1. IntroductionThe silicon heterojunction (SHJ) solar cell is one of the most promising candidates for photovoltaic applications due to high performance, high reliability and low cost. This type of cell features ultra-thin intrinsic and doped hydrogenated amorphous silicon (a-Si:H) layers on an n- or a p-type c-Si wafer.[1] Panasonic has achieved a record conversion efficiency of 25.6% in an SHJ structure based on n-type solar-grade c-Si.[2] It has been reported that the n-type SHJ solar cell can produce a higher efficiency because it has an inherent advantage over the p-type SHJ solar cell, due to a higher minority carrier band offset at the interface, which can result in a higher open circuit voltage.[3] In spite of this fact, the p-type c-Si wafer still dominates current industrial photovoltaics production. Moreover, the minority carrier diffusion length in p-type c-Si is higher than that in n-type material.[4] So the exploitation of SHJ solar cells on p-type c-Si is of great practical interest. In recent years, many research groups have conducted the investigation on developing high performance SHJ solar cells on p-type substrates.[5–7] So far the best efficiency of the p-type SHJ solar cell has been reached at 21.4%.[5] However, efforts are still needed to further improve the performance of the p-type SHJ solar cell.
It is well known that a very important loss mechanism of the p-type SHJ solar cells is parasitic absorption of photons in the n-type a-Si:H layer. Thus one promising way is to use a silicon-based material with wide bandgap as the window layer of the SHJ solar cell. Hydrogenated amorphous silicon carbide (a-SiC:H) with wide optical bandgap (1.8 eV–2.3 eV), low absorption in visible wavelength region, and good thermal stability is an appealing alternative for the window layer. In addition, a good interface passivation can be obtained by a high-quality a-SiC:H(n) layer.[8] Despite encouraging results having been obtained by incorporating a-SiC:H into p-type SHJ solar cells, the
and FF are relatively low. Moreover, S-shaped I–V curves are often observed.[4,9–11] The phenomena mentioned above are closely linked to the interface states, increased heterojunction band offset and TCO/a-SiC:H front contact. A poor interface with high interface states density (
) can lead to high interface recombination. This will sharply increase dark tunneling leakage current and deteriorate
dramatically. Considering minority carrier electrons transport across the a-SiC:H(n)/c-Si(p) heterojunction to the front contact, the minority carrier band offset (
) and the work function of transparent conductive oxide (
) are of clear importance. They play crucial roles in determining the minority carriers transfer property. However, it is still a matter of debate how the
,
, and
control the performance of a-SiC:H(n)/c-Si(p) SHJ solar cells. Thus we addressed this question by means of numerical computer simulation method in order to improve the performance of p-type SHJ solar cells. The purpose of this work is to systematically study the influence mechanisms of these parameters.
In this paper, the SHJ solar cells with the structure of TCO/a-SiC:H(n)/c-Si(p) (textured)/a-Si:H(i)/a-Si:H(p+)/TCO were modeled by using AFORS-HET software. We discussed in detail the influence of the interface states density, conduction band offset and front contact work function on the photovoltaic properties of solar cells. The influence mechanisms that interface states, conduction band offset, and front contact impact on the carrier transport, interface recombination and cell performance were studied in detail. Finally, by optimizing
,
, and
, the p-type a-SiC:H(n)/c-Si(p) SHJ solar cell with efficiency of 23.06% was achieved.
2. Modeling2.1. Physical modelThe SHJ solar cell structure TCO/a-SiC:H(n)/c-Si(p)(textured)/a-Si:H(i)/a-Si:H(p)/TCO was modeled by AFORS-HET. Within the bulk of each layer, AFORS-HET solves one-dimensional Poisson equation and the electron and hole continuity equations for a given defect state distribution using Shockley–Read–Hall recombination statistics,[12,13]
where
φ,
q,
, and
n/
p are the electric potential, electron charge, defect density of charged defects, and electron/hole density, respectively.
and
denote the electron/hole current and the donor/acceptor concentration.
and
represent the electron/hole optical generation rate and recombination rate, respectively.
