In the development of hydrogen-based technology, a key challenge is the sustainable production of hydrogen in terms of energy consumption and environmental aspects. However, existing methods mainly rely on fossil fuels due to their cost efficiency, and as such, it is difficult to be completely independent of carbon-based technology. Electrochemical hydrogen production is essential, since it has shown the successful generation of hydrogen gas of high purity. Similarly, the photoelectrochemical (PEC) method is also appealing, as this method exhibits highly active and stable water splitting with the help of solar energy. We discuss the exceptional optical and electrical characteristics of perovskite materials which often dictate PEC performance. We further extend our discussion to the material limit of perovskite under a hydrogen production environment, i.e., that PEC reactions often degrade the contact between the electrode and the electrolyte.
The general molecular formula of perovskite is ABO
3
; perovskite materials have a cubic lattice-nested octahedral layered structure
These materials have superior magnetic, ferroelectric, electrical, and optical properties, attracting attention for potential use in optoelectronic devices. At present, many research groups are actively investigating these characteristic properties, and recently, nuclear magnetic resonance (NMR) spectroscopy has been significantly used for analyzing the structure of these perovskite materials
[1]
. Perovskite materials also have low recombination probability and high carrier lifetimes and diffusion lengths
. Therefore, metal-organic halide perovskites have been amongst the most interesting subjects in optoelectronic materials research, and are utilized in various fields such as light-emitting diodes, solar cells, lasers, and photodetectors
.
The metal-organic halide perovskite has a molecular structure of type ABX
3
, where A and B are cations (A is larger than B), and X is the anion. The general crystal structure of perovskite is shown in Figure 1a–c
. Perovskites commonly have unit cells consisting of five atoms in a cubic structure (α phase), where the A cation (methylammonium, CH
3
NH
3+
, MA
+
, or formamidinium, CH(NH
2
)
2+
, FA
+
) is surrounded by twelve X anions (Cl
−
, Br
−
, or I
−
, or a coexistence of several halogens) to form a cuboctahedron, and the B cation (Pb
2+
, Sn
2+
, etc.) is located at the octahedral site of X. The B cation–X anion octahedra are joined together to form stable three-dimensional network structures
. To achieve a cubic structure with a high degree of high symmetry, i.e., an ideal perovskite crystal structure, the radius ratio of A, B, and X should be such that tolerance factor {t = (R
A
+ R
B
)/ R
B
+ R
X
)} is close to 1
[11]
. For the tolerance factor to approach 1, the A cation must be much larger than the B cation. In metal-organic halide perovskite, the A site must be occupied by a very large atom because the B site is usually occupied with a large atom such as Pb or Sn. Tolerance factors of cubic structures are generally between 0.89 and 1, where a tolerance factor lower than 0.89 could induce a tetragonal (β phase) or orthorhombic (γ phase) structure, whereas higher tolerance factors could induce a two-dimensional (2D) layer structure due to the unstable three-dimensional (3D) B–X bonding
. Actually, the transverse phonon easily displaces the X anion from the B–B intermediate location of the cubic structure
. The non-perovskite δ phase appears in perovskite materials such as HC(NH
2
)
2
PbI
3
, FAPbI
3
, CsPbI
3
, and CsSnI
3
, which, unlike the β and γ phases, is caused not by B–X–B angle distortion in the α phase, but by the breaking of the B–X bond
.
Figure 1.
Crystal structure of perovskite materials. (a) A typical perovskite crystal structure. (b) Crystal structure of B cation and X anion (B cation is located at the octahedral site of X anion). (c) Crystal structure of A cation and X anion (A cation is located at the octahedron site of X anion.
