Chem Soc Rev

View Article Online

REVIEW ARTICLE View Journal | View Issue
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Chemical approaches for electronic doping in

photovoltaic materials beyond crystalline silicon
Cite this: Chem. Soc. Rev., 2022,
51, 10016
Xueyuan Wei,a Pengxiang Zhang,a Tailai Xu,a Huanping Zhou, b
Yang Bai *a
and Qi Chen *a

Electronic doping is applied to tailor the electrical and optoelectronic properties of semiconductors,
which have been widely adopted in information and clean energy technologies, like integrated circuit
fabrication and PVs. Though this concept has prevailed in conventional PVs, it has achieved limited
success in the new-generation PV materials, particularly in halide perovskites, owing to their soft lattice
nature and self-compensation by intrinsic defects. In this review, we summarize the evolution of the
theoretical understanding and strategies of electronic doping from Si-based photovoltaics to thin-film
technologies, e.g., GaAs, CdTe and Cu(In,Ga)Se2, and also cover the emerging PVs including halide
perovskites and organic solar cells. We focus on the chemical approaches to electronic doping,
emphasizing various chemical interactions/bonding throughout materials synthesis/modification to
device fabrication/operation. Furthermore, we propose new classifications and models of electronic
Received 3rd July 2022 doping based on the physical and chemical properties of dopants, in the context of solid-state
DOI: 10.1039/d2cs00110a chemistry, which inspires further development of optoelectronics based on perovskites and other hybrid
materials. Finally, we outline the effects of electronic doping in semiconducting materials and highlight
rsc.li/chem-soc-rev the challenges that need to be overcome for reliable and controllable doping.

1. Introduction promising candidates for replacing fossil fuels to address the

energy crisis. The Sun provides more than enough energy to
Energy depletion and environmental pollution are two primary meet the whole world’s energy needs. Photovoltaics directly
bottlenecks for global development. Solar energy is one of the convert solar energy into electricity, which is an important
sustainable energy source. However, approximately only 1.7%
a
Experimental Centre for Advanced Materials, School of Materials Science and of the global electricity demand has been met by electricity
Engineering, Beijing Institute of Technology, Beijing, 100081, P. R. China. generated by PVs up to now. The only limitation of solar energy
E-mail: [email protected], [email protected]
b utilization is its higher cost for electricity generation compared
School of Materials Science and Engineering, Peking University, Beijing, 100871,
P. R. China to other technologies.1

Xueyuan Wei received his BS Pengxiang Zhang received his

degree (2019) from the School of master degree in 2022 from the
Materials Science & Engineering, School of Materials Science &
the University of Science and Engineering, Beijing Institute of
Technology Beijing, and received Technology. He was a master
his MS degree (2022) at Beijing student under the supervision of
Institute of Technology, under Prof. Qi Chen in the School of
Prof. Qi Chen’s supervision. Now Materials Science & Engineering,
he focuses on the perovskite opto- Beijing Institute of Technology.
electronic device design, pre- His research interest is deve-
paration and applications. loping high-efficiency and stable
perovskite solar cells while
Xueyuan Wei Pengxiang Zhang learning in Beijing Institute of
Technology.

10016 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

In PV technology, semiconductor materials are usually used contact regions severely limits the PCEs of these devices.
as light absorbers. Electronic doping is a process that enables Through heavy doping, BSF was established in the metal–
the local manipulation of charge carrier density and conduc- semiconductor contact interface, promoting the efficiency of
tivity in a semiconductor, and it is essential for semiconductor- commercial Si solar cells to B20%, thus reducing the genera-
based electronic devices, in particular solar cells (Table 1). tion cost significantly.5
Through electronic doping, p–n junctions can be formed, Even so, Si solar cells remain expensive and still have a high
which are the elementary ‘‘building blocks’’ of solar cells. Based EPBT.30 New-generation PV materials feature weak chemical
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

on the p-type and n-type wafers, c-Si solar cells were produced bonding, low cost, and solution processability, which are
and used for the first time in a prominent application when promising for achieving PV grid parity. However, it is difficult
they were equipped on the Vanguard satellite in 1958.2 to achieve reliable and controllable doping, especially for
However, the cost of generating electricity from solar cells halide perovskite materials, limiting their device performance
was staggering due to their high manufacturing cost and low for commercialization.31–33 The fundamental understanding
conversion efficiency at that time, which was still as high as and modulation of the underlying properties of these materials
$100 per watt in 1971. In c-Si solar cells, the Schottky barrier, a are still limited, which hinders the future material/device
region with high resistance for carriers, can be formed at design. In addition, typical physical approaches that are suc-
the interface between the metal and the semiconductor.3,4 cessfully adopted in conventional PVs, like ion implantation,
Recombination of photo-generated electrons and holes in these may cause severe crystal damage to materials featuring soft

Tailai Xu received his bachelor Huanping Zhou received her PhD

degree in renewable energy at in inorganic chemistry from
Beijing Institute of Technology Peking University in 2010. After
and is subsequently studying for that, she joined the University of
a master’s degree in materials California, Los Angeles, as a
science. He had evaluated the postdoctoral researcher from
bottlenecks for the application 2010 to 2015. Now she is a
of perovskite/crystalline silicon professor at Peking University in
tandem solar cells. Now he the School of Materials Science
focuses on the manufacturing of and Engineering. She is a
the functional layers of perovskite materials chemist with expertise
solar cells while also gives in the fields of nanoscience, thin
Tailai Xu consideration to the applic- Huanping Zhou film optoelectronics, organic/
ability of the developed methods inorganic interface engineering,
on continuous fabrication. and the development and fabrication of related devices, such as
photovoltaic cells, LEDs, etc. Currently, her group focuses on thin
film optoelectronics, e.g., perovskite materials and solar cells.

Yang Bai received his BS degree Qi Chen holds both his BS and
(2013) from the School of MS degrees from Tsinghua
Chemistry and Environment, University, and received his PhD
Beihang University, and received degree from the University of
his PhD degree (2017) at The California, Los Angeles (UCLA).
Hong Kong University of Science From 2013 to 2016, he worked
and Technology (HKUST). He as a postdoc fellow at California
worked as a postdoc fellow at Nanosystem Institute (CNSI),
the Department of Chemistry, UCLA. Now he is a professor at
HKUST, under Prof. Shihe Beijing Institute of Technology.
Yang’s supervision. Now he is an His research focuses on hybrid
associate professor at the Beijing materials design, processing and
Yang Bai Institute of Technology. His Qi Chen applications in opto-electronics
research focuses on perovskite and energy harvesting and
solar cells and nanomaterials for opto-electronics. He has storage. He has published more than 150 peer-reviewed papers,
published more than 60 peer-reviewed articles, with his total with total citation over 20 000 and a H-Index of 53.
citation over 5000 and a H-Index of 35.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10017
 View Article Online

Chem Soc Rev Review Article

Table 1 Performance and electronic doping type of different photovoltaic technologies achieved using intrinsic and extrinsic dopants

Technology PCEa (%) Doping type Intrinsic dopants Extrinsic dopants Ref.
Si 26.7 n or p Si dangling bonds, VSi B, P, Al 6–8
GaAs 29.1 n or p GaAs, AsGa, Gai Si, Zn 9 and 10
CdTe 22.1 n or p VCd, VTe, TeCd Cu, Cl, P 11 and 12
CIGS 23.4 p VCu, InCu, VSe, VSe–VCu Na, K 13–17
CZTSSe 13.0 p VCu, ZnCu, CuZn — 18–20
QDSC 18.1 n or p — Ni(tfd)2 21
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

OPV 18.2 n or p — F4TCNQ, Mo(tfd)3 22–24

PSC 25.7 n or p VPb, VI, VMA, Ii Ag, Bi, F4TCNQ 25–29
a
The listed power conversion efficiencies are for the record highest light-to-electricity conversion achieved in a photovoltaic device confirmed by
National Renewable Energy Laboratory.

lattice and ionic behavior.34,35 Thus, chemical approaches have The electron configuration in the semiconductor under thermo-
been developed and intensively investigated.36,37 However, dynamic equilibrium follows the Fermi–Dirac statistics. But the
systematic investigation is still required to reveal their doping introduced localized electrons from dopants do not follow the
validity and impacts on materials properties and consequent original electronic distribution, and were in temporary violation of
carrier behaviors. the Fermi–Dirac statistics.44 The dopants must have the statistical
‘‘Doping’’ in new-generation PV materials focuses on the tendency to move to a new thermodynamic equilibrium that
introduction of certain kind of impurities into the lattice for obeys the Fermi–Dirac statistics, resulting in the electron trans-
controlling crystal growth and/or tuning the device perfor- port between dopants and the semiconductor.
mance (efficiency and stability).38–40 In this review, the ‘‘elec- The doping type is determined by the introduced localized
tronic doping’’ techniques applied in PV technologies are energy level. If an isolated full energy level appears after doping
summarized. We review the electronic doping mechanisms, and it is higher than the original full energy level, that is, VB of
strategies, and impacts from Si-based photovoltaics to thin-film the semiconductor, the electrons in this energy level are more
technologies (e.g., CdTe, CIGS), to the latest generation of PSCs likely to enter an empty band, that is, the CB of the semi-
and OPV materials finally. To understand the evolution of conductor and become free electrons. This doping is called
electronic doping, we make two comparisons: (1) conventional n-doping. In contrast, if an isolated vacant energy level appears
semiconductors and halide perovskites, and (2) physical and after doping, as long as it is lower than the CB of the semi-
chemical approaches, in which their applications in PVs are conductor, it is more likely to accept electrons and leave a hole
extensively discussed. In particular, we discuss chemical in the VB. This doping is called p-doping.
approaches to electronic doping, emphasizing various chemical When the dopant is atomic, the positions of the introduced
interactions/bonding throughout materials synthesis/modifica- energy levels are usually related to the number of valence
tion to device fabrication/operation from a chemistry perspec- electrons, so we can simply use the atomic valence to analyze
tive. New classifications (atomic doping, molecular doping and the doping type.45,46 When the dopant is molecular, the positions
self-doping) and quantitative models of electronic doping are of empty and full bands are related to the valence electrons of the
proposed based on the physical and chemical properties of molecule, the doping type is determined by the HOMO and/or
dopants in the context of solid-state chemistry, which hopes to LUMO energy levels of the molecules as compared to the semi-
inspire further developments of optoelectronics for perovskites conductor energy levels.47,48 In addition, some semiconductors
and other hybrid materials. We hope our contribution will shed with low defect formation energy and high defect concentration
light on the understanding of electronic doping and develop- exhibit unusually observed self-doping.49,50 So far, numerous
ment of effective approaches to modulating the electronic studies have explored the introduction of free charge carriers into
properties of materials. semiconductors under equilibrium conditions by atomic doping,
molecular doping, or self-doping.

2. The working principles of electronic

doping 3. Atomic doping
Different from a single, isolated atom with discrete energy The most prevalent strategy for semiconductor electronic doping
levels, a semiconductor crystal consists of a huge number of involves the introduction of charged impurities compensated by
atoms (N - N) whose atomic orbitals overlap and split into a excess charge carriers.51–53 Unlike molecular doping, this strategy
near-continuous energy band. The dopants can be incorporated has been widely adopted in the industry to yield highly stable n- or
into the semiconductor lattice as substitutions or interstitials, p-type doped semiconductors.54 Conventional electronic doping is
or they may be segregated at lattice defects, grain boundaries realized by ion implantation via the bombardment of semicon-
and interfaces.41–43 Usually, dopants are dispersed in a low ductors with energetic ions (the dopants) followed by thermal
concentration without an orbital overlap, which forms a series annealing, which has been successfully utilized in Si, GaAs, CdTe,
of isolated energy levels scattered in the energy bands. and oxides.55–58 However, atomic doping remains a challenge in

10018 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

some systems in which only a small amount of excess charge

carriers can be introduced from the charged defects, especially
for perovskite materials.59 In most cases, charged defects are
compensated by localized counter charges at the grain boundaries
or surfaces.54,59

3.1 Doping mechanism

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

According to the position of foreign atoms in the lattice,

impurities can be categorized into interstitial or substitutional
defects. Substitutional impurities are formed when the original
atom in the lattice site of a crystalline solid is replaced by a Fig. 2 Schematic bond diagram for n-type doping of Si with P as dopants
(a), and for p-type doping with B (b).
different type of atom, as shown in Fig. 1(a). Foreign atoms that
occupy the interstitial position are called interstitial impurities,
as shown in Fig. 1(b). Both substitution and interstitial 3.2 Doping strategy
impurities can introduce impurity energy levels into a semi-
conductor, resulting in p- or n-type doping. The ionization energy and electron affinity energy of an isolated
When foreign atoms occupy the lattice site of the crystal, atom are primarily determined by the number of electrons in
depending on their valence (valence electrons number), the its outermost shell (valence electron), which determine whether
dopants may give electrons as free charge carriers to form an the dopants give or take electrons from the semiconductor.
n-type doping, or accept electrons from the semiconductor, According to the valence of doped ions, the doping strategy can
leaving positively charged holes in the VB, to form a p-type be divided into heterovalent doping and isovalent doping.
doping. Taking Si materials as an example, P atoms will occupy 3.2.1 Heterovalent doping. In heterovalent doping, hetero-
the crystal lattice of Si atoms and generate impurity defects, the valent atoms are incorporated into host semiconductors, and
reaction involved can be represented using the Kröger–Vink excess free charge carriers are introduced according to the
notation formula:60 difference between the valence of dopants and the semi-
conductor. Heterovalent doping is the most common strategy
P + SixSi " PxSi + Si (1) for semiconductor electronic doping, which has been success-
fully applied to Si, GaAs, CdTe, CIGS and other semiconductors.
where SixSi represents one occupied Si site in the perfect crystal Crystalline Si is the earliest and the most widely used
and PxSi represents the P atoms that occupy the Si lattice site. photovoltaic material with a diamond crystal structure, which
A P atom has five valence electrons but can form only four is composed of two interpenetrating face-centered cubic primi-
covalent bonds with the surrounding Si atoms. One of the tive lattices with each atom in a tetrahedral coordination
valence electrons does not participate in the formation of a environment (Fig. 3(a)). Crystalline Si has an indirect bandgap
covalent bond, which is bounded to the P atom loosely. And it is of 1.12 eV at 300 K, leading to a high theoretical efficiency limit
easier for this electron to gain energy to become one of the free according to the SQ limit. As the p–n homojunction is the basic
electrons: structure for Si solar cells, both p-type and n-type Si semicon-
ductors are needed. The B atom, which has a high diffusion
P 
Si Ð PSi þ e
0
(2) coefficient in the Si, has been widely used in p-type wafer
production, with the advantage of obtaining a spatially homo-
where PSi represents a positively charged substitution defect geneous doping concentration.6 And for monocrystal Si solar
and e 0 is a negatively charged electron. PSi is immobilized due cells, P and Al can also be used to make n- or p-type regions,
to covalent bonds, but the electron can move freely in the Si respectively, by the solid-state diffusion method.61 After decades
crystal. As a result, the electron concentration increases. of research on Si doping techniques, the correlation between the
When B atoms are doped into the Si lattice site, B0Si and free resistivity and dopant concentration has been thoroughly investi-
holes are formed, where B0Si cannot move but holes can move gated and documented (Fig. 3(b)).62
freely, leading to the increase of hole concentration. This GaAs is another frontrunner in solar cell technologies. With
doping process is shown schematically in Fig. 2. a crystal structure similar to Si, GaAs has a direct bandgap of
1.44 eV at 300 K. With higher electron mobility and larger light
absorption coefficient, the GaAs based solar cell has obtained
the best PCE among single-junction solar cells of 29.1%. When
Si atoms replace some of the trivalent Ga atoms in the lattice,
shallow donor levels (EC  0.0058 eV) and additional valence
electrons are introduced.63,64 However, Si atoms can also
replace parts of As atoms at high Si concentration conditions
Fig. 1 Schematic diagram of foreign impurity for substitutional impurity and act as acceptors.65 Interestingly, preferred substitutions in
(a) and interstitial impurity (b). the GaAs lattice that can be altered by doping techniques have

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10019
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 3 (a) The crystal structure of Si. (b) Room-temperature resistivity in n- and p-type silicon as a function of impurity concentration. (c) Si doping levels
in different GaAs nanowires; positive value corresponds to n-type doping and negative value corresponds to p-type doping. Reproduced from ref. 66
with permission from American Chemical Society, copyright 2020. (d) Partial charge density plots for the CBM of CIGS with Li impurity. Yellow and blue
isosurfaces correspond to the density accumulation and depletion, respectively. Blue, pink, and green spheres correspond to the Cu, In, and Se atoms,
respectively. White spheres denote Cu sites left vacant. Reproduced from ref. 79 with permission from American Chemical Society, copyright 2017. (e)–
(g) Resistivity (e), efficiency (f), and minority carrier lifetime (g) of CdCl2 treated CdTe thin films on glass doped with different amounts of Cu. Reproduced
from ref. 73 with permission from Springer Nature, copyright 2013.

been demonstrated, as shown in Fig. 3(c).66 By HVPE, the Si lifetime of CdTe solar cells can be optimized. In addition, P and
doping in GaAs was close to nominal due to the near-equili- As dopants can form p-type doping when coordinated
brium character of this growth method, which could incorpo- tetrahedrally.72,74,75 However, defects of PTe and AsTe could
rate Si in the Ga sites and lead to n-type doping. However, the transform into deep donors, which limits the doping efficiency.
incorporating of Si in the Ga sites can be almost completely Chloride treatment, typically with CdCl2, has been widely
suppressed, and the As sites are predominantly replaced by Si used to form n-type CdTe.76 Cl ions may be concentrated at the
in the strongly non-equilibrium MBE technique for construc- grain boundaries and form ClTe defects, which can offer free
ting p-type conductivity. Moreover, the IVA group of elements, electrons and become positively charged.77 After Cl doping, the
Ge and Sn, can also be used as a heterovalent cation to prepare quantum efficiency of CdTe quantum dots dramatically
n-type GaAs.67 In addition, Zn was used as a p-type dopant, enhanced from 10% to 90% and the carrier lifetime increased
which can replace Ga and introduce a shallow-acceptor level from 1.2 ns to 2.3 ns. Moreover, MgCl2 could also be used
(EV + 0.031 eV).68 to replace CdCl2 for low-cost and environment friendly
CdTe has a direct bandgap of 1.48 eV at 300 K and a Zinc processing.78
blende crystal structure similar to that of GaAs. With a more CuInSe2 has a bandgap of 1.01 eV at 300 K, which can be
optimal bandgap and lower carrier effective masses, it has a tuned by the substitution of In to Ga and/or Se to S to construct
higher theoretical PCE than Si. The CdTe champion device has CIGS. Alkali metals, such as Li, Na, K, Rb, and Cs, are also
an architecture with a p-type CdTe/n-type CdS heterojunction, commonly used as the post-deposition treatment dopants after
thus its p-doping is important. CIGS growth.79 DFT calculations suggested that alkali atoms
Group IA and group VA elements have been used to sub- tend to occupy the In sites and these substitutional defects have
stitute Cd and Te, respectively, for p-type doping, such as Cu, P a lower formation energy and act as acceptors (Fig. 3(d)).
and As.69,70 The first-principles calculation proposes that Besides, alkali metals could increase the intrinsic defect
the Cu dopants in CdTe can occupy the Cd sites and act as formation energy and passivate the CIGS layer, which decrease
acceptors, or occupy the interstitial positions as donors.71,72 the deep-level defect density and increase the net carrier
Under Te-rich conditions, Cu dopants can be used to increase concentration.80,81 In brief, local p–n junctions can be formed
the hole concentration due to their larger preference of sub- by electronic doping to facilitate charge separation.
stitution over interstitial positions. The resistance of CdTe Halide perovskites have a tunable bandgap and can be
layers decreased rapidly at first and then increased slowly with adjusted through chemical substitution. The structure of halide
the increase of doped Cu atom concentration, leading to the perovskites is a framework of corner-sharing lead iodide octa-
same evolution of efficiency and minority carrier lifetime of hedra (PbI3) with a cation, such as Cs, MA and/or FA, at the
CdTe (Fig. 3(e)–(g)).73 The formation of compensating donor- center of each cuboctahedral cage (Fig. 5(a)). Theoretically,
type defects may be responsible for the increase in resistivity. heterovalent heteroatoms can be inserted into the A, B, or X
At moderate doping concentration, the efficiency and carrier site of the perovskite lattice to tailor its carrier concentration.

10020 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 4 The possible heterovalent doping elements of halide perovskites and their doping type.

However, the orbitals of A-site cations tend to fall much deeper B107 O1 cm1 (undoped MAPbI3) to B106 O1 cm1
within the bands, which has limited effect on carrier concen- (Na-doped MAPbI3).86 In addition, DFT calculations indicated
tration.82 B and X site ions majorly contribute to the CBM and that In and Tl were promising candidates to form p-type doping
VBM in perovskites, so it is easier to tune the carrier concen- in perovskites materials (Fig. 5(b)).27
tration and EF by the B and/or X site ion doping (Fig. 4). In addition to monovalent cations, trivalent cations were
Replacing Pb with monovalent or trivalent cations is the also used for the heterovalent doping of perovskites. The first-
commonly used doping method to adjust carrier concentration principles calculations implied that the substituting of Pb with
in halide perovskite semiconductors. Generally, substituting Pb Sb and Bi could form an n-type doping (Fig. 5(c)).27 Due to a
with monovalent cations could increase hole concentration and similar ionic radius to that of Pb (1.18 Å), Bi (1.02 Å) was the
achieve p-type doping. The effect of monovalent cation halide mostly-used heterovalent doping ion, which can be easily
additives such as CuBr, CuI, NaI, and AgI was explored in doped into lead halide perovskites. As shown in Fig. 5(d), the
MAPbI3 perovskite films.83 These results showed that a better free carrier concentration measured by the Hall effect increased
PCE was achieved in NaI- and CuBr-based films, but the JSC of from B109 cm3 (undoped MAPbBr3 single crystals) to
the AgI-based films decreased. KPFM measurement identified B1011–1012 cm3 after Bi doping. And the conductivity mea-
that the work function of additive-based perovskites signifi- sured by the four point probe method also increased by 4 orders
cantly shifted (about 0.1 V) toward the work function of Au, of magnitude, from B108 O1 cm1 for the undoped crystal to
which was responsible for efficiency improvement. In contrast, B104 O1 cm1 for the Bi-doped crystal.26 The conductivity of
it is also reported that Ag incorporation resulted in an enhance- Bi-doped perovskite polycrystalline film increased from 3.02 
ment of the PCE from 16.0% to 18.4% in MAPbI3 based devices, 106 to 9.76  106 O1 cm1.87 Based on Sb doping, the
and from 11.2% to 15.4% in MAPbI3xClx based devices.28 The electron density in the CB of MAPbI3 was increased and the
EF of the Ag-doped perovskite shifted to the VBM, indicative quasi-Fermi energy level was elevated.88 Furthermore, Al has a
of increased hole concentration. Interestingly, Ag doping can higher valence and smaller ionic radius than Pb, which can
realize not only p-type doping, but also n-type doping. Zhang substantially reduce the microstrain of the MAPbI3 film and
et al. found that an n-type perovskite film was obtained by enhance the device performance.37
adding AgI to the perovskite precursor solution, which was Other methods like X-site doping were also applied to
attributed to the crystallization control with aligned orientation manipulate the carrier concentration. Oxygen can act as an
of MA cations.84 Therefore, electron mobility was improved acceptor dopant, in which the O atoms replaced the I atoms or
from 7.9  103 cm2 V1 s1 (pristine MAPbI3) to 1.5  accommodated interstitially.89,90 Experimental results showed
101 cm2 V1 s1 after AgI doping, which enhanced the PCE that the concentration of holes and iodine vacancies increased
from 17.59% to 20.02%. with oxygen pressure under illumination. Surprisingly, the
The copper(I) complexes can be incorporated into the per- concentration of majority carriers increased at low oxygen pressure,
ovskite film to modulate carrier concentration. Incorporating but decreased when the pressure is high, which may be related
Cu(thiourea)I into perovskites was found to form a p–n bulk to the increase of the defect state concentration.88
heterojunction, which increased the depletion width from Besides the heterovalent ion substitution, the electronic
126 to 265 nm, accelerating hole transportation.85 Besides, it doping can be also modulated by interstitial impurities. In Si
was proved that Na could form substitutional defects in the solar cells, the monocrystalline silicon surfaces have abundant
Pb-site of the perovskite lattice. A 0.5% Na doping significantly Si dangling bonds, which introduce a half-occupied deep
improved the electronic conductivity of perovskite, from energy level state in the bandgap.91,92 Hydrogen atoms can

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10021
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 5 (a) Crystal structure of a cubic ABX3 halide perovskite. (b) and (c) Partial charge density of In (b) and Sb (c) doped CsPbI3. Reproduced from ref. 27
with permission from Elsevier, copyright 2019. (d) Bismuth doping modulates the conductivity (blue line) and majority carrier concentration (red line) of
MAPbBr3. Reproduced from ref. 26 with permission from American Chemical Society, copyright 2016.

chemically passivate the Si dangling bonds, and act as accep- impurity level into the forbidden band realizing isoelectronical
tors in n-type Si or as donors in p-type Si.6,93 Surface H could doping.98,99 The N atom replaces P atom, which could capture
further migrate into the bulk region as interstitial H, which can an electron and become a negative charge center with an
neutralize shallow dopants, so it can act as a compensating ionization energy of 0.008 eV. Conversely, Bi atom that replaces
donor or acceptor in p-type Si or n-type Si, respectively.94 P atom can capture a hole and become a positive center, and
With a small ionic radius, Li+ cations may locate on inter- the excess hole has an ionization energy of 0.038 eV. Therefore,
stitial lattice sites and act as n-dopants in halide perovskite in addition to the valence number of dopants, the electro-
materials. DFT calculations indicate that interstitial occupancy negativity and atomic radius are also crucial for the doping
is energetically favorable for Li+ compared to A-site substitu- selection.
tions. After Li-doping, the measured EF in MAPbI3 shifted up by In addition, isovalent ions can influence carrier concen-
440 meV, reducing the non-radiative process by trap filling.95 tration through regulating and passivating defects. This pheno-
Besides, Li+ can do n-doping for CsPbBr3 electrochemically, menon is not significant in conventional PVs due to their high
which improves the electron injection efficiency.96 defect formation energy and low defect density. However,
3.2.2 Isovalent doping. Isovalent doping means the doped it cannot be neglected in halide perovskite materials that have
ions have the same valence as the lattice ions they replace. Take a high defect density. As an isovalent dopant, CaCl2 realized a
Sn-doped MAPbI3 as an example, the defect reaction equation controllable n-type doping of perovskite material. The EF of
can be described as: the Ca-doped CsPbI2Br perovskite film was closer to the CB
compared with that in the control film, achieving an enhanced
MAPbI3
SnI2 ! Sn 
Pb þ 2II
(3) built-in electric field and higher VOC.42 Other isovalent cations,
such as Mg2+, Ni2+, Mn2+ and Sr2+ have also been reported to
where Sn Pb represents the Sn atom that occupies the Pb lattice passivate the surface/interface defects and/or increase the grain
site and II represents the one occupying the I site in the perfect size, which are capable for electronic doping.100–105
crystal. It can be seen from eqn (3) that the incorporation of the Furthermore, doping unstable isovalent ions can modulate
isovalent ions will not generate extra electrons and holes, so it carrier density through redox reactions. Compared with Pb2+,
will not affect the carrier concentration. However, the carrier Sn2+ is much easier to be oxidized to Sn4+.106,107 The E0 of
concentration of a semiconductor can be significantly changed Sn2+/Sn4+ is +0.15 V vs. +1.67 V for the Pb2+/Pb4+, indicating that
if there are considerable discrepancies of electronegativity Sn2+ is thermodynamically unstable.108 In addition, Sn2+ has
and/or atomic radius between dopants and semiconductors.97 the same valence as and similar ionic radius to Pb2+, suggesting
For example, doping N or Bi atoms in GaP will introduce that Sn2+ is a suitable electronic dopant with a redox reaction.

