐U, Hermle M, Glunz SW.
52. Kessler M, Ohrdes T, Altermatt PP, Brendel R. The effect of sample
Analysis of the diffused front surface field of n‐type silicon solar cells
edge recombination on the averaged injection‐dependent carrier life-
with a‐Si/c‐Si heterojunction rear emitter. Energy Procedia.
time in silicon. J Appl Phys. 2012;111(5):54508.
2011;8:185‐192.
53. Kerr M, Cuevas A. Generalized analysis of the illumination intensity vs.
42. Fong KC, Padilla M, Fell A, et al. Perimeter recombination charac- open‐circuit voltage of solar cells. Solar Energy. 2004;76(1–3):263‐267.
terization by luminescence imaging. IEEE J Photovoltaics. 2016;
6(1):244‐251. 54. Sinton RA, Cuevas A. A quasi‐steady‐state open‐circuit voltage method
for solar cell characterization, In: Proceedings of 16st European
43. Hermle M, Granek F, Schultz O, Glunz SW. Analyzing the effects of Photovoltaic Specialists Conference and Exhibition, Glasgow, 2000,
front‐surface fields on back‐junction silicon solar cells using the pp. 1152–1155.
charge‐collection probability and the reciprocity theorem. J Appl Phys.
55. Pysch D, Mette A, Glunz SW. A review and comparison of different
2008;103(5):54507
methods to determine the series resistance of solar cells. Sol Energy
44. Schumacher JO, Altermatt PP, Heiser G, Aberle AG. Application of an Mater Sol Cells. 2007;91(18):1698‐1706.
improved band‐gap narrowing model to the numerical simulation of
56. Wolf M, Rauschenbach H. Series resistance effects on solar cell mea-
recombination properties of phosphorus‐doped silicon emitters. Sol.
surements. Advanced Energy Conversion. 1963;3(2):455‐479.
Energy Mater. Sol. Cells. 2001;65(1–4):95‐103.
57. Bivour M, Schröer S, Hermle M, Glunz SW. Silicon heterojunction rear
45. Kane D.E., Swanson R.M. Measurement of the emitter saturation cur- emitter solar cells: less restrictions on the optoelectrical properties of
rent by a contactless photoconductivity decay method (silicon solar front side TCOs. Sol. Energy Mater Sol. Cells. 2014;122:120‐129.
cells), in: Proceedings of 18th IEEE Photovoltaic Specialists Conference,
Las Vegas, 1985, pp. 578–583.
46. Kimmerle A, Greulich J, Wolf A. Carrier‐diffusion corrected J0‐analysis How to cite this article: Richter A, Benick J, Müller R, et al.
of charge carrier lifetime measurements for increased consistency. Sol. Tunnel oxide passivating electron contacts as full‐area rear
Energy Mater. Sol. Cells. 2015;142:116‐122.
emitter of high‐efficiency p‐type silicon solar cells. Prog
47. Sinton RA, Cuevas A. Contactless determination of current–voltage
Photovolt Res Appl. 2017;1–8. https://doi.org/10.1002/
characteristics and minority‐carrier lifetimes in semiconductors from
quasi‐steady‐state photoconductance data. Appl Phys Lett. pip.2960
1996;69(17):2510‐2512.

You might also like