As one of the unique properties of metal-organic halide perovskites, the optical properties of photo-generated charge carriers have been researched. The specific excitonic absorption peaks of the metal-organic halide perovskite could be transited to various absorption spectra, and it changed significantly in visible light through the adjustment of metal atoms and halogens (Figure 2a, b). Converting the metal atom directly changes the M–X bonding, resulting in changes in valence band maximum and conduction band minimum, as the band edge is determined by the metal orbitals of the B site. In fact, in perovskite, the valence band is composed of 5p orbitals of I and the s-antibonding states of Pb 6s. The conduction band is composed of 5s orbital of I and the s-antibonding states of Pb 6p. For example, when the B site is changed from Pb to Sn in MAPbI
3
, the bandgap changes from 1.57 to 1.17 eV, and therefore, the bandgap between 1.57 eV and 1.17 eV can be controlled by adjusting the ratio of Pb and Sn
. Additionally, converting the A organic cations changes the length and angle of the M–X–M bonding, which, in turn, changes the bandgap but does not affect the valence band maximum. In fact, MAPbI
3
shows a rapid absorption rise at 825 nm (1.5 eV) and has a large absorption coefficient of 1.8 × 104 cm
−1
. By substituting the methylammonium (MA) cation with formamidinium (FA), the energy bandgap could be lowered to 1.48 eV, which reduces the bandgap by 0.09 eV and allows for additional light absorption
[22]
. Metal-organic halide perovskites are widely used in the optoelectronics field due to their excellent and wide-ranging absorbance in the visible light range
[23]
. Furthermore, metal-organic halide perovskites could exhibit amplified spontaneous emission due to the low defect density and slow Auger recombination, even in the presence of electron and hole extinctions
. The absorption coefficient of perovskite is affected by temperature, tending to decrease at lower temperatures. This is associated with the interaction of excitons and phonons at low temperatures and the phase transition of MAPbI
3
at about 160 K (Figure 2c)
[26]
. Similar to light absorption, the photoluminescence (PL) of metal-organic halide perovskite could be tuned by changing organic cations or anions. In the case of MAPbI
3
quantum dot (QD), the PL spectrum could be tuned from 407 to 734 nm by changing the composition of anions (Figure 2d)
[27]
. Due to its properties to tune the strong PL spectrum, perovskite is utilized in a variety of light-emitting applications including lasers, light-emitting diodes, and optical sensors
.
Figure 2.
Optical properties of perovskite materials. (a) UV-vis spectra for the APbI3 perovskites formed, where A is either cesium (Cs), methylammonium (MA) or formamidinium (FA). (b) UV-vis absorption spectra FAPbI
y
Br
3-y
perovskites. (c) Temperature dependence of the absorption coefficient of methylammonium lead iodide (MAPbI
3
) extracted from PL spectra. (d) Near-bandgap absorption and photoluminescence spectra at room temperature mixed-organic perovskites
.
A time-dependent photoluminescence (TDPL) measurement was performed to determine the carrier lifetime (τ) of single crystal FAPbI
3
. As a result, a peak was observed at 820 nm. The two-exponential decay showed fast (τ
1
= 32 ns) and slow (τ
2
= 484 ns) carrier lifetimes. A short lifetime represents a high trap density on the crystal surface, while a long lifetime represents carrier transporting in bulk with fewer defects. Generally, FAPbI
3
has a lower carrier lifetime than single-crystal MAPbI
3
. In the case of carrier mobility, it shows change with the perovskite phase transition. The mobility was measured at 35 cm at room temperature; the mobility tends to decrease with increasing temperature at all three steps. Most experimental values follow the T
−3/2
dependency without discontinuity in phase transitions, and show 150 cm2 V
−1
s
−1
at 80 K. Mobility is relatively high when the temperature is above room temperature or when entering the cubic phase
. In addition, the change of carrier mobility is shown by the perovskite layer thickness. Zhang et al.
[38]
characterized the photovoltaic power conversion efficiency (PCE) with different MAPbI
3
layer thicknesses to investigate the effect of film thickness on perovskite solar cell devices. The PCE increases as the film thickness approach 300 nm and tends to decrease rapidly in the 300–530 nm range. The PCE has a maximum of about 10.16% at 300 nm thickness and decreases to 5.29% at 530 nm. This allowed us to determine the optimal thickness of the MAPbI
3
layer. When thin, pinholes result in low V
OC
and shunt resistance, while increasing the thickness of the film eliminates this contact, thereby improving V
OC
. J
SC
is affected by light-harvesting efficiency(η
lh
) and carrier injection efficiency(η
inj
). The increase in J
SC
with increasing thickness up to 300 nm indicates that the increase in the absorber layer thickness results in higher absorption and higher η
lh
.
The instability of metal-organic halide perovskite in humid environments is the biggest obstacle to its applicability in devices, given its other notable properties. The perovskite film is highly sensitive to the presence of water, which affects the stability of devices in which it is used. In addition, perovskite can degrade at the polar solvent. The degradation of perovskite has been reported due to the highly hygroscopic property of amine salts
[42]
. Due to the high hygroscopicity of organic compounds, when perovskite is exposed to water, organic compounds detach from the crystal structure, and the perovskite crystallinity is changed. This results in changes in the optical and electrical properties causing variations in the performance of perovskite-based devices
[43]
. In other words, due to the degradation of perovskite by water, the noteworthy optical and electrical properties of perovskite are lost, and the device performance is degraded. The degradation mechanism due to moisture is shown in Figure 3
[44]
.