10022 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Due to the lack of detailed and clear explanation of the Taking these three main factors into account, the effective
mechanism of Sn2+ doping, we thus consider various free carrier concentration n generated by doping can be expressed
carrier formation processes for the electronic doping of as follows:
Sn-based perovskites. We assume that Sn2+ occupies the lattice
n = ndopant  ZIP  (1  ZSelf-compensation), (7)
site of Pb2+, and the Sn-related defects are activated by lattice
instability and defect-mediated processes as endogenous where ndopant is the dopant concentration, which is limited by
defects. For p-doped halide perovskites, the possible intrinsic solubility, ZIP is the donor or acceptor ionization probabilities,
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

defect electronic disorder reactions read as follows: and ZSelf-compensation is the self-compensation probability.
3.3.1 Solubility. The factors that affect solid solubility are
Sn2+ " Sn4+ + 2e 0 (4)
quite complicated and have not been completely unveiled by
Null Ð V00Sn þ 2h ; (5) experiments. Hume–Rothery and his collaborators proposed a
semi-empirical law that illustrates the factors and rules govern-
where V00Sn represents a Sn vacancy with double negative ing the solid solubility in solid solutions. For substitutional
charges, and h is a positively charged hole. Consider the tin solid solutions, the Hume–Rothery rules are as follows:
oxidation in halide perovskites as exogenous, i.e., Sn2+ is (1) The atomic radius of the solute (rsolute) and solvent
oxidized to Sn4+ via the chemical reaction with O2, which may (rsolvent) must not differ by more than 15%:
 
mostly occur at/near grain boundaries and surfaces.109,110 The rsolute  rsolvent
d¼  100%  15% : (8)
overall reaction equations can be written as follows: rsolvent

Sn2+ + O2 " SnO2 + 2h (6) (2) The solute and solvent should have similar electronega-
tivity. If the electronegativity difference is too great, the metals
As the Sn-doped perovskites were found to be p-type doped tend to form intermetallic compounds instead of solid solutions.
(Table 2) and Sn2+ is easily oxidized, reaction (5) and (6) may It is possible to obtain large solid solubility, when the electro-
be the dominant doping mechanisms for Sn-doped halide negativity difference is less than 0.4–0.5.
perovskites.111 However, the SnO2 phase has not been detected (3) The crystal structures of solute and solvent must be
by XRD in an oxidized perovskite. It may exist as an amorphous similar.
material or in small ‘‘crystalline’’ aggregates, which was sepa- (4) Complete solubility occurs when the solvent and solute
rated and confirmed by high-resolution XPS.112–114 Note that have the same valency.
tin oxidation should be further distinguished between the Moreover, for interstitial solid solutions, solute atoms
preliminary oxidation of tin in the solution precursors, i.e., should have a radius smaller than 59% of the radius of solvent
before the perovskite nucleation process (discussed in Section atoms to achieve a considerable solid solubility.
4.2), and after the perovskite crystal formation (Sn2+ occupied Fundamentally, Hume–Rothery rules are based on the geo-
in the lattice sites of Pb2+). And it implies that not only the Sn metrical restraints, which is suitable for systems that form
oxygenation/reduction of O2, but also V00Sn (eqn (5), at least in either substitutional or interstitial solid solutions. In addition,
part) can contribute to p-type doping. The doping mechanism the Darken–Gurry diagrams are applied to qualitatively assess
of Sn is complicated, which requires further investigation. the solid solubility of dopants, which comprehensively consi-
der the influence of geometry factors and electronegativity
3.3 Doping efficiency (0.4) on solubility.116
The effective carrier concentration produced by impurities is As shown in Fig. 6(a) and Table 3, the analysis of binary
affected by many factors. Among them, three main factors may silica-based systems is presented by using the Hume–Rothery
further determine/limit the doping efficiency:115 (1) the desired rules, indicating that P, As, Al, Ga, Ni, Co, Ge, Cu, Fe and Cr
dopant may have a low solubility. (2) The desired dopant has have larger solid solubility as substitutional impurities and H
good solubility, but the defect transition energy level may be can enter interstitial sites of the Si lattice, ignoring the effect of
too deep, thus, the defect cannot be ionized at normal operat- valence electron concentration and the relative price.
ing temperatures. (3) The desired dopant has good solubility Similarly, we assess solubility of dopants in perovskite
and proper ionizability, but as the EF shifts due to the increase materials by using the Hume–Rothery rules. However, perovs-
of carrier density, which may also lead to the formation of kites are ionic crystals, and the model needs to be modified to
oppositely charged native defects or defect complexes of the accommodate the perovskites. We reasonably adopt the ionic
dopant, thus limiting the change of the EF. radius to replace the atomic radius and critical atomic ratio (d)

Table 2 Carrier concentration and mobility achieved by Sn(II) defect doping

Material Bandgap (eV) Type p (cm3) n (cm3) Mobility Technique Ref.

MAPbI3 1.54 N — 1.5  109 66 Resistivity, Hall 111
MASn0.5Pb0.5I3 — N — 4.98  1013 270 Resistivity, Hall 111
MASnI3 1.21 P 7.94  1014 — 2320 Resistivity, Hall 111

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10023
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 6 The Darken–Gurry diagram for dopants doped at the Si (a) and B (b) sites of the perovskite.

Table 3 The AR and EN of silicon and its dopants Table 4 The IR and EN of a Pb-based perovskite and its B-site dopants

Atom AR (Å) ENa Atom AR (Å) ENa Atom AR (Å) ENa Cation IR (Å) EN Cation IR (Å) EN Cation IR (Å) EN
Si 1.11 1.98 Te 1.36 2.12 Zn 1.25 1.65 Ba2+ 1.35 0.89 Lu3+ 0.86 1.27 Ag2+ 0.94 1.93
Li 1.23 0.98 Ta 1.34 1.51 Ga 1.26 1.81 Sr2+ 1.18 0.95 Pu3+ 1 1.28 Ag+ 1.15 1.93
Be 0.89 1.57 W 1.3 2.36 Ge 1.22 2.01 Ca2+ 1 1 Mg2+ 0.72 1.31 Sn2+ 1.15 1.96
B 0.82 2.04 Pt 1.3 2.28 As 1.2 2.18 Yb2+ 1.02 1.26 Np2+ 1.1 1.36 Hg2+ 1.02 2
C 0.77 2.55 Au 1.34 2.54 Se 1.17 2.55 La3+ 1.03 1.11 Ti2+ 0.86 1.54 Ge2+ 0.73 2.01
N 0.74 3.04 Hg 1.49 2 Mo 1.3 2.16 Ce3+ 1.01 1.12 Mn2+ 0.83 1.55 Bi3+ 1.03 2.02
Na 1.54 0.93 Tl 1.48 1.62 Pd 1.28 2.2 Pr3+ 0.99 1.13 Be2+ 0.45 1.57 Sb3+ 0.76 2.05
Mg 1.36 1.31 Pb 1.47 1.87 Ag 1.34 1.93 Nd3+ 0.98 1.14 Tl+ 1.5 1.62 Pd2+ 0.86 2.2
Al 1.18 1.61 Bi 1.46 2.02 Cd 1.48 1.69 Sm3+ 0.96 1.17 V2+ 0.79 1.63 Pt2+ 0.8 2.28
P 1.06 2.19 Fe 1.17 1.83 In 1.44 1.78 Sm2+ 1.22 1.17 Zn2+ 0.74 1.65 Pb2+ 1.19 1.87
S 1.02 2.58 Co 1.16 1.88 Sn 1.4 1.96 Gd3+ 0.94 1.21 Cr2+ 0.8 1.66 Au3+ 0.85 2.54
Ti 1.32 1.54 Ni 1.15 1.92 Sb 1.4 2.05 Eu3+ 0.95 1.19 Cd2+ 0.95 1.69 Au+ 1.37 2.54
V 1.22 1.63 Cu 1.17 1.9 Mn 1.17 1.55 Eu2+ 1.17 1.19 Fe2+ 0.78 1.83 K+ 1.38 0.82
Cr 1.18 1.66 Dy3+ 0.91 1.22 Co2+ 0.75 1.88 Na+ 1.02 0.93
a
Dy2+ 1.07 1.22 Cu2+ 0.73 1.9 Tm2+ 1.03 1.25
The electronegativity was calculated using Pauling scale. Er3+ 0.89 1.24 Cu+ 0.77 1.9 Ni2+ 0.69 1.92
Tm3+ 0.88 1.25

of interstitial solid solution modifies to 0.434, the radius ratio

of tetrahedral and octahedral gaps. ellipse with the center of I, which is rarely studied in PSCs.
As shown in Fig. 6(b) and Table 4, Sn2+, Ag+, Bi3+ and Hg2+ Theoretically, I can be replaced by S2 and/or Se2, which acts
fall into the ellipse with the center of the Pb cation, indicating as acceptors and realizes p-type doping. In this respect, sulfides
that those cations have unlimited solubility in perovskites as and selenides have the potential to be used as p-type dopants in
substitutional impurities. Taking the relative valence effect into halide perovskite materials.
account, Ag+ and Bi3+ may have limited solubility in perovskite. Darken–Gurry diagram considers the main parameters of
Be2+ has a smaller ionic radius than 62% of the radius of Pb2+ to metal doping, which can be used for rapid screening of possible
form the interstitial impurities. electronic dopants. However, it may fail in complex systems.
The same analysis is applied to X-site doping in perovskites, DFT calculation was developed to evaluate the doping effects of
as shown in Fig. 7(a) and Table 5. Br can be infinitely semiconductors quantitatively.115 And the influences of sec-
substituted for I in perovskites, which has been reported in ondary phases must be considered, if impurities could form
the literature.117–119 Surprisingly, S2 and Se2 also fall into the strong bonds with the host elements. To avoid the formation of

10024 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 7 (a) The Darken–Gurry diagram for dopants doped at the X-site of the perovskite. (b) The calculated available equilibrium chemical potential region
for CdTe:Na in the two dimensional (mCd, mNa) plane. Under equilibrium growth conditions, there are some thermodynamic limits for Na doping CdTe on
the achievable values of the chemical potentials, as shown in the equation in the blank area. The yellow area is thermodynamically forbidden under
equilibrium growth conditions. Reproduced from ref. 120 with permission from American Physical Society, copyright 2002.

Table 5 The IR and EN of an I-based perovskite and its X-site dopants 3.3.2 Ionization probability. For the impurity level, each
energy level can contain two electrons with opposite spin
Anion IR (Å) EN Anion IR (Å) EN Anion IR (Å) EN
directions, while the donor and acceptor impurity levels can
  2
F 1.33 3.98 I 2.2 2.66 Se 1.98 2.55 only be: (1) occupied by one electron with either spin direction,
Cl 1.81 3.16 O2 1.4 3.44 Te2 2.21 2.12
Br 1.96 2.96 S2 1.84 2.58 or (2) be empty. According to Fermi–Dirac probability function,
the ionization probability of the donors and acceptors without
considering self-compensating effects can be obtained:
secondary phases between the dopants and the host elements,
the chemical potentials m should be limited,115 1
 
ZD ¼ 1  fD ðE Þ ¼ 1  (11)
1 ED  EF
nmA + mmX r DHf (AnXm), (9) 1þ exp
gD kT
where A and X are the host and doping elements, respectively.
DHf (AnXm) is the formation energy of AnXm. If the host (AB) is
1
composed of elements A and B, an additional limited formula ZA ¼ 1  fA ðE Þ ¼ 1    (12)
1 EF  EA
needs to be added,115 1 þ exp
gA kT
nmB + mmX r DHf (BnXm). (10)
where ZD and ZA are the ionization probability of the donors
Take the Na-doped CdTe as an example, Fig. 7(b) plots the and acceptors, fD(E) and fA(E) are the density of electrons and
calculated chemical potential region accessible under equili- holes occupying the impurity level, respectively, ED and EA are
brium growth condition, as defined by eqn (9) and (10).120 Na the energy of the donor and acceptor levels, respectively, and gD
can form a very stable compound Na2Te with Te, with a and gA are called degeneracy factors, which are a direct result of
calculated formation energy of 2.84 eV. To avoid the for- the spin factor. For Ge, Si and GaAs semiconductor materials,
mation of Na2Te, mNa should be limited in a certain range. gD = 2 and gA = 4.
Under Te-rich conditions (mTe = 0, mCd = DHf (CdTe) = 0.79 eV), 3.3.3 Self-compensation. Ideally, charged defects need to
Na2Te is easy to form, which has a lowest chemical potential be compensated by the opposite charge carriers (electrons
of 1.42 eV. Under the Cd-rich conditions (mCd = 0, mTe = DHf or holes) to keep a strict stoichiometric ratio. However, the
(CdTe) = 0.79 eV), Na2Te is difficult to form with a high stoichiometric ratio of some compounds varies with the change
chemical potential of 1.02 eV. When mTe varies between of the surrounding atmosphere and the partial pressure.90,121
0.79 eV and 0, mNa changes between 1.42 eV and 1.02 eV. It is possible to compensate the ionized impurity charges by
Beyond these chemical potential limits, Na2Te will be formed, intrinsic defects instead of the opposite charge carriers.
and the doping process will stop, as shown in the yellow area Take the Bi3+ doping in perovskite as an example. The ideal
of Fig. 7(b). defect reaction equation (based on positive ions) can be

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10025
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 8 Formation energies of AsTe, AX-center, VCdAsCd and VCdAsTe under Cd-rich (a) and Te-rich (b) conditions. Reproduced from ref. 125 with
permission from Springer Nature, copyright 2017.

described as follows: donor, can also compensate the doping effect (for example,
MAPbI3 As-doped CdTe, Fig. 8).125
! BiPb þ 3I
BiI3  I þe
0; (13)
The self-compensation remains a challenge for efficient
where BiPb represents the Bi atom that occupies the Pb electronic doping in certain materials. Unfortunately, the halide
lattice site. perovskite materials with a ‘‘soft lattice’’ exhibit low defect
However, negatively charged defects may be compensated by formation energy (Section 4), which probably leads to a
localized positive charges, strong self-compensation effect. The investigation of the self-
compensating effect and the modulating strategies in halide
MAPbI3
! BiPb þ 2I
BiI3  0
I þ Ii
(14) perovskite materials are still lack. The control of self-
compensating effect is the key to applying atomic doping on
MAPbI3 semiconductor photovoltaic materials featuring weak chemical
! BiPb þ 3I
BiI3  0
I þ VMA
(15)
bonds and soft lattices.
MAPbI3
! 2BiPb þ 6I
2BiI3  00 ;
I þ VPb
(16)

where I0i , V0MA and V00Pb represent I interstitial, MA vacancy and 4. Self-doping
Pb vacancy, respectively. Side reactions (reaction (14)–(16))
compete with the desired reaction (reaction (13)), which Different from the conventional PVs materials with strong
reduces majority carrier concentration. The reaction coefficient chemical bonding, halide perovskite materials have weak ionic
depends on the defect formation energy (Section 4), that is, the and secondary bonding, which makes it easy to modulate the
defect’s intrinsic characteristics and the position of EF. electronic doping.40,126,127 However, the ‘‘soft’’ nature and low
Si and GaAs have large formation energies of vacancy and defect formation energy of this new-generation PV materials
interstitial defects, and their side reactions can hardly may lead to the severe compensation effects from the intrinsic
proceed.122,123 The doping concentration can be presumed to defects, leading to the challenge of efficiently and reliably
be equal to the majority carrier concentration. However, the tuning the carrier density.51,128 In addition, typical physical
side reactions are significant for CdTe due to the low formation doping approaches, which have been widely adopted in con-
energy of Te vacancy.12,124 The increase in the hole concen- ventional PVs, like ion implantation, may cause severe crystal
tration from doping can be neutralized by self-doping damage in halide perovskites.129 Therefore, chemical doping
(Te vacancy), indicating a serious self-compensation effect. strategies should be developed according to the material
Also, the incorporation of the donor (acceptor) will cause the features. New deposition approaches, which can tune the
upshift (downshift) of EF, leading to the formation energy materials properties by the easy chemical methods, are
decrease of the negatively (positively) charged defect, which required, such as controlling the stoichiometric ratio in pre-
further facilitates self-compensating. In addition, the formation of cursor solution for manipulating the intrinsic defect properties
the AX center, which is an acceptor-induced defect that acts as a (self-doping, Section 4) and introducing dopants (in precursor

10026 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

solution and/or surface modification) to collect/donate electron 4.2 Doping strategy

from/to the semiconductor. (molecular doping, Section 5). 4.2.1 Conventional solar cell. CdTe solar cells have lower
doping efficiency and higher non-radiative electron–hole recom-
4.1 Doping mechanism
bination rates than Si owing to their defect physics.72,130–132
In an ideal crystal, the ordered arrangement of atoms in a By controlling the growth conditions, the composition of CdTe
crystal extends indefinitely in three dimensions strictly and could be off-stoichiometric and EF could be modulated (Fig. 10).12
periodically. However, the crystal growth is always flawed. VTe has a low formation energy, which is the dominating defect
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Frenkel and Schottky defects are the two types of intrinsic type in CdTe films. Ionized VTe can offer unbounded electrons,
point defects in ionic crystals. Take NaCl as an example, ions which can induce heavy n-type doping. VCd is an intrinsic acceptor
might be transferred from the normal position to the inter- defect, whose level is deep and located at the level of B0.4 eV
stitial site nearby or the crystal surface, as shown in Fig. 9(b) higher than the VB according to DFT calculations.72,133,134
and (c). The reaction in NaCl crystal can be written as follows: In principle, higher hole concentrations can be achieved by
Na  0 reducing the Cd/Te ratio under CdTe growth conditions. However,
Na Ð Nai þ VNa (17)
DFT calculations show that the hole density is always limited
below B1015 cm3, regardless of the growth temperature or
Null Ð VCl þ V0Na ; (18)
conditions, owing to the relatively deep VCd acceptor level.72
where Nai , V0Na and VCl represent Na interstitial, Na vacancy Non-equilibrium growth strategies, such as quenching from high
and Cl vacancy, respectively. In reality, some non-stoichio- temperature to room temperature, have been used to further
metric compounds, whose composition deviates obviously from enhance the doping efficiency.12 In addition, extrinsic acceptor
stoichiometric composition depending on the atmosphere and dopants need to be used when a heavy p-type doping for CdTe is
pressure, will become n-type and p-type semiconductors. Still, required.
take NaCl as example (Fig. 9(d)–(f)), the doping reactions are as CIGS solar cells can be self-doped for p-type by intrinsic
follows: defects. The VCu has a low formation energy associated with a
shallow acceptor level for p-type doping.135–137 The relevant
Null Ð V0Na þ h (19) Kröger–Vink notation formula is shown as follows:

Null Ð VCl þ e0 (20) Cu 0 

Cu Ð VCu þ h þ CuðsÞ ; (22)

Null Ð Nai þ e0 (21) where Cu 0

Cu and VCu represent Cu perfect crystal and Cu
vacancy, respectively. DFT calculations show that Cu antisites
(CuIn and CuGa) substantially contribute to p-type doping.17,138
In short, if the defect is negatively charged (cation vacancy or Although CuIn and CuGa may have a deeper acceptor level than
anion interstitial), the concentrations of electron and negatively VCu, they were predicted to be more abundant in CIGS thin
charged defects are suppressed and the concentrations of hole films only under Cu-rich conditions. Other antisite defects,
and positively charged defects are augmented, which leads to such as InCu and GaCu, have deep donor levels that would trap
p-type doping. The opposite is true for positively charged electrons.135,136 Moreover, defect–defect interactions occur and
dopants (anion vacancy or cation interstitial). form defect complexes in CIGS. VCu tends to interact with

Fig. 9 Schematic diagram of perfect and defective lattices of an ionic crystal. (a) Perfect lattice. (b)–(f) defective lattices. Frenkel (b) and Schottky (c)
defects, cation vacancy (d), anion vacancy (e), and cation-interstitial (f).

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10027
 View Article Online

Chem Soc Rev Review Article

observed in this material. Compared with CIGS, CZTS has a

larger number of antisite defects as it is easier for Cu to replace
the group IIA (Zn) elements than the group IIIA elements
(In and Ga).19,140,143 Similarly, CZTS has a mass of charge-
compensated defect complexes, such as VCu–ZnCu, VZn–SnZn
and CuSn–SnCu. which may passivate deep donor levels and
decrease the hole concentration.19,140,143–145 Owing to the
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

‘‘soft’’ lattice and low defect formation energy, the formation

of different types of defects in CTZS materials is highly depen-
dent on their atomic chemical potentials, thus the carrier
concentration of CTZS can be controlled by modulating the
growth conditions (Fig. 11).19 Experimental investigation also
suggests that CZTS is an intrinsic p-type semiconductor. The
competitive acceptor defects, CuZn and VCu, may be responsible
for the self-doping of CZTS. CuZn has a low formation energy
under Cu-rich conditions, which contributes free holes to the
Fig. 10 Formation energy of intrinsic point defects in CdTe under differ- VB. While under Cu-poor conditions, the VCu contributes more
ent growth conditions. The formation energy of intrinsic point defects is to hole concentrations.19,140,146
dependent on whether CdTe is grown under Te-rich (a) or Cd-rich (b) 4.2.2 Perovskite solar cells. Different from the above thin
conditions. Reproduced from ref. 12 with permission from American film solar cells, the synthesis/processing of halide perovskite
Physical Society, copyright 2014.
films is mostly based on the solution methods.126 Various
interactions of multiple perovskite ions and the polar aprotic
solvents in solution environment offer more abundant chemical
antisite defects, turning the deep-level defects into electrically pathways to nucleation and crystallization, such as clusters of
benign defect complexes with shallow energy levels. The VSe–VCu intermediate (adducts and complexes).147 Compared with the
complex might act as either an acceptor or donor depending on classic nucleation process based on the physical approach, it
the position of the EF in the bandgap.139 provides an almost endless chemical space to tailor material
CZTS, CZTSe, and CZTSSe are the emerging PV technologies properties (Fig. 12).148 However, it also brings challenges for
due to their environmental friendly nature, abundance, reliable and controllable doping. Composition, atmosphere, sol-
and promising PCEs.140–142 Similar to the above thin-film PV vent, and additives must be carefully controlled for reproducibly
materials, they have high defect density due to their low defect modulating the carrier concentration of halide perovskites.51,128
formation energy and complex component. Take CZTS as an
example, the Cu, Zn and Sn cations have similar coordination 4.2.2.1 Lead-base halide perovskites. Due to the ‘‘soft’’ lattice
environments, so sublattice cation disorders are commonly and weak chemical bonds, halide perovskite materials have a

Fig. 11 The formation energy (DH) of low energy point defects and defect complexes in CZTS as a function of the chemical potential: (a) mCu from 2.10
to 0 eV (Cu poor to rich), mZn = 0, and mSn = 0, (b) mCu = 0 and mSn = 0, mZn from 4.21 to 0 eV (Zn poor to rich), (c) mCu = 0, and mZn= 0, mSn from 4.21 to
0 eV (Sn poor to rich). Reproduced from ref. 19 with permission from American Physical Society, copyright 2010.

10028 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

deficiency of PbI2, which increases the hole concentration for

p-type doping (Fig. 14(c)).150 A similar analysis could also be
applied to the PbI2-rich film. In addition, the composition
engineering can be also applied to other perovskite materials,
such as FAPbI3 and MAPbBr3, to modulate self-doping.153–155
The atmosphere also determines the atomic chemical potential,
which plays an important role in the self-doping of perovskites
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

(Table 6). The evolution of EF of perovskite on TiO2 and PEDOT:PSS

in different atmospheres is probed by SPV measurements,
indicating that the EF varies with the atmosphere and the
deposited substrate.156 Oxygen and water molecules in air
could form superoxide or a hydrate phase, which may be the
origin of the EF shift.157–159 The ‘‘self-healing’’ effect, moving EF
to the mid bandgap, is also reported in the perovskite crystal
structure restoration.157 Oxygen and water initially adsorbed in
Fig. 12 Pathways to crystallization by classical ‘‘monomer’’ (gray curve) the film may desorb in high vacuum, which restores the initial
and nonclassical particle attachment. Reproduced from ref. 148 with defective crystal structure and makes EF drastically upshift.160
permission from Science Publishing Group, copyright 2015. The sample will reabsorb water and oxygen, and EF will be
restored gradually when stored in the air. However, the initial
states cannot be fully recovered due to the formation of the
high defect density (41016 cm3), and their carrier concen- hydrate phase on the perovskite surface, and prevent the
tration can be regulated by defect concentration. There are oxygen diffusion into the film. Heating the sample, instead,
many types of defects in perovskite materials (Fig. 13), and we could restore the EF completely, in which the water–film barrier
take the point defects of MAPbI3 as a model. All the possible on the surface was fully evaporated, resulting in the intercala-
point defects are considered in MAPbI3 including three vacan- tion of oxygen into the bulk.156
cies (VMA, VPb, and VI), three interstitials (MAi, Pbi, and Ii), The iodine pressure also influences the electrical properties
two cation substitutions (MAPb and PbMA), and four anti-site of perovskites (Table 6). It is proposed that, with the increase of
substitutions (MAi, Pbi, IMA, and IPb), whose energy levels and the partial pressure of I2, the concentration of VI decreases and
formation energy are shown in Fig. 14(a).25,141,149–151 The total the concentration of VMA increases. This results in an increase
energy change (DE) between a pristine crystal and defective in the concentration of holes and a decrease in the concen-
crystal, the atomic chemical potentials (mi), and the EF together tration of electrons.86,161 This model needs to be verified by
determine the defect formation energy as well as the type of experiment results, especially the transition from p- to n-type
doping.115 electronic conductivity at a certain iodine pressure. And indeed,
Composition engineering, such as changing the PbI2/MAI this occurrence was experimentally observed when exposing the
ratio in the precursor solution, is one proper way to change the samples to an I2 sink.162
atomic chemical potentials and realizes self-doping in perovs- The different decomposition rates between PbI2 and organic
kite preparation (Table 6). It shows that PbI2-rich films are n- cations also lead to the self-doping of perovskite. The formation
type and PbI2-deficient films are p-type doping (Fig. 14(b)).152 energy of Schottky defects can be calculated by applying the law
When the MAI/PbI2 ratio is much larger than 1, the formed of mass action to the decomposition reaction. The calculated
perovskite may possess a lot of Pb vacancies due to the results showed that MAI had a remarkably low Schottky defect

Fig. 13 Defect illustration in halide perovskite materials.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10029
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 14 (a) The transition energy levels of intrinsic acceptors and intrinsic donors in MAPbI3. Reproduced from ref. 141 with permission from John Wiley
& Sons, copyright 2014. (b) The carrier concentration (red symbols) and mobility (black symbols) for electrons (circle) and holes (square) in the perovskite
films extracted by Hall measurements. The area colours of orange and blue represent the p-type and n-type doping, respectively, for different PbI2/MAI
ratios. Reproduced from ref. 152 with permission from Springer Nature, copyright 2019. (c) Schematic conductivity type conversions in perovskite films
by excess MAI (left side) or excess PbI2 (right side). Reproduced from ref. 150 with permission from AIP Publishing, copyright 2014. (d) The energetic levels
of the corresponding perovskite films. The column colors of orange and blue represent the p-type and n-type, respectively. Reproduced from ref. 164
with permission from John Wiley & Sons, copyright 2017.

formation energy (0.08 eV) compared with that of PbI2 (0.22 eV) Mayer bond order) of typically used solvents for perovskite
and MAPbI3 (0.14 eV), thereby making the MAI and PbI2 ratio fabrication in solution processing are listed in Table 7.165 The
deviate from the stoichiometric ratio during decomposition.163 dissolution of PbI2 in solvents involves the interaction between
The annealing process can exacerbate non-stoichiometry and the lone pair electrons of the solvent/additive and the Pb atoms,
thus the self-doping process (Table 6). Perovskite films with a whose strength affects the path of the chemical reactions,
precursor ratio (PbI2 : MAI) of 0.3 had an EF of 0.08 eV below the mainly the intermediate phases (the adducts, complexes,
middle bandgap, indicating weak p-doping properties. After clusters or colloidal particles).166,167 Many intermediate phases,
annealing at 150 1C for 45 min, the perovskite films were i.e., PbI2DMF, PbI22DMSO, MAIPbI2DMF, and (MA)2Pb3I8
converted from weak p-type to heavily n-type doping, whose 2DMSO, have been investigated, which are crucial for modulat-
EF moved up to 0.25 eV below the CBM.150 Furthermore, ing the nucleation and crystallization processes.168–172 In the
annealing temperature was a key factor in regulating the initial future, electrical property modulation of perovskite materials
defect state of the MAPbI3 thin film (Fig. 14(d)).164 The EF of the needs to be intensively investigated for efficient and stable
perovskite film was 0.35 eV above the VBM without annealing, optoelectronic device design with the focus on precursor
indicating a p-type doping film. By increasing the annealing chemistry.
temperature from 60 1C to 100 1C, the EF shifted upwards, and
the doping type changed from the p-type to the intrinsic type. 4.2.2.2 Tin-based halide perovskite. Although lead-based per-
After further increasing the annealing temperature to 150 1C, ovskites have achieved impressive development, the toxicity of
the EF located at 0.53 eV below the CBM, showing obvious Pb limits their commercialized applications. Therefore, lead-
n-type doping properties. free perovskite materials are proposed and developed.109,113
Solvent and adduct intermediated processes are vital to So far, various perovskite materials based on halide single and
solution processing, which determinate the chemical reaction double perovskites have been studied, such as the substitution
path and film quality, such as perovskite composition, grain of Pb by Sn, Bi, Sb, Ge, and Cu.176 Theories and experiments
boundaries and defect densities, and thus the carrier concen- have shown that the tin-based perovskites are the most promis-
tration. The chemical properties (the enthalpy of solvation and ing candidates as alternatives due to their similar outer

10030 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Table 6 Doping densities and Fermi level position achieved by self-doping

Materiala Method Typeb n (cm3) p (cm3) Ec  Ef (eV) Ef  Ev (eV) Processing Technique Ref.
MAPbI3b PbI2/MAI = 0.3 p — 4.0  1016
0.85 0.7 One-step UPS; Hall 150
PbI2/MAI = 0.65 n 8.1  1016 — 0.75 0.8 Two-step UPS; Hall
PbI2/MAI = 1.0 n 2.8  1017 — 0.65 0.9 Two-step UPS; Hall
PbI2/MAI = 1.7 n 3.5  1018 — 0.35 1.2 Two-step UPS; Hall
PbI2/MAI = 0.3 p — — 0.87 0.68 One-step STS 153
PbI2/MAI = 0.65 p — — 0.81 0.75 One-step STS
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

PbI2/MAI = 1.0 n — — 0.63 0.93 One-step STS

PbI2/MAI = 1.2 n — — 0.57 0.99 One-step STS
PbI2/MAI = 1.7 n — — 0.39 1.18 One-step STS
PbI2/MAI = 0.91 p — 8.0  109 1.05 0.5 Two-step UPS; Hall 152
PbI2/MAI = 1.0 n 1.2  1012 — — — One-step UPS; Hall
PbI2/MAI = 1.05 n 3.1  1012 — 0.48 1.07 One-step UPS; Hall
PbI2/MAI = 1.10 n 3.64  1012 — 0.32 1.23 One-step UPS; Hall
PbI2/MAI = 1.15 n 8.31  1012 — — — One-step UPS;Hall
PbI2/MAI = 1.0 n B1.3  1011 — — — One-step CRPH 173
PbI2/MAI = 1.05 n B2.7  1012 — — — One-step CRPH
I-rich/Pb-poor p — — 1.4 0.1 — DFT 25
Moderate i — — 0.72 0.78 — DFT
I-poor/Pb-rich n — — 0.2 1.3 — DFT
No annealing p — 2.3  1011 1.0 0.55 One-step UPS; Hall 164
60 1C annealing p — 3.1  108 0.81 0.74 One-step UPS; Hall
100 1C annealing n 2.8  1010 — 0.71 0.84 One-step UPS; Hall
130 1C annealing n 4.9  1010 — 0.59 0.96 One-step UPS; Hall
150 1C annealing n 1.5  1011 — 0.53 1.02 One-step UPS;Hall
No annealing p — 5.0  1017 1.1 0.4 Two-step XPS; M–S 174
60 1C annealing p — 1.5  1017 0.9 0.6 Two-step XPS; M–S
100 1C annealing p — 3.4  1016 0.8 0.7 Two-step XPS; M–S
130 1C annealing n 2.8  1016 — 0.7 0.8 Two-step XPS; M–S
160 1C annealing n 6.3  1016 — 0.55 0.95 Two-step XPS; M–S
No I2 exposure p — B1  1012 — 0.5 One-step Conductivity; UPS 161
I2 exposure p — B9  1012 — 0.35 One-step Conductivity; UPS
b
MAPbI3(Cl) Air n — — 0.66 1.04 One-step SPV, CBD 175
O2 n — — 0.68 1.02 One-step SPV, CBD
Vacuum n — — 0.14 1.56 One-step SPV, CBD
N2 n — — 0.23 1.47 One-step SPV, CBD
FAPbI3b PbI2/FAI = 0.3 p — — 1.07 0.57 One-step STS 153
PbI2/FAI = 0.65 p — — 0.76 0.70 One-step STS
PbI2/FAI = 1.0 n — — 0.66 0.80 One-step STS
PbI2/FAI = 1.2 n — — 0.58 0.88 One-step STS
PbI2/FAI = 1.7 n 0.41 1.06 One-step STS
MAPbBr3a PbBr2/MABr = 0.2 p 1.3  107 1.20 0.96 AVC PYS; Hall 154
7
PbBr2/MABr = 0.6 n 3.5  10 1.08 1.08 AVC PYS; Hall
PbBr2/MABr = 1.0 n 4.1  107 1.02 1.14 AVC PYS; Hall
PbBr2/MABr = 1.5 n 4.9  107 0.99 1.17 AVC PYS; Hall
MAPbBr3a p 4.0  107 ITC TOF; Hall 155
MAPbBr2.94Cl0.06a i B3.1  10 6
B3.1  106 ITC TOF; Hall
MAPbBr2.85Cl0.15a n 1.1  107 ITC TOF; Hall
MAPbBr2.4Cl0.6a n 3.6  108 ITC TOF; Hall
MAPbCl3a n 6.7  108 ITC TOF; Hall
a a
Single crystal; bpolycrystalline thin film. b
n, n-type doping; i, intrinsic doping; p, p-type doping.

