Han 2017

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

View Article Online

View Journal

Journal of
Materials Chemistry A
Materials for energy and sustainability
Accepted Manuscript

This article can be cited before page numbers have been issued, to do this please use: D. Han, M. H. Du,
C. Dai, D. Sun and S. Chen, J. Mater. Chem. A, 2017, DOI: 10.1039/C6TA10377D.

Volume 4 Number 1 7 January 2016 Pages 1–330 This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
Journal of accepted for publication.
Materials Chemistry A Accepted Manuscripts are published online shortly after
Materials for energy and sustainability

acceptance, before technical editing, formatting and proof reading.


www.rsc.org/MaterialsA

Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.

You can find more information about Accepted Manuscripts in the


author guidelines.

Please note that technical editing may introduce minor changes


to the text and/or graphics, which may alter content. The journal’s
ISSN 2050-7488
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
PAPER
Kun Chang, Zhaorong Chang et al.
Bubble-template-assisted synthesis of hollow fullerene-like
MoS2 nanocages as a lithium ion battery anode material
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.

rsc.li/materials-a
Page 1 of 22 Journal of Materials Chemistry A

View Article Online


DOI: 10.1039/C6TA10377D
Influence of Defects and Dopants on the Photovoltaic Performance of Bi 2S3:
First-Principles Insights

Journal of Materials Chemistry A Accepted Manuscript


Dan Han1,2, Mao-Hua Du3, Chen-Min Dai2, Deyan Sun1*, and Shiyou Chen2,3*
1
Department of Physics and 2Key Laboratory of Polar Materials and Devices
(MOE), East China Normal University, Shanghai 200241, China
3
Materials Science and Technology Division, Oak Ridge National Laboratory,
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Oak Ridge, TN 37831, USA


*Email - chensy@ee.ecnu.edu.cn

Bi2S3 attracted intensive attention recently as light-absorber, sensitizer or electron


acceptor materials in various solar cells. Using first-principles calculations, we find that
the photovoltaic efficiency of Bi2S3 solar cells is limited by its intrinsic point defects,
i.e., both S vacancy and S interstitial can have high concentration and produce deep
defect levels in the bandgap, leading to non-radiative recombination of electron-hole
carriers and reduced minority carrier lifetime. Unexpectedly most of the intrinsic
defects in Bi2S3, including even the S interstitial, act as donor defects, explaining the
observed n-type conductivity and also causing the high p-type conductivity impossible
thermodynamically. The doping in Bi2S3 by a series of extrinsic elements is studied,
showing that most of dopant elements such as Cu, Br and Cl make the material even
more n-type and only the Pb doping makes it weakly p-type. Based on this, we propose
that the surface region of n-type Bi2S3 nanocrystals in p-PbS/n-Bi2S3 nano-
heterojunction solar cells may be type-inverted into p-type due to the Pb doping, with
a buried p-n junction formed in the Bi2S3 nanocrystals, which provides a new
explanation to the longer carrier lifetime and higher efficiency [Nature Photonics 6, 529
(2012)]. Considering the relatively low conduction band and high n-type conductivity,
we predict that Cu, Br and Cl doped Bi2S3 may be an ideal n-type electron acceptor or
counter electrode material, while the performance of Bi2S3 as the light-absorber or
sensitizer material is intrinsically limited.
Journal of Materials Chemistry A Page 2 of 22

View Article Online


1. Introduction DOI: 10.1039/C6TA10377D

The search of novel photovoltaic semiconductor materials has drawn intensive


attention in the past decade 1-4. The emergence of the perovskite semiconductor

Journal of Materials Chemistry A Accepted Manuscript


CH3NH3PbI3 represented an important breakthrough, with the CH3NH3PbI3-based solar
cell efficiency increasing rapidly to over 22% 5-9. One important reason for the high
performance is that CH3NH3PbI3 has low carrier effective masses and thus superior
carrier mobility, which result partially from the large relativistic effects of Pb,
increasing the conduction band width through a significant lowering of the conduction
band minimum (CBM), and the hybridization of the Pb 6s states with I 5p states in the
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

top valance band, increasing the valence band width 3, 10. As the neighbor in the
elemental periodic table, the Bi3+ cation has similar electron configuration to Pb2+
(5d106s26p0) and thus can have the similarly large relativistic effects and strong
hybridization of 6s states with the anion p states in the valence band11, so the Bi related
compound semiconductors may also possess superior carrier mobility. Furthermore, Bi
is a non-toxic and environment-friendly element. As a result, the search of candidate
solar cell materials in the Bi based compounds attracts more and more attention 11-13.
Bi2S3 has a band gap around 1.3-1.45 eV 14, close to the optimal bandgap for the
light-absorber material in the single-junction solar cells 15, 16, and very high optical
absorption coefficient, >104 cm−1 for λ < 780 nm and >105 cm−1 for λ < 500 nm.
Therefore Bi2S3 attracted intensive attention as a photovoltaic material in the past
decade 17-26, e.g., as the sensitizer in the dye-sensitized TiO2 solar cells (replacing the
expensive Ruthenium dye) 22, 23, light absorber material 26, counter electrode in the
CdSe quantum dot-sensitized solar cells 24, active layer in the organic/inorganic bulk
heterojunction (BHJ) solar cells (blending Bi2S3 and polymers) 27, 28, n-type electron
acceptor in the p-PbS/n-Bi2S3 nano-heterojunction and polymer/nanocrystal
heterojunction solar cells 17, 25, 29.
However, the absolute power-conversion efficiency (PCE) of the Bi2S3 based solar
cells is still low relative to the commercialized Si, CdTe, Cu(In,Ga)Se2 or perovskite
CH3NH3PbI3 solar cells. The reason depends on the specific roles of Bi2S3 in these solar
cells, e.g., a short lifetime and diffusion length of minority carriers is detrimental if it
is used as the light-absorber or sensitizer, and low n-type conductivity is detrimental if
used as the electron acceptor. Since the intrinsic defects and extrinsic dopants influence
both the minority carrier lifetime and electrical conductivity significantly, a
systematical study on the possible defects and dopants is critically important for
understanding the current low efficiency of Bi2S3 based solar cells.
Motivated by the optimization of the thermoelectric performance, Chmielowski et al.
studied four intrinsic defects (S and Bi vacancy, SBi and BiS antisites), finding that S
vacancy and BiS antisite make Bi2S3 weakly n-type. However, other possible defects
such as S and Bi interstitials were neglected and their influences are still unknown 30.
Doping Bi2S3 using extrinsic elements including Cu, Ag, Sb, Si, Ge and Cl had been
tried experimentally 31-35, but their influence on the electrical conductivity is still not
clear. Chmielowski et al. studied the Si, Ge, Sn, Hf and Cl doping theoretically, but
none of them leads to p-type conductivity 30. Therefore, a clear understanding on how
the intrinsic defects and extrinsic dopants influence the carrier lifetime, electrical
Page 3 of 22 Journal of Materials Chemistry A