For amorphous layers, it has been assumed that there are both exponential band tail states and Gaussian mid-gap states (associated to silicon dangling bonds) in the bandgap. The valence and conduction band tail states are usually given as follows
where
and
are the densities per energy range for tail states at the band edge energies
and
.
and
stand for characteristic parameters for the conduction and valence band tail states, respectively.
The mid-gap states are described by Gaussian distributions of acceptor-like states and donor-like states,i.e.
where
and
are the peak energy position.
and
denote the effective density of states.
and
represent the standard energy deviation of the Gaussian acceptor and donor levels, respectively.
Furthermore, the voltaged controlled boundary condition is selected, and the potential at the front contact (x = 0) is fixed at zero. Then at the front contact (x = 0), the boundary condition is
Accordingly, the boundary condition at the back contact(
x =
L) is
where
ϕ and
S are the metal work function and surface recombination velocity, respectively.
2.2. Input parameters for simulationThe basic parameters for the modeled structure TCO/a-SiC:H(n)/c-Si(p) (textured)/a-Si:H(i)/a-Si:H(p)/TCO are listed in Table 1. The gap state densities of all kinds of amorphous layers were taken from Refs. [14] and [15]. Figure 1 shows the gap state distribution in the a-SiC:H(n) and a-Si:H(p) layers. The energy difference between the donor-type and acceptor-type defect is 0.3 eV. As for the a-Si:H(i) layer, the acceptor-like and donor-like defect are symmetrically positioned 0.15 eV from the midgap with a much lower density(not shown). The interface defects density at the front a-SiC:H(n)/c-Si(p) interface was modeled by an additional defect rich layer 5-nm thick. Then the volume defect density range of
–
in the 5-nm thick layer corresponds to the surface defect density range of
–
. The recombination velocities at the front and back contact were both fixed at 1 × 107 cm/s. The operation temperature was set at 300 K, and the solar illumination was set at AM 1.5 (a power density of 100 mW/cm2). The light reflection of the front and the back contacts was set at 0.1 and 1, respectively. Moreover, in Subsections 3.1 and 3.2, the influence of TCO/a-SiC:H(n) front contact was ignored by setting the front contact as an ohmic contact.
Table 1.
Table 1.
 | Table 1.
Basic parameters adopted for layers used in the simulation.
. |
3. Results and discussion3.1. Influence of interface statesFigure 2 shows the calculated J–V characteristics of the a-SiC:H(n)/c-Si(p) cells with different interface state densities. It can be seen that the increase of
leads directly to a loss in
. When the
is above
,
drops sharply. This result is in agreement with the Refs. [14], [16]–[18]. The influence of the interface states on
can be described by the following formula proposed by Jensen et al.[19]
where
and
denote the effective barrier height in c-Si and interface recombination velocity, respectively. Obviously, a high
means a high
at the a-SiC:H(n)/c-Si(p) interface. It can be seen from Eq. (
14) that the increase of
leads to a logarithmic decrease in
.
To further understand the sensitivity of
to the interface states, we calculated the electric field distribution in the a-SiC:H(n)/c-Si(p) heterojunction for the two cases of low
and high
, as shown in Fig. 3(a). It can be seen that the interface with high
has a very strong electric field at the a-SiC:H(n)/c-Si(p) interface but a weak field on the c-Si(p) side. This can be attributed to the high density of interface states that can trap most of the electrons during the p-n junction formation. When an n-type a-SiC:H layer is in contact with a p-type c-Si wafer, electrons will flow from the n-side to the p-side to equilibrate their Fermi levels. In the high interface state density case, most electrons will be trapped in the states. These occupied acceptor-like interface states will become a high density of negative space charges on the c-Si(p) surface, leading to a localized space charge region on the c-Si(p) side. As a result, a very high electric field at the interface and a collapsed one on the c-Si(p) side come into being. Furthermore, figure 3(b) indicates that for the high
case, the band bending on the c-Si(p) wafer side is flattened. This increases the interface recombination velocity, giving rise to a striking fall in
. Therefore, in order to obtain high
, it is very critical to ensure a good surface passivation for SHJ solar cells. The interface states’ density should be controlled below
.