Table 7 The commonly used solvents in the solution-processing of lead- and high background hole densities, aggravating the SRH
based PSCs and their enthalpy of solvation (Pb2+) along with calculated recombination processes and limiting the efficiency of the
MBO
corresponding solar cells.108,178,179
Solvent DHsolv:Pb2+ (kcal mol1) MBO DFT calculations showed that VSn dominated among all point
DMSO 412 1.5 defects under I-rich and Sn-poor conditions. So, most tin-based
DMF 403 1.88 perovskites, such as CsSnI3, FASnI3 and MASnI3, showed heavy
NMP 401 1.90 p-doping properties.51 Other studies suggested that bulk Sn4+
GBL 384 1.99
ACN 374 3.03
spontaneously transforms into Sn2+, which released two holes to
the VB and is responsible for p-doping of perovskites (Fig. 15(a)).110
Although the doping mechanism of tin-based perovskite is still
controversial, increasing the content of Sn2+ in the precursor
electronic configuration (ns2 np2) and ionic radius.177 However, solution has been proved to be an effective way to reduce the hole
the easy oxidation of Sn2+ to Sn4+ results in heavy p-type doping concentration.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10031
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 15 (a) Possible mechanism and defective models of the perovskite self-p-doping. Reproduced from ref. 110 with permission from American
Chemical Society, copyright 2020. (b) Schematic illustration of the Sn(IV) scavenging method. TM-DHP is added to the tin perovskite precursor solution.
Sn(0) nanoparticles formed by the reduction of SnF2 by TM-DHP scavenge residual Sn(IV) impurities. Reproduced from ref. 181 with permission from
Springer Nature, copyright 2020. (c) Scheme of the reducing vapor atmosphere process of the perovskite device structure. (d) Possible mechanism of the
hydrazine vapor reaction with Sn-based perovskite materials. Reproduced from ref. 185 with permission from American Chemical Society, copyright
2017.

Reducing agents can be used to convert Sn4+ to Sn2+ in tin- spin-coating the hydrazine solution on the substrates, which
based perovskites during the solution process. Initially, excess can suppress the formation of the A2SnI6 phase via the possible
Sn powder is used in a perovskite precursor solution to restore redox reaction path (Fig. 15(c) and (d)),185
Sn4+:180
N2H4 + 2SnI62 - 2SnI42 + N2 + 4HI. (24)
Sn 4+
+ Sn - 2Sn .2+
(23)
As a consequence, the Sn4+/Sn2+ ratios were reduced by
Gu et al. demonstrated that after adding the Sn power, B20% compared to the device without N2H4 treatment.
the color of SnI2 precursor solution turned from dark red to However, N2H4 was sufficient enough to reduce Sn2+ to metallic
light yellow, indicating that the comproportionation reaction Sn, which could further destroy the perovskite lattice.
(eqn (23)) occurred. Except for excess Sn powder, a highly Therefore, the hydrazinium chloride (N2H5Cl) salt, with a
chemical selective organosilicon reductant, TMDHP, was added moderate reducibility together with SnF2, was adopted to pre-
into the precursor solution to restore SnF2 to Sn(0) nano- vent the overreduction of Sn2+.186
particles (Fig. 15(b)), which was reported to eliminate the In addition, Sn source supplement was another effective
residual Sn4+ impurities and get Sn(IV)-free perovskite method to attenuate or to eliminate self-doping of Sn4+. The
films.181 By adding 1 mol% TMDHP, the Sn4+ on the surface increase of Sn2+ concentration effectively decreased the concen-
and in the bulk of the film was reduced by 10% and B100%, tration of VSn defects, avoiding the transformation of the
respectively. Similarly, Sn powders can also reduce Sn4+ in perovskites to heavy p-type doping.187
mixed Sn–Pb perovskites.182 After adding Sn powders, the SnF2, which was first reported in CsSnI3, has been proved to
MA0.3FA0.7Pb0.5Sn0.5I3 precursor solution can keep the color be an effective additive to prevent the heavy p-doping from self-
unchanged in ambient air for 2 weeks. doped Sn4+.188 By adding 20% SnF2, the hole density measured
Apart from Sn power and organosilicon reductant, ionic by Hall and capacitance measurements decreased from
reducing agents were also applied to tin-based perovskites. B1019 to B1017 cm3.189 This result was also confirmed by
Reductants, such as AHP and HPA, were adopted to suppress TA measurements, in which the hole concentration of CsSnI3
the oxidation of Sn2+. It thus reduced VSn density and conse- increased from 45.7  1017 cm3 to B2.8  1016 cm3 after the
quently improved the VOC and PCE of PSCs substantially.183,184 introduction of 20 mol% SnF2.190 Surprisingly, reducing the
Recently, a new reductant with higher reducibility, hydrazine background hole density through SnF2 is also effective to other
(N2H4), was used to provide a reducing atmosphere by tin-based perovskite materials.111,191–194 For example, the hole

10032 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

density of FASnI3 decreased from 2.2  1020 cm3 to 7.2  which included a DEF:DMPU mixture was proposed, and the
1018 cm3 with the addition of 10 mol% of SnF2. In addition to solar cell performance was improved compared with the one
SnF2, other Sn halides, such as SnCl2, were also demonstrated based on DMSO solvent.
to reduce VSn effectively.195,196 Similar to SnF2, the SnCl2
additive could offer a Sn-rich environment to restrain the 4.3 Doping efficiency
generation of VSn. 4.3.1 Method of calculation. In modern first-principles
Halide perovskite films are mostly fabricated via solution calculation, the defect formation energy DHf (a,q) can be
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

process, and the chemical interactions, such as the Lewis acid– expressed as a function of three factors: (1) the total energy
base coordination and hydrogen bonding interaction in per- change (DE) between a pristine crystal and defective crystal,
ovskite precursors, are vitally important for defect density and (2) the atomic chemical potentials mi, and (3) the electron Fermi
electronic doping of perovskite devices. Additives can form energy EF:115,202–204
intermediates with SnI2 to regulate the defect density and P
DHf ða; qÞ ¼ DE ða; qÞ þ ni mi þ qEF ; (25)
self-doping of the resultant film. AA was introduced into i
MA0.5FA0.5Pb0.5Sn0.5I3 precursor solutions, whose CQC double P
where DE ða; qÞ ¼ E ða; qÞ  E ðhostÞ þ ni E ðiÞ  qeVBM . E(host)
bonds can quench the singlet oxygen to avoid the oxidization of i
Sn2+.197 In addition, TP can suppress the oxidation of Sn2+ in is the total energy of the host for the same supercell in the absence
CsPb0.5Sn0.5I2Br perovskite solution as the hydroxybenzene of the defect. ni is the number of elements, E(i) is the elemental
groups act as an antioxidant.198 Besides, SnI2–TP–PbI2 compo- solid/gas energy, q is the number of electrons transferred from the
sites based on the coordination between the ester group and supercell to the reservoirs in forming the defect cell, and eVBM is the
metal ions can improve crystallization and film quality. defect transition energy level. Combined with the formula and the
Moreover, solvent for precursor solution is critical for device boundary conditions for charge neutrality (the total charge of
fabrication. DMF and DMSO mixtures are the common solvents electrons and negatively charged defects equals the total charge
for high-efficiency tin-based perovskite device fabrication. of holes and positively charged defects), the defect formation
However, DMSO can oxidize Sn2+ to Sn4+, which might be the energies and the equilibrium concentrations of charge carriers
primary source of tin oxidation in the precursor solution.199,200 can be calculated self-consistently.205
New solvents or solvent mixtures need to be explored for 4.3.2 Doping limit rule. Under equilibrium growth conditions,
preparing high-performance and stable tin-based perovskites. there are some thermodynamic limits of the chemical potentials
Thirteen solvents have been examined in the fabrication of (mi). First, to avoid precipitation of the host elements, a negative mi
FASnI3 (Fig. 16).201 Considering the solubility, stability (no Sn2+ must be satisfied:115
oxidation) and the ‘‘perovskite-ability’’ (i.e., the possibility to
form the perovskite structure) (Table 8), novel solvent system, mi r 0 (26)

Fig. 16 Molecular structures of several possible solvents. The different colors highlight the division in classes.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10033
 View Article Online

Chem Soc Rev Review Article

Table 8 Physicochemical properties of the solvents

Solvent Mw (g mol1) d (g cm3) Tm (1C) Tb (1C) er m (D) DN (kcal mol1) dD dP dH EREL (eV)
DMF 73.09 0.94 61 153 38.25 3.86 26.6 17.39 13.7 11.25 0.848
DEF 101.15 0.91 — 177 29.02 3.99 30.9 16.4 11.4 9.2 0.841
DMAC 87.12 0.94 20 165 37.78 3.72 27.8 16.8 11.5 10.2 0.824
TMU 116.16 0.97 1 177 24.46 3.47 29.6 17.3 6.0 10.5 0.880
NMAC 73.09 0.96 27 206 179 4.12 — 16.9 18.7 13.9 0.855
GBL 86.09 1.05 44 204 40.96 4.27 18 19 16.6 7.4 0.607
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

NMP 99.13 1.03 24 204 32 4.02 27.3 18 12.27 7.16 0.810
DMI 114.15 1.06 8.2 224 37.6 2.64 27.3 18 10.5 9.7 0.758
DMPU 128.18 1.06 20 247 36.12 4.17 33 17.8 9.5 9.3 0.893
3MOx 101.11 1.17 15 259 77.5 5.0 — 18.0 12.1 9.1 0.753
HMPA 179.20 1.03 7 230 31.3 5.38 38.8 18.5 8.6 8.6 0.961
Sulfolane 120.17 1.26 25 285 44 4.68 14.8 20.3 18.2 10.9 0.473
DMSO 78.13 1.10 19 189 47.24 3.96 29.8 18.4 16.36 10.23 0.976
2-ME 76.09 0.965 85.1 124.5 16.93 2.04 — — — — —
*Mw, molar mass; d, density; Tm, melting point; Tb, boiling point; er, dielectric constant; m, dipole moment; DN, donor number; dD, dispersion
interactions; dP, dipolar interactions; dH, hydrogen bond; Erel, the relative formation energy of SnI2–solvent complexes, calculated using DFT.

Second, to maintain a stable host, take binary compounds the equilibrium growth of MAPbI3, which indicates the growth
(AB) for example, mi are bounded by:115 conditions that can be used to tune the n-type/p-type properties
of the materials.
mA + mB = DHf (AB), (27) In addition, the above equations can be used to deduce the
where DHf (AB) is the formation enthalpy of AB. Finally, to formation energies of intrinsic defects at different points in the
exclude the possible secondary phases (AnBm) of host elements, phase diagram. Fig. 17(b)–(d) show the formation energies of
mi should be limited by:115 vacancy and interstitial defects in MAPbI3 calculated at three
different points, I (I-rich/Pb-poor), II (moderate), and III (I-poor/
nmA + mmB r DHf (AnBm) (28) Pb-rich) in the phase diagram.25 The formation energies of
the considered point defects are listed in Table 9. Also, the
Take MAPbI3 as an example, the existence of MAPbI3
equilibrium carrier concentration and EF in self-doping can be
without secondary phases of MAI and PbI2 should satisfy the
obtained.
later equations under thermodynamic equilibrium growth
conditions:206

mMA + mPb + 3mI = DHf (MAPbI3) (29)

5 Molecular doping
New-generation semiconductor materials have been developed
mMA + mI r DHf (MAI) (30)
with low-cost, and solution-processability, and the corres-
mPb + 2mI r DHf (PbI2) (31) ponding devices exhibit unique mechanical, optoelectronic,
and chemical properties. Different from the classical electronic
Fig. 17(a) plots the calculated chemical potential region doping strategy (atomic doping), the solution process enables
accessible under equilibrium growth conditions for MAPbI3, a new electronic doping strategy, the molecular doping. The
as defined by eqn (29)–(31).25 The narrow but long region molecular dopants can be designed and customized through
marked in red color is the thermodynamic stable range for heteroatom substitution (such as F4TCNQ and F6TCNQ) and

Fig. 17 (a) The calculated phase diagram of MAPbI3. (b)–(d) The formation energies of intrinsic point defects in MAPbI3 at chemical potentials I (b), II (c),
and III (d). Defects with much high formation energies are displayed as dashed lines. Reproduced from ref. 25 with permission from AIP Publishing,
copyright 2014.

10034 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Table 9 The formation energies (unit: eV) of twelve neutral defects in MAPbl3 at chemical potential points I, II, and III shown in Fig. 17

Condition Ii MAPb VMA VPb IMA IPb MAi PbMA VI Pbi MAI PbI
I 0.23 0.28 0.81 0.29 1.96 1.53 1.39 2.93 1.87 4.24 3.31 5.54
II 0.83 1.15 1.28 1.62 3.01 3.45 0.93 2.51 1.27 2.91 2.25 3.62
III 1.42 1.47 2.01 2.68 4.34 5.1 0.2 1.74 0.67 1.85 0.93 1.97
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

structural modifications.24,207,208 Furthermore, the distribu- IEsemiconductor  EA-dopant represents the free energy for elec-
tion, diffusion, and concentration of the dopants in the semi- trons to transfer between molecules and the semiconductor.
conductor precursor solution and films can be precisely Moreover, the distance between electrons and holes is quite
controlled to modulate the electronic properties of the short, thus the electronic interaction between dopants and the
material.207,209 semiconductor can’t be ignored. Considering the electronic
interaction, the energy of an ion pair can be written as follows:212

5.1 Doping mechanisms e2

E ðdÞ ¼ ðIE  EA Þd  k d2 ; (32)
r
The mechanism of molecular doping is illustrated in Fig. 18 by
comparing the energy level differences between perovskites and where d is the degree of charge transfer, k is Coulomb’s constant, e
dopants. The dopants possess unoccupied and/or occupied is the electric charge, and r is the distance between the electron
molecular orbitals for electrons. As shown in Fig. 18, if the and the hole. The value of d which minimizes E(d) determines the
LUMO level of the molecular dopant (e.g. F4TCNQ) is near the amount of charge transferred.
VB of the semiconductor (perovskite), electrons will transfer In addition to the ion pairs, the dopants and semiconductor
from the VB of semiconductors to the empty electron acceptor, may form CTCs by the hybridization of an unoccupied orbital
leaving holes in the semiconductor. As a result, the doped from an acceptor and an occupied frontier molecular orbital
molecules become negatively charged and build up a space– from a donor, resulting in the creation of new local HOMO and
charge layer with semiconductors. This process in turn raises LUMO levels.213–215 Electrons rearrange in the new hybrid
the LUMO and HOMO (or electrochemical potential) of the orbitals, and both molecular dopants and the semiconductor
molecule till thermodynamic equilibrium. If the HOMO of the share the charge, so fractional charge transfer may occur
molecular dopant (e.g. CoCp2) is higher than the CB of semi- between the semiconductor and dopants in the CTCs.24
conductors, electrons will transfer from the dopant to semi-
conductors, resulting in n-type doping. 5.2 Doping strategies
However, to establish a model to predict the free charge 5.2.1 Bulk molecular doping. Benefiting from the chemical
density qualitatively, the interaction between electrons must be processability and hybrid features of halide perovskite materials,
considered.210,211 Take the p-type doping as an example, the IE molecular doping has been successfully applied. Molecular
is the amount of energy required to remove an electron from an additives have various chemical interactions/bonding with the
isolated atom or molecule. And the EA is the amount of energy solvent/perovskite, so they can adjust the crystallinity, regulate the
liberated when an electron is added to a neutral atom or orientation and stress, and form doping gradients. More impor-
molecule. They are referenced to the vacuum level, so the value tantly, the properties of molecular dopants can be easily tuned

Fig. 18 Simplified model of the semiconductor doping process exemplified by the halide perovskite doped with F4TCNQ or CoCp2.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10035
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 19 Molecular structures of common p-type molecular dopants.

through heteroatom substitution and molecular design, which Table 10 Energy levels of common p-and n-type molecular dopants and
provide endless approach to manipulating the electronic doping. their application

Variety of molecules have been studied as dopants in HOMO LUMO Doping

organic semiconductors. Fig. 19 and Table 10 show the molecular Molecular (eV) (eV) type Application Ref.
structure and energy level of the commonly used p-dopants. F4TCNQ — 5.2 p OPV, PSC, 221
TCNQ, a quinone material with a conjugated structure, can form QDSC
charge transfer salts with electron donor molecules.209,216–218 The TCNQ — 4.3 p OPV, PSC 218 and 219
F6TCNNQ — 5.6 p OPV, PSC 219 and 224
resulting organic salts showed an increased conductivity of up to F2HCNQ — 5.6 p OPV, PSC 220
1010, which was a big progress in this field.216 F2TCNQ — 5.03 p OPV, PSC 220
Nevertheless, the EA (4.23 eV) of TCNQ is not sufficient to FTCNQ — 4.56 p OPV 210
HAT-CN — 4.7 p OPV, PSC 24
effectively withdraw the electrons from most OPV materials.24,219 Mo(tfd)3 — 5.37 p OPV, PSC, 233
Introducing an electron withdrawing group into the dopant QDSC
molecule, such as fluorine, –CF3 or nitrile groups, has been Mo(tfd-COCF3)3 — 5.48 p OPV, PSC, 209
QDSC
proposed to raise its EA. A series of derivatives based on the Mo(tfd-CO2Me)3 — 5.21 p OPV, PSC 233
parent compound TCNQ have been synthesized for organic CN6-CP — 5.87 p OPV, PSC 230
semiconductor doping, as shown in Fig. 19. Fluorine is the DDQ — 4.81 p OPV 227
CoCp2 3.47 — n OPV, PSC 252
conventional electron withdrawing group to exchange the CoCp*2 2.9 — n OPV, PSC 253
hydrogen in TCNQ. Interestingly, the EA of TCNQ does not (RuCp*Mes)2 2.76 — n OPV, PSC, 242
showing a linear relationship with the replacing number of QDSC
(RhCp*Cp)2 2.7 — n OPV, PSC 243
fluorine.219 For example, the estimated values of EA of F2TCNQ
and F4TCNQ are 5.03 and 5.2 eV, respectively.220,221
Nitrile groups, a stronger electron withdrawing group com- good processability, leading to easy and efficient doping.225–227
pared with fluorine, was used to further increase the EA. The In contrast, HAT-CN had poor solution processability due
F2HCNQ, in which two fluorines are substituted by nitrile to its larger and rigid planar structure.228 However, comparing
groups, showed a higher EA, which was B0.4 eV higher than with the above micromolecule dopants, HAT-CN was relatively
that of F4TCNQ.216,222 In addition, with six fluorine groups, the stable without noticeable diffusion. The success of DDQ and
EA of F6TCNNQ, was further elevated to 5.60 eV.219,223,224 HAT-CN utilization has encouraged the development of higher
DDQ was also widely used as a p-dopant. Though its EA was EA molecular dopants. With an EA of 5.87 eV, CN6-CP has
lower than that of TCNQ,24 the DDQ had high solubility with become the strongest p-type dopant in the recent literature

10036 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

for OPV.227,229,230 Although this neutral molecule has been solubility, they can only be processed under vacuum deposi-
reported to be insoluble and highly reactive, they appeared to tion. Meanwhile, 3D structure dopants occupy a large volume in
be stable and soluble when mixed with conductive polymers, OPV layers, limiting the doping energy levels. And their large
such as the PDPP(6-DO)2TT, in CH2Cl2. And the conductivity size could prevent them from being inserted into the crystalline
of PDPP(6-DO)2TT was promoted from 106 O1 cm1 to lattice, except for the amorphous materials with a loosen
30–70 O1 cm1 after doping by CN6-CP.230 structure.239,240
The organo-metal dopant, Mo(tfd)3, with a high EA of In the early stage, the n-type dopants used for organic
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

5.37 eV and air stability, has also been investigated.231–233 semiconductors were atomic dopants.241 However, the atomic
However, due to the poor solubility in solutions, it can only be dopants are easy to diffuse, limiting the stability and commer-
processed by using vacuum deposition.24 To improve its solu- cial applications of OPV devices. In principle, n-type dopants
bility that enabled the solution process and prevented the with a low HOMO level (2.8–3.6 eV) could achieve a high doping
unwanted diffusion, structural modifications were adopted. efficiency for OPV.207 The widely used n-type molecular dopants
Substituting the nitrile functional groups with either methyl and their energy levels are shown in Fig. 20 and Table 10.
or octyl esters was applied to increase the solubility. For, Organometallic dopants, with strong electron-donating proper-
example, Mo(tfd-CO2Me)3 has an improved solubility, which ties, such as mesitylene pentamethylcyclopentadienyl ruthe-
is 20 times higher than that of Mo(tfd)3.234,235 Similarly, a series nium dimer (RuCp*mes)2, dimers of rhodocene, (RhCp2)2,
of F4TCNQ analogues, with nitrile functional groups substi- and pentamethylcyclopentadienyl cyclopentadienyl rhodium
tuted by either methyl or octyl esters, showed an increased dimer, (RhCp*Cp)2, are widely used as n-type molecular
solubility of more than 100 times higher and without change dopants.209,242,243 Besides, metallic sandwich compounds, such
in EA.236 According to these molecular design principles, a high as CoCp2 and CoCp*2, were also used as n-type dopants, but
EA soluble molecular dopant, TMCN3-CP, was proposed and they are unstable when exposed to air or water.244–246 The metal
synthesized.208 And groups like –CN3 and –COCF3 were also sandwich dimers, which cleaved upon doping, can resolve the
applied to increase the solubility of dopants.208,209 The F4-R- air-instability issues. And the dimer can extended the storage
TCNQ was synthesized to adjust its diffusion properties by lifetime of the dopant, providing a strategy for stabilizing the
replacing the nitrile with the tailorable ester group, which low IE dopants.247–249 It needs to be reminded that with the
significantly increases the dopants’ solubility and reduces its increase of EA, strong P dopants are difficult to be stabilized.230
diffusion rate in the OPV layer.24 We expect that tremendous Bulk molecular doping was also investigated in perovskite
efforts need to be devoted for customizing the molecular solar cells. F4TCNQ was doped in the perovskite film by solid
dopants such as Mo(tfd-R)3, F3-R-TCNQ, and F4-RTCNQ to diffusion via heating the F4TCNQ powder on MAPbI3, which
enable a controllable solution processability. modified the ITO/perovskite interface and enabled a HTL-free
To minimize the diffusion of the dopants in the composite PSC device. After F4TCNQ doping, the device efficiency was
film, three-dimensional (3D) dopants were proposed, such as significantly improved, from 11.0% to 20.2%.29 F4TCNQ was
the C60 molecules. Replacing the hydrogen atoms with the also used to modulate the electronic properties of CsPbI3
electron-adsorbing fluorine atoms in C60 molecules increases perovskites.250 After incorporating F4TCNQ into the perovskite
the EA, tuning the C60 molecules to a p-dopant and derivatives nanocrystals, the hole mobility was improved from 1.7 
such as C60F36 and C60F48 were synthesized.237,238 With higher 105 cm2 V1 s1 to 5.9  105 cm2 V1 s1, and the con-
EA than that of F4TCNQ, they showed excellent stability against ductivity was enhanced by 250%. However, benzyl viologen or
thermal diffusion in experiments. However, due to their poor cobaltocene dopants significantly reduce the conductivity

Fig. 20 Molecular structures of common n-type molecular dopants.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10037
 View Article Online

Chem Soc Rev Review Article

owing to their n-dope effect on the CsPbI3 perovskite, resulting in QDs. (RuCp*mes)2, with an effective IE o 3.0 eV, donated
in the increasing of hole injection barriers. Besides, DMBI electrons to the QDs, thus resulting in a strong n-type doping
derivatives were introduced into MAPbI3 materials, which effect (Fig. 22(d)).242,259 Note that the molecular dopant concen-
upshifted the EF by 520 meV, leading to n-type doping.251 tration should be moderate, high doping concentrations might
Furthermore, the molecular dopants that are used in OPV can create larger surface dipoles, thus hindering the carrier
also be applied to PSCs given the composition and ‘‘soft’’ transport.21
nature of perovskites; however, intensive studies are needed. Organic PV materials are usually bonded via weak inter-
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

5.2.2 Surface transfer doping. Surface transfer doping molecular interactions, such as hydrogen-bonding, van der Waals
relies on the electron exchange between a semiconductor and interaction force, or electrostatic interactions, making it easier
dopants situated at its surface, which is a meritorious doping to introduce defects through the traditional doping process.129
strategy for nondestructive doping of ‘‘soft’’ semiconductors.254 Surface transfer doping has received broad interest as a ‘‘gentle’’
For the conventional doping process as ion implantation, the method for organic semiconductor doping. SAMs can be used to
energetic ions may destroy the lattice of ‘‘soft’’ semiconductors.129 modify the surface of an organic semiconductor. By designing the
However, as a low cost method, the surface transfer doping avoids functional groups (electron-withdrawing or electron-donating
the utilization of energetic ions, which can effectively dope the groups), the direction of electron transfer between the semicon-
semiconductors without lattice damage. ductor and molecules can be adjusted.260–262 CF3-SAM and
The molecular doping process and the schematic energy 4-methyl-benzenethiol (CH3-SAM) are usually used to modify
levels are illustrated in Fig. 21; dopants situated at the semi- OPVs, e.g., CuPc.263–265 CF3-SAM has a strong electron with-
conductor surface could exchange electrons with the semi- drawing –CF3 group, making its HOMO level lower than that of
conductor, causing the electron or hole accumulation. CH3-SAM. After deposition of 30 Å CuPc on the SAM/Au(1 1 1), a
QDs have a large surface-to-volume ratio, which can be larger downward vacuum level shift (1.05  0.05 eV) was
utilized to control the passivation and net doping by surface observed on CF3-SAM/Au(1 1 1) compared to CH3-SAM/Au(1 1 1)
ligands.255–257 Surface transfer doping in QDSC was first (0.25  0.05 eV).
observed by Shim and Guyot-Sionnest in CdSe QDs.258 Through Other SAMs, such as FTS and OTS, were also used for surface
sodium biphenyl surface treatment, the CdSe QDs exhibited modification of OPV materials.266 On exposure of organic
n-type doping due to the electron transfer from sodium biphe- molecular crystals to their SAMs, the surface conductivity of
nyl to the LUMO of colloidal semiconductor. It opened the door the organic crystals increases by four to five orders of magni-
to a successful doping method for QDSC without atomic doping tude for FTS, by about two orders of magnitude for fluoro-
in the QD bulk. Moreover, other colloidal semiconductor alkyltriethoxysilane and by a factor of ten for OTS treatment.
nanocrystals, such as CdS and ZnO, can also be modulated to Organosilanes with a higher fluorine content have a stronger
n-type by sodium biphenyl. electron-withdrawing ability, which induce a higher conduc-
The EA of the molecular dopants determines the type of tivity and more effective p-type doping. Owing to solution
surface transfer doping. For a metal–organic complex with deep processability, the surface transfer doping is also applied to tune
EA, it can act as the p-dopants. As shown in Fig. 22(a), Mo(tfd- the carrier concentration of PSCs via chemical approaches.
COCF3)3 is an oxidizing agent, which has a high EA and can Transition metal oxides, such as MoO3, was a widely used
accept electrons from the VB of QDs.233,243 When comparing its p-dopant due to their deep-lying electronic energy states.267,268
electrochemistry with that of Mo(tfd)3, Ni(tfd)2 was predicted to With 64 Å of the n-type MoO3 (EA of about 6.7 eV) deposited
have an EA of 5.6 eV, which might also act as a p-dopant to oxidize upon MAPbI3, the energy levels of MoO3 shifted toward a lower
QDs (Fig. 22(b)). Mo(PhBz-dt)3 with a lower EA (4.8 eV) compared binding energy by 0.30 eV, while the energy levels of MAPbI3
to Mo(tfd)3 was being used as a weaker p-dopant (Fig. 22(c)).232 shifted toward a higher binding energy by 0.25 eV, indicating
Meanwhile, the utilization of complexes (reductants) with a the electron transfer from MAPbI3 to MoO3.269 Mo4+ and Mo5+
low effective IE may result in the shifting of EF towards the CB oxidation states, with 3d5/2 binding energies at 230.1 eV and

Fig. 21 Schematic diagram of the surface transfer doping process. The band diagram of an intrinsic semiconductor (a), a semiconductor by p-type
surface transfer doping (b), and a semiconductor by n-type surface transfer doping (c).