View Article Online


conductivity and thus the photovoltaic performance of Bi2S3 is still not available,
DOI: 10.1039/C6TA10377D
limiting the knowledge-based improvement of the Bi2S3 based solar cells.
In this paper, we investigated all the common point defects and a series of dopants in

Journal of Materials Chemistry A Accepted Manuscript


Bi2S3 using first-principles calculations. We found that not only the vacancy and antisite
defects, but also the interstitial defects can have high concentration in Bi2S3.
Unexpectedly most of these intrinsic defects, including even the S interstitial, act as
donors, contributing to the observed n-type conductivity and also causing the high p-
type conductivity impossible. These high-concentration defects can produce deep levels
in the band gap and act as electron-hole recombination centers, decreasing the minority
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

carrier lifetime, which should be an important reason for the current low efficiency of
the Bi2S3 based solar cells. The doping of in Bi2S3 by extrinsic elements such as Cu, Br
and Cl makes the materials even more n-type and only the Pb doping makes it weakly
p-type. Based on this, we propose that the surface region of n-type Bi2S3 nanocrystals
in the p-PbS/n-Bi2S3 nano-heterojunction solar cells may be type-inverted to p-type due
to the Pb doping, with a buried p-n junction formed in the Bi2S3 nanocrystals, which
provides a new explanation to the observed longer carrier lifetime and higher efficiency
29
. With a relatively low conduction band and high n-type conductivity, Cu, Br and Cl
doped Bi2S3 may be an ideal n-type electron acceptor or counter electrode material,
while its performance as the light-absorber or sensitizer material is intrinsically limited.

2. Calculation Methods and Models


2.1 Defect Formation Energies and Ionization Energy Levels
The defect formation energy ∆Hf (α,q) is a function of the elemental chemical
potentials as well as the electron Fermi level 36, 37, and can be calculated using the
supercell model as defined,
∆Hf (α, q) = E(α, q) − E (host) + ∑ ni (μi + Ei ) + q(EF + EVBM ) (1)
where E(α, q) is the calculated total energy of the supercell with a defect α in its
charge state q, E(host) is the total energy of the host supercell without defects, ni is
the number of atoms removed and q is the number of electrons removed from the
supercell when the defect α (in the charge state q) is formed in the supercell, Ei is the
total energy per atom of the element i in its elemental solid or gas phase, μi is the
chemical potential of the element i referenced to elemental phase with the energy Ei,
EVBM is the valence band maximum (VBM) level of the host Bi2S3 and EF is Fermi
level referenced to the VBM level (ranging from 0 to the band gap).
Under the equilibrium growth condition, there are a series of thermodynamic limits
to the chemical potentials of the component elements. Firstly, the leftover of the
elemental phases of Bi, S, and the dopant X (X=Pb, Ag, Cu, F, Cl, Br, I) should be
avoided, so μi should satisfy,
μBi < 0, μS < 0, μX < 0
Journal of Materials Chemistry A Page 4 of 22

View Article Online


Secondly, the Bi2S3 compound should be stable (under equilibrium), which requires,
DOI: 10.1039/C6TA10377D

2μBi + 3μS = ∆Hf (Bi2 S3 ) (2)


where ∆Hf (Bi2 S3 ) is the calculated formation energy of Bi2S3 (-2.01 eV).

Journal of Materials Chemistry A Accepted Manuscript


Finally, the possible secondary compounds that can be formed by the dopant
elements X and the host elements should be avoided, thus,
nμX + mμS < ∆Hf (Xn Sm ) (3)
nμBi + mμX < ∆Hf (Bin Xm ) (4)
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

For example, the formation of PbS (with a calculated formation energy -0.91 eV/f. u.)
should be avoided, so the highest possible μPb is -0.24 eV at Bi-rich condition (μBi =
0, μS = −0.67 eV), and -0.91 eV under the S-rich condition (μBi = −1.00 eV, μS =
0 ). With these chemical potential requirements satisfied, there will be no leftover of
the secondary compounds such as PbS in the Pb-doped Bi2S3 samples.
After the range of the elemental chemical potentials are determined by these
requirements, we can now calculate the formation energy ∆Hf (α, q) as a function of
the electron Fermi level (as shown in Figs. 2, 5 and 7). The transition energy level
ϵ(q⁄q′ ) is defined as the Fermi level at which the formation energies of the defect α in
the charge state q and q’ are equal, i.e., the Fermi level at the cross points in Figs. 2, 5
and 7. It can be calculated as,
ϵ(q⁄q′ ) = [∆Hf (α, q) − ∆Hf (α, q′ )]/(q − q′ ) (5)
where q′ and q are different charge states, and ∆Hf (α, q) and ∆Hf (α, q′ ) are given
by Eq. (1).38 When the Fermi level is located at the transition energy levels of a defect,
the defect can accept (or donate) electrons to get ionized to the negatively (positively)
charged state.
The total energy and electronic eigenstates are calculated using the density functional
theory 39, 40 (DFT) methods as implemented in the VASP code 41, with the local density
approximation42 (LDA) to the exchange correlation functional. The cutoff energy for
the plane-wave basis is 400 eV. Since Bi is a heavy element, the spin-orbit coupling
(SOC) is considered in our calculations. To calculate the formation energy of a defect,
a 60 atom supercell and a Γ point centered 4×4×4 k-point mesh were used. For a charged
defect, a uniform background charge is added to keep the global charge neutrality of
the periodic supercell, and the Makov–Payne correction is included to account for the
interaction between the image charges. Since LDA tends to underestimate the band gap
(usually underestimates the CBM energy and overestimates the VBM energy), we
corrected our calculated formation energies by correcting the band edge energies 43, 44.
The correction to the band edge energies are determined by comparing the band edge
energies from the LDA and hybrid functional (HSE45) calculations. Test calculations
showed that the corrected formation energies agree well with the formation energies
calculated directly using the hybrid functional 44.
Page 5 of 22 Journal of Materials Chemistry A

View Article Online


2.2 Crystal Structure and Doping Sites DOI: 10.1039/C6TA10377D

The crystal structure of Bi2S3 is orthorhombic with a space group Pbnm and has 20
atoms in the primitive cell. As shown in Fig. 1, the crystal structure of Bi2S3 can be

Journal of Materials Chemistry A Accepted Manuscript


considered as aligned (Bi4S6)n ribbons that are bound with each other by the van der
Waals force, i.e., the structure is a quasi-one-dimensional structure. In the (Bi4S6)n
ribbons there are strong covalent Bi-S bonds along [010] direction, whereas only weak
van der Waals bonds along the [001] and [100] directions between the ribbons. A
series of possible intrinsic defects, including the vacancy, antisite and interstitial defects,
are studied. For a certain defect (dopant), we consider all the possible defect (doping)
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

sites, e.g., the Bi vacancy has two different Bi sites, VBi1 and VBi2, which are both
studied, and the results are reported for the site that has the lowest formation energy.
For the interstitial defects (dopants), there are many possible interstitial sites in the
lattice. We constructed six different interstitial sites, among which the lowest-energy
sites for the cation and anion interstitials are shown in Fig. 1.