Figure 3. Degradation pathway of metal-organic halide perovskite materials in water.
CH₃NH₃PbI₃ (s) → CH₃NH₃PbI (aq) + PbI2 (s) |
(3) |
CH₃NH₃PbI (aq) → CH₃NH2 (aq) + HI (aq) |
(4) |
4HI (aq) + O2 (g) → 2I2 (s) + 2H2O (l) |
(5) |
2HI (aq) → H2 (g) + I2 (s) |
(6) |
In fact, perovskite readily transforms to monohydrate phase MAPbI
3
∙H
2
O in moderate humidity (RH < 60%), and to dihydrate phase (MA)
4
PbI
6
∙2H
2
O in high humidity (RH > 80%). This can be explained by the fact that the hydrogen bonding interaction between the lead iodide framework and organic MA
+
cations is weakened by hydration. As a result, MA is diffused and detached from PbI
6
octahedra, whereby MAPbI
3
degrades rapidly. The activation barrier for vacancy-mediated MA
+
migration is reduced from 1.18 eV for MAPbI
3
to 0.38 eV for water-intercalation, and 1.14 eV for the monohydrated phase. When MAPbI
3
is exposed to an aqueous solution, it degrades to PbI
2
precipitate, iodide anion, and methylammonium cation
.
Kye et al.
[44]
identified perovskite degradation in water through a point defect process by density-functional theory (DFT) calculation. Due to the kinetic barrier for I
−
, ion migration becomes very low when hydrated, and V
PbI2
formation occurs spontaneously. In the hydrous compound, the formation of V
I
and V
MA
is preferred to the formation of V
MAI
, so that during the MAPbI
3
degradation, the formation of I
2
or CH
3
NH
2
or HI is higher than that of MAI (Figure 4a). Unlike bulk MAPbI
3
, all vacancy defects form deep transition levels through electrostatic interaction with water molecules (Figure 4b).
Several methods have been identified to reduce the degradation of perovskite solar cells (PSC) by water. The first is to add a thin blocking layer like Al
2
O
3
between the perovskite and hole transporting material (HTM)
. The second is to use moisture blocking HTM. The last method is to use hydrophobic carbon electrode
.
Figure 4.
DFT calculation of perovskite’s degradation pathway in water. (a) Formation enthalpies of vacancy point and pair defects as a function of the Fermi energy (EF) under I-poor (Pb-, MA-rich) conditions (left) and I-rich (Pb-, MA-poor) conditions (right). (b) Band alignment and thermodynamic transition levels in MAPbI
3
, water-intercalated MAPbI
3
_H
2
O, and monohydrate MAPbI
3
·H
2
O, where deep-lying Pb 5d levels are used as a reference for the VBM and CBM of each phase
[44]
.
The electrochemical reaction is based on the junction of the electrode and the electrolyte. In electrochemical water splitting, a hydrogen evolution reaction (HER) occurs at the cathode, and an oxygen evolution reaction (OER) occurs at the anode. The oxidation and reduction reactions for water splitting are as follows.
2H2O + 2e−→H2(g) + 2OH−, Eo = 0.00 V (HER) |
(1) |
2OH−→1/2 O2(g) + H2O + 2e−, E = 1.23 V (OER) |
(2) |
The theoretical potential required for electrochemical water splitting is 1.23 V. However, due to the activation energy required for the reaction, more than 1.23 V is required in practice. In other words, the actual response requires a theoretical value of 1.23 V or higher. The additional potential required is called “overpotential”. Researchers are continuously endeavoring to reduce the overpotential
.
The mechanism by which photoelectrochemical water splitting occurs is shown in Figure 5. Minority carriers generated by semiconductor light absorption are induced into the solution by the electric field at the junction. The electrons and holes generated by the semiconductor are transferred to the molecular catalyst to induce the OER and HER reactions, where, typically, p-type semiconductors are used for water reduction and n-type semiconductors are used for water oxidation. Water reduction is caused by one or two electron steps at the metal center M
n+
, which is followed by the reduction of M
n+
and, subsequently, followed by protonation, to give an intermediate hydride. Monometallic and bimetallic pathways occur independently or in parallel, and depend on the catalyst properties, its reduction potential, pK
a
for deprotonation, and pH. In water oxidation, O–O-forming mechanisms are