10038 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 22 General molecular dopants for quantum dots. Reproduced from ref. 21 with permission from American Chemical Society, copyright 2016.

232.0 eV respectively, appear in the XPS core level spectrum, The reduction effect of surface band bending (0.16 eV) on a
revealing that a chemical reaction takes place at the perovskite/ high defect density perovskite surface is much lower than that
MoO3 interface.270,271 Furthermore, the DFT calculations on a low defect density perovskite surface (0.60 eV) after the
showed that Mo–I covalent bonds and N–H  O hydrogen HAT-CN modification, which may be probably due to the Fermi
bonds were formed, leading to the increase of work function level pinning effect by surface defects.274
by B0.52 eV.272 Interestingly, the MAPbIBr2/MoO3 contact was In addition, FDC-2-5Cl with an electron-withdrawing penta-
revealed to form a hole–electron recombination region through chlorophenyl group was explored as acceptors and passivators
in situ UPS and XPS measurements (Fig. 23(a)), which provided (Fig. 23(d)).275 After the FDC-2-5Cl surface treatment by
a new pathway for hole transfer.270 In addition to MoO3, other solution coating, a p-type doping surface occurred and a high
oxides, such as WO3 and V2O5, were expected to modulate the PCE (21.16%) was achieved with an enhanced VOC of 1.14 V.
electronic characteristics of perovskites. Moreover, by using the spin-coating method an interfacial
Small molecules such as the F16CuPc and HAT-CN are dipole layer was formed after F4TCNQ interface modification,
commonly used dopants in organic materials, while they can and the electrons transferred from the perovskite to F4TCNQ,
be deposited onto perovskites by physisorption or chemi- leaving positive charges in the perovskite side. The p-type
sorption.270 F16CuPc has a larger EA (B4.7 eV) than MAPbIBr2, doping enhanced the built-in electric field and increased the
which can downshift the vacuum level of MAPbIBr2 by 0.7 eV energy barrier to electron diffusion, improving the PCE from
(Fig. 23(b)). Compared to F16CuPc, HAT-CN is a stronger 15.3% to 18.1%.276
electron accepting molecule with a deep LUMO of 5.2 eV. CoCp2 can be used as an n-type surface dopant to change
However, the deep-lying HOMO level may mismatch energy electrical properties of perovskite thin films.277 By varying the
levels and cause a high hole extraction barrier at the interface depositing quantity of CoCp2, a tunable shift in the EF was
of MAPbIBr2/HAT-CN, which aggravated electron–hole recom- observed, indicating that the interfacial energy levels could be
bination (Fig. 23(c)). Surprisingly, HAT-CN is a suitable electron controlled by surface transfer doping. HZ-TIPS and its homo-
accepting molecule for 2D perovskites due to their deep VB.273 logue, OZ-TIPS, formed an interfacial dipole at the interface
Introducing HAT-CN molecules into 2D perovskites enabled the due to the van der Waals interaction between molecules and
charge transfer between the 2D metal halide perovskites and the first few layers of the CsPbBr3 surface. The HZ-TIPS/
HAT-CN, which was proved by the external quantum efficiency CsPbBr3 interfacial dipole made the energy band bend down-
spectra. However, no charge transfer process was observed ward. However, the upward band bending occurred in the case
when the 2D perovskite contained large organic cations, such of the OZ-TIPS/CsPbBr3 interface.278 In addition, polymeric
as phenethylammonium (PEA+), and the large cations may have materials, such as b-PEI, PVP, and PAA, were used for passivat-
prevented the contact between the 2D perovskite metal halide ing the perovskite surfaces and changing the surface EF.279
sheet and the HAT-CN molecule layer. Interestingly, the surface Coordination bonds were formed between polymeric materials
defect density of perovskites was greatly affected by doping. and the perovskite, because of the interaction between the lone

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10039
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 23 (a)–(c) Schematic energy level diagrams of MAPbIBr2/MoO3 (a), MAPbIBr2/F16CuPc (b), and MAPbIBr2/HAT-CN (c) interfaces. Reproduced from
ref. 270 with permission from John Wiley & Sons, copyright 2015. (d) Illustration of the proposed energy band bending at the surface of the perovskite
due to the charge transfer from the perovskite to FDC-2-5Cl. Reproduced from ref. 275 with permission from Elsevier, copyright 2021.

electron pairs at N or O atoms and the empty orbits of Pb. 5.3 Doping efficiency
Also, they prevented the water and oxygen from eroding the Molecular doping is achieved through electron transfer
perovskites, thus improving the operational stability of the between the doping molecules and the semiconductor. When
corresponding devices. doping molecules are incorporated into the semiconductor,
In addition, benefiting from the ionic nature of perovskites, the doped molecules are generally in low concentration and
the lead halide perovskite surface can convert to a thin, dispersed in the semiconductor, which provide enough
compact material layer through in situ reaction with selected surface to donate/accept electrons within the semiconductor.
ions, leading to carrier redistribution at the interface.280 The For surface transfer doping, there is a high probability for
organic cations (MAI and FAI) are volatile, which can leave the molecules within several monolayers to have electron
undercoordinated Pb2+ ions and/or PbI2 at the perovskite exchanged at the semiconductor surface. With the increase of
surface.281–283 Cationic species have been studied for their the thickness of the molecular layer, the probability for electron
compensation or replacement of cation for elimination of sur- exchange between molecules and semiconductors is greatly
face defects. MAI and FAI are used to eliminate the surface lead reduced.
iodide, which downshift the EF and reduces the inner series 5.3.1 Doping efficiency of doped molecules. When accep-
resistance from 30.77 O to 26.76 O.284 Recently, lead–sulfur tor molecules are used to modify semiconductor interfaces, the
bonds (Pb–S) were introduced at the perovskite/ETL interface electrons transfer from the semiconductor to the acceptor due
through the chemical reaction between PbPyA2 and TMS, which to the higher EA of the acceptor molecule. By only considering
upshifted the EF and offered a BSF for electron extraction, the monolayer of acceptor molecules that is directly in contact
resulting in a PCE enhancement from 21.8% to 24.3% in with the semiconductor, the distribution of electrons in the
inverted PSCs.285 Based on the ionic nature of perovskites, acceptor’s molecular orbital follows the Fermi Dirac distribu-
and with the endless space for chemical materials exploration, tion at thermodynamic equilibrium:129
the chemical approaches, like ion exchange, in situ reaction,
interfacial modification, etc., are the most prospective strate-
nA nA
gies for achieving controllable doping in the material and n¼ ¼ FLUMO EF
; (33)
DEþUS
device design. 1þ e kT 1þ e kT

10040 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

where n and nA are the areal electron density and the density of EF). For acceptor molecules that have a low LUMO, such as
surface acceptors, respectively. DE is the activation energy F4TCNQ, the transfer of electrons between the semiconductor
of acceptor molecules, which is a constant that equals the and the acceptor is significant, which may cause a large upward
difference between EV and surface acceptors LUMO (FLUMO). energy band bending that leads to the EF to fall below the VB at
US is the surface potential, which is equal to the difference the surface. Thus, the semiconductor surface becomes degen-
between EV and EF of the semiconductors. US is variable, which erated, and the classical Boltzmann distribution is no longer
is related with the hole density (ps) in the semiconductor and as applicable. Fermi distribution function is used as a step func-
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

a function of acceptor coverage nA. tion to describe the areal hole density:288
In eqn (35), we set the degeneracy factor to 1, assuming that
only one unoccupied LUMO state per molecule can accommo- rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
2kTee0 NV Us 8 Us
date an additional electron charge even in the case of a ps ðUs Þ ¼  1 þ þ p ffiffiffi : (37)
e2 kT 15 p kT
degenerate LUMO. This hypothesis is reasonable, because the
electron–electron repulsion is very strong when more than one
electron enter an isolated molecule.286,287 It is important to emphasize that the above calculation of
According to eqn (35), the doping efficiency (Z) of the surface doping efficiency for semiconductors and dopant molecules is
acceptors can be described as follows: based on ideal conditions. In practice, the interaction between
n 1 1 doping molecules and semiconductors is complex. The shape,
Z¼ ¼ ¼ : (34) size, orientation, concentration and the distribution of dopants
nA DEþUS FLUMO EF
1þ e kT 1þ e kT and the solution processing techniques may affect the real
carrier concentration. Also, doping may induce the formation
Similarly, when the interface is modified with donor mole-
of defects or defect complexes in semiconductors, which
cules, the doping efficiency of the surface donor is
may locally compensate the charge and/or trap free charges.
1 : Moreover, the CTC state formation and coulombic binding of
Z¼ EF FLUMO (35)
1þ e kT ion pairs can influence the distribution of charge carriers.
At present, it is difficult to predict the doping efficiency
The relative position of the LUMO (HOMO) in doped mole- of molecular doping accurately. Future research needs to
cules and EF in the semiconductor reflects the probability of the establish the quantitative relationship between doping concen-
hole (electron) occupying the molecule energy level. When the tration and carrier concentration, which may require improving
LUMO level is much lower than the EF (EF  FLUMO c kT), such chemical imaging and energetic mapping techniques. Also,
as F4TCNQ, the doping efficiency is close to 100%, theoreti- new models, such as delocalized electronic states, are needed
cally, and it is same for surface donors. to understand how the dopant’s orientation and shape affect
5.3.2 Doping efficiency for doped semiconductors. The the magnitude and delocalization of the disorder in the
first several layers of the semiconductor at the surface will semiconductor. In addition, controllable solution processing
generate an electric field when the semiconductor surface techniques need to be developed, which can modulate the
comes into contact with other materials or adsorbs molecules distribution, diffusion, and concentration of dopants within
(charged ion). Driven by the electric field, electrons will be the semiconductor.
redistributed to form a space–charge region near the contact In conclusion, when dopants are atoms, the valence, radius,
interface. The distribution of the electric field intensity and and electronegativity of the atoms determine bond formation
electrostatic potential in the space–charge region due to and carrier generation. Generally, a negatively charged dopant
the existence of surface adsorbates or surface states can be leads to p-type doping, and the opposite is true for positively
obtained quantitatively by solving Poisson’s equation. charged dopants. When dopants are molecules, the HOMO and
Let us assume that the semiconductor surface is modified by LUMO levels of dopants determine the direction of electron
acceptor molecules. For wide bandgap semiconductors, such transfer between semiconductors and dopants. Usually, a high
as diamond, assuming that the semiconductor has unipolar EA molecule can accept electrons from the VB of a semi-
transport characteristics (holes only), using the Boltzmann conductor for p-type doping and a low IE molecule gives
distribution, the ps on the semiconductor surface can be simply electrons to the CB of a semiconductor for n-type doping.
estimated as follows:288 In addition, some semiconductors have low defect formation
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   energy and high intrinsic defect density, showing the unusual
2kTee0 NV US
p s ðU S Þ ¼  exp ; (36) defect physics of self-doping. If the intrinsic defect is effectively
e2 2kT
and negatively charged (cation vacancy or anion interstitial),
where k is the Boltzmann constant, T is the temperature, e and the hole concentration increases, which leads to p-type doping.
e0 are relative dielectric constant for semiconductor and If the intrinsic defect is positively charged (anion vacancy or
vacuum permeability, respectively, and NV is the effective cation interstitial), it may show an intrinsic n-type semi-
density of states for the VB. conductor. It is expected to reveal new chemistry and physics
It should be noted that eqn (38) is only applicable to non- in halide perovskites with defect profiles that are distinguished
degenerate cases (e.g. the VB of the semiconductor is below the with conventional PV materials.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10041
 View Article Online

Chem Soc Rev Review Article

6. Effects of electronic doping o1012 cm3 to (3.8  0.6)  1014 cm3 were achieved.73 Also,
the conductivity of the perovskite crystal was increased from
The essence of electronic doping is the redistribution of B108 O1 cm1 to B104 O1 cm1 after Bi doping.26 The
electrons in the CB and VB, which leads to macroscopic same result could also be obtained in Bi-doped perovskite
differences in conductivity, charge transport, carrier recombi- polycrystalline films, which increased from 3.02  106 to
nation, device performance, etc. The desired electrical charac- 9.76  106 cm1 s1.87 In addition, 0.5% Na doping signifi-
teristics of a semiconductor can be obtained through electronic cantly improved the electronic conductivity of the perovskite,
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

doping, which further determines the performance of photo- from B107 O1 cm1 to B106 O1 cm1 for the MAPbI3.86
voltaic devices and related optoelectronic devices. During the operation of photovoltaic devices, a depletion
layer, also called Schottky barrier, is established at the contact
6.1 Conductivity
interface due to their work function differences, which leads to
Under an electric field, charge carriers (electron and/or hole) a high resistance for photon-generated carrier, blocks carrier
can drift and contribute to the current, and their concentration transport, and is harmful for device performance.5,51 The
determines the internal resistance of conventional PVs. Owing theoretical calculation indicates that the width of the depletion
to the high intrinsic defect density and ionic behavior, facile layer is inversely proportional to the doping concentration:292
migration of intrinsic defects of new-generation PV materials sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
could also contribute to the current, especially for the perovs- 2er e0 Vbi
w¼ ; (39)
kite material. According to different conducting mechanisms, qND=A
conductors can be divided into ionic conductors, electronic
conductors, and mixed conductors. Electronic conductors, where er and e0 are the relative and vacuum permittivity of the
such as metals, graphite, and metal compounds are conductive semiconductor, respectively. Vbi is the built-in voltage, which is
initiated from the movement of free electrons and no chemical given by the difference between the Fermi energies in the
changes occur during the conduction process. Similarly, the n-doped (EF,n) and the p-doped (EF,p) regions in a p–n junction.
electrons in the CB of semiconductors flow along a direction ND/A are donor ND densities for n-doped semiconductors or
driven by the electric field, which is also the electronic conduction. acceptor NA densities for p-doped semiconductors.
Ionic conductors, such as electrolyte solutions or molten electro- To overcome the Schottky barrier, heavily doped strategy, or
lytes, rely on the oriented movement of ions (ion migration) to metallized electrodes, is proposed in crystalline silicon photo-
conduct electricity, which involve chemical changes. In ionic voltaics, and the Al-BSF technology is mature and predominant
solids, the ionic conduction occurs mainly through the migration in industrial processing (Fig. 24(a)).5,293 BSF can decrease the
of ions through lattice defects. Previous studies proved that new- dark current and the reflection of minority carriers, which
generation PV materials are mixed conductors, which have both results in high-efficiency commercial photovoltaics.294 However,
ion and electron transport properties.289,290 metallized electrodes produce a high defect density at the metal–
6.1.1 Electronic conductivity. There are two kinds of semiconductor interfaces, causing severe recombination at the
charge carriers in semiconductors, the positively charged holes rear contact and constraining the PCE to B20% for commercial
and the negatively charged electrons. These charge carriers can silicon cells.5 To limit their impact, reducing the metal–Si contact
drift under an electric field and contribute to the current. The area was applied to minimize the rear-side recombination, lead-
conductivity (s) of the semiconductor can be expressed as ing to increased VOC and JSC.295–297 Even so, classic metallization
follows:291 still results in significant nonradiative recombination, limiting
the VOC of photovoltaics. Inspired by the structure of Schottky
s = nqmn + pqmp, (38)
diodes, the metal–insulator–semiconductor structure was used in
where n and p are the electron and hole concentration, respec- solar cells to substitute the metal–semiconductor structure.298
tively; q stands for elementary charge; and mn and mp stand for A thin layer of insulator, such as silicon oxide (SiO2), silicon
the migration rate of electrons and holes, respectively. There nitride (SiNx), or aluminum oxide (AlOx), was inserted between the
are no numerous differences in the order of magnitude wafer and electrode, which balances the optimization between
between the electron mobility and hole mobility of semi- contact conductivity and passivation.5,127,293,299 This device struc-
conductor materials used in solar cells. As a result, carrier ture is shown in Fig. 24(b), and benefiting from this method on
concentration is a key factor in determining the conductivity. PCE development, the PERC (passivated emitter and rear cell)
The conductivity of the semiconductor can be controlled by technology becomes the mainstream in current production.293
electronic doping. As previously discussed, electronic doping Besides, replacing heavily doped Si by a doped transport layer
was first used in silicon semiconductors, and a continuous (selectively conducts a polarity of charges), usually hydrogenated
conductivity variation from 104 O cm1 to 104 O cm1 was amorphous silicon (a-Si:H) with a low interface defect density is
achieved by changing the dopant (P or B) concentration from another option to minimize rear-side recombination, which are
1012 atoms per cm3 to 1021 atoms per cm3.62 Cu is used for called silicon heterojunction (SHJ) devices (Fig. 24(c)).293,300–302
p-type doping in CdTe, and after adding 0.8  1015 Cu atoms Based on heterojunction structures, the Si photovoltaics achieve
per cm2 to a CdTe layer, an abrupt decrease in resistivity by the current world record PCE of 26.7% and the PCE on production
three orders of magnitude and an increase in hole density from lines is in the range of 23.5–24.5%.293,303

10042 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 24 (a)–(c) Schematic representation of typical Si solar cell architectures, Al-BSF (a), PERC (b) and SHJ (c) structures. Reproduced from ref. 293 with
permission from Springer Nature, copyright 2022. (d) Simplified graph of planar-structure PSCs with BSF. (e)–(f) Schematic diagram of band shift of
heterostructures without (e) and with (f) BSF. Reproduced from ref. 304 with permission from Elsevier, copyright 2021. (g) Schematic diagram of SST.
(h) BSF formation at the perovskite surface with SST. Reproduced from ref. 285 with permission from Science Publishing Group, copyright 2022.

Similar to Si technology, the device structure of PSCs is also is more than 5 mm, which is much larger than B0.5 mm in the
based on the heterojunction structure, that is, the n–i–p and Si p–n junction and B0.026 mm in the monolayer WSe2–MoS2
p–i–n structures. However, electronic doping is rarely used in lateral p–n junction, and more importantly, much larger than
PSCs. Recently, the BSF concept was applied to halide perovs- the thickness of PSC devices of B0.8 mm.272,305–307 Therefore,
kites, in which F4TCNQ was introduced at the two interfaces the perovskite has a higher FF and VOC loss compared with Si
(perovskite/HTL and HTL/electrode interface) to bridge the solar cells. In addition, the perovskite/transport layer interface
perovskite to the electrode with the desired band alignment presents a larger lattice and thermal expansion constant
(Fig. 24(d)).304 As shown in Fig. 24(e) and (f), introducing mismatch, bringing severe recombination and poor long-term
F4TCNQ realized the p-type doping of perovskite and HTL stability.308–310 Perovskite-based ETL and HTL with heavy
surfaces, which created a BSF at the perovskite/HTL/electrode doping should be designed and developed to minimize the
interface with a low band mismatch. Similarly, Li et al. con- mismatch and optimize charge transport.
structed heterojunctions and down-shifted EF through surface 6.1.2 Ionic conductivity. Ion transport is also important for
sulfidation treatment (SST) for inverted PSCs (Fig. 24(g)).285 The electronic devices, such as memristors, sensors, optoelectronic
extra BSF led to inhibited hole transport and improved electron devices, batteries etc.90 When a bias voltage is applied between
transport (Fig. 24(h)), achieving an increased PCE of 24.3% with two electrodes, ion migration will occur, with the electric field
a high VOC of 1.19 V and a low voltage loss of 0.36 V (bandgap of as the driving force. Eventually, the ion migration will reach an
1.55 eV). equilibrium state. Assuming that ion transport processes are
Despite the above successes, the electronic doping in per- random and uncorrelated, the ionic conductivity can be esti-
ovskite materials is still limit. Compared with Si technology, mated through hopping in an ionic conductor:311
the transport layers and perovskite layers have low doping    
1 cq2 2 DS DH
concentration, resulting in a broader depletion region and s¼ a n 0 exp exp  : (40)
6 kT k kT
lower built-in electric field. The calculated depletion region
value of monocrystalline MAPbBr3 solar cells is B7 mm and the Here the factor of 1/6 results from the assumed 3D isotropic
measured depletion region thickness of the MAPbI3 nanosheet motion as in the random walk model (i.e., b = 6), c is the mobile

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10043
 View Article Online

Chem Soc Rev Review Article

ion concentration and q is the charge of mobile ion, a is the the ionic conductivity of SrTiO3 is expected to be modulated
hopping distance between two neighboring sites or free path, through self-doping.
n0 is the attempt frequency, and includes both successful Besides, ion transport is observed in halide perovskite
jumps that can lead to macroscopic diffusion and unsuccessful materials, such as photoinduced phase separation and large
ones, and DS and DH are the activation entropy and activation dielectric constants, which contributes to the serious J–V hys-
enthalpy of elementary jumps, respectively. teresis and poor stability problems of PSCs.313–316 Similar to
According to eqn (40), defect thermodynamics are curial for oxide perovskites, self-doping dominates the ionic conductivity
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

evaluating the ionic conductivity, especially the defect concen- modulation of halide perovskite materials. Iodine atmosphere
tration. The concentration of ionic and electronic defects upon treatment of halide perovskite film could decrease the concen-
deviation from the perfect crystal is the function of stoichio- tration of iodine vacancy, with a low migration activation
metry (d) and temperature, thus, ionic conductivity can be energy between 0.08 and 0.58 eV, which eliminated possible
accommodated through self-doping. Considering an oxide, ion migration paths.162,317–319 As expected, the ionic conduc-
MO, its stoichiometry and conductivity vary with P(O2). Accord- tivity reversibly decreases upon increasing the iodine partial
ing to the related mass-action law, the reaction is described as pressure unambiguously (Fig. 25(a) and (d)). Besides, PbI2 was
follows: proved to passivate defects and eliminate hysteresis in PSCs.
Annealing can accelerate the volatilization of MAI and generate
1 
O2 ðgÞ þ V
O ! OO þ 2h
 (41) PbI2. As the annealing time increases from 5 min to 100 min at
2
150 1C, hysteresis index for the Cs0.1FA0.9PbI3 perovskite was
reduced from 0.384 to 0.146.320
½h 2 ; In addition, atomic doping has been demonstrated to be an
KO ¼   1 (42)
V
O P ð O 2 Þ 2 available approach to regulating ionic conductance. For oxide
perovskites, La0.8Sr0.2Ga0.85M0.15O3d (LSGM), substitution

where VO represents an O vacancy and OO represents one of La with alkaline-earth cations (Sr, Ca, Ba) and/or Ga with
occupied O site in the perfect crystal. The change of the defect bivalent metal cations (Mg, Ni) can yield high oxygen vacancy
concentrations with stoichiometry changes can be predicted by concentration and ionic conductivities.321–325 In contrast,
combining eqn (44) and electric neutrality principle. Based substitution of Ga with higher valence cations (Ti, Cr) yields
on this method, defect concentrations upon stoichiometry lower levels of oxygen-ion conductivity, increasing the migra-
changes can be obtained, known as Kröger–Vink diagrams. tion activation energy from 60 (5%-doped) to 230 kJ mol1
For a classical oxide perovskite, SrTiO3, its Sr/Ti ratio is mani- (40%-doped).324
pulated by the reaction pathway, such as synthetic conditions, In addition to oxide perovskites, atomic doping is also
atmosphere and high-temperature treatments.312 Therefore, applied for halide perovskite to regulate ionic conductivity,

Fig. 25 (a)–(c) Defect chemical modeling applied to MAPbI3 to reveal the defect concentration dependences as a function of iodine partial pressure (a),
acceptor and donor doping contents (b), and oxygen partial pressure (c). (d)–(f) Ionic and electronic conductivity of MAPbI3 as a function of iodine partial
pressure (d), acceptor (Na) dopant content (e), and oxygen partial pressure (f). Reproduced from ref. 90 with permission from American Chemical Society,
copyright 2019.