Figure 1. The crystal structure of a 60-atom Bi2S3 supercell, with the yellow balls
showing the S anions and the purple balls showing the Bi cations. The most stable
interstitial sites for the cations (in blue) and anions (in red) are also shown.

3. Intrinsic Defects
To study the properties of intrinsic defects in Bi2S3, we calculated their formation
energies as a function of the Fermi level under (a) Bi-rich and S-poor and (b) Bi-poor
and S-rich conditions, as plotted in Fig. 2 where the slope of a line shows the charge
state of a defect. According to Eq. (1), the positive slope of a line means that the defect
is a donor and can donate electrons to get ionized to a positively charged state, while
the negative slope means the defect is an acceptor and can accept electrons to get
Journal of Materials Chemistry A Page 6 of 22

View Article Online


ionized to a negatively charged state. The slope equaling to zero means that the defect
DOI: 10.1039/C6TA10377D
is neutral (not ionized). The crosspoints of the lines determine the position of the
ionization energy levels in the band gap, which are also plotted in Fig. 3.

Journal of Materials Chemistry A Accepted Manuscript


Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Figure 2. Calculated formation energies of intrinsic defects as a function of Fermi levels


under (a) Bi-rich, S-poor and (b) S-rich, Bi-poor condition. The Fermi level is
referenced to the VBM level, i.e., 0 is at the VBM level, and 1.4 eV (calculated bandgap)
is at the CBM level.

Unexpectedly most of the intrinsic defects act as the donor defects, even the anion
interstitial Si, which is usually an acceptor in compound semiconductors46, acts as a
donor. Only VBi and BiS can be acceptor, but their acceptor levels are both very deep
(high), close to the conduction band minimum (CBM) level, so they cannot get ionized
effectively and generate free hole carriers. The absence of effective acceptor defects
indicates that the undoped Bi2S3 cannot show p-type conductivity.
As Fig. 2 shows, whether under the Bi-rich and S-poor or under the Bi-poor and S-
rich condition, the donor defects such as VS and Si have much lower formation energies
and thus higher concentration than the acceptor defects, so the undoped Bi2S3 should
exhibit intrinsic n-type conductivity. This is consistent with the experimental
observation of the n-type conductivity 47-51. Furthermore, since the formation energies
of the donor defects decrease to negative as the Fermi level is close to the VBM level
(0), they will form spontaneously if the material is doped to p-type and thus pin the
Fermi level at the middle of the bandgap, so a high hole concentration (thus high p-type
conductivity) is thermodynamically impossible in Bi2S3.
Under the Bi-rich and S-poor condition, the neutral sulfur vacancy (VS) has the lowest
formation energy, only 0.89 eV (when the Fermi level is at CBM) as shown in Fig. 2.
Page 7 of 22 Journal of Materials Chemistry A

View Article Online


This is normal since the anion vacancy usually has low energy in the binary compound
DOI: 10.1039/C6TA10377D
semiconductors 46, 52, 53, especially under an S-poor condition. The ionization energy
level of VS is relatively deep, with the (0/2+) level 0.46 eV below CBM as shown in
Fig. 3. This deep level can be understood through analyzing the component of electronic

Journal of Materials Chemistry A Accepted Manuscript


band structure. In Bi2S3, the top part of the valence band is mainly composed of the S
3p states and Bi 6s and 6p states (a mixture of the bonding state of the S 3p+Bi 6p
hybridization and the antibonding state of the S 3p+Bi 6s hybridization), and the bottom
part of the conduction band is composed of the Bi 6p and S 3p states (the antibonding
state of the S 3p+Bi 6p hybridization). When an S vacancy VS is formed, the Bi-S bonds
are broken, so the anti-bonding states in the conduction band shift down and produce a
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

deep level in the bandgap. This deep level results from the Bi dangling bond, and is
thus composed mainly of the Bi 6p states, localized near the broken Bi-S bond. Because
the S anion is small while Bi cation is large, the neighboring Bi cations get closer and
have Bi 6p+Bi 6p hybridization after the S is removed, which pushes the bonding state,
(0/2+) level, downward.

Figure 3. Calculated transition (ionization) energy levels of intrinsic point defects in the
bandgap of Bi2S3.

Among all the defects, BiS antisite (Bi replacing S) has the second lowest formation
energy, always lower than 1.6 eV. It is abnormal that an antisite defect has such a low
formation energy, since the cation-substitute-anion or anion-substitute-cation antisites
usually have high formation energies in three-dimensional covalent semiconductors
such as CdTe and Cu2ZnSnSe4.44, 54 This abnormal behavior of BiS results from the
quasi-one-dimensional structure of Bi2S3, which has more tolerance to the antisite
defects than three-dimensional covalent semiconductor. As shown by the calculated
Journal of Materials Chemistry A Page 8 of 22

View Article Online


formation energy of BiS in Fig. 2, the neutral (unionized) state of BiS is not stable. When
DOI: 10.1039/C6TA10377D
the Fermi level is close to the CBM, it acts as an acceptor defect with the negatively
charged state as the most stable state, and when the Fermi level is more than 0.32 eV
below the CBM, it acts as a donor defect with the positively charged states as the most

Journal of Materials Chemistry A Accepted Manuscript


stable state. Its (+/-) and (3+/+) ionized energy levels are located 0.32 and 0.87 eV
below the CBM, respectively, as shown in Fig. 3. When S is replaced by Bi, the Bi-S
bonds are also broken, so the Bi 6p states in the conduction band drop down and
produce the defect levels in the bandgap. The (+/-) level is relatively shallow (close to
the CBM), while the (3+/+) level is very deep, because the Bi-Bi bonds are formed
when S is replaced by Bi and the strong Bi 6p+Bi 6p hybridization pushes the bonding
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

state downward dramatically into the bandgap.


Since both VS and BiS produce deep defect levels in the bandgap and have high
concentration in the Bi-rich and S-poor Bi2S3 samples, they are effective recombination
centers of the photo-generated electron-hole carriers and thus diminish the lifetime of
the minority carriers in this n-type semiconductor. Therefore we can predict that the
photovoltaic efficiency should be limited if the Bi-rich and S-poor Bi2S3 is used as the
light-absorber material.