Figure 4 shows the dependence of
and
on the interface states density for the a-SiC:H(n) emitter and a-Si:H emitter, respectively. It can be seen that both
and
show the same varying trend with the increase of
for the two emitters. What is more, much higher
was obtained in the a-SiC:H(n)/c-Si(p) structures. We attributed this to the larger bandgap of a-SiC:H(n), which decreases optical absorption losses in the layer. Figure 5 shows the IQE curves for the a-SiC:H(n)/c-Si(p) and a-Si:H(n)/c-Si(p) structures. It clearly shows that the a-SiC:H(n)/c-Si(p) structure has a higher blue response than the a-Si:H(n)/c-Si(p) structure. This means that there are more holes generated in the a-SiC:H(n) layer that are able to pass the junction and then to enhance the short circuit current density.
3.2. Influence of the conduction band offsetCompared to a-Si:H(n), the a-SiC:H(n) emitter with a larger bandgap may give rise to a higher conduction band offset (
) at the a-SiC:H/c-Si interface. Considering the
can vary with the deposition conditions, it is essential to carefully inspect the impacts of
on the performance of solar cells.[20,21]
Figure 6 shows the simulated dependence of the solar cell performance on the conduction band offset. Here, two cases of the interface defect density,
and
, were assumed. Simulation reveals that the open circuit voltage and fill factor are very sensitive to
. Moreover,
is relatively high when
falls in the range of 0.15 eV∼0.2 eV. On the other hand, FF is in a low level when
is above 0.15 eV. It can be clearly understood through the equilibrium energy band diagram for the a-SiC:H(n)/c-Si(p) heterojunction with different conduction band offsets, as shown in Fig. 7(a). It can be seen that with the increase of
, the band bending in the c-Si absorber is enhanced. As shown in Fig. 7(a), the built-in voltage is enhanced from 530 mV to 700 mV as
increases from 0.05 eV to 0.25 eV. This leads to a strong inversion layer forming at the a-SiC:H(n)/c-Si(p) interface. Then the interface recombination is suppressed due to fewer majority carrier holes available there for recombination.
Figure 7(b) illustrates the simulated electric field distribution at the a-SiC:H(n)/c-Si(p) heterojunction. As can be seen from Fig. 7(b), the higher
results in a stronger electric field at both the a-SiC:H(n) and c-Si(p) sides. This further suppresses the interface recombination, leading to an increase in the
. However, when
is above 0.2 eV, the enhanced band offset gives rise to a transport barrier for electrons. Hence, electrons will accumulate at the a-SiC:H(n)/c-Si(p) interface. This not only causes electrons trapped in the acceptor interface states, but also encourages the holes in the absorber c-Si(p) to back diffuse and recombine with electrons. All this offsets the suppressing effect of the inversion layer on the interface recombination. The net result is that
and FF strongly decrease when
is in the range of 0.20 eV∼0.25 eV. Therefore, higher efficiency can be obtained in the a-SiC:H/c-Si SHJ solar cells with
in the range of 0.10 eV∼0.15 eV.
3.3. Influence of TCO/a-SiC:H(n) front contactThe work function of transparent conductive oxide (
) is usually higher than that of a-SiC:H(n) layer. Therefore, when TCO and a-SiC:H(n) are brought into contact, a Schottky barrier develops at the TCO/a-SiC:H(n) interface, which significantly influences the performance of SHJ solar cells.