10044 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

especially for alkali metal cations. LiI, NaI, KI, RbI, and CsI was gap of PV materials, JSC could increase due to the broadening
incorporated into halide perovskite to reduce the intrinsic ion spectral absorption range; however, VOC, which is positively
migration.326 Compared with other alkali metal ions, including correlated with the band gap, will decrease. The ideal band gap
Li+, Na+, Rb+ and Cs+, K+ has a moderate ionic radius, which of a single-junction solar cell is 1.3–1.4 eV through the trade-off
could form 6 coordinations with I and fit the interstitial site between JSC and VOC.331 The PV materials should be thick
of perovskite lattice perfectly, thus, K+ is the best dopant to enough to maximize light absorption, and at the same time
eliminate hysteresis and suppress ion migration. The hysteresis be thin enough for efficient charge carrier transfer. The carrier
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

of PSCs is used to evaluate the ionic conductivity of halide extraction efficiency (ZC) can be described using the following
perovskite materials. For the mixed perovskite (FA0.85MA0.1- exponential equation:113
Cs0.05PbI2.7Br0.3), the control film showed a significant hyster- x
esis with a PCE of 17.99% for the reverse scan and 15.08% for ZC ðlÞ ¼ exp ; (43)
L
the forward scan. After KI doping, hysteresis of PSCs was
significantly eliminated, where a PCE of 18.20% for the reverse where L is the diffusion length of the minor carriers and x is
scan and 17.97% for the forward scan were achieved. KI doping the distance into the semiconductor layer. In detail, L is
also showed the same tendency for MAPbI3 and FAPbI3, indi- proportional to the lifetime of the excited carrier (t), the charge
cating a promising strategy to suppress ion transport. More- carrier mobility (m) and the thermal voltage (VT):
over, potassium ions have also been used to modify the grain pffiffiffiffiffiffiffiffiffiffiffi
L ¼ mtVT (44)
boundaries and surfaces of halide perovskites, resulting in the
decreased defect density and mitigated photo-induced ion
migration of perovskite films.327 Interestingly, the alkali metal 6.2.1 Charge transport. Charge carriers could be scattered
doping could also increase ionic conductance. Na-doping will during transport which might influence the lifetime and mobi-
generate a substitutional defect of Na on the Pb-site (NaPb), lity of carriers. Considering the main scattering mechanism,
which can induce the generation of iodine vacancies and the effective semiconductor mobility m could be approximated
increase ionic conductivity (Fig. 25(b) and (e)).86 by Matthiessen’s rule:51
Furthermore, O2 is expected to behave as an acceptor dopant 1 1 1 1
that increases both ionic and electronic conductivity (Fig. 25(c)).89 ¼ þ þ ; (45)
m mimpurity mphonon mdefect
By treating the device under illumination (Fig. 25(f)) at a
low oxygen pressures (P(O2) o 102 bar), the ionic conductivity where mimpurity, mphonon and mdefect are individual mobilities,
appeared invariant as the oxygen partial pressure changes, limited by coulomb scattering by impurities, phonon scattering
while the ionic conduction increased when the P(O2) reached by the lattice and surface or bulk defects, respectively. The
higher pressures (P(O2) 4102 bar). This behavior indicated introduction of a dopant can optimize the carrier concen-
that MAPbI3 had a certain oxygen solubility and kept the state tration, but offer an ionized center to scatter carriers at the
stable only for low oxygen partial pressures. Also, superoxide same time (Fig. 26(a) and (b)).
ions could cause such an effect in principle but is thermo- Extrinsic impurity and surface defects dominate charge
dynamically not favorable if incorporated interstitially. transport in single crystal materials. Through diffusing P and
Normally, DFT calculation is a helpful tool to evaluate the Al in Si materials, the n+–p–p+ structure is established on Si
activation energy and understand the anomalous ion transport; PVs. Owing to the effects of the BSF, P–Al gettering, and the
however, the experimentally measured activation energy is reflection of Al, the bulk conductivity and carrier diffusion
often lower than the value predicted by DFT calculation.90,280 length of Si PVs are significantly improved, thus greatly improving
These differences are not surprising because the theoretical the PCE of Si PVs.332
method only calculates the ionic conductance in a bulk crystal Dangling bond is the main surface electrical defect in
(only considers bulk point defects), while the crystal surface conventional PVs, whose energy is about B0.55 eV above the
and grain boundaries play as major channels for ion migra- VB in Si PV materials. Chemical passivation and surface recon-
tion.328,329 In addition, there are many other possible ion struction could repair it to shallow defect. By passivating the
migration paths, such as strain induced extended defects and channel/oxide interface of a Si MOSFET with hydrogen, the
soften lattice caused by light illumination.310,330 New theore- defect density was reduced by up to 12 times, which reduced
tical models (common assumptions in DFT calculations) the gate leakage current as well as the tunneling barriers.
should be proposed to predict ionic conductivity by considering Chemical passivation is also widely used in the polycrystalline
these possible paths as more as possible. new-generation solar cells, as well as other electronic devices, such
as transistors, memory transistors, and photodetectors.285,333–337
6.2 Carrier dynamics In polycrystalline material, bulk defects, e.g., grain boundary
The effective carrier concentration determines the PCE of solar and dislocation, also dominate the charge transport process.
cells, which is limited by two factors: (1) the total carrier Self-doping is a common method to regulate bulk defect.
concentration and (2) the carrier extraction efficiency. The total MOCVD-grown GaInNAs have been reported to have a long
carrier concentration is mainly determined by the bandgap and hole diffusion length of 0.9 mm, but the electron diffusion
the thickness of the absorber. By narrowing the optical band lengths are quite short. Through the MBE-grown strategy, a

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10045
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 26 Carrier dynamics associated with dopants in semiconductors. (a) and (b) Scattering of a charge carrier near an ionizing donor (a) and ionization
acceptor (b). Red and blue solid circles represent holes and electrons, respectively. (c) and (d) Electron (c) and hole (d) trapping by a defect of energy ED
between the CB and VB. (e) and (f) Electron (e) and hole (f) de-trapping by a defect of energy ED between the CB and VB. (g) Hole and electron
recombination assisted by defects. The arrow highlights the direction of the transition from the initial state to the final state.

higher electron diffusion length of 0.5 mm was achieved in before it can be excited if the activation energy is large, as
GaInNAs.338 This strategy has also been applied in new- illustrated in Fig. 26(e).
generation solar cells.150,205 In addition, extrinsic dopants The recombination of carriers in semiconductors are usually
may command nucleation and crystallization during film divided into 3 parts: defect-assisted recombination, radiative
growth, thus controlling the bulk defect, especially for films recombination and Auger recombination,347
fabricated by the solution process. Initially, MAPbI3 prepared
dn
using Cl-based precursors has the carrier diffusion coefficients ¼ k1 n  k2 n2  k3 n3 ; (46)
dt
of 1–5  102 cm2 s1, carrier lifetimes of hundreds of
nanoseconds, and diffusion lengths of over a micrometer.339,340 where n is the photo-generated carrier density, and k1, k2, and
Through improving film fabrication techniques, such as additive k3 are the first-order, second-order, and third-order rate con-
engineering and composition engineering, the lifetimes increased stants associated with monomolecular (defect-assisted), bi-
to tens of microseconds and the corresponding diffusion lengths molecular (radiative), and three-body (Auger) recombination,
increased to tens of micrometers.341–344 respectively. When the carrier concentration is relatively low,
In addition to ionized center scattering, neutral defects also the recombination is dominated by the defect-assisted recom-
contribute to scattering.345 Neutral defects can be introduced bination. As carrier concentration increases, the defect-assisted
by isovalent doping or low-ionized-degree impurities and its recombination gradually saturates and radiative recombination
scattering behavior is similar to the scattering of slow electrons gradually increases, which becomes the dominant recombination.
by hydrogen.346 In general, neutral impurity scattering is much At a higher carrier concentration (41017 cm3), the role of Auger
weaker than that of ionized impurities, which can be ignored at recombination becomes non-negligible.149,348 Through the experi-
low doping concentrations. In addition, scattering between mental measurements, such as pump-probe spectroscopy and
carriers should be considered under a high carrier concentration. photoluminescence decay, the recombination rate constants of
6.2.2 Carrier recombination. Under continuous sunlight halide perovskites (MAPbI3 and FAPbI3) have been estimated,
illumination, electrons from the VB will be excited into the where k1 = B107 s1 for Shockley–Read–Hall (SRH) recombina-
CB with a relatively high energy, which tends to shift to a low tion, k2 = B1010 cm3 s1 for radiative recombination, and
energy state such as the impurity energy level. Impurity levels, k3 = B1028 cm6 s1 for the Auger recombination.349
introduced by impurities to the forbidden band of the semi- The carrier recombination can be modulated by electronic
conductor, where an approaching electron or hole is likely to be doping in PV materials. The recombination center is formed
captured or trapped, are spatially localized. Also, there is a when impurities are incorporated into the semiconductors.
likelihood that the trapped electrons (or holes) with a low Through altering the types of dopants, such as Si and In, GaN
activation energy can de-trap to the CB (or VB). These processes LEDs can be manufactured to produce light with colors ranging
are presented in Fig. 26(c)–(f). It is more likely that the trapped from the low end of visible spectrum up to ultraviolet.350
carrier will annihilate or recombine with an opposite carrier Interestingly, a moderate concentration of dopants could

10046 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

reduce the carrier recombination and increase minority carrier of the donor and acceptor which determine carrier concen-
life time. Co-diffusion of P, Al, and/or B in Si materials tration, and ni is the intrinsic carrier concentration. Eqn (49)
significantly enhanced the minority-carrier diffusion length, and (50) reveal that high concentration doping is essential for
and this concept was also applied in OPV for increasing obtaining a larger VOC.
the exciton diffusion lengths.351–353 Furthermore, dopants can As mentioned above, self-doping is an effective method
reduce the concentration of intrinsic defects of polycrystalline to modulate electronic doping. The composition dependent
films. Studies have shown that the carrier lifetime decreased carrier concentration and carrier mobility in perovskite films
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

in PSCs at high doping concentrations. However, low levels are shown in Fig. 27(a).150 By verifying the MAI concentrations,
of doping (Sr, Ca, Ag, Mg, Bi, and Sb, etc.) could increase the J–V curve evolution with the VOC and JSC changes of
the carrier lifetime, which might be related to the reduction perovskite devices were modulated, which are summarized in
of defect density by crystallization control and/or passi- Fig. 27(b) and (c). As the MAI concentration increases, the
vation.28,87,88,100,102,103,354 device VOC first decreases and then increases. According to
In conclusion, carrier transport and recombination are key eqn (49) and (50), the variation of VOC is related to the carrier
factors that determine the carrier properties of a material, such concentration and doping type. Perovskite films made using
as carrier lifetime, diffusion length, and mobility. If the defect 4.0% MAI solution were weakly p-doped with a hole concen-
states are energetically deep within the bandgap, a proportion tration of 4  1014 cm3. Further increasing the MAI content in
of carriers are trapped, reducing the overall conductivity, and the films by increasing the MAI solution concentration to 7%
altering the coulombic or neutral scattering rates of the remain- promoted the hole concentration, which results in the device
ing free carriers. Besides, defect density and carrier properties VOC enhancement. Reducing MAI content converts films from
have great impact on each other, and finally determine the the weak p-type to moderately doped n-type, which explains the
efficiency and stability of the device (Table 11). larger VOC in these devices.
However, the doping concentration also need to be well
6.3 Device performance controlled. As the doping concentration increased, the scatter-
6.3.1 Device efficiency. Electronic doping has a great ing and recombination centers gradually increased, thereby
impact on the VOC of the device, and the VOC is determined reducing the lifetime of the carriers and facilitating the non-
by quasi-Fermi energy splitting and the relationship of doping radiative recombination, so an appropriate doping level is
level:373 crucial. In silicon PVs, a moderate p-type doping concentration
  is about 1015 cm3 (in intrinsic crystalline silicon, there are
kT JSC
VOC ¼ ln þ1 (47) approximately 5  1022 atoms per cm3) and the moderate n-type
q J0
doping concentration is about 1017 cm3.61,374,375 Doping
    concentration above about 1018 cm3 is considered degenerate
qDh qDe
J0 ¼ þ ni2 ; (48) at room temperature.
Lh Nd L e Na
FF is determined by many factors, of which the most
where J0 is the dark saturation current, Dh and De are the important ones are the RS and RSH resistance.376,377 At a high
diffusivity of holes and electrons, Lh and Le are the diffusion current density, the FF is determined by the RS (Fig. 27(e)). The
length of holes and electrons, Nd and Na are the concentration Rs is influenced by both ohmic elements, such as the perovskite

Table 11 Comparison of defect density and carrier properties in photovoltaic technologies

Materials Defect density (cm3) Mobility (cm2 V1 s1) Diffusion length (mm) Lifetime of a minority carrier Ref.
a 9 10
Si 10 –10 1350 (electron) 100–300 8.8 ms 303
480 (hole)
GaAsa 1013–1015 8500 (electron) 10–70 5–3000 ns 355 and 356
400 (hole)
CdTeb 1015–1017 1100a (electron) 0.4–1.6 1–30 ns 357–360
100a (hole)
CIGSb 1015–1016 0.02–0.7 (hole) 0.56–5.6 10–100 ns 16, 361 and 362
OPVb 1018–1020 103–101 B0.1 — 363
MAPbI3a 1.4  1010 67.2 1.8–10 100–1000 ns 339, 341 and 364–372
MAPbBr3a 3.0  1011 24 1.3–4.3
MAPbCl3a 3.1  1010 42 3.0–8.5
MAPbI3(Cl)xa 1.8  109 167 —
MAPbI3x(Cl)xa — 167 1.07 (electron)
1.21 (hole)
MA0.45FA0.55PbI3a 2.6  109 271 254
CsPbBr3a 1.1  109 52 5.5
FAPbBr3a 9.6  109 62 10.5
FAPbI3a 1.13  1010 35 1.7
a b
Single crystal thin film. Polycrystalline thin film.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10047
 View Article Online

Chem Soc Rev Review Article

Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Fig. 27 (a) Composition dependent carrier concentration (circle) and carrier mobility (square) in perovskite films formed by one-step pre-mixed
precursor deposition (black) and the two-step interdiffusion method (blue). (b) and (c) Photocurrent curves (b) and the corresponding VOC and JSC (c) of
the perovskite devices calculated by the interdiffusion method. Reproduced from ref. 150 with permission from AIP Publishing, copyright 2014.
(d) A typical J–V curve with the maximum power point. (e) and (f) Dependence of FF on RS (e) and RSH (f). Reproduced from ref. 376 with permission from
Elsevier, copyright 2020.

layer, the conductive substrate, transport layers, and the metal During the development of Si technology, increasing the
back contact, as well as the interface resistance, such as current density contributes greatly to the efficiency improve-
Schottky barrier. The electronic conductivity of bulk and sur- ment at the preliminary stage. Afterwards, the efficiency is
face can be improved to reduce RS, thus to optimize the carrier improved through passivated emitter and rear cells, which
transport and to get a high FF. Under high internal electric field accounts for enhanced FF and VOC.293,378 And the development
conditions, RSH has a pronounced effect on the shape of J–V of the efficiency of PSCs appeared in the same trend (Table 12).
curves (Fig. 27(f)). It arises from the current leakage through the As JSC gradually approaches the SQ limit of JSC ( JSQ), the
device, such as the carrier recombination in bulk and interface. improvement of VOC and FF is the key to further enhancing
Electronic doping can minimize the recombination as well as the PCE of PSCs. For GaAs, the ratios of the achieved perfor-
reduce the space–charge buildup to increase the RSH value. mance parameters to the limit values of the S–Q limit are 88%
Moreover, carrier lifetime determines the charge extraction and for PCE/PCESQ, 99% for VOC/VSQ, 91% for JSC/JSQ and 97% for
the steepness of the J–V curve under low-internal field conditions FF/FFSQ.376,379 However, PSCs exhibit low FF and VOC ratios
or near VOC. Appropriate doping concentration can increase the compared to GaAs. The FF of most reported PSCs is around
carrier lifetime and thus increase FF. 0.7–0.8 and even certified devices in ‘‘Solar Cell Efficiency Tables’’

Table 12 Photovoltaic parameters of high-efficiency PSCs

Composition JSC (mA cm2) VOC (V) FF (%) PCE (%) Strategy Processing Organizationb (year) Ref.
a
MAPb(I1xBrx)3 19.64 1.11 74.2 16.2 Intermediate phase One-step KRICT (2013) 382
(FAPbI3)1x(MAPbBr3)x 21.84 1.11 73.6 17.9a Narrow bandgap One-step KRICT (2015) 383
(FAPbI3)12x(MAPbBr3)x 23.76 1.13 78.0 21.0a PMMA One-step EPFL (2016) 384
(FAPbI3)0.95(MAPbBr3)0.05 24.92 1.14 79.6 22.7a 2D+P3HT One-step KRICT/UNIST (2017) 380
FA0.92MA0.08PbI3 25.2 1.18 78.4 23.3a PEAI Two-step ISCAS (2018) 381
FAPbI3 25.6 1.15 80.1 23.5 MACl One-step UNIST (2018) 385
FAPbI3 26.1 1.15 79.0 23.7 MDACl2 One-step UNIST (2018) 386
FAPbI3 26.2 1.16 80.1 24.4 Cs+MDACl2 One-step UNIST (2019) 387
FAPbI3 26.2 1.18 79.6 24.6 Spiro-F One-step UNIST/KU (2019) 388
FAPbI3 26.25 1.17 81.8 25.2a FAHCOO One-step KU (2019) 389
FAPbI3 25.14 1.18 84.8 25.2a CBD SnO2+MAPbBr3 One-step KRICT/MIT (2019) 390
FAPbI3 25.74 1.19 83.2 25.5a FASnClx One-step UNIST (2020) 334
a
Certified value. ‘‘Solar Cell Efficiency Tables’’ published in Progress in Photovoltaics: Research and Applications. b Institutes: KRICT, Korea
Research Institute of Chemical Technology; EPFL, Swiss Federal Institute of Technology Lausanne; KRICT, Korea Research Institute of Chemical
Technology; UNIST: Ulsan National Institute of Science and Technology; ISCAS: Institute of Semiconductors of Chinese Academy of Sciences;
KU, Korea University; and MIT, Massachusetts Institute of Technology.

10048 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

with PCEs of 22.7% and 23.3% show low FF values less than 0.8 transition metal dichalcogenide semiconductors, has been adopted
(FF/FFSQ o 90%).380,381 Therefore, the optimization of perfor- in halide perovskites, which constructed highly stable and tun-
mance parameters VOC and FF will be the key factors for further able lateral epitaxial heterostructures, multiheterostructures and
PCE improvements. superlattices.393 These 2D heterostructures exhibited a suitable
The homojunction and heterojunction structures, which band alignment and good diode behaviors, with a rectification
could minimize the band mismatch and increase the built-in ratio of around 102 without hysteresis. These findings and
electric field at the interface to accelerate the carrier transfer, methodologies may inspire further developments of opto-
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

are the major technologies for a high efficiency device in the electronics for perovskites and other hybrid materials.
commercial success of solar cells, especially for VOC and FF Homojunction structure can be constructed for high effi-
enhancement.5,293 Inspired by commercial success of solar cell ciency PSCs by compositional regulation. The MAPbI3 films
technologies, heterojunctions were also constructed in PSCs with MAI rich precursors are n-doped and those with PbI2 rich
by electronic doping to promote VOC and FF of the devices. precursors are p-doped.150 The homojunction structure com-
Trivalent cations, such as Sb3+ and In3+, were spontaneously bining a thermally evaporated p-type perovskite layer on a
distributed in the crystal with concentration gradient due to the solution-processed n-type perovskite layer was integrated in a
difference in the ionic size and electrostatic interaction between planar PSC, which formed a built-in electric field and promoted
the dopants and host atoms, as shown in Fig. 28(a).391 A hetero- the oriented transport of the photo-induced carriers (Fig. 28(d)
junction perovskite layer with well-matched band-edge to the and (e)).152 As a result, the homojunction PSC showed a better
ETL was formed based on solution manipulation, which yielded performance than the standard p–i–n PSC with an increased
a high FF of 0.84 and a PCE of 21.04%. In addition, Pb–Sn PCE from 17.76% to 20.80%. Furthermore, through a robust
gradient doping was achieved by feeding a continuous flow p–n homojunction, a simplified PSC in the absence of both ETL
of precursors of different compositions (Fig. 28(b) and (c)).392 and HTL was constructed, yielding a PCE of over 8% without
Surprisingly, the graded MAPb0.5+xSn0.5xI3 exhibited a relatively using any interfacial modification layers. In addition, capsaicin
high VOC and showed the best overall performance among the was applied as an additive in perovskite precursors, which
three absorbers (MAPbI3, MAPb0.5Sn0.5I3 and MAPb0.5Sn0.5I3). realized the defect passivation and energetic modification for
The graded bandgap of the MAPb0.5+xSn0.5xI3, which was perovskites at the same time, making a complete transforma-
similar to the built-in field of a p–n junction, facilitated carrier tion of the MAPbI3 surface from p- to n-type (Fig. 28(f)).394 After
separation, transport and collection, thus improving the carrier defect passivation by capsaicin, the p–n homogeneous junction
lifetime. Furthermore, epitaxial growth, an unique material (B100 nm) spontaneously formed, resulting in a high effi-
deposition strategy for oxide perovskites and III–V, II–VI and ciency of 21.88% and an excellent FF of 83.81%.

Fig. 28 (a) Reconstructed 3D images and elemental 3D maps of Sb3+ doped and In3+ doped devices traced in the depth profile. Reproduced from
ref. 391 with permission from John Wiley & Sons, copyright 2019. (b) Scanning electron microscope images and the corresponding electron-beam-
induced current mapping of cross-sections of three different single-crystal perovskites: MAPbI3 (top), graded MAPb0.5+xSn0.5xI3 (middle) and
MAPb0.5Sn0.5I3 (bottom). (c) A PL image (left) of the graded single-crystal MAPb0.5+xSn0.5xI3 grown on a MAPbBr3 substrate. Reproduced from
ref. 392 with permission from Springer Nature, copyright 2020. (d) and (e) Carrier generation and transport in PSCs without (d) and with (e) the
homojunction. Reproduced from ref. 152 with permission from Springer Nature, copyright 2019. (f) Energy level alignment schematics of the PSC
comprising capsaicin. Reproduced from ref. 394 with permission from Elsevier, copyright 2021.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10049
 View Article Online

Chem Soc Rev Review Article

6.3.2 Device stability. The device stability of solar cells DFT calculation showed that perovskites were chemical
depends on two aspects: intrinsic material stability and extrin- unstable, whose enthalpy of formation (with respect to MAI
sic stability against environment. The encapsulation of solar and PbI2) can be negligible.206 Intrinsic degradation would
cells can prevent moisture and oxygen invasion effectively and occur in perovskite materials upon ageing treatments such as
some external encapsulation materials have already been heat, oxygen, light and moisture. On one hand, Pb2+ can be
applied to solar cells for commericilization.395 However, the easily reduced to metallic Pb0 upon heating and/or illumina-
intrinsic material stability, which is affected by complex factors tion, which was probed in the halide perovskite film by
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

such as the phase and structure stability, chemical stability, XPS.400–402 Besides, the nonstoichiometric composition and/
and ion migration, is difficult to improve significantly, severely or the asynchronous degradation of halide perovskites could
limiting the further enhancement of long-term stability and leave undercoordinated Pb2+ ions on the film surface and grain
leading to the reduction of EPBT of solar cells, especially for boundaries, which had higher reduction or oxidation (redox)
PSCs.40,396 activity.280 On the other hand, I was prone to be oxidized to I0,
Defect density of PV materials has huge influence on the which triggered the chain reactions and accelerated the degra-
intrinsic device stability. Material degradation is generally dation of perovskite films.403 Sacrificial agents such as HPA,
initialized from defect sites, which are more reactive with water EtOH and BQ were used to eliminate either Pb0 or I0
and oxygen.397–399 In addition, defects are channels for water defects.402,404,405 However, these additives diminished soon
and oxygen capture and rapid diffusion, which further accel- after taking effects. To solve this problem, the ‘‘redox shuttle’’
erates the degradation of bulk materials.395 For covalent-bond concept was proposed to constantly eliminate Pb0 and I0
based PV materials like silicon, structural defects are usually simultaneously in PSCs over their life span by using the ion
electrically neutral. Therefore, eliminating structural defects pair Eu3+–Eu2+ on the basis of the following chemical reactions
mainly involves a chemical reaction with primary chemical (Fig. 29(a)):406
bonding. For Si solar cells, structural defects are usually
2Eu3+ + Pb0 - 2Eu2+ + Pb2+ (49)
eliminated by forming chemically stable oxides and nitrides,
i.e., SiO2, SiNx, or AlOx, through thermal oxidization and/or Eu2+ + I0 - Eu3+ + I. (50)
chemical vapor deposition.127 Benefiting from the ionic nature
of new-generation PV materials, especially for perovskites, the
structural defects are generally charged defects, which enable Although the reaction is thermodynamically favored with
multifarious elimination methods including redox shuttle and a Gibbs formation energy (DG) of 173.6 kJ mol1, metallic
chemical bond enhancement.280 Pb and I2 powder only led to limited formation of PbI2.

Fig. 29 (a) Diagram of the proposed mechanism of cyclic elimination of Pb0 and I0 defects and regeneration of Eu3+–Eu2+ metal ion pairs. Reproduced
from ref. 406 with permission from Science Publishing Group, copyright 2019. (b) A schematic illustration of enhancing the hydrogen bond between the
halogen and MA/FA ions, and strengthening the ionic bond between the halogen and metal ions through increasing the electronegativity of halogen.
Reproduced from ref. 407 with permission from Springer Nature, copyright 2019. (c) Double-layered halide architecture (DHA) of P3HT-based perovskite
solar cells. Left, the structure of an n–i–p perovskite solar cell based on a DHA using P3HT as the hole-transport material. Right, schematic structure of
the interface between the WBH and P3HT. Reproduced from ref. 380 with permission from Springer Nature, copyright 2019. (d) Schematic illustration of
the protection of perovskites through in situ formation of a lead sulfate top layer on the perovskite surface. Reproduced from ref. 127 with permission
from Science Publishing Group, copyright 2019.

10050 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Interestingly, this reaction can be easily triggered by applying migration issues, methods have been developed including
ion pairs with appropriate redox potentials. Eu3+ could easily alkali metal ion doping, introducing functionalized additives,
oxidize the Pb0 defects and Eu2+ could reduce the I0 defects, adjusting perovskite components, and forming a passivation
which was the best candidate of ion pairs. The resultant device layer on the perovskite surface, which is described in detail in
(unencapsulated) retained 91% of the original PCE after 500 h Section 6.1.2.
of MPP tracking. In addition, the devices maintained 89% and The band mismatch and defect states at the perovskite/
92% of the original PCE after heating at 85 1C or under 1 sun transport interfaces would cause carrier accumulation and
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

continuous illumination for 1500 h, respectively. undesired nonradiative recombination, which damaged the
Moreover, chemical bond enhancement, incorporating the long-term stability of the device, especially for MPP tracking.419–421
dopants with strong interactions with the perovskite crystal The Lewis acid–base theory is widely adopted for chemical
lattice, can immobilize ions to reduce intrinsic defects and passivation and doping in perovskite surface. Lewis base-based
increase chemical stability. Fluoride ions with extremely high deep traps, such as the undercoordinated I and antisite PbI3
electronegativity were introduced into the perovskite film, can be eliminated by the Lewis acid, such as C60 and its
which formed strong ionic bonds with Pb2+ and strong hydro- derivatives (PCBM, ICBA, etc.) with the capability to accept a
gen bonds (N–H  F) with organic cations (MA/FA), leading to lone pair of electrons.422–424 Similarly, Lewis acid-based traps,
the reduction of both the organic cation and halide anion such as the undercoordinated Pb2+ and Pb clusters, can be
vacancies (Fig. 29(b)).407 The devices maintained 90% of the eliminated by the Lewis base, such as the molecules containing
original PCE after 1000 h of MPP tracking under continuous functional groups like nitrogen (N) or sulphur (S) with a pair of
one sun illumination. In addition, the devices retained 90% nonbonding electrons.425–428
and 95% of the original PCE after heating at 85 1C or under The charged nature of the defects in the perovskite surface
1 sun continuous illumination for 1000 h, respectively. enables the neutralization through ionic bonding, such as
Recently, fluorine-containing molecules and organic salts, such conversion of surface imperfections to wide bandgap materials.
as FEAI and amphiphilic TFMBAI, had demonstrated great PbI2 is a wide bandgap material with a band gap of 0.8 eV
improvement on the stability of PSCs.304,408,409 greater than that of MAPbI3. After annealing at 150 1C for
Under electric field, illumination, and thermal treatment, 60 min, enriched PbI2 formed at the perovskite/TiO2 interface,
the charged defects migrate, which could lead to the J–V leading to n-doping and suppressing the recombination at the
hysteresis, phase segregation, band bending or even material surface.429 Besides, the formation of a layered 2D perovskite at
degradation.410–414 When an electric field was applied to the the 3D perovskite surface can create type-I band alignment,
perovskites, the negatively defects, i.e., VMA and Ii, migrated which is another effective approach to improve the stability.
along the opposite direction of the electric field. At the Through in situ reaction between n-hexyl trimethyl ammonium
same time, the positively defects, i.e., VI and MAi, migrated in bromide and the perovskite, a WBH perovskite was established
the direction of the electric field, which caused the charge on the NBH perovskite layer (Fig. 29(c)).380 The modified
accumulation in HTL/perovskite/CTL interface and uninten- devices without any encapsulation kept nearly 80% of its
tional doping effect.413,415 According to the electric neutrality original efficiency after 1008 h under 85% RH at room tem-
principle, the negative charged defects migrated and accumu- perature. The operational stability of modified devices signifi-
lated in the perovskite surface which introduced equivalent cantly increased, where the modified devices maintained over
numbers of holes to chemically p-dope the interface. Similarly, 95% of their initial PCE for 1370 h and the control devices
the migration of negative defects in the perovskite surface led quickly degraded within 100 h. Other chemicals, such as
to a chemically n-doped perovskite surface. The chemical butylammonium (BA+), PEA+, or AVAI were reported to form
doping at the interface led to the generation of an additional different layered perovskites at the interface.430–432
electric field in the perovskite absorber that compensated Due to the weak secondary bonding between the passivation
the built-in electric field. It hindered the photogenerated molecules and the perovskite surface, inorganic anions, which
carrier transport, which was harmful for device efficiency can form strong chemical bonds with the perovskite, are
and stability.318,319 In addition, the migrating ions have the investigated. Similar to the Si passivation strategies, the chemi-
potential to react with the transport layer or metal electrode. cally stable lead oxysalt layer can be established on the per-
It was proved that the oxidized spiro-OMeTAD+ can chemically ovskite surface, which can provide sufficient protection for
react with the mobile I to form an electroneutral spiro- extreme conditions such as thermal ageing and humidity aging
MeOTAD–I complex under light illumination or thermal stress (Fig. 29(d)).127 Considering the operation conditions of PVs,
(85 1C aging).416–418 Besides, the chemical reaction between the lead salts that are insoluble in water are ideal candidates.
mobile I and metal electrodes, such as silver, was observed in PbSO4 and Pb3(PO4)2, whose solubility is merely 0.0443 and
the experiment.414 After aging the MAPbI3-based device, silver 0.00014 g L1, respectively, at 20 1C, were formed through the
electrodes would change from white to yellow due to the surface reaction by using (C8H17NH3)2SO4 and (C8H17NH3)3PO4
formation of AgI, which originated from the chemical reaction salts. Devices with the lead oxysalt layers (encapsulation)
between the migrated MAI and Ag. The chemical reactions are retained 96.8% of their original efficiency after 1200 MPP tracking
also observed in the perovskite/TiO2 interface, which may form under 1 sun continuous illumination at 65 1C, whereas the control
Ti–I–Pb bonds between TiO2 and MAPbI3.416 To solve these ion device showed severe degradation within 474 h. Recently, this