Under the Bi-poor and S-rich condition, VS and BiS increase their formation energy
significantly, while SBi and Si have formation energies decreased to lower than 1.5 eV,
becoming the dominant defects, as shown in Fig. 2. The formation energy for neutral Si
is only 0.84 eV, and it acts as a donor defect, which is quite abnormal because the anion
interstitial is usually an acceptor defect in compound semiconductors such as in ZnS
and ZnO 52, 53. Fig. 4 shows the atomic structure around the S interstitial. As we can see,
the interstitial S not only bonds with the neighboring Bi (Bi-S bond length 2.67 Å), but
also bonds with a neighboring S (S-S bond length 2.06 Å), which is similar to the split-
anion interstitials in other compounds semiconductors 8, 55. The p-p hybridization
between the two S atoms decreases its energy, thus making Si the dominant defect.
Furthermore, the S 3p+S 3p hybridization pushes one anti-bonding state upward,
producing an occupied (0/2+) donor level in the band gap, about 0.6 eV above VBM as
shown in Fig. 3.
Neutral SBi has a formation energy slightly lower than 1.4 eV. Compared to Si, it is a
relatively shallow donor, with the (0/+) level about 0.1 eV below CBM. When an S
atom replaces a Bi atom, a large structural relaxation happens too, with a 2.06 Å S-S
bond formed (as shown in Fig. 4 for the +2 charged Si) and the strong S 3p+S 3p
hybridization pushes one anti-bonding state upward, producing the localized but
shallow (0/+) donor level near CBM.
Page 9 of 22 Journal of Materials Chemistry A

View Article Online


DOI: 10.1039/C6TA10377D

Journal of Materials Chemistry A Accepted Manuscript


Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Figure 4. Relaxed atomic structure around +2 charged S interstitial, +1 charged SBi


antisite and +1 charged Bi vacancy.

As we can see, the formation of S-S bonds makes the anion interstitial Si and anion-
substitute-cation antisite SBi become donor defects, which is abnormal because they are
usually acceptor defects52, 53. For the same reason, the cation vacancy VBi becomes a
donor defect in Bi2S3. As shown in Fig. 4, an S-S bond is also formed near the +1
charged VBi after relaxation, and the strong S 3p+S 3p hybridization pushes the
occupied bonding states downward into the valence band while the unoccupied anti-
bonding states upward to close to the CBM, which decreases the total energy
significantly and stabilizes the +1 charged donor state of VBi. This is the case when the
Fermi level is lower than 1.1 eV. However, when the Fermi level is high and the anti-
bonding states are occupied by four electrons (VBi becomes -3 charged and thus an
Journal of Materials Chemistry A Page 10 of 22

acceptor), the S 3p+S 3p hybridization cannot decrease the total energy DOI: theView Article Online
and10.1039/C6TA10377D
formation of the S-S bond is energetically not favorable. The transition from the +1
charged donor state to -3 charged acceptor state happens at the Fermi level 1.1 eV above
VBM (0.3 eV below CBM), corresponding to the (3-/+) level as shown in Fig. 3.

Journal of Materials Chemistry A Accepted Manuscript


Considering the common mechanism that the easy formation of S-S bonds stabilizes
the donor state of Si, SBi and VBi, we can understand why most of the intrinsic defects
act as donor defects in Bi2S3. The easy formation of S-S bonds results from the quasi-
one-dimensional structure of Bi2S3, which has more tolerance to the structural distortion.
Since the low-energy (high-concentration) defect Si produces deep levels in the
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

bandgap, the photovoltaic efficiency of the Bi-poor and S-rich Bi2S3 samples is also
limited. Therefore the intrinsic Bi2S3 should not be an ideal photovoltaic absorber
materials no matter synthesized under a Bi-rich, S-poor or a Bi-poor, S-rich condition.

4. Extrinsic Doping (Impurities)


From the discussion above, we can see that the undoped Bi2S3 always shows intrinsic
n-type conductivity. In the solar cell devices, the absorber or buffer layer materials are
usually doped to control its electrical conductivity and form p-n junction. To predict
whether it is possible to achieve the bipolar (both p-type and n-type) doping in Bi2S3
which is important for designing p-n junction based devices, we studied a series of
extrinsic dopant elements in Bi2S3, including Pb, Cu, Ag, I, Br, Cl and F. We found that
only Pb doping under Bi-poor and S-rich condition can make the sample show weak p-
type conductivity, while other dopants lead to good n-type conductivity.

4.1 Pb doping
When Pb is doped into Bi2S3, it tends to occupy the PbBi and Pbi sites. Their formation
energy dependence on the Fermi level is plotted in Fig. 5. Overall PbBi has lower
formation energy than Pbi, which indicates that PbBi is easier to form than Pbi. This can
be understood considering that Pb2+ has large ionic radius and the ionic radius disparity
between Bi3+ and Pb2+ is small.
The PbBi substitution is an acceptor with the (-/+) ionization energy level 0.51 eV
above VBM, as shown in Fig. 7. When Pb replaces Bi in the lattice, the Bi-S bonds are
replaced by Pb-S bonds. Because the Pb 6s levels are higher in energy than the Bi 6s
levels, the hybridization between Pb 6s states and S 3p states is stronger than that
between Bi 6s and S 3p, pushing the acceptor level up relative to the top valance bands
(which are mainly the antibonding states of the Bi 6s and S 3p hybridization), and thus
producing a deep level in the gap. In contrast, the Pbi interstitial is a donor with the
(+/2+) ionization level 0.25 eV below CBM. This deep level is composed mainly of the
Pb 6p and Bi 6p orbitals, because a Pb-Bi bond is formed around the Pbi interstitial and
the hybridization between Pb 6p states and the Bi 6p states pushes the donor levels
Page 11 of 22 Journal of Materials Chemistry A

View Article Online


downward from the conduction band into the bandgap. DOI: 10.1039/C6TA10377D

Journal of Materials Chemistry A Accepted Manuscript


Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Figure 5. Calculated formation energies of PbBi and Pbi as a function of Fermi levels
under (a) Bi-rich and (b) S-rich conditions. The formation energies of the intrinsic
defects are also included in order to show the competition between intrinsic and
extrinsic defects.