Figure 8 shows the J–V characteristics of the a-SiC:H(n)/c-Si(p) cells with different values of
. The doping concentration of a-SiC:H(n) (
) was set as
, corresponding to the
. The values of
and
were fixed at
and 0.15 eV, respectively. It can be seen that
and FF are strongly influenced by the
. When the
is higher than 4.3 eV,
and FF decrease considerably. What is more, the S-shaped J–V curves can be clearly observed when
. Similar results were also reported by Zhao et al. and Dao et al.[16,17]
Figure 9 shows the equilibrium energy band diagram of the a-SiC:H(n)/c-Si(p) cells for the two cases of
and
. As shown in Fig. 9, when
is 4.2 eV, the TCO/a-SiC:H(n) front contact is in flat band condition. However, when
is high, the conduction band near the TCO/a-SiC:H(n) interface distinctly bends upward, forming an inverted junction to the a-SiC:H(n)/c-Si(p) junction. Consequently, the Schottky front contact with high
severely hinders the electron transport by forming a resistive bed. This induces the generation of S-shaped J–V curves. Furthermore, for the case of
, the Fermi level has an obvious shift towards the valence band, and meanwhile a strongly flattened band bending occurs on the absorber side. This is mainly because when
is 4.6 eV, the built-in potential
in the TCO/a-SiC:H(n) Schottky contact is as high as 0.4 eV. The high
not only drives the thin a-SiC:H(n) emitter layer into depletion but also, more importantly, additional electrons in the absorber have to be used to shield the electric field. As a result, the width of the space charge region and the band bending on the absorber side are drastically reduced.
Figure 10 indicates that the flattened band bending in the case of high
results in a significantly weakened electric field on the c-Si(p) side. This enhances the interface recombination, leading to a sharp decrease in both
and FF. So it is necessary to carefully tune the value of
to obtain high efficiency a-SiC:H(n)/c-Si(p) solar cells. Also, our simulation results show that the influence of the TCO/a-SiC:H(n) front contact is negligible when the
in the front contact TCO/a-SiC:H(n) is no higher than 0.2 eV. This means that
should be as low as possible for the p-type SHJ solar cells.
Finally, according to the analyses and optimum parameters provided above, we modeled the J–V characteristics for the a-SiC:H(n)/ c-Si(p) SHJ solar cell by choosing
as 4.2 eV,
as
and
as 0.15 eV. As we expected, high performance with
,
, FF = 0.814, and η = 23.06% was achieved. For p-type SHJ solar cell the improved efficiency is 1.7 points over the record efficiency by experiment. However, it is still lower than the best efficiency achieved on n-type SHJ solar cell. This is mainly due to the different energy band structure between the p-type and n-type SHJ solar cells. N-type SHJ solar cell has the higher minority carrier band offset at the interface, which can suppress the interface recombination effectively. On the other hand, the efficiency of p-type SHJ solar cell can be further improved by optimizing its back c-Si(p)/a-Si:H(p+)/TCO structure.[22] The related research is right going on.
4. ConclusionsIn summary, by numerical computer modeling we have simulated the p-type SHJ solar cells with the structure of TCO/a-SiC:H(n)/c-Si(p) (textured)/a-Si:H(i)/a-Si:H(p+)/TCO. The influence of interface states, conduction band offset, and front contact on the performance of the p-type SHJ solar cells was studied in detail. It is shown that by using wide bandgap a-SiC:H(n) instead of a-Si:H(n) as the window layer, the short-circuit current can be remarkably increased. What is more, the open circuit voltage and fill factor are very sensitive to the interface states, conduction band offset and work function of TCO. Firstly, a poor a-SiC:H(n)/c-Si(p) interface with high
can result in a striking fall in the open circuit voltage due to a flattened band bending on the c-Si(p) wafer side. In order to achieve high
the interface state density should be lower than 1
. Secondly, the increase of conduction band offset may induce a strong inversion layer forming at the a-SiC:H(n)/c-Si(p) interface, which suppresses the surface recombination. On the other hand, it gives rise to a transport barrier for electrons, which enhances carrier recombination due to electrons accumulation at the interface. The trade-off between the inversion layer and transport barrier leads to the
in the range of 0.1 eV–0.15 eV. Thirdly, the
should be as low as possible. It is shown that when the
is high, it can drive the J–V curves into an S-shape, owing to an inverted junction forming at the front contact. Meanwhile, high
can also reduce
and FF drastically due to the enhanced interface recombination. Finally, by employing the optimum parameters for
,
, and
, the p-type a-SiC:H(n)/c-Si(p) SHJ solar cell with the high efficiency of 23.06% was achieved.