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10051
 View Article Online

Chem Soc Rev Review Article

strategy was applied to inverted PSCs, where the PbPyA2 and TMS homojunction and gradient heterojunction is effective to opti-
were introduced to form strong lead–sulfur bonds (Pb–S) that not mize carrier transport and improve the efficiency in silicon
only are chemically stable, with an equilibrium constant (Ksp) solar cells, and they are expected to be feasible for the emerging
of 1.0  1028, but also upshifted the EF and offered a BSF for PVs. Indeed, p–n homojunctions and gradient heterojunction
electron extraction.285 In addition to the above lead compounds, constructed by the solution process have been reported in OPV
other lead compound with low solubility (Ksp), such as PbCO3 and PSC.24,152,391,392,394 However, facile ion migration and
(1.46  1013), PbO (1.26  1015), Pb5(PO4)3Cl (3.98  1085), defect interdiffusion in these PV materials remain a problem
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Pb5(PO4)3Br (7.94  1079) and Pb5(PO4)3F (2.51  1072), await to for the stability and effectiveness of the homojunction and
be developed and investigated. heterojunction structures, which need further research.51
Electronic doping at the surface is an effective strategy
which can be accomplished by facile in situ chemical/wet
7. Conclusions and outlook processing. It could modify the electrode–semiconductor
interfaces for efficient carrier extraction, which is potentially
Electronic doping is crucial to modulate the electrical proper- feasible for perovskite solar cells without ETL or HTL.280,366
ties of photovoltaic materials. As the material composition and/ However, for new-generation PV materials, like halide perovs-
or molecular structure extend beyond conventional materials kites, it is likely to react with metal electrodes, thus the
systems, like the new-generation PV materials, the classical investigation of inert conducting materials to retard their
understanding of electronic doping based on the crystalline corrosion is highly demanded.127,285,414
silicon cannot be directly applied. In some cases, these arguments Furthermore, self-compensation is a common phenomenon
are even controversial which require further investigation. New in new semiconductors with a soft lattice due to their low lattice
theories and techniques need to be developed to investigate formation energy, which limits the concentration of electronic
electronic doping in new PV materials. carriers.25,51,150 Non-equilibrium growth, such as quenching
In this review, we have first introduced conventional electronic and rapid solvent evaporation, which has been widely used in
doping (atomic doping using physical approaches mainly in CdTe, may be applicable to overcome the limitations of self-
crystalline silicon). We emphasize that physicochemical pro- compensation.12
perties of atoms such as the valence electron, electronegativity, New electronic models and assumptions for DFT calcula-
radius, and the redox properties determine the doping type. tions need to be developed to interpret experiment results,
Subsequently, we turned to chemical approaches, which are which are important to guide the processing for effective
developed to tackle with the new-generation of PV materials electronic doping in new-generation PVs. For atomic dopants,
featuring weak chemical bonding and a soft lattice. New self-compensating effects and defect complexes are worth care-
electronic doping strategies (molecular doping and self- ful consideration. Also, the location of impurity atoms in the
doping) have been summarized with the focus on chemical crystal lattice affects the doping significantly, and precise
approaches. We concluded that by adjusting the HOMO or location control is desired. For molecular dopants, the dopant
LUMO levels of dopant molecules (molecular doping) and solubility, configuration, and size can affect charge delocalization
the intrinsic defects in polycrystalline semiconductor films and disorder. To further control electronic doping, the develop-
(self-doping), controllable electronic doping can be achieved ment of new electronic models and calculation methods are
in new-generation PV materials. required to compute charge and non-localized systems more
In addition, the effects of electronic doping are discussed efficiently. At the same time, time-dependent defect model
and summarized. Electronic doping can bring slight or steep instead of static defect model must be developed to simulate
shifts in electronic band structure across semiconductor bulks influences caused by dynamic effects such as aggregation, diffu-
and interfaces. It thus modulates carrier concentration, con- sion, and drift.
ductivity, and thus carrier dynamics, which further improves The new-generation solar cells, especially for PSCs, have
device performance. To assess the effects of electronic doping shown rapid development and have the potential to realize PV
quantitatively for new-generation solar cells, new models based grid parity. However, the emerging solar cells face significant
on the physicochemical properties of individual dopants and challenges. In particular, the fundamental understanding is
the formation and transformation of intrinsic defects are still quite limited, which is far behind the development in
proposed in this review, which can help to better control the device efficiency. On one hand, novel materials and doping
carrier concentration of semiconductors and guide the design methods need to be developed to facilitate the further com-
of devices. mercialization of new-generation solar cells. On the other hand,
Finally, we have proposed a few parameters of essence that theories applicable for new materials are urgently required
play an important role in different doping strategies, including to design and conduct effective electronic doping. It is impor-
the chemical properties of materials, defect density, processing tant that the theoretical and experimental investigators, and
conditions, etc. academic and industrial researchers work closely together to
Reviewing the development of electronic doping, the tech- have a comprehensive consideration of electronic doping, and
nologies in conventional semiconductors may shed light on the develop more realistic and a reliable process for new-generation
design and optimization of new solar cells. Constructing p–n or next-generation solar cells.

10052 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

Abbreviations TP Tea polyphenol

IE Ionization energy
PVs Photovoltaics EA Electron affinity
c-Si Crystalline silicon CTCs Charge transfer complexes
PCEs Power conversion efficiencies TCNQ Tetracyanoquinodimethane
BSF Back surface field F2TCNQ 2,5-Difluoro-7,7,8,8-
QDSC Quantum dot solar cell tetracyanoquinodimethane
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

OPV Organic photovoltaic solar cells TMS Hexamethyldisilathiane

PSC Perovskite solar cell F2HCNQ 2,5-Difluoro-3,6,7,7,8,8-
i Interstitial hexacyanoquinodimethane
V Vacancy F6TCNNQ 2,2 0 -(Perfluoronaphthalene-2,6-
MA Methylammonium diylidene)dimalononitrile
FA Formamidinium DDQ 2,3-Dichloro-5,6-dicyanobenzoquinone
Ni(tfd)2 Nickel bis(1,2-bis(trifluoromethyl)ethane- HAT-CN 1,4,5,8,9,11-
1,2-dithiolene) Hexaazatriphenylenehexacarbonitrile
F4TCNQ 2,3,5,6-Tetrafluoro-7,7,8,8- CN6-CP Hexacyano-trimethylene-cyclopropane
tetracyanoquinodimethane PDPP(6-DO)2TT Poly[3,6-(dithiophene-2-yl)-2,5-di(6-dode-
Mo(tfd)3 Molybdenum tris(1,2-bis(trifluoromethyl) cyloctadecyl)-pyrrolo[3,4-c]pyrrole-1,4-dione-
ethane-1,2-dithiolene) alt-thieno[3,2-b]thiophene]
EPBT Energy payback time Mo(tfd-CO2Me)3 Molybdenum tris(1-(methoxycarbonyl)-2-(tri-
VB Valence band fluoromethyl)ethane-1,2-dithiolene)
CB Conduction band TMCN3-CP Trimethyl 2,2 0 ,200 -(cyclopropane-1,2,3-
HOMO Highest occupied molecular orbital triylidene)-tris(cyanoacetate)
LUMO Lowest unoccupied molecular orbital CoCp*2 Decamethylcobaltocene
SQ limit Shockley–Queisser limit ITO Indium tin oxide
HVPE Hydride vapor phase epitaxy HTL Hole transport layer
MBE Molecular beam epitaxy DMBI 1,3-Dimethyl-2-phenyl-2,3-dihydro-1H-
DFT Density functional theory benzoimidazole
EF Fermi level QDs Quantum dots
JSC Short-circuit current density SAMs Self-assembled monolayer
KPFM Kelvin probe force microscopy CF3-SAM 4-Trifluoromethyl-benzenethiol
VOC Open circuit voltage CH3-SAM 4-Methyl-benzenethiol
E0 Standard redox potential CuPc Copper(II) phthalocyanine
XRD X-ray diffraction FTS (Tridecafluoro-1,1,2,2-
XPS X-ray photoelectron spectroscopy tetrahydrooctyl)trichlorosilane
AR Atomic radius OTS N-Octyltrichlorosilane
EN Electronegativity F16CuPc Copper-hexadecafluoro-phthalocyanine
IR Ionic radius 2D Two-dimension
ED The energy of the donor levels FDC-2-5Cl Benzopentafulvalenes compound
EA The energy of the acceptor levels HZ-TIPS Heptazethrene
VTe Te vacancy OZ-TIPS Octazethrene
VCd Cd vacancy b-PEI Branched polyethyle-neimine
VCu Cu vacancy PVP Polyvinylpyrrolidone
VSe Se vacancy PAA Polyacrylic acid
CZTS Cu2ZnSnS4 ETL Electron transfer layer
CZTSe Cu2ZnSnSe4 PbPyA2 Pyridine-2-carboxylic lead
CZTSSe Cu2ZnSn(S,Se)4 TMS Hexamethyldisilathiane
SPV Surface photovoltage SST Surface sulfidation treatment
MBO Mayer bond order MOSFET Metal-oxide–semiconductor field-effect
2-ME 2-Methoxyethanol transistor
TMDHP 1,4-Bis-(trimethylsilyl)-2,3,5,6-tetramethyl- MOCVD Metal–organic chemical vapor deposition
1,4-dihydropyrazine LEDs Light-emitting diodes
AHP Ammonium hypophosphite RS Series
HPA Hypophosphorous acid RSH Shunt
TA Transient absorption PMMA Poly(methyl methacrylate)
AA Ascorbic acid P3HT Poly(3-hexylthiophene)

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10053
 View Article Online

Chem Soc Rev Review Article

PEAI Phenethylammonium iodide 9 J. E. Northrup and S. Zhang, Phys. Rev. B: Condens. Matter
MACl Methylammonium chloride Mater. Phys., 1993, 47, 6791.
Spiro-F Fluorinated isomeric analogs of 10 R. Ahrenkiel, D. Dunlavy, J. Benner, R. Gale, R. McClelland,
Spiro-OMeTAD J. Gormley and B. King, Appl. Phys. Lett., 1988, 53, 598–599.
MDACl2 Methylenediammonium dichloride 11 J.-H. Yang, W.-J. Yin, J.-S. Park, W. Metzger and S.-H. Wei,
FAHCOO Formamidine formate J. Appl. Phys., 2016, 119, 045104.
CBD Chemical bath deposition 12 J.-H. Yang, J.-S. Park, J. Kang, W. Metzger, T. Barnes and
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

MAPbBr3 Methylammonium lead bromide S.-H. Wei, Phys. Rev. B: Condens. Matter Mater. Phys., 2014,
HPA Hypophosphorous acid 90, 245202.
EtOH Ethanol 13 D. W. Niles, K. Ramanathan, F. Hasoon, R. Noufi, B. J.
BQ Benzoquinone Tielsch and J. E. Fulghum, J. Vac. Sci. Technol., A, 1997, 15,
MPP Maximum power point 3044–3049.
FEAI 2-(2,3,4,5,6- 14 F. Pianezzi, P. Reinhard, A. Chirilă, B. Bissig, S. Nishiwaki,
Pentafluorophenyl)ethylammonium iodide S. Buecheler and A. N. Tiwari, Phys. Chem. Chem. Phys.,
TFMBAI Amphiphilic 4-(trifluoromethyl)benzylam- 2014, 16, 8843–8851.
monium iodide 15 M. Lammer, U. Klemm and M. Powalla, Thin Solid Films,
C60 Fullerene 2001, 387, 33–36.
WBH Wide-bandgap halide 16 A. Redinger, S. Levcenko, C. J. Hages, D. Greiner, C. A.
NBH Narrow-bandgap halide Kaufmann and T. Unold, Appl. Phys. Lett., 2017, 110,
AVAI Aminovaleric acid iodide 122104.
PbPyA2 Pyridine-2-carboxylic lead 17 J. Pohl and K. Albe, Phys. Rev. B: Condens. Matter Mater.
AVC anti-solvent vapor-assisted crystallization Phys., 2013, 87, 245203.
method 18 S.-H. Wei and S. Zhang, J. Phys. Chem. Solids, 2005, 66,
CIGS Cu(In,Ga)(S,Se)2 1994–1999.
CPD Kelvin-probe contact potential difference 19 S. Chen, J.-H. Yang, X.-G. Gong, A. Walsh and S.-H. Wei,
CRPH carrier-resolved photo-Hall Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 81, 245204.
ITC inverse temperature crystallization method 20 A. Nagoya, R. Asahi, R. Wahl and G. Kresse, Phys. Rev. B:
M-S Mott-Schottky analysis Condens. Matter Mater. Phys., 2010, 81, 113202.
PYS photoelectron yield spectroscopy 21 A. R. Kirmani, A. Kiani, M. M. Said, O. Voznyy, N. Wehbe,
STS scanning tunneling spectroscopy G. Walters, S. Barlow, E. H. Sargent, S. R. Marder and
ToF time-of-flight technique A. Amassian, ACS Energy Lett., 2016, 1, 922–930.
UPS ultraviolet photoelectron spectroscopy 22 M. L. Tietze, P. Pahner, K. Schmidt, K. Leo and B. Lüssem,
Adv. Funct. Mater., 2015, 25, 2701–2707.
23 J. Gao, B. W. Stein, A. K. Thomas, J. A. Garcia, J. Yang,
Conflicts of interest M. L. Kirk and J. K. Grey, J. Phys. Chem. C, 2015, 119,
16396–16402.
There are no conflicts to declare.
24 I. E. Jacobs and A. J. Moulé, Adv. Mater., 2017, 29, 1703063.
25 W.-J. Yin, T. Shi and Y. Yan, Appl. Phys. Lett., 2014,
References 104, 063903.
26 A. L. Abdelhady, M. I. Saidaminov, B. Murali, V. Adinolfi,
1 N. Kannan and D. Vakeesan, Renewable Sustainable Energy O. Voznyy, K. Katsiev, E. Alarousu, R. Comin, I. Dursun and
Rev., 2016, 62, 1092–1105. L. Sinatra, J. Phys. Chem. Lett., 2016, 7, 295–301.
2 C. J. Rhodes, Sci. Prog., 2010, 93, 37–112. 27 Y. Chen, H. Jing, F. Ling, W. Kang, T. Zhou, X. Liu,
3 R. McKee, F. Walker, M. B. Nardelli, W. Shelton and W. Zeng, Y. Zhang, L. Qi and L. Fang, Chem. Phys. Lett.,
G. Stocks, Science, 2003, 300, 1726–1730. 2019, 722, 90–95.
4 O. Madani Ghahfarokhi, K. von Maydell and C. Agert, Appl. 28 Q. Chen, L. Chen, F. Ye, T. Zhao, F. Tang, A. Rajagopal,
Phys. Lett., 2014, 104, 113901. Z. Jiang, S. Jiang, A. K.-Y. Jen and Y. Xie, Nano Lett., 2017,
5 T. G. Allen, J. Bullock, X. Yang, A. Javey and S. De Wolf, Nat. 17, 3231–3237.
Energy, 2019, 4, 914–928. 29 W.-Q. Wu, Q. Wang, Y. Fang, Y. Shao, S. Tang, Y. Deng,
6 C. Battaglia, A. Cuevas and S. De Wolf, Energy Environ. Sci., H. Lu, Y. Liu, T. Li and Z. Yang, Nat. Commun., 2018, 9, 1–8.
2016, 9, 1552–1576. 30 M. M. de Wild-Scholten, Sol. Energy Mater. Sol. Cells, 2013,
7 P. Śpiewak and K. J. Kurzydłowski, Phys. Rev. B: Condens. 119, 296–305.
Matter Mater. Phys., 2013, 88, 195204. 31 E. H. Sargent, Nat. Photonics, 2009, 3, 325–331.
8 E. Poindexter, G. Gerardi, M. E. Rueckel, P. Caplan, 32 M. Yoshimura, J. Mater. Res., 1998, 13, 796–802.
N. Johnson and D. Biegelsen, J. Appl. Phys., 1984, 56, 33 Y. Vaynzof, Adv. Energy Mater., 2020, 10, 2003073.
2844–2849. 34 J. Williams, Mater. Sci. Eng., A, 1998, 253, 8–15.

10054 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

35 L. A. Larson, J. M. Williams and M. I. Current, Reviews Of 59 A. M. Schimpf, S. T. Ochsenbein, R. Buonsanti, D. J.

Accelerator Science And Technology: Volume 4: Accelerator Milliron and D. R. Gamelin, Chem. Commun., 2012, 48,
Applications in Industry and the Environment, 2011, pp. 11–40. 9352–9354.
36 A. Kausar, A. Sattar, C. Xu, S. Zhang, Z. Kang and Y. Zhang, 60 T. Norby, J. Korean Ceram. Soc., 2010, 47, 19–25.
Chem. Soc. Rev., 2021, 50, 2696–2736. 61 P. M. Fahey, P. Griffin and J. Plummer, Rev. Mod. Phys.,
37 J. T.-W. Wang, Z. Wang, S. Pathak, W. Zhang, D. W. 1989, 61, 289.
deQuilettes, F. Wisnivesky-Rocca-Rivarola, J. Huang, 62 W. R. Thurber, The relationship between resistivity and
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

P. K. Nayak, J. B. Patel and H. A. M. Yusof, Energy Environ. dopant density for phosphorus-and boron-doped silicon, US
Sci., 2016, 9, 2892–2901. Department of Commerce, National Bureau of Standards,
38 C. Zhang, Y. Wang, X. Lin, T. Wu, Q. Han, Y. Zhang and 1981.
L. Han, J. Mater. Chem. A, 2021, 9, 1372–1394. 63 M. Pashley and K. Haberern, Phys. Rev. Lett., 1991, 67,
39 Y. Chen and H. Zhou, J. Appl. Phys., 2020, 128, 060903. 2697.
40 N. Li, X. Niu, Q. Chen and H. Zhou, Chem. Soc. Rev., 2020, 64 X. Tang, H. Lochs, P. Hageman, M. De Croon, L. Giling and
49, 8235–8286. A. Bons, J. Cryst. Growth., 1989, 98, 827–837.
41 Y. Feng, J. Wu, Q. Chi, W. Li, Y. Yu and W. Fei, Chem. Rev., 65 J. Neave, P. Dobson, J. Harris, P. Dawson and B. Joyce, Appl.
2020, 120, 1710–1787. Phys. A: Solids Surf., 1983, 32, 195–200.
42 Y. Han, H. Zhao, C. Duan, S. Yang, Z. Yang, Z. Liu and 66 V. G. Dubrovskii, H. Hijazi, N. I. Goktas and R. R. LaPierre,
S. Liu, Adv. Funct. Mater., 2020, 30, 1909972. J. Phys. Chem. C, 2020, 124, 17299–17307.
43 L. Loh, Z. Zhang, M. Bosman and G. Eda, Nano Res., 2021, 67 A. Cho, J. Appl. Phys., 1975, 46, 1733–1735.
14, 1668–1681. 68 F. Yang, M. E. Messing, K. Mergenthaler, M. Ghasemi,
44 P. P. Altermatt, J. O. Schumacher, A. Cuevas, M. J. Kerr, J. Johansson, L. R. Wallenberg, M.-E. Pistol, K. Deppert,
S. W. Glunz, R. R. King, G. Heiser and A. Schenk, J. Appl. L. Samuelson and M. H. Magnusson, J. Cryst. Growth.,
Phys., 2002, 92, 3187–3197. 2015, 414, 181–186.
45 S. Miao, K. Liang, J. Zhu, B. Yang, D. Zhao and B. Kong, 69 J. Pautrat, J. Francou, N. Magnea, E. Molva and
Nano Today, 2020, 33, 100879. K. Saminadayar, J. Cryst. Growth., 1985, 72, 194–204.
46 J. Zhang, R.-J. Zhou, Q.-Y. Chang, Z.-J. Sui, X.-G. Zhou, 70 M. Said and M. Kanehisa, J. Cryst. Growth., 1990, 101,
D. Chen and Y.-A. Zhu, Catal. Today, 2021, 368, 46–57. 488–492.
47 C. Yang, H. Wang, Y. Miao, C. Chen, M. Zhai, Q. Bao, 71 J. D. Poplawsky, N. R. Paudel, C. Li, C. M. Parish,
X. Ding, X. Yang and M. Cheng, ACS Energy Lett., 2021, 6, D. Leonard, Y. Yan and S. J. Pennycook, Adv. Energy Mater.,
2690–2696. 2014, 4, 1400454.
48 Q. Jiang, Z. Ni, G. Xu, Y. Lin, P. N. Rudd, R. Xue, Y. Li, Y. Li, 72 J.-H. Yang, W.-J. Yin, J.-S. Park, J. Ma and S.-H. Wei,
Y. Gao and J. Huang, Adv. Mater., 2020, 32, 2001581. Semicond. Sci. Technol., 2016, 31, 083002.
49 Y. Deng, Z. Ni, A. F. Palmstrom, J. Zhao, S. Xu, C. H. Van 73 L. Kranz, C. Gretener, J. Perrenoud, R. Schmitt, F. Pianezzi,
Brackle, X. Xiao, K. Zhu and J. Huang, Joule, 2020, 4, F. La Mattina, P. Blösch, E. Cheah, A. Chirilă and
1949–1960. C. M. Fella, Nat. Commun., 2013, 4, 1–7.
50 S. H. Cho, J. Byeon, K. Jeong, J. Hwang, H. Lee, J. Jang, 74 J. M. Burst, J. N. Duenow, D. S. Albin, E. Colegrove,
J. Lee, T. Kim, K. Kim and M. Choi, Adv. Energy Mater., M. O. Reese, J. A. Aguiar, C.-S. Jiang, M. Patel, M. M. Al-
2021, 11, 2100555. Jassim and D. Kuciauskas, Nat. Energy, 2016, 1, 1–8.
51 J. Euvrard, Y. Yan and D. B. Mitzi, Nat. Rev. Mater., 2021, 6, 75 J.-H. Yang, W.-J. Yin, J.-S. Park, J. Burst, W. K. Metzger,
531–549. T. Gessert, T. Barnes and S.-H. Wei, J. Appl. Phys., 2015,
52 Q. Zhang, J. Hou, J. Fan, S. Chen, W. Fan, H. Zhang, 118, 025102.
W. Wang, Y. Wu and B. Xu, Phys. Chem. Chem. Phys., 76 S. A. Razi, N. K. Das, S. F. U. Farhad and M. A. M. Bhuiyan,
2020, 22, 7012–7020. Int. J. Renewable Energy Res., 2020, 10, 1012–1020.
53 A. Zunger and O. I. Malyi, Chem. Rev., 2021, 121, 77 C. Li, Y. Wu, J. Poplawsky, T. J. Pennycook, N. Paudel,
3031–3060. W. Yin, S. J. Haigh, M. P. Oxley, A. R. Lupini and M. Al-
54 A. M. Schimpf, K. E. Knowles, G. M. Carroll and D. R. Jassim, Phys. Rev. Lett., 2014, 112, 156103.
Gamelin, Acc. Chem. Res., 2015, 48, 1929–1937. 78 S. Patel, H. Pandey, A. Purohit, S. Chander, M. Kannan and
55 D. J. Rowe, J. S. Jeong, K. A. Mkhoyan and U. R. M. Dhaka, AIP Conf. Proc., 2020, 2265, 030331.
Kortshagen, Nano Lett., 2013, 13, 1317–1322. 79 M. Malitckaya, H.-P. Komsa, V. Havu and M. Puska, J. Phys.
56 J. M. Luther, P. K. Jain, T. Ewers and A. P. Alivisatos, Nat. Chem. C, 2017, 121, 15516–15528.
Mater., 2011, 10, 361–366. 80 K. Kim, I. Jeong, Y. Cho, D. Shin, S. Song, S. K. Ahn,
57 D. Dorfs, T. Härtling, K. Miszta, N. C. Bigall, M. R. Kim, Y.-J. Eo, A. Cho, C. Jung and W. Jo, Nano Energy, 2020,
A. Genovese, A. Falqui, M. Povia and L. Manna, J. Am. 67, 104201.
Chem. Soc., 2011, 133, 11175–11180. 81 W. Wang, C. Zhang, B. Hu, W. Su, S. Xu, M. Ma, Y. Feng,
58 A. Schimpf, E. Runnerstrom, S. Lounis, D. Milliron and W. Li, M. Chen and C. Yang, Mater. Lett., 2021, 302,
D. Gamelin, J. Am. Chem. Soc., 2015, 137, 518–524. 130410.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10055
 View Article Online

Chem Soc Rev Review Article

82 R. Prasanna, A. Gold-Parker, T. Leijtens, B. Conings, 106 N. K. Noel, S. D. Stranks, A. Abate, C. Wehrenfennig,

A. Babayigit, H.-G. Boyen, M. F. Toney and M. D. S. Guarnera, A.-A. Haghighirad, A. Sadhanala, G. E.
McGehee, J. Am. Chem. Soc., 2017, 139, 11117–11124. Eperon, S. K. Pathak and M. B. Johnston, Energy Environ.
83 M. Abdi-Jalebi, M. I. Dar, A. Sadhanala, S. P. Senanayak, Sci., 2014, 7, 3061–3068.
M. Franckevičius, N. Arora, Y. Hu, M. K. Nazeeruddin, 107 Y. Takahashi, R. Obara, Z.-Z. Lin, Y. Takahashi, T. Naito,
S. M. Zakeeruddin and M. Grätzel, Adv. Energy Mater., T. Inabe, S. Ishibashi and K. Terakura, Dalton Trans., 2011,
2016, 6, 1502472. 40, 5563–5568.
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

84 Y. Zhang, C. C. Zhang, C. H. Gao, M. Li, X. J. Ma, 108 S. Gupta, D. Cahen and G. Hodes, J. Phys. Chem. C, 2018,
Z. K. Wang and L. S. Liao, Sol. RRL, 2019, 3, 1800269. 122, 13926–13936.
85 S. Ye, H. Rao, Z. Zhao, L. Zhang, H. Bao, W. Sun, Y. Li, 109 B. Li, B. Chang, L. Pan, Z. Li, L. Fu, Z. He and L. Yin, ACS
F. Gu, J. Wang and Z. Liu, J. Am. Chem. Soc., 2017, 139, Energy Lett., 2020, 5, 3752–3772.
7504–7512. 110 D. Ricciarelli, D. Meggiolaro, F. Ambrosio and F. De
86 A. Senocrate, T.-Y. Yang, G. Gregori, G. Y. Kim, M. Grätzel Angelis, ACS Energy Lett., 2020, 5, 2787–2795.
and J. Maier, Solid State Ionics, 2018, 321, 69–74. 111 C. C. Stoumpos, C. D. Malliakas and M. G. Kanatzidis,
87 C. Chen, D. Liu, B. Zhang, W. Bi, H. Li, J. Jin, X. Chen, Inorg. Chem., 2013, 52, 9019–9038.
L. Xu, H. Song and Q. Dai, Adv. Energy Mater., 2018, 112 Z. Chen, M. Liu, Z. Li, T. Shi, Y. Yang, H.-L. Yip and Y. Cao,
8, 1703659. iScience, 2018, 9, 337–346.
88 J. Zhang, M.-h Shang, P. Wang, X. Huang, J. Xu, Z. Hu, 113 T. Wu, X. Liu, X. Luo, X. Lin, D. Cui, Y. Wang, H. Segawa,
Y. Zhu and L. Han, ACS Energy Lett., 2016, 1, 535–541. Y. Zhang and L. Han, Joule, 2021, 5, 863–886.
89 A. Senocrate, T. Acartürk, G. Y. Kim, R. Merkle, U. Starke, 114 J. Cao and F. Yan, Energy Environ. Sci., 2021, 14, 1286–1325.
M. Grätzel and J. Maier, J. Mater. Chem. A, 2018, 6, 115 S.-H. Wei, Comput. Mater. Sci., 2004, 30, 337–348.
10847–10855. 116 Y. Mae, Metall. Mater. Trans. A, 2016, 47, 6498–6506.
90 A. Senocrate and J. Maier, J. Am. Chem. Soc., 2019, 141, 117 N. Li, C. Shi, L. Li, Z. Zhang and C. Ma, Superlattices
8382–8396. Microstruct., 2017, 104, 445–450.
91 P. Broqvist, A. Alkauskas and A. Pasquarello, Phys. Rev. B: 118 L. Xie, P. Song, L. Shen, J. Lu, K. Liu, K. Lin, W. Feng,
Condens. Matter Mater. Phys., 2008, 78, 075203. C. Tian and Z. Wei, J. Mater. Chem. A, 2020, 8, 7653–7658.
92 B. George, J. Behrends, A. Schnegg, T. Schulze, M. Fehr, 119 Y. Lv, Y. Shi, X. Song, J. Liu, M. Wang, S. Wang, Y. Feng,
L. Korte, B. Rech, K. Lips, M. Rohrmüller and E. Rauls, S. Jin and C. Hao, ACS Appl. Mater. Interfaces, 2018, 10,
Phys. Rev. Lett., 2013, 110, 136803. 31755–31764.
93 A. G. Aberle, Prog. Photovoltaics Res. Appl., 2000, 8, 473–487. 120 S.-H. Wei and S. Zhang, Phys. Rev. B: Condens. Matter
94 K. J. Chang and D. Chadi, Phys. Rev. B: Condens. Matter Mater. Phys., 2002, 66, 155211.
Mater. Phys., 1989, 40, 11644. 121 L. Zhang, H. Wang, J. Liu, Q. Zhang and H. Yan, J. Mater.
95 Z. Fang, H. He, L. Gan, J. Li and Z. Ye, Adv. Sci., 2018, Sci.: Mater. Electron., 2020, 31, 861–873.
5, 1800736. 122 X. Liu, A. Prasad, J. Nishio, E. Weber, Z. Liliental-Weber
96 Q. Jiang, M. Chen, J. Li, M. Wang, X. Zeng, T. Besara, J. Lu, and W. Walukiewicz, Appl. Phys. Lett., 1995, 67, 279–281.
Y. Xin, X. Shan and B. Pan, ACS Nano, 2017, 11, 1073–1079. 123 E. Kamiyama, K. Sueoka and J. Vanhellemont, ECS J. Solid
97 C. Londos, E. Sgourou, D. Hall and A. Chroneos, J. Mater. State Sci. Technol., 2013, 2, P104.
Sci.: Mater. Electron., 2014, 25, 2395–2410. 124 V. Lyahovitskaya, L. Chernyak, J. Greenberg, L. Kaplan and
98 S. Pyshkin, J. Ballato, M. Bass and G. Turri, J. Electron. D. Cahen, J. Cryst. Growth, 2000, 214, 1155–1157.
Mater., 2009, 38, 640–646. 125 T. Ablekim, S. K. Swain, W.-J. Yin, K. Zaunbrecher, J. Burst,
99 A. Assali, F. Kanouni, Q. Zou and R. Khenata, Phys. Lett. A, T. M. Barnes, D. Kuciauskas, S.-H. Wei and K. G. Lynn, Sci.
2020, 384, 126147. Rep., 2017, 7, 1–9.
100 D. Pérez-del-Rey, D. Forgács, E. M. Hutter, T. J. Savenije, 126 M. Jung, S.-G. Ji, G. Kim and S. I. Seok, Chem. Soc. Rev.,
D. Nordlund, P. Schulz, J. J. Berry, M. Sessolo and 2019, 48, 2011–2038.
H. J. Bolink, Adv. Mater., 2016, 28, 9839–9845. 127 S. Yang, S. Chen, E. Mosconi, Y. Fang, X. Xiao, C. Wang,
101 Y. Jia, Y. Zeng, X. Li and L. Meng, Int. J. Energy Res., 2020, Y. Zhou, Z. Yu, J. Zhao and Y. Gao, Science, 2019, 365,
44, 5765–5778. 473–478.
102 F. Yang, M. A. Kamarudin, G. Kapil, D. Hirotani, P. Zhang, 128 E. Amerling, H. Lu, B. W. Larson, A. E. Maughan,
C. H. Ng, T. Ma and S. Hayase, ACS Appl. Mater. Interfaces, A. Phillips, E. Lafalce, L. Whittaker-Brooks, J. J. Berry,
2018, 10, 24543–24548. M. C. Beard and Z. V. Vardeny, ACS Energy Lett., 2021, 6,
103 C. F. J. Lau, M. Zhang, X. Deng, J. Zheng, J. Bing, Q. Ma, 1104–1123.
J. Kim, L. Hu, M. A. Green and S. Huang, ACS Energy Lett., 129 W. Chen, D. Qi, X. Gao and A. T. S. Wee, Prog. Surf. Sci.,
2017, 2, 2319–2325. 2009, 84, 279–321.
104 S. Das Adhikari, A. K. Guria and N. Pradhan, J. Phys. Chem. 130 J. Ma, D. Kuciauskas, D. Albin, R. Bhattacharya, M. Reese,
Lett., 2019, 10, 2250–2257. T. Barnes, J. V. Li, T. Gessert and S.-H. Wei, Phys. Rev. Lett.,
105 M. Islam, M. Hadi and J. Podder, AIP Adv., 2019, 9, 125321. 2013, 111, 067402.