Under the Bi-rich and S-poor condition, the most favorable doping site PbBi has a
formation energy lower than 1.1 eV, but it is not as low as those of the intrinsic donor
defects such as VS (see Fig. 5), so the Pb doping cannot make the Bi2S3 sample p-type.
The competition between the PbBi acceptor and the Vs donor sets the Fermi level
located above the middle of the band gap, about 0.5 eV below the CBM level, i.e., the
sample is weakly n-type under this condition.
Under the Bi-poor and S-rich condition, PbBi decreases its formation energy
significantly, making its formation energy even lower than those of the lowest-energy
intrinsic defects such as Si and SBi. (when the Fermi level is high). The competition
between the acceptor PbBi and the donor Si and SBi sets the Fermi level located below
the middle of the band gap, about 0.5 eV above the VBM level, so the sample becomes
weakly p-type. However, it should be noted that the Fermi level is still quite high, so
the hole concentration is still low.

4.2 F, Cl, Br, I, Ag and Cu doping


In contrast with the Pb doping which makes Bi2S3 weakly n-type and even p-type,
the F, Cl, Br, I, Cu, and Ag doping enhances the n-type conductivity significantly. The
Journal of Materials Chemistry A Page 12 of 22

calculated formation energies for these dopants at different sites are plotted inDOI:
Fig. 6,View Article Online
10.1039/C6TA10377D
and their ionization (transition) energy levels in the bandgap are plotted in Fig. 7.
We discuss the results about the F, Cl, Br and I doping first. They all prefer

Journal of Materials Chemistry A Accepted Manuscript


substituting the S anion as the energetically favorable doping sites, i.e., the formation
energies of the FS, ClS, BrS and IS substitution are much lower than those of the Fi, Cli,
Bri and Ii interstitials, which can be understood considering that the substitution causes
small structural distortion. Among FS, ClS, BrS and IS, ClS has the lowest formation
energy and the FS and IS have almost degenerate energy, which should result from the
smallest size difference between Cl and S.
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

The ionization energy levels of the FS, ClS, BrS and IS donors are all resonant in the
conduction band, with no impurity levels in the bandgap, indicating that these dopants
are all ionized and generate free electron carriers once they are doped into the Bi 2S3
lattice. Since their formation energies are only 0.2-0.5 eV under the Bi-rich, S-poor
condition and 0.5-0.8 eV under the Bi-poor, S-rich condition (when Fermi level is at
CBM), their concentration (solubility) can be very high in the Bi2S3 lattice. Therefore
very high concentration of electron carriers may be achieved in Bi2S3 through Cl, Br, F
and I doping.

Figure 6. Calculated formation energies of F, Cl, Br, I, Cu, and Ag dopants at different
sites, as a function of Fermi levels under (a) Bi-rich and (b) S-rich conditions.
Page 13 of 22 Journal of Materials Chemistry A

View Article Online


DOI: 10.1039/C6TA10377D

Journal of Materials Chemistry A Accepted Manuscript


Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Figure 7. Calculated transition energy levels of different dopants in Bi2S3.

Although the Fi, Cli, Bri and Ii interstitials have higher formation energies (lower
concentration) in Bi2S3, they all show an abnormal trend, i.e., they prefer acting as
donors when the Fermi level is low, in contract with the anion interstitials in many other
semiconductors which usually act as acceptors46, 52, 53. This abnormal trend is similar to
that of the intrinsic S interstitial which acts also as a donor. When these interstitials act
as donors, they all undergo dramatic structural distortion. As shown in Fig. 8, the +1
charged Ii forms two I-S bonds with the neighboring S atoms. Bri actually takes the
place of one S atom and displaces the S atom to an interstitial site, thus splitting into
BrS and Si (with an S-S bond formed near the Si, as discussed in Section 3). Because of
the formation of S-I and S-S bonds, the p-p hybridization pushes the partially occupied
antibonding state (the donor level) to close to CBM for Ii and into the conduction band
for Bri, so the (0/+) transition energy level of Ii is close to CBM, and that of Bri is
resonant in the conduction band (absent in the bandgap). Therefore, only in a very small
Fermi level range close to the CBM level, Ii can be stable in its neutral state, and in the
other range Ii is ionized to the +1 charged donor state. Bri is always ionized to the +1
charged donor state in the whole Fermi level range.
Journal of Materials Chemistry A Page 14 of 22

View Article Online


DOI: 10.1039/C6TA10377D

Journal of Materials Chemistry A Accepted Manuscript


Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Figure 8. Structures of F, Cl, Br and I interstitials at different charged states: (a) +1


charged Fi, (b) -1 charged Fi, (c) +1 charged Cli, (b) -1 charged Cli, (e) +1 charged Bri,
(f) +1 charged Ii. The interstitial dopants are shown in blue, green and olive.
In contrast, Fi acts as an acceptor in a much wider Fermi level range, and acts as a
donor only when the Fermi level is close to VBM, as shown in Fig. 6. Fig. 8 (a) and (b)
show the local structures around Fi. It forms three F-Bi bonds with the neighboring Bi
and causes small structural distortion when it acts as a -1 charged acceptor. However, it
displaces one neighboring S significantly and forms an F-S pair (bond) which takes the
place of the S atom, when it acts as a +1 charged donor. The (-/+) transition energy level
of Fi is deep in the bandgap, resulting from the local antibonding state of F-S p-p
hybridization. Since the formation energy of the -1 charged Fi decreases as the Fermi
level increases, the formation energies of Fi decreases to close to or even lower than
that of FS when the Fermi level is close to CBM, so its concentration in n-type Bi2S3 is
not negligible. Considering that Fi may be a deep-level recombination center, we predict
that F doping is not a good choice for improving the n-type conductivity and
photovoltaic performance.
The local structure of Cli is similar to that of Fi. When the Fermi level is low, Cli is
stable as a +1 charged donor, but when the Fermi level is high, Cli can be stable in its
neutral state or the -1 charged acceptor state. Its (0/+) and (-/0) levels are close to the
CBM level. The formation energy of Cli is always high in n-type Bi2S3, so its influence
on the electric or photovoltaic performance can be neglected. Considering that both Br
Page 15 of 22 Journal of Materials Chemistry A

and Cl doping increases only the electron carrier concentration and does not cause highView Article Online
DOI: 10.1039/C6TA10377D
concentration of recombination centers, we predict that Br and Cl doping should be a
good choice for improving the n-type conductivity and photovoltaic performance of
Bi2S3. Since the formation energy of ClS and BrS can be very low under the Bi-rich, S-

Journal of Materials Chemistry A Accepted Manuscript


poor condition, very high concentration of electron carriers and high n-type
conductivity may be achieved under this condition.