10056 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

131 T. a Gessert, S.-H. Wei, J. Ma, D. Albin, R. Dhere, 157 R. Brenes, C. Eames, V. Bulović, M. S. Islam and S. D.
J. Duenow, D. Kuciauskas, A. Kanevce, T. Barnes and Stranks, Adv. Mater., 2018, 30, 1706208.
J. Burst, Sol. Energy Mater. Sol. Cells, 2013, 119, 149–155. 158 X.-X. Ma and Z.-S. Li, Appl. Surf. Sci., 2018, 428, 140–147.
132 M. O. Reese, C. L. Perkins, J. M. Burst, S. Farrell, T. M. 159 A. M. Leguy, Y. Hu, M. Campoy-Quiles, M. I. Alonso,
Barnes, S. W. Johnston, D. Kuciauskas, T. A. Gessert and O. J. Weber, P. Azarhoosh, M. Van Schilfgaarde, M. T.
W. K. Metzger, J. Appl. Phys., 2015, 118, 155305. Weller, T. Bein and J. Nelson, Chem. Mater., 2015, 27,
133 A. Shepidchenko, B. Sanyal, M. Klintenberg and S. Mirbt, 3397–3407.
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

Sci. Rep., 2015, 5, 1–6. 160 H.-H. Fang, S. Adjokatse, H. Wei, J. Yang, G. R. Blake,
134 A. Lindström, S. Mirbt, B. Sanyal and M. Klintenberg, J. Huang, J. Even and M. A. Loi, Sci. Adv., 2016, 2, e1600534.
J. Phys. D: Appl. Phys., 2015, 49, 035101. 161 A. Zohar, I. Levine, S. Gupta, O. Davidson, D. Azulay,
135 S.-H. Wei, S. Zhang and A. Zunger, Appl. Phys. Lett., 1998, O. Millo, I. Balberg, G. Hodes and D. Cahen, ACS Energy
72, 3199–3201. Lett., 2017, 2, 2408–2414.
136 S. Zhang, S.-H. Wei, A. Zunger and H. Katayama-Yoshida, 162 G. Y. Kim, A. Senocrate, T.-Y. Yang, G. Gregori, M. Grätzel
Phys. Rev. B: Condens. Matter Mater. Phys., 1998, 57, 9642. and J. Maier, Nat. Mater., 2018, 17, 445–449.
137 C. Stephan, S. Schorr, M. Tovar and H.-W. Schock, Appl. 163 A. Walsh, D. O. Scanlon, S. Chen, X. Gong and S. H. Wei,
Phys. Lett., 2011, 98, 091906. Angew. Chem., Int. Ed., 2015, 54, 1791–1794.
138 M. Malitckaya, H. P. Komsa, V. Havu and M. J. Puska, Adv. 164 P. Cui, D. Wei, J. Ji, D. Song, Y. Li, X. Liu, J. Huang,
Electron. Mater., 2017, 3, 1600353. T. Wang, J. You and M. Li, Sol. RRL, 2017, 1, 1600027.
139 S. Lany and A. Zunger, J. Appl. Phys., 2006, 100, 113725. 165 J. Stevenson, B. Sorenson, V. H. Subramaniam, J. Raiford,
140 S. Chen, A. Walsh, X. G. Gong and S. H. Wei, Adv. Mater., P. P. Khlyabich, Y.-L. Loo and P. Clancy, Chem. Mater.,
2013, 25, 1522–1539. 2017, 29, 2435–2444.
141 W. J. Yin, T. Shi and Y. Yan, Adv. Mater., 2014, 26, 166 R. L. Davidovich, V. Stavila, D. V. Marinin, E. I. Voit and
4653–4658. K. H. Whitmire, Coord. Chem. Rev., 2009, 253, 1316–1352.
142 H. Zhou, W.-C. Hsu, H.-S. Duan, B. Bob, W. Yang, T.-B. 167 K. Yan, M. Long, T. Zhang, Z. Wei, H. Chen, S. Yang and
Song, C.-J. Hsu and Y. Yang, Energy Environ. Sci., 2013, 6, J. Xu, J. Am. Chem. Soc., 2015, 137, 4460–4468.
2822–2838. 168 A. Wakamiya, M. Endo, T. Sasamori, N. Tokitoh, Y. Ogomi,
143 S. Chen, X. Gong, A. Walsh and S.-H. Wei, Appl. Phys. Lett., S. Hayase and Y. Murata, Chem. Lett., 2014, 43, 711–713.
2010, 96, 021902. 169 H. Miyamae, Y. Numahata and M. Nagata, Chem. Lett.,
144 M. Paris, L. Choubrac, A. Lafond, C. Guillot-Deudon and 1980, 663–664.
S. Jobic, Inorg. Chem., 2014, 53, 8646–8653. 170 F. Hao, C. C. Stoumpos, Z. Liu, R. P. Chang and M. G.
145 M. Dimitrievska, A. Fairbrother, E. Saucedo, A. Perez- Kanatzidis, J. Am. Chem. Soc., 2014, 136, 16411–16419.
Rodriguez and V. Izquierdo-Roca, Appl. Phys. Lett., 2015, 171 Y. Guo, K. Shoyama, W. Sato, Y. Matsuo, K. Inoue,
106, 073903. K. Harano, C. Liu, H. Tanaka and E. Nakamura, J. Am.
146 D. Han, Y. Sun, J. Bang, Y. Zhang, H.-B. Sun, X.-B. Li and Chem. Soc., 2015, 137, 15907–15914.
S. Zhang, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 172 Y. Rong, Z. Tang, Y. Zhao, X. Zhong, S. Venkatesan,
87, 155206. H. Graham, M. Patton, Y. Jing, A. M. Guloy and Y. Yao,
147 W. A. Dunlap-Shohl, Y. Zhou, N. P. Padture and D. B. Mitzi, Nanoscale, 2015, 7, 10595–10599.
Chem. Rev., 2018, 119, 3193–3295. 173 J. Euvrard, O. Gunawan and D. B. Mitzi, Adv. Energy Mater.,
148 J. J. De Yoreo, P. U. Gilbert, N. A. Sommerdijk, R. L. Penn, 2019, 9, 1902706.
S. Whitelam, D. Joester, H. Zhang, J. D. Rimer, A. Navrotsky 174 D. Song, P. Cui, T. Wang, D. Wei, M. Li, F. Cao, X. Yue, P. Fu,
and J. F. Banfield, Science, 2015, 349, aaa6760. Y. Li and Y. He, J. Phys. Chem. C, 2015, 119, 22812–22819.
149 J. M. Ball and A. Petrozza, Nat. Energy, 2016, 1, 1–13. 175 A. Zohar, N. Kedem, I. Levine, D. Zohar, A. Vilan, D. Ehre,
150 Q. Wang, Y. Shao, H. Xie, L. Lyu, X. Liu, Y. Gao and G. Hodes and D. Cahen, J. Phys. Chem. Lett., 2016, 7, 191–197.
J. Huang, Appl. Phys. Lett., 2014, 105, 163508. 176 F. Giustino and H. J. Snaith, ACS Energy Lett., 2016, 1,
151 S. Yun, X. Zhou, J. Even and A. Hagfeldt, Angew. Chem., Int. 1233–1240.
Ed., 2017, 56, 15806–15817. 177 S. Gu, R. Lin, Q. Han, Y. Gao, H. Tan and J. Zhu, Adv.
152 P. Cui, D. Wei, J. Ji, H. Huang, E. Jia, S. Dou, T. Wang, Mater., 2020, 32, 1907392.
W. Wang and M. Li, Nat. Energy, 2019, 4, 150–159. 178 Z. Xiao, Z. Song and Y. Yan, Adv. Mater., 2019, 31, 1803792.
153 G. Paul, S. Chatterjee, H. Bhunia and A. J. Pal, J. Phys. 179 J. Pascual, M. Flatken, R. Félix, G. Li, S. H. Turren-Cruz,
Chem. C, 2018, 122, 20194–20199. M. H. Aldamasy, C. Hartmann, M. Li, D. Di Girolamo and
154 Z. Su, Y. Chen, X. Li, S. Wang and Y. Xiao, J. Mater. Sci.: G. Nasti, Angew. Chem., Int. Ed., 2021, 60, 21583–21591.
Mater. Electron., 2017, 28, 11053–11058. 180 F. Gu, S. Ye, Z. Zhao, H. Rao, Z. Liu, Z. Bian and C. Huang,
155 H. Wei, D. DeSantis, W. Wei, Y. Deng, D. Guo, T. J. Savenije, Sol. RRL, 2018, 2, 1800136.
L. Cao and J. Huang, Nat. Mater., 2017, 16, 826–833. 181 T. Nakamura, S. Yakumaru, M. A. Truong, K. Kim, J. Liu,
156 Z. Hu, Z. Liu, L. K. Ono, M. Jiang, S. He, D. Y. Son and S. Hu, K. Otsuka, R. Hashimoto, R. Murdey and T. Sasamori,
Y. Qi, Adv. Energy Mater., 2020, 10, 2000908. Nat. Commun., 2020, 11, 1–8.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10057
 View Article Online

Chem Soc Rev Review Article

182 R. Lin, K. Xiao, Z. Qin, Q. Han, C. Zhang, M. Wei, 205 J. S. Park, S. Kim, Z. Xie and A. Walsh, Nat. Rev. Mater.,
M. I. Saidaminov, Y. Gao, J. Xu and M. Xiao, Nat. Energy, 2018, 3, 194–210.
2019, 4, 864–873. 206 W. Ming, S. Chen and M.-H. Du, J. Mater. Chem. A, 2016, 4,
183 W. Li, J. Li, J. Li, J. Fan, Y. Mai and L. Wang, J. Mater. Chem. A, 16975–16981.
2016, 4, 17104–17110. 207 W. Zhao, J. Ding, Y. Zou, C.-A. Di and D. Zhu, Chem. Soc.
184 J. Cao, Q. Tai, P. You, G. Tang, T. Wang, N. Wang and Rev., 2020, 49, 7210–7228.
F. Yan, J. Mater. Chem. A, 2019, 7, 26580–26585. 208 J. Saska, G. Gonel, Z. I. Bedolla-Valdez, S. D. Aronow,
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

185 T.-B. Song, T. Yokoyama, C. C. Stoumpos, J. Logsdon, D. H. N. E. Shevchenko, A. S. Dudnik, A. J. Moulé and
Cao, M. R. Wasielewski, S. Aramaki and M. G. Kanatzidis, M. Mascal, Chem. Mater., 2019, 31, 1500–1506.
J. Am. Chem. Soc., 2017, 139, 836–842. 209 F. Zhang, H. L. Smith and A. Kahn, Appl. Phys. Rev., 2021,
186 M. E. Kayesh, T. H. Chowdhury, K. Matsuishi, R. Kaneko, 8, 041301.
S. Kazaoui, J.-J. Lee, T. Noda and A. Islam, ACS Energy Lett., 210 H. Méndez, G. Heimel, S. Winkler, J. Frisch, A. Opitz,
2018, 3, 1584–1589. K. Sauer, B. Wegner, M. Oehzelt, C. Röthel and S. Duhm,
187 P. Xu, S. Chen, H.-J. Xiang, X.-G. Gong and S.-H. Wei, Nat. Commun., 2015, 6, 1–11.
Chem. Mater., 2014, 26, 6068–6072. 211 I. Salzmann, G. Heimel, M. Oehzelt, S. Winkler and
188 I. Chung, B. Lee, J. He, R. P. Chang and M. G. Kanatzidis, N. Koch, Acc. Chem. Res., 2016, 49, 370–378.
Nature, 2012, 485, 486–489. 212 J. Torrance, J. Vazquez, J. Mayerle and V. Lee, Phys. Rev.
189 M. H. Kumar, S. Dharani, W. L. Leong, P. P. Boix, Lett., 1981, 46, 253.
R. R. Prabhakar, T. Baikie, C. Shi, H. Ding, R. Ramesh 213 A. Panda, K. Ding, X. Liu and S. R. Forrest, Phys. Rev. B,
and M. Asta, Adv. Mater., 2014, 26, 7122–7127. 2016, 94, 125429.
190 G. Xing, M. H. Kumar, W. K. Chong, X. Liu, Y. Cai, H. Ding, 214 E. F. Aziz, A. Vollmer, S. Eisebitt, W. Eberhardt, P. Pingel,
M. Asta, M. Grätzel, S. Mhaisalkar and N. Mathews, Adv. D. Neher and N. Koch, Adv. Mater., 2007, 19, 3257–3260.
Mater., 2016, 28, 8191–8196. 215 I. Salzmann, G. Heimel, S. Duhm, M. Oehzelt, P. Pingel,
191 S. Gupta, T. Bendikov, G. Hodes and D. Cahen, ACS Energy B. M. George, A. Schnegg, K. Lips, R.-P. Blum and
Lett., 2016, 1, 1028–1033. A. Vollmer, Phys. Rev. Lett., 2012, 108, 035502.
192 R. L. Milot, M. T. Klug, C. L. Davies, Z. Wang, H. Kraus, 216 J. B. Torrance, Acc. Chem. Res., 1979, 12, 79–86.
H. J. Snaith, M. B. Johnston and L. M. Herz, Adv. Mater., 217 J. B. Torrance, B. A. Scott and F. Kaufman, Solid State
2018, 30, 1804506. Commun., 1975, 17, 1369–1373.
193 W. Liao, D. Zhao, Y. Yu, C. R. Grice, C. Wang, A. J. 218 K. Kanai, K. Akaike, K. Koyasu, K. Sakai, T. Nishi,
Cimaroli, P. Schulz, W. Meng, K. Zhu and R. G. Xiong, Y. Kamizuru, T. Nishi, Y. Ouchi and K. Seki, Appl. Phys.
Adv. Mater., 2016, 28, 9333–9340. A: Mater. Sci. Process., 2009, 95, 309–313.
194 Y. Dang, Y. Zhou, X. Liu, D. Ju, S. Xia, H. Xia and X. Tao, 219 H. Méndez, G. Heimel, A. Opitz, K. Sauer, P. Barkowski,
Angew. Chem., Int. Ed., 2016, 55, 3447–3450. M. Oehzelt, J. Soeda, T. Okamoto, J. Takeya and J. B. Arlin,
195 K. Marshall, M. Walker, R. Walton and R. Hatton, Nat. Angew. Chem., Int. Ed., 2013, 125, 7905–7909.
Energy, 2016, 1, 1–9. 220 P. Ru, E. Bi, Y. Zhang, Y. Wang, W. Kong, Y. Sha, W. Tang,
196 K. P. Marshall, R. I. Walton and R. A. Hatton, J. Mater. P. Zhang, Y. Wu and W. Chen, Adv. Energy Mater., 2020,
Chem. A, 2015, 3, 11631–11640. 10, 1903487.
197 X. Xu, C.-C. Chueh, Z. Yang, A. Rajagopal, J. Xu, S. B. Jo and 221 W. Gao and A. Kahn, Appl. Phys. Lett., 2001, 79, 4040–4042.
A. K.-Y. Jen, Nano Energy, 2017, 34, 392–398. 222 G. M. Rangger, O. T. Hofmann, B. Bröker and E. Zojer,
198 H. Ban, Q. Sun, T. Zhang, H. Li, Y. Shen and M. Wang, Sol. Synth. Met., 2010, 160, 1456–1462.
RRL, 2020, 4, 1900457. 223 P. K. Koech, A. B. Padmaperuma, L. Wang, J. S. Swensen,
199 M. I. Saidaminov, I. Spanopoulos, J. Abed, W. Ke, J. Wicks, E. Polikarpov, J. T. Darsell, J. E. Rainbolt and D. J. Gaspar,
M. G. Kanatzidis and E. H. Sargent, ACS Energy Lett., 2020, Chem. Mater., 2010, 22, 3926–3932.
5, 1153–1155. 224 F. Zhang and A. Kahn, Adv. Funct. Mater., 2018,
200 J. Pascual, G. Nasti, M. H. Aldamasy, J. A. Smith, 28, 1703780.
M. Flatken, N. Phung, D. Di Girolamo, S.-H. Turren- 225 K. M. Al-Ahmary, M. M. El-Kholy, I. A. Al-Solmy and
Cruz, M. Li and A. Dallmann, Mater. Adv., 2020, 1, M. M. Habeeb, Spectrochim. Acta, Part A, 2013, 110,
1066–1070. 343–350.
201 D. Di Girolamo, J. Pascual, M. H. Aldamasy, Z. Iqbal, G. Li, 226 R. Kroon, D. Kiefer, D. Stegerer, L. Yu, M. Sommer and
E. Radicchi, M. Li, S.-H. Turren-Cruz, G. Nasti and C. Müller, Adv. Mater., 2017, 29, 1700930.
A. Dallmann, ACS Energy Lett., 2021, 6, 959–968. 227 D. Zhou, H. Zhang, H. Zheng, Z. Xu, H. Xu, H. Guo, P. Li,
202 G. Baraff and M. Schlüter, Phys. Rev. Lett., 1985, 55, 1327. Y. Tong, B. Hu and L. Chen, Small, 2022, 2200679.
203 S. Zhang and J. E. Northrup, Phys. Rev. Lett., 1991, 67, 2339. 228 J. Niederhausen, P. Amsalem, J. Frisch, A. Wilke,
204 D. Laks, C. Van de Walle, G. Neumark, P. Blöchl and A. Vollmer, R. Rieger, K. Müllen, J. Rabe and N. Koch,
S. Pantelides, Phys. Rev. B: Condens. Matter Mater. Phys., Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84,
1992, 45, 10965. 165302.

10058 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

229 T. Fukunaga, M. D. Gordon and P. J. Krusic, J. Am. Chem. 250 E. A. Gaulding, J. Hao, H. S. Kang, E. M. Miller,
Soc., 1976, 98, 611–613. S. N. Habisreutinger, Q. Zhao, A. Hazarika, P. C. Sercel,
230 Y. Karpov, T. Erdmann, I. Raguzin, M. Al-Hussein, M. Binner, J. M. Luther and J. L. Blackburn, Adv. Mater., 2019, 31,
U. Lappan, M. Stamm, K. L. Gerasimov, T. Beryozkina and 1902250.
V. Bakulev, Adv. Mater., 2016, 28, 6003–6010. 251 H. Chen, Y. Zhan, G. Xu, W. Chen, S. Wang, M. Zhang, Y. Li
231 Y. Qi, T. Sajoto, M. Kröger, A. M. Kandabarow, W. Park, and Y. Li, Adv. Funct. Mater., 2020, 30, 2001788.
S. Barlow, E.-G. Kim, L. Wielunski, L. Feldman and 252 N. G. Connelly and W. E. Geiger, Chem. Rev., 1996, 96,
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

R. A. Bartynski, Chem. Mater., 2010, 22, 524–531. 877–910.

232 Y. Qi, T. Sajoto, S. Barlow, E.-G. Kim, J.-L. Brédas, 253 C. N. Valdez, M. Braten, A. Soria, D. R. Gamelin and
S. R. Marder and A. Kahn, J. Am. Chem. Soc., 2009, 131, J. M. Mayer, J. Am. Chem. Soc., 2013, 135, 8492–8495.
12530–12531. 254 J. Ristein, Science, 2006, 313, 1057–1058.
233 S. K. Mohapatra, Y. Zhang, B. Sandhu, M. S. Fonari, 255 D. Bozyigit, W. M. Lin, N. Yazdani, O. Yarema and
T. V. Timofeeva, S. R. Marder and S. Barlow, Polyhedron, V. Wood, Nat. Commun., 2015, 6, 1–10.
2016, 116, 88–95. 256 K. Katsiev, A. H. Ip, A. Fischer, I. Tanabe, X. Zhang,
234 A. Dai, A. Wan, C. Magee, Y. Zhang, S. Barlow, S. R. Marder A. R. Kirmani, O. Voznyy, L. R. Rollny, K. W. Chou and
and A. Kahn, Org. Electron., 2015, 23, 151–157. S. M. Thon, Adv. Mater., 2014, 26, 937–942.
235 A. Dai, Y. Zhou, A. L. Shu, S. K. Mohapatra, H. Wang, 257 A. R. Kirmani, G. H. Carey, M. Abdelsamie, B. Yan, D. Cha,
C. Fuentes-Hernandez, Y. Zhang, S. Barlow, Y. L. Loo and L. R. Rollny, X. Cui, E. H. Sargent and A. Amassian, Adv.
S. R. Marder, Adv. Funct. Mater., 2014, 24, 2197–2204. Mater., 2014, 26, 4717–4723.
236 J. Li, G. Zhang, D. M. Holm, I. E. Jacobs, B. Yin, P. Stroeve, 258 M. Shim and P. Guyot-Sionnest, Nature, 2000, 407, 981–983.
M. Mascal and A. J. Moulé, Chem. Mater., 2015, 27, 259 S. Guo, S. B. Kim, S. K. Mohapatra, Y. Qi, T. Sajoto,
5765–5774. A. Kahn, S. R. Marder and S. Barlow, Adv. Mater., 2012,
237 O. Solomeshch, Y. J. Yu, A. A. Goryunkov, L. N. Sidorov, 24, 699–703.
R. F. Tuktarov, D. H. Choi, J. I. Jin and N. Tessler, Adv. 260 Y. Jang, J. H. Cho, D. H. Kim, Y. D. Park, M. Hwang and
Mater., 2009, 21, 4456–4460. K. Cho, Appl. Phys. Lett., 2007, 90, 132104.
238 A. Tadich, M. T. Edmonds, L. Ley, F. Fromm, Y. Smets, 261 H. Sakai, Y. Takahashi and H. Murata, Appl. Phys. Lett.,
Z. Mazej, J. Riley, C. I. Pakes, T. Seyller and M. Wanke, 2007, 91, 113502.
Appl. Phys. Lett., 2013, 102, 241601. 262 W. Chen, X. Y. Gao, D. C. Qi, S. Chen, Z. K. Chen and
239 N. C. Cates, R. Gysel, Z. Beiley, C. E. Miller, M. F. Toney, A. T. S. Wee, Adv. Funct. Mater., 2007, 17, 1339–1344.
M. Heeney, I. McCulloch and M. D. McGehee, Nano Lett., 263 V. De Renzi, R. Rousseau, D. Marchetto, R. Biagi,
2009, 9, 4153–4157. S. Scandolo and U. Del Pennino, Phys. Rev. Lett., 2005,
240 A. Mayer, M. F. Toney, S. R. Scully, J. Rivnay, C. J. Brabec, 95, 046804.
M. Scharber, M. Koppe, M. Heeney, I. McCulloch and 264 H. Ahn, M. Zharnikov and J. E. Whitten, Chem. Phys. Lett.,
M. D. McGehee, Adv. Funct. Mater., 2009, 19, 1173–1179. 2006, 428, 283–287.
241 R. Haddon, A. Hebard, M. Rosseinsky, D. Murphy, 265 H. Ishii, K. Sugiyama, E. Ito and K. Seki, Adv. Mater., 1999,
S. Duclos, K. Lyons, B. Miller, J. Rosamilia, R. Fleming 11, 605–625.
and A. Kortan, Nature, 1991, 350, 320–322. 266 M. Calhoun, J. Sanchez, D. Olaya, M. Gershenson and
242 S. K. Mohapatra, A. Fonari, C. Risko, K. Yesudas, V. Podzorov, Nat. Mater., 2008, 7, 84–89.
K. Moudgil, J. H. Delcamp, T. V. Timofeeva, J. L. Brédas, 267 K. Xing, Y. Xiang, M. Jiang, D. L. Creedon, G. Akhgar,
S. R. Marder and S. Barlow, Chem. – Eur. J., 2014, 20, S. A. Yianni, H. Xiao, L. Ley, A. Stacey and J. C. McCallum,
15385–15394. Appl. Surf. Sci., 2020, 509, 144890.
243 S. A. Paniagua, J. Baltazar, H. Sojoudi, S. K. Mohapatra, 268 Y. Liu, Q. Zhang, X. Zhang, N. Gao and H. Li, Adv. Theory
S. Zhang, C. L. Henderson, S. Graham, S. Barlow and Simul., 2022, 2100460.
S. R. Marder, Mater. Horiz., 2014, 1, 111–115. 269 P. Liu, X. Liu, L. Lyu, H. Xie, H. Zhang, D. Niu, H. Huang, C. Bi,
244 Z. Liang and B. A. Gregg, Adv. Mater., 2012, 24, 3258–3262. Z. Xiao and J. Huang, Appl. Phys. Lett., 2015, 106, 193903.
245 C. K. Chan, W. Zhao, S. Barlow, S. Marder and A. Kahn, 270 Q. K. Wang, R. B. Wang, P. F. Shen, C. Li, Y. Q. Li, L. J. Liu,
Org. Electron., 2008, 9, 575–581. S. Duhm and J. X. Tang, Adv. Mater. Interfaces, 2015,
246 Y. Zhang, B. de Boer and P. W. Blom, Phys. Rev. B: Condens. 2, 1400528.
Matter Mater. Phys., 2010, 81, 085201. 271 P. Schulz, J. O. Tiepelt, J. A. Christians, I. Levine, E. Edri,
247 S. Guo, S. K. Mohapatra, A. Romanov, T. V. Timofeeva, E. M. Sanehira, G. Hodes, D. Cahen and A. Kahn, ACS Appl.
K. I. Hardcastle, K. Yesudas, C. Risko, J. L. Brédas, S. R. Mater. Interfaces, 2016, 8, 31491–31499.
Marder and S. Barlow, Chem. – Eur. J., 2012, 18, 14760–14772. 272 Q. Ou, Y. Zhang, Z. Wang, J. A. Yuwono, R. Wang, Z. Dai,
248 Y. Qi, S. K. Mohapatra, S. Bok Kim, S. Barlow, S. R. Marder W. Li, C. Zheng, Z. Q. Xu and X. Qi, Adv. Mater., 2018,
and A. Kahn, Appl. Phys. Lett., 2012, 100, 54. 30, 1705792.
249 P. Wei, J. H. Oh, G. Dong and Z. Bao, J. Am. Chem. Soc., 273 L. Zhao, Y. L. Lin and B. P. Rand, Adv. Funct. Mater., 2018,
2010, 132, 8852–8853. 28, 1802060.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10059
 View Article Online