The effect of Cu and Ag doping is similar to that of F, Br and I doping, although the
favorable doping site is different. Cu and Ag dopants prefer the interstitial sites (Cu i,
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Agi), rather than substituting the cation Bi (CuBi, AgBi), with formation energies at the
substitution sites much higher than those of the interstitial sites. The formation energies
of Cui and Agi are only 0.2 and 0.8 eV respectively under the Bi-rich, S-poor condition,
and 0.7 and 0.8 eV under the Bi-poor, S-rich condition (when the Fermi level is close
to CBM). They both act as the donors, with the ionization energy levels resonant in the
conduction band, so these interstitial dopants all generate free electron carriers once
they are doped into Bi2S3.
Considering their low formation energies and resonant donor levels, we predict that
Cu and Ag doping can also make the Bi2S3 sample highly n-type and push the Fermi
level upward. Especially under the Bi-rich, S-poor condition, Cu doping can produce
very high concentration of electron carriers and pin the Fermi level close to CBM.
When the Fermi level is determined to be close to CBM (n-type) by Cui and Agi, CuBi
and AgBi both act as acceptors, and their ionization energy levels are deep in the
bandgap. Their formation energies can be low (concentration can be high) in n-type
Bi2S3 sample under the Bi-poor, S-rich condition, but is very low (concentration can be
very low) under the Bi-rich, S-poor condition. So a Bi-rich, S-poor condition should be
better for increasing the n-type conductivity through Cu or Ag doping and meanwhile
suppressing the formation of recombination centers.
Journal of Materials Chemistry A Page 16 of 22

View Article Online


DOI: 10.1039/C6TA10377D

Journal of Materials Chemistry A Accepted Manuscript


Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

Figure 9. The band edge positions (relative to the vacuum level) of different
photovoltaic semiconductors, collected from Ref. 56-63. Note that the band edge position
and band gaps depend on the morphology of these semiconductors.

Summarizing the effect of the F, Cl, Br, I, Cu and Ag doping, we can see that these
dopants all prefer acting as donors and producing electron carriers in Bi2S3. This seems
to be similar to the case of the intrinsic defects, where most of them act as donors and
make Bi2S3 intrinsically n-type. The favorability of the n-type conductivity may result
from the low conduction band edge position of Bi2S3, which is 4.27 eV below the
vacuum level 17. According to the doping limit rule 46, 64, it is easier to dope a
semiconductor to n-type if its conduction band is low and it is difficult to dope it to p-
type if its valence band is low. As shown in Fig. 9, both the conduction and valence
band edges of Bi2S3 are low compared to other photovoltaic semiconductors, so the
favorability of the n-type conductivity can be understood.

5. Discussion
The calculated results showed that Pb doping in Bi2S3 is unique among all the dopant
elements that we studied, because it can make the intrinsically n-type Bi2S3 type-
inverted into p-type. Therefore a p-n junction may be formed between the Pb-doped
Bi2S3 and undoped Bi2S3. Interestingly, Rath et al. realized an over threefold efficiency
increase in the p-PbS/n-Bi2S3 nano-heterojunction solar cell with a 3-absorber-layer
structure (120 nm PbS quantum dots layer, 180 nm PbS/Bi2S3 blend nanocomposite
layer and 40 nm Bi2S3 nanocrystals layer), relative to the conventional bilayer
heterojunction solar cell (210 nm PbS quantum dots and 120 nm Bi2S3 nanocrystals) 29.
Page 17 of 22 Journal of Materials Chemistry A

The explanation to the increased efficiency is the more efficient charge separation andView Article Online
DOI: 10.1039/C6TA10377D
transport in the nano-heterojunction structure, with the recombination in the n-Bi2S3
largely suppressed since the photo-generated holes in the PbS quantum dots and
PbS/Bi2S3 nanocomposite layers are repelled from the n-type Bi2S3 layer by the built-

Journal of Materials Chemistry A Accepted Manuscript


in field 29. However, one open question is that the relatively narrow band gap of Bi2S3
around 1.3-1.45 eV makes the visible-light absorption strong and there is still a high
concentration of photo-generated holes in the Bi2S3 layer, although the photo-generated
holes in the PbS quantum dots and PbS/Bi2S3 nanocomposite layers are repelled from
the Bi2S3 layer by the built-in field. Since the intrinsic defects such as VS and Si have
high concentration and deep levels, the recombination in the Bi2S3 layer is still serious
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

and cannot be suppressed. Our current calculations show that the Pb doping is unique
in Bi2S3 and can make the semiconductor weakly p-type. During the fabrication of the
p-PbS/n-Bi2S3 nano-heterojunction solar cell, some of the Pb atoms may be doped into
the surface layer of the n-type Bi2S3 nanocrystals, with the surface layer type-inverted
into weakly p-type, then a buried p-n junction may be formed in the Bi2S3 layer, which
can facilitate the electron-hole separation and thus increase the carrier lifetime as well
as the photovoltaic efficiency. If this new explanation based on our dopant calculations
is correct, the efficiency increase should be not so obvious if the PbS layer is replaced
by other materials free of Pb (since many other elements cannot dope Bi2S3 to p-type).
We call for future experimental investigation.
Bi2S3 plays different roles in a series of solar cells, such as the sensitizer in the dye-
sensitized TiO2 solar cells 22, 23, light absorber material 26, counter electrode in CdSe
quantum dot-sensitized solar cells 24, the active layer in the organic/inorganic bulk
heterojunction (BHJ) solar cells 27, 28, n-type electron acceptor in p-PbS/n-Bi2S3
heterojunction solar cells 17, 25, 29. With the calculated properties of defects and dopants,
we can analyze the performance of Bi2S3 in these solar cells. Firstly, because of the high
concentration of deep-level intrinsic defects (serious non-radiative recombination), the
performance of the solar cells using Bi2S3 as the light absorber, sensitizer or active layer
should be limited. On the other hand, the quasi-one-dimensional structure of Bi2S3
indicates that the electron or hole carriers may have a large effective mass and thus a
poor mobility along the direction perpendicular to the (Bi4S6)n ribbons, which limits the
diffusion length of minority carriers. This is another limitation to the application of
Bi2S3 as a light absorber, sensitizer or active layer, resulting from the electronic
dimensionality as introduced by Xiao et al.4, and indicates that the orientation control
of Bi2S3 is necessary for these applications.
Secondly, although most of the intrinsic defects are donors, their formation energies
are not very low and thus the concentration is not very high, so the n-type conductivity
of undoped Bi2S3 and its performance as the n-type electron acceptor or counter
electrode is also limited. Cl, Br and Cu doping can increase the n-type conductivity
dramatically, especially under the Bi-rich condition. Meanwhile, the deep-level
recombination centers (mainly the compensating acceptors) are suppressed under the
Bi-rich condition. Therefore, we predict that good performance may be achieved using
the Cl, Br and Cu doped Bi2S3 synthesized under the Bi-rich condition as the electron
acceptor or counter electrode.
Journal of Materials Chemistry A Page 18 of 22

View Article Online


The band alignment in Fig. 9 shows that the conduction band edge of Bi2S3 isDOI:
slightly
10.1039/C6TA10377D
lower than those of CuInSe2 and Cu2ZnSnSe4 (CZTSe), between those of MAPbI3 and
MAPbBr3, and very close to that of PbS. Considering the high n-type conductivity
achievable in Cl, Br and Cu doped Bi2S3, we predict that a thin layer of n-type Bi2S3

Journal of Materials Chemistry A Accepted Manuscript


may be a good candidate as the n-type electron acceptor or counter electrode material
in CuInSe2, Cu2ZnSnSe4, MAPb(I,Br,Cl)3 and PbS based solar cells.

6. Conclusions
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

A series of intrinsic point defects and dopants in Bi2S3 solar cell are studied using
first-principles calculations. Most of the intrinsic defects in Bi2S3 including even Si are
found to be donor defects, explaining the observed n-type conductivity and also limiting
the p-type doping. Significant structural distortion with the formation of S-S bonds
makes Si, SBi and VBi (which are usually acceptor defects in other semiconductors)
become donors in Bi2S3. No matter under Bi-rich or Bi-poor condition, there are a high
concentration of deep level defects (BiS, VS, or SBi, Si), so the minority carrier lifetime
and photovoltaic efficiency of solar cells using Bi2S3 as the light absorber is limited.
Most of dopant elements such as Cu, Br and Cl make Bi2S3 more n-type, except Pb
which makes it weakly p-type. Based on this, we propose that the surface region of n-
type Bi2S3 nanocrystals in p-PbS/n-Bi2S3 nano-heterojunction solar cells may be type-
inverted into p-type due to the Pb doping, with a buried p-n junction formed in the Bi2S3
nanocrystals, which provides a new explanation to the observed longer carrier lifetime
and higher efficiency. Considering its relatively low conduction band and high n-type
conductivity, we predict that the Cu, Br and Cl doped Bi2S3 may be an ideal n-type
electron acceptor or counter electrode material, especially under the Bi-rich condition,
while its performance as the light-absorber or sensitizer material is intrinsically limited.

Acknowledgements
This work was supported by National Natural Science Foundation of China (NSFC)
under grant No. 61574059, Shanghai Rising-Star Program (14QA1401500), Shu-
Guang program (15SG20), National Key Research and Development Program of China
(2016YFB0700700) and CC of ECNU. This research was supported in part by an
appointment to the Higher Education Research Experience for Faculty at Oak Ridge
National Laboratory Program.

References:
1. M. Graetzel, R. A. J. Janssen, D. B. Mitzi and E. H. Sargent, Nature, 2012, 488, 304-
312.
2. Y. Zhou, M. Leng, Z. Xia, J. Zhong, H. Song, X. Liu, B. Yang, J. Zhang, J. Chen, K.
Zhou, J. Han, Y. Cheng and J. Tang, Adv. Energy Mater., 2014, 4.
3. A. M. Ganose, C. N. Savory and D. O. Scanlon, Chemical Communications, 2017,
53, 20-44.
Page 19 of 22 Journal of Materials Chemistry A

View Article Online


4. Z. Xiao, W. Meng, J. Wang, D. B. Mitzi and Y. Yan, Materials Horizons, 2017, DOI:DOI:
10.1039/C6TA10377D
10.1039/C6MH00519E.
5. F. Zhang, W. Ma, H. Guo, Y. Zhao, X. Shan, K. Jin, H. Tian, Q. Zhao, D. Yu, X. Lu,

Journal of Materials Chemistry A Accepted Manuscript


G. Lu and S. Meng, Chemistry of Materials, 2016, 28, 802-812.
6. N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo and S. I. Seok, Nature,
2015, 517, 476-480.
7. Y. Zhou and K. Zhu, ACS Energy Letters, 2016, DOI:
10.1021/acsenergylett.6b00069, 64-67.
8. M.-H. Du, J. Phys. Chem. Letters, 2015, 6, 1461-1466.
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

9. W.-J. Yin, T. Shi and Y. Yan, Adv. Mater., 2014, 26, 4653-4658.
10. A. Walsh, The Journal of Physical Chemistry C, 2015, 119, 5755-5760.
11. A. M. Ganose, K. T. Butler, A. Walsh and D. O. Scanlon, Journal of Materials
Chemistry A, 2016, 4, 2060-2068.
12. J. T. R. Dufton, A. Walsh, P. M. Panchmatia, L. M. Peter, D. Colombara and M. S.
Islam, Physical Chemistry Chemical Physics, 2012, 14, 7229-7233.
13. H. Shi, W. Ming and M.-H. Du, Physical Review B, 2016, 93, 104108.
14. O. Madelung, Semiconductors: Data Handbook, Springer, New York, 2004.
15. E. Pineda, M. Elena Nicho, P. K. Nair and H. Hu, Solar Energy, 2012, 86, 1017-
1022.
16. G.-Y. Liu, L.-Y. Xu, F. Zhou, Y. Zhang, H. Li, Q. F. Xu and J. M. Lu, Physical
Chemistry Chemical Physics, 2013, 15, 11554-11558.
17. L. Martinez, M. Bernechea, F. P. Garcia de Arquer and G. Konstantatos, Advanced
Energy Materials, 2011, 1, 1029-1035.
18. H.-C. Liao, M.-C. Wu, M.-H. Jao, C.-M. Chuang, Y.-F. Chen and W.-F. Su,
Crystengcomm, 2012, 14, 3645-3652.
19. S. ten Haaf, H. Sträter, R. Brüggemann, G. H. Bauer, C. Felser and G. Jakob, Thin
Solid Films, 2013, 535, 394-397.
20. S. K. Saha, A. Bera and A. J. Pal, ACS Applied Materials Interfaces, 2015, 7, 8886-
8893.
21. D. B. Salunkhe, D. P. Dubal, J. V. Sali and B. R. Sankapal, Materials Science in
Semiconductor Processing, 2015, 30, 335-342.
22. L. M. Peter, K. G. U. Wijayantha, D. J. Riley and J. P. Waggett, Journal of Physical
Chemistry B, 2003, 107, 8378-8381.
23. W. X. Li, J. Y. Yang, Q. H. Jiang, Y. B. Luo, Y. R. Hou, S. Q. Zhou, Y. Xiao, L. W.
Fu and Z. W. Zhou, Journal of Power Sources, 2016, 307, 690-696.
24. H. J. Yu, H. L. Bao, K. Zhao, Z. L. Du, H. Zhang and X. H. Zhong, J. Phys. Chem.
C, 2014, 118, 16602-16610.
25. A. K. Rath, M. Bernechea, L. Martinez and G. Konstantatos, Adv. Mater., 2011, 23,
3712-3717.
26. H. B. Song, X. J. Zhan, D. B. Li, Y. Zhou, B. Yang, K. Zeng, J. Zhong, X. S. Miao
and J. Tang, Solar Energy Materials and Solar Cells, 2016, 146, 1-7.
27. Z. J. Wang, S. C. Qu, X. B. Zeng, J. P. Liu, F. R. Tan, L. Jin and Z. G. Wang, Applied
Surface Science, 2010, 257, 423-428.
28. L. Whittaker-Brooks, J. Gao, A. K. Hailey, C. R. Thomas, N. Yao and Y. L. Loo,
Journal of Materials Chemistry A Page 20 of 22

View Article Online


Journal of Materials Chemistry C, 2015, 3, 2686-2692. DOI: 10.1039/C6TA10377D
29. A. K. Rath, M. Bernechea, L. Martinez, F. P. G. de Arquer, J. Osmond and G.
Konstantatos, Nature Photonics, 2012, 6, 529-534.