Chem Soc Rev Review Article

274 F. Zu, P. Amsalem, M. Ralaiarisoa, T. Schultz, R. Schlesinger 302 H. Keppner, P. Torres, R. Flückiger, J. Meier, A. Shah,
and N. Koch, ACS Appl. Mater. Interfaces, 2017, 9, C. Fortmann, P. Fath, G. Willeke, K. Happle and H. Kiess,
41546–41552. Sol. Energy Mater. Sol. Cells, 1994, 34, 201–209.
275 X. Guo, J. Su, Z. Lin, X. Wang, Q. Wang, Z. Zeng, J. Chang 303 K. Yoshikawa, H. Kawasaki, W. Yoshida, T. Irie, K. Konishi,
and Y. Hao, iScience, 2021, 24, 102276. K. Nakano, T. Uto, D. Adachi, M. Kanematsu and H. Uzu,
276 A. N. Cho and N. G. Park, ChemSusChem, 2017, 10, Nat. Energy, 2017, 2, 1–8.
3687–3704. 304 H. Zai, J. Su, C. Zhu, Y. Chen, Y. Ma, P. Zhang, S. Ma,
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

277 E. E. Perry, J. G. Labram, N. R. Venkatesan, H. Nakayama X. Zhang, H. Xie and R. Fan, Joule, 2021, 5, 2148–2163.
and M. L. Chabinyc, Adv. Electron. Mater., 2018, 4, 1800087. 305 J. B. Lee, Y. R. Lim, A. K. Katiyar, W. Song, J. Lim, S. Bae,
278 H. Pan, A. Xie, S. Hou, X. Yin, C. S. Tang, N. T. Hoa, T. W. Kim, S. K. Lee and J. H. Ahn, Adv. Mater., 2019,
M. D. Birowosuto, H. Wang, C. Dang and A. Rusydi, Nano 31, 1904194.
Res., 2019, 12, 77–84. 306 M. Rudan, Physics of semiconductor devices, Springer, 2015.
279 L. Zuo, H. Guo, D. W. deQuilettes, S. Jariwala, N. De Marco, 307 W. Peng, L. Wang, B. Murali, K. T. Ho, A. Bera, N. Cho,
S. Dong, R. DeBlock, D. S. Ginger, B. Dunn and M. Wang, C. F. Kang, V. M. Burlakov, J. Pan and L. Sinatra, Adv.
Sci. Adv., 2017, 3, e1700106. Mater., 2016, 28, 3383–3390.
280 B. Chen, P. N. Rudd, S. Yang, Y. Yuan and J. Huang, Chem. 308 N. Rolston, K. A. Bush, A. D. Printz, A. Gold-Parker,
Soc. Rev., 2019, 48, 3842–3867. Y. Ding, M. F. Toney, M. D. McGehee and R. H.
281 A. F. Akbulatov, V. M. Martynenko, L. A. Frolova, N. N. Dauskardt, Adv. Energy Mater., 2018, 8, 1802139.
Dremova, I. Zhidkov, S. A. Tsarev, S. Y. Luchkin, E. Z. 309 H.-S. Kim and N.-G. Park, NPG Asia Mater., 2020, 12, 1–14.
Kurmaev, S. M. Aldoshin and K. J. Stevenson, Sol. Energy 310 D. Liu, D. Luo, A. N. Iqbal, K. W. Orr, T. A. Doherty,
Mater. Sol. Cells, 2020, 213, 110559. Z.-H. Lu, S. D. Stranks and W. Zhang, Nat. Mater., 2021,
282 S. Olthof and K. Meerholz, Sci. Rep., 2017, 7, 1–10. 20, 1337–1346.
283 K. Ho, M. Wei, E. H. Sargent and G. C. Walker, ACS Energy 311 Y. Gao, A. M. Nolan, P. Du, Y. Wu, C. Yang, Q. Chen, Y. Mo
Lett., 2021, 6, 934–940. and S.-H. Bo, Chem. Rev., 2020, 120, 5954–6008.
284 Z. Hu, Q. An, H. Xiang, L. Aigouy, B. Sun, Y. Vaynzof and 312 T. Shi, Y. Chen and X. Guo, Prog. Mater. Sci., 2016, 80,
Z. Chen, ACS Appl. Mater. Interfaces, 2020, 12, 54824–54832. 77–132.
285 X. Li, W. Zhang, X. Guo, C. Lu, J. Wei and J. Fang, Science, 313 J.-W. Lee, S.-G. Kim, J.-M. Yang, Y. Yang and N.-G. Park,
2022, 375, 434–437. APL Mater., 2019, 7, 041111.
286 M. Knupfer, Surf. Sci. Rep., 2001, 42, 1–74. 314 E. Bi, Z. Song, C. Li, Z. Wu and Y. Yan, Trends Chem., 2021,
287 R. Hettich, R. Compton and R. Ritchie, Phys. Rev. Lett., 3, 575–588.
1991, 67, 1242. 315 T. Zhang, C. Hu and S. Yang, Small Methods, 2020,
288 J. Ristein, J. Phys. D: Appl. Phys., 2006, 39, R71. 4, 1900552.
289 W. Chen, Z. Gan, M. A. Green, B. Jia and X. Wen, Small 316 H. Zhang, X. Fu, Y. Tang, H. Wang, C. Zhang, W. W. Yu,
Methods, 2021, 5, 2000731. X. Wang, Y. Zhang and M. Xiao, Nat. Commun., 2019, 10,
290 A. Walsh and S. D. Stranks, ACS Energy Lett., 2018, 3, 1–8.
1983–1990. 317 J. Haruyama, K. Sodeyama, L. Han and Y. Tateyama, J. Am.
291 Y. Li, Y. Gao, N. Xiao, P. Ning, L. Yu, J. Zhang, P. Niu, Y. Ma Chem. Soc., 2015, 137, 10048–10051.
and C. Gao, AIP Adv., 2018, 8, 115202. 318 C. Eames, J. M. Frost, P. R. Barnes, B. C. O’regan, A. Walsh
292 Z. Zhang and J. T. Yates Jr, Chem. Rev., 2012, 112, and M. S. Islam, Nat. Commun., 2015, 6, 1–8.
5520–5551. 319 J. M. Azpiroz, E. Mosconi, J. Bisquert and F. De Angelis,
293 C. Ballif, F.-J. Haug, M. Boccard, P. J. Verlinden and Energy Environ. Sci., 2015, 8, 2118–2127.
G. Hahn, Nat. Rev. Mater., 2022, 1–20. 320 Z. Ni, C. Bao, Y. Liu, Q. Jiang, W.-Q. Wu, S. Chen, X. Dai,
294 O. Von Roos, J. Appl. Phys., 1978, 49, 3503–3511. B. Chen, B. Hartweg and Z. Yu, Science, 2020, 367,
295 M. A. Green, Appl. Phys. Lett., 1975, 27, 287–288. 1352–1358.
296 R. Swanson, Sol. Cells, 1986, 17, 85–118. 321 P. n Huang and A. Petric, J. Electrochem. Soc., 1996, 143,
297 M. A. Green, Prog. Photovoltaics Res. Appl., 2009, 17, 1644.
183–189. 322 J. Stevenson, T. Armstrong, D. McCready, L. Pederson and
298 M. Green and A. Blakers, Sol. Cells, 1983, 8, 3–16. W. Weber, J. Electrochem. Soc., 1997, 144, 3613.
299 Y.-Y. Huang, Y.-W. Ok, K. Madani, W. Choi, A. Upadhyaya, 323 K. Huang, R. S. Tichy and J. B. Goodenough, J. Am. Ceram.
V. Upadhyaya, B. Rounsaville, V. Chandrasekaran and Soc., 1998, 81, 2565–2575.
A. Rohatgi, Sol. Energy Mater. Sol. Cells, 2021, 230, 111183. 324 T. Wu, W. Dai, M. Ke, Q. Huang and L. Lu, Adv. Sci., 2021,
300 J. Pankove and M. Tarng, Appl. Phys. Lett., 1979, 34, 8, 2100774.
156–157. 325 T. Ishihara, H. Matsuda and Y. Takita, J. Am. Chem. Soc.,
301 M. Tanaka, M. Taguchi, T. Matsuyama, T. Sawada, 1994, 116, 3801–3803.
S. Tsuda, S. Nakano, H. Hanafusa and Y. Kuwano, Jpn. 326 D.-Y. Son, S.-G. Kim, J.-Y. Seo, S.-H. Lee, H. Shin, D. Lee
J. Appl. Phys., 1992, 31, 3518. and N.-G. Park, J. Am. Chem. Soc., 2018, 140, 1358–1364.

10060 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

327 T. Bu, J. Li, F. Zheng, W. Chen, X. Wen, Z. Ku, Y. Peng, 351 B. Hartiti, S. Sivoththaman, R. Schindler, J. Nijs, J. Muller
J. Zhong, Y.-B. Cheng and F. Huang, Nat. Commun., 2018, and P. Siffert, IEEE, 1994, 2, 1519–1522.
9, 1–10. 352 A. Wadsworth, Z. Hamid, J. Kosco, N. Gasparini and
328 W. Zhu, S. Wang, X. Zhang, A. Wang, C. Wu and F. Hao, I. McCulloch, Adv. Mater., 2020, 32, 2001763.
Small, 2022, 2105783. 353 M. C. Scharber and N. S. Sariciftci, Prog. Polym. Sci., 2013,
329 L. McGovern, I. Koschany, G. Grimaldi, L. A. Muscarella 38, 1929–1940.
and B. Ehrler, J. Phys. Chem. Lett., 2021, 12, 2423–2428. 354 C. Chen, Y. Xu, S. Wu, S. Zhang, Z. Yang, W. Zhang, H. Zhu,
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

330 Q. Dong, J. Mendes, L. Lei, D. Seyitliyev, L. Zhu, S. He, Z. Xiong, W. Chen and W. Chen, J. Mater. Chem. A, 2018, 6,
K. Gundogdu and F. So, ACS Appl. Mater. Interfaces, 2020, 7903–7912.
12, 48845–48853. 355 T. Sajoto, M. Santos, J. Heremans, M. Shayegan, M. Heiblum,
331 G. Nasti and A. Abate, Adv. Energy Mater., 2020, 10, 1902467. M. Weckwerth and U. Meirav, Appl. Phys. Lett., 1989, 54,
332 K. Mahfoud, M. Loghmarti, J. Muller and P. Siffert, Mater. 840–842.
Sci. Eng., B, 1996, 36, 63–67. 356 M. de Murcia, D. Gasquet, A. Elamri, J.-P. Nougier and
333 R. Azmi, E. Ugur, A. Seitkhan, F. Aljamaan, A. S. Subbiah, J. Vanbremeersch, IEEE Trans. Electron Devices, 1991, 38,
J. Liu, G. T. Harrison, M. I. Nugraha, M. K. Eswaran and 2531–2539.
M. Babics, Science, 2022, eabm5784. 357 L. Tarricone, N. Romeo, G. Sberveglieri and S. Mora, Sol.
334 H. Min, D. Y. Lee, J. Kim, G. Kim, K. S. Lee, J. Kim, Energy Mater., 1982, 7, 343–350.
M. J. Paik, Y. K. Kim, K. S. Kim and M. G. Kim, Nature, 358 S. A. Jensen, J. M. Burst, J. N. Duenow, H. L. Guthrey,
2021, 598, 444–450. J. Moseley, H. R. Moutinho, S. W. Johnston, A. Kanevce,
335 M. Wang, H. Sun, F. Cao, W. Tian and L. Li, Adv. Mater., M. M. Al-Jassim and W. K. Metzger, Appl. Phys. Lett., 2016,
2021, 33, 2100625. 108, 263903.
336 M. Gedda, E. Yengel, H. Faber, F. Paulus, J. A. Kreß, M. C. 359 M. Veale, J. Kalliopuska, H. Pohjonen, H. Andersson,
Tang, S. Zhang, C. A. Hacker, P. Kumar and D. R. Naphade, S. Nenonen, P. Seller and M. Wilson, J. Instrum., 2012,
Adv. Mater., 2021, 33, 2003137. 7, C01035.
337 A. Liu, H. Zhu, S. Bai, Y. Reo, T. Zou, M.-G. Kim and 360 C.-H. Huang and W.-J. Chuang, Vacuum, 2015, 118, 32–37.
Y.-Y. Noh, Nat. Electron., 2022, 5, 78–83. 361 S. Dinca, E. A. Schiff, B. Egaas, R. Noufi, D. Young and
338 A. Tukiainen, A. Aho, V. Polojarvi, R. Ahorinta and W. Shafarman, Phys. Rev. B: Condens. Matter Mater. Phys.,
M. Guina, J. Environ. Eng., 2015, 5, 113–132. 2009, 80, 235201.
339 S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou, 362 K. Decock, J. Lauwaert and M. Burgelman, Energy Procedia,
M. J. Alcocer, T. Leijtens, L. M. Herz, A. Petrozza and 2010, 2, 49–54.
H. J. Snaith, Science, 2013, 342, 341–344. 363 C. Longeaud, A. F. Allah, J. Schmidt, M. El Yaakoubi,
340 G. Xing, N. Mathews, S. Sun, S. S. Lim, Y. M. Lam, S. Berson and N. Lemaitre, Org. Electron., 2016, 31,
M. Grätzel, S. Mhaisalkar and T. C. Sum, Science, 2013, 253–257.
342, 344–347. 364 Y. Yang, M. Yang, Z. Li, R. Crisp, K. Zhu and M. C. Beard,
341 D. W. DeQuilettes, S. Koch, S. Burke, R. K. Paranji, J. Phys. Chem. Lett., 2015, 6, 4688–4692.
A. J. Shropshire, M. E. Ziffer and D. S. Ginger, ACS Energy 365 D. Shi, V. Adinolfi, R. Comin, M. Yuan, E. Alarousu,
Lett., 2016, 1, 438–444. A. Buin, Y. Chen, S. Hoogland, A. Rothenberger and
342 M. M. Tavakoli, L. Gu, Y. Gao, C. Reckmeier, J. He, K. Katsiev, Science, 2015, 347, 519–522.
A. L. Rogach, Y. Yao and Z. Fan, Sci. Rep., 2015, 5, 1–9. 366 J.-F. Liao, W.-Q. Wu, Y. Jiang, J.-X. Zhong, L. Wang and
343 M. Peng, W. Wen, S. Chen, B. Chen, K. Yan, H. Hu, D.-B. Kuang, Chem. Soc. Rev., 2020, 49, 354–381.
B. Dong, X. Gao, X. Yu and X. Jiang, Sci. China: Chem., 367 W.-G. Li, H.-S. Rao, B.-X. Chen, X.-D. Wang and
2016, 59, 653–658. D.-B. Kuang, J. Mater. Chem. A, 2017, 5, 19431–19438.
344 Z. Yu, X. Chen, S. P. Harvey, Z. Ni, B. Chen, S. Chen, C. Yao, 368 M. I. Saidaminov, A. L. Abdelhady, B. Murali, E. Alarousu,
X. Xiao, S. Xu and G. Yang, Adv. Mater., 2022, 2110351. V. M. Burlakov, W. Peng, I. Dursun, L. Wang, Y. He and
345 T. McGill and R. Baron, Phys. Rev. B: Solid State, 1975, G. Maculan, Nat. Commun., 2015, 6, 1–6.
11, 5208. 369 G. Maculan, A. D. Sheikh, A. L. Abdelhady, M. I.
346 G. L. Pearson and J. Bardeen, Phys. Rev., 1949, 75, 865. Saidaminov, M. A. Haque, B. Murali, E. Alarousu,
347 W. d Dane, K. Frohna, D. Emin, T. Kirchartz, V. Bulovic, D. S. O. F. Mohammed, T. Wu and O. M. Bakr, J. Phys. Chem.
Ginger and S. D. Stranks, Chem. Rev., 2019, 119, 11007–11019. Lett., 2015, 6, 3781–3786.
348 D. Luo, R. Su, W. Zhang, Q. Gong and R. Zhu, Nat. Rev. 370 Z. Lian, Q. Yan, T. Gao, J. Ding, Q. Lv, C. Ning, Q. Li and
Mater., 2020, 5, 44–60. J.-L. Sun, J. Am. Chem. Soc., 2016, 138, 9409–9412.
349 M. B. Johnston and L. M. Herz, Acc. Chem. Res., 2016, 49, 371 M. I. Saidaminov, M. A. Haque, J. Almutlaq, S. Sarmah,
146–154. X. H. Miao, R. Begum, A. A. Zhumekenov, I. Dursun,
350 T. Sugahara, H. Sato, M. Hao, Y. Naoi, S. Kurai, S. Tottori, N. Cho and B. Murali, Adv. Opt. Mater., 2017, 5, 1600704.
K. Yamashita, K. Nishino, L. T. Romano and S. Sakai, Jpn. 372 A. A. Zhumekenov, M. I. Saidaminov, M. A. Haque,
J. Appl. Phys., 1998, 37, L398. E. Alarousu, S. P. Sarmah, B. Murali, I. Dursun,

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10061
 View Article Online

Chem Soc Rev Review Article

X.-H. Miao, A. L. Abdelhady and T. Wu, ACS Energy Lett., 396 M. Li, H. Li, J. Fu, T. Liang and W. Ma, J. Phys. Chem. C,
2016, 1, 32–37. 2020, 124, 27251–27266.
373 S. O. Kasap, Principles of electronic materials and devices, 397 Q. Wang, B. Chen, Y. Liu, Y. Deng, Y. Bai, Q. Dong and
McGraw-Hill, New York, 2006. J. Huang, Energy Environ. Sci., 2017, 10, 516–522.
374 S. Solmi, A. Parisini, R. Angelucci, A. Armigliato, D. Nobili 398 Z. Fan, H. Xiao, Y. Wang, Z. Zhao, Z. Lin, H.-C. Cheng,
and L. Moro, Phys. Rev. B: Condens. Matter Mater. Phys., S.-J. Lee, G. Wang, Z. Feng and W. A. Goddard III, Joule,
1996, 53, 7836. 2017, 1, 548–562.
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

375 R. Baron, G. Shifrin, O. Marsh and J. W. Mayer, J. Appl. 399 N. Aristidou, C. Eames, I. Sanchez-Molina, X. Bu, J. Kosco,
Phys., 1969, 40, 3702–3719. M. S. Islam and S. A. Haque, Nat. Commun., 2017, 8, 1–10.
376 C. Ma and N.-G. Park, Chem, 2020, 6, 1254–1264. 400 S. R. Raga, M.-C. Jung, M. V. Lee, M. R. Leyden, Y. Kato and
377 N. Mundhaas, Z. J. Yu, K. A. Bush, H. P. Wang, J. Häusele, Y. Qi, Chem. Mater., 2015, 27, 1597–1603.
S. Kavadiya, M. D. McGehee and Z. C. Holman, Sol. RRL, 401 Y. Li, X. Xu, C. Wang, B. Ecker, J. Yang, J. Huang and
2019, 3, 1800378. Y. Gao, J. Phys. Chem. C, 2017, 121, 3904–3910.
378 M. A. Green, Sol. Energy Mater. Sol. Cells, 2015, 143, 190–197. 402 W. Zhang, S. Pathak, N. Sakai, T. Stergiopoulos, P. K.
379 N.-G. Park and H. Segawa, ACS Photonics, 2018, 5, Nayak, N. K. Noel, A. A. Haghighirad, V. M. Burlakov,
2970–2977. D. W. Dequilettes and A. Sadhanala, Nat. Commun., 2015,
380 E. H. Jung, N. J. Jeon, E. Y. Park, C. S. Moon, T. J. Shin, T.-Y. 6, 1–9.
Yang, J. H. Noh and J. Seo, Nature, 2019, 567, 511–515. 403 S. Wang, Y. Jiang, E. J. Juarez-Perez, L. K. Ono and Y. Qi,
381 Q. Jiang, Y. Zhao, X. Zhang, X. Yang, Y. Chen, Z. Chu, Nat. Energy, 2016, 2, 1–8.
Q. Ye, X. Li, Z. Yin and J. You, Nat. Photonics, 2019, 13, 404 Z. Liu, J. Hu, H. Jiao, L. Li, G. Zheng, Y. Chen, Y. Huang,
460–466. Q. Zhang, C. Shen and Q. Chen, Adv. Mater., 2017,
382 N. J. Jeon, J. H. Noh, Y. C. Kim, W. S. Yang, S. Ryu and 29, 1606774.
S. I. Seok, Nat. Mater., 2014, 13, 897–903. 405 C. Qin, T. Matsushima, T. Fujihara and C. Adachi, Adv.
383 N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo Mater., 2017, 29, 1603808.
and S. I. Seok, Nature, 2015, 517, 476–480. 406 L. Wang, H. Zhou, J. Hu, B. Huang, M. Sun, B. Dong,
384 D. Bi, C. Yi, J. Luo, J.-D. Décoppet, F. Zhang, S. M. G. Zheng, Y. Huang, Y. Chen, L. Li, Z. Xu, N. Li, Z. Liu,
Zakeeruddin, X. Li, A. Hagfeldt and M. Grätzel, Nat. Energy, Q. Chen, L.-D. Sun and C.-H. Yan, Science, 2019, 363, 265.
2016, 1, 1–5. 407 N. Li, S. Tao, Y. Chen, X. Niu, C. K. Onwudinanti, C. Hu,
385 M. Kim, G.-H. Kim, T. K. Lee, I. W. Choi, H. W. Choi, Y. Jo, Z. Qiu, Z. Xu, G. Zheng and L. Wang, Nat. Energy, 2019, 4,
Y. J. Yoon, J. W. Kim, J. Lee and D. Huh, Joule, 2019, 3, 408–415.
2179–2192. 408 X. Wang, K. Rakstys, K. Jack, H. Jin, J. Lai, H. Li,
386 H. Min, M. Kim, S.-U. Lee, H. Kim, G. Kim, K. Choi, C. S. K. Ranasinghe, J. Saghaei, G. Zhang and P. L. Burn,
J. H. Lee and S. I. Seok, Science, 2019, 366, 749–753. Nat. Commun., 2021, 12, 1–10.
387 G. Kim, H. Min, K. S. Lee, D. Y. Lee, S. M. Yoon and 409 H. Zhu, Y. Ren, L. Pan, O. Ouellette, F. T. Eickemeyer,
S. I. Seok, Science, 2020, 370, 108–112. Y. Wu, X. Li, S. Wang, H. Liu and X. Dong, J. Am. Chem.
388 M. Jeong, I. W. Choi, E. M. Go, Y. Cho, M. Kim, B. Lee, Soc., 2021, 143, 3231–3237.
S. Jeong, Y. Jo, H. W. Choi and J. Lee, Science, 2020, 369, 410 D. H. Kang and N. G. Park, Adv. Mater., 2019, 31, 1805214.
1615–1620. 411 N. Li, Y. Luo, Z. Chen, X. Niu, X. Zhang, J. Lu, R. Kumar,
389 J. Jeong, M. Kim, J. Seo, H. Lu, P. Ahlawat, A. Mishra, J. Jiang, H. Liu and X. Guo, Joule, 2020, 4, 1743–1758.
Y. Yang, M. A. Hope, F. T. Eickemeyer and M. Kim, Nature, 412 Y.-H. Lin, N. Sakai, P. Da, J. Wu, H. C. Sansom,
2021, 592, 381–385. A. J. Ramadan, S. Mahesh, J. Liu, R. D. Oliver and J. Lim,
390 J. J. Yoo, G. Seo, M. R. Chua, T. G. Park, Y. Lu, Science, 2020, 369, 96–102.
F. Rotermund, Y.-K. Kim, C. S. Moon, N. J. Jeon and J.-P. 413 Y. Yuan, T. Li, Q. Wang, J. Xing, A. Gruverman and
Correa-Baena, Nature, 2021, 590, 587–593. J. Huang, Sci. Adv., 2017, 3, e1602164.
391 H. W. Qiao, S. Yang, Y. Wang, X. Chen, T. Y. Wen, 414 S. Wu, R. Chen, S. Zhang, B. H. Babu, Y. Yue, H. Zhu,
L. J. Tang, Q. Cheng, Y. Hou, H. Zhao and H. G. Yang, Z. Yang, C. Chen, W. Chen and Y. Huang, Nat. Commun.,
Adv. Mater., 2019, 31, 1804217. 2019, 10, 1–11.
392 Y. Lei, Y. Chen, R. Zhang, Y. Li, Q. Yan, S. Lee, Y. Yu, 415 H. Yu, H. Lu, F. Xie, S. Zhou and N. Zhao, Adv. Funct.
H. Tsai, W. Choi and K. Wang, Nature, 2020, 583, 790–795. Mater., 2016, 26, 1411–1419.
393 E. Shi, B. Yuan, S. B. Shiring, Y. Gao, Y. Guo, C. Su, M. Lai, 416 J. Carrillo, A. Guerrero, S. Rahimnejad, O. Almora,
P. Yang, J. Kong and B. M. Savoie, Nature, 2020, 580, I. Zarazua, E. Mas-Marza, J. Bisquert and G. Garcia-
614–620. Belmonte, Adv. Energy Mater., 2016, 6, 1502246.
394 S. Xiong, Z. Hou, S. Zou, X. Lu, J. Yang, T. Hao, Z. Zhou, 417 I. M. Hermes, Y. Hou, V. W. Bergmann, C. J. Brabec and
J. Xu, Y. Zeng and W. Xiao, Joule, 2021, 5, 467–480. S. A. Weber, J. Phys. Chem. Lett., 2018, 9, 6249–6256.
395 S. Ma, G. Yuan, Y. Zhang, N. Yang, Y. Li and Q. Chen, 418 S. Kim, S. Bae, S.-W. Lee, K. Cho, K. D. Lee, H. Kim, S. Park,
Energy Environ. Sci., 2022, 15, 13–55. G. Kwon, S.-W. Ahn and H.-M. Lee, Sci. Rep., 2017, 7, 1–9.

10062 | Chem. Soc. Rev., 2022, 51, 10016–10063 This journal is © The Royal Society of Chemistry 2022
 View Article Online

Review Article Chem Soc Rev

419 P. Chen, Y. Bai and L. Wang, Small Struct., 2021, 2, 426 W. Shen, Y. Dong, F. Huang, Y.-B. Cheng and J. Zhong,
2000050. Mater. Rep.: Energy, 2021, 1, 100060.
420 M. Jamal, S. Shahahmadi, M. A. A. Wadi, P. Chelvanathan, 427 G. Liu, H. Zheng, H. Xu, L. Zhang, X. Xu, S. Xu and X. Pan,
N. Asim, H. Misran, M. Hossain, N. Amin, K. Sopian and Nano Energy, 2020, 73, 104753.
M. Akhtaruzzaman, Optik, 2019, 182, 1204–1210. 428 X. Wei, M. Xiao, B. Wang, C. Wang, Y. Li, J. Dou, Z. Cui,
421 Z. Sun, Y. Kang, G. Wang, M. Liang and S. Xue, J. Phys. J. Dou, H. Wang and S. Ma, Angew. Chem., Int. Ed., 2022,
Chem. C, 2020, 124, 12912–12921. e202204314.
Published on 18 November 2022. Downloaded by National Tsing Hua University on 5/23/2024 6:03:51 AM.

422 B. M. Sachith, T. Okamoto, S. Ghimire, T. Umeyama, 429 Q. Chen, H. Zhou, T.-B. Song, S. Luo, Z. Hong, H.-S. Duan,
Y. Takano, H. Imahori and V. Biju, J. Phys. Chem. Lett., L. Dou, Y. Liu and Y. Yang, Nano Lett., 2014, 14, 4158–4163.
2021, 12, 8644–8651. 430 A. Krishna, S. Gottis, M. K. Nazeeruddin and F. Sauvage,
423 Y. Zhong, M. Hufnagel, M. Thelakkat, C. Li and S. Huettner, Adv. Funct. Mater., 2019, 29, 1806482.
Adv. Funct. Mater., 2020, 30, 1908920. 431 M. A. Mahmud, T. Duong, J. Peng, Y. Wu, H. Shen,
424 C. Zhang, Y.-N. Lu, W.-Q. Wu and L. Wang, Nano Energy, D. Walter, H. T. Nguyen, N. Mozaffari, G. D. Tabi and
2021, 81, 105634. K. R. Catchpole, Adv. Funct. Mater., 2022, 32, 2009164.
425 Y. Wu, D. Wang, J. Liu and H. Cai, Nanomaterials, 2021, 432 A. Mei, Y. Sheng, Y. Ming, Y. Hu, Y. Rong, W. Zhang, S. Luo,
11, 775. G. Na, C. Tian and X. Hou, Joule, 2020, 4, 2646–2660.

This journal is © The Royal Society of Chemistry 2022 Chem. Soc. Rev., 2022, 51, 10016–10063 | 10063