Journal of Materials Chemistry A Accepted Manuscript


30. R. Chmielowski, D. Péré, C. Bera, I. Opahle, W. Xie, S. Jacob, F. Capet, P. Roussel,
A. Weidenkaff, G. K. H. Madsen and G. Dennler, Journal of Applied Physics, 2015,
117, 125103.
31. Z. H. Ge, B. P. Zhang, Y. Liu and J. F. Li, Phys. Chem. Chem. Phys., 2012, 14,
4475-4481.
32. L.-J. Zhang, B.-P. Zhang, Z.-H. Ge, C.-G. Han, N. Chen and J.-F. Li, Intermetallics,
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

2013, 36, 96-101.


33. Y. Kawamoto and H. Iwasaki, Journal of Electronic Materials, 2013, 43, 1475-
1479.
34. D. Becerra, M. T. S. Nair and P. K. Nair, Journal of The Electrochemical Society,
2011, 158, H741.
35. X. Du, F. Cai and X. Wang, Journal of Alloys and Compounds, 2014, 587, 6-9.
36. S. B. Zhang and J. E. Northrup, Phys. Rev. Lett., 1991, 67, 2339-2342.
37. D. B. Laks, C. G. Van de Walle, G. F. Neumark, P. E. Blochl and S. T. Pantelides,
Phys. Rev. B, Condens. Matter, 1992, 45, 10965-10978.
38. S. B. Zhang, S. H. Wei and A. Zunger, Phys. Rev. B, 2001, 63, 075205.
39. P. Hohenberg and W. Kohn, Phys. Rev., 1964, 136, B864-871.
40. W. Kohn and L. J. Sham, Physcial Review 1965, 140, A1133.
41. G. Kresse and J. Furthmuller, Physical Review B, 1996, 54, 11169-11186.
42. D. M. Ceperley and B. J. Alder, Physical Review Letters, 1980, 45, 566-569.
43. S. Lany and A. Zunger, Physical Review B, 2008, 78, 235104.
44. S. Chen, A. Walsh, X. G. Gong and S. H. Wei, Advanced Materials, 2013, 25, 1522.
45. J. Heyd, G. E. Scuseria and M. Ernzerhof, J. Chem. Phys., 2003, 118, 8207-8215.
46. Y. Yan and S. H. Wei, physica status solidi (b), 2008, 245, 641-652.
47. K. Biswas, L.-D. Zhao and M. G. Kanatzidis, Adv. Energy Mater., 2012, 2, 634-
638.
48. H. T. Shaban, M. M. Nassary and M. S. El-Sadek, Physica B: Condensed Matter,
2008, 403, 1655-1659.
49. H. Mizoguchi, H. Hosono, N. Ueda and H. Kawazoe, Journal of Applied Physics,
1995, 78, 1376.
50. B. X. Chen, C. Uher, L. Iordanidis and M. G. Kanatzidis, Chem. Mater., 1997, 9,
1655-1658.
51. S. C. Liufu, L. D. Chen, Q. Yao and C. F. Wang, Appl. Phys. Lett., 2007, 90, 112106.
52. A. M. Diamond, L. Corbellini, K. R. Balasubramaniam, S. Chen, S. Wang, T. S.
Matthews, L.-W. Wang, R. Ramesh and J. W. Ager, physica status solidi (a), 2012,
209, 2101-2107.
53. A. Janotti and C. G. Van de Walle, Physical Review B, 2007, 76, 165202.
54. S.-H. Wei and S. Zhang, Physical Review B, 2002, 66, 155211.
55. D. O. Scanlon, B. J. Morgan, G. W. Watson and A. Walsh, Physical Review Letters,
2009, 103, 096405.
56. S. Chen, A. Walsh, J.-H. Yang, X. G. Gong, L. Sun, P.-X. Yang, J.-H. Chu and S.-
Page 21 of 22 Journal of Materials Chemistry A

View Article Online


H. Wei, Phys. Rev. B, 2011, 83, 125201. DOI: 10.1039/C6TA10377D
57. S. Chen and L.-W. Wang, Chemistry of Materials, 2012, 24, 3659-3666.
58. C.-H. M. Chuang, P. R. Brown, V. Bulović and M. G. Bawendi, Nature Materials,

Journal of Materials Chemistry A Accepted Manuscript


2014, 13, 796-801.
59. N. Guijarro, T. Lutz, T. Lana-Villarreal, F. O’Mahony, R. Gómez and S. A. Haque,
J. Phys. Chem. Letters, 2012, 3, 1351-1356.
60. H. S. Jung and N.-G. Park, Small, 2015, 11, 10-25.
61. L. Martinez, M. Bernechea, F. P. G. de Arquer and G. Konstantatos, Adv. Energy
Mater., 2011, 1, 1029-1035.
Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

62. C. E. Patrick and F. Giustino, Advanced Functional Materials, 2011, 21, 4663-4667.
63. Z.-K. Yuan, S. Chen, H. Xiang, X.-G. Gong, A. Walsh, J.-S. Park, I. Repins and S.-
H. Wei, Advanced Functional Materials, 2015, 25, 6733-6743.
64. S.-H. Wei, Comput. Mater. Sci., 2004, 30, 337-348.
Journal of Materials Chemistry A Page 22 of 22
View Article Online
DOI: 10.1039/C6TA10377D

Table of Contents

Journal of Materials Chemistry A Accepted Manuscript


Published on 23 February 2017. Downloaded by University of Newcastle on 23/02/2017 22:34:05.

First-principles calculations show that the photovoltaic efficiency of the solar cells
using Bi2S3 as the light absorber is intrinsically limited by its point defects, while Cu,
Br and Cl doped Bi2S3 may be ideal n-type electron acceptor or counter electrode
material.

You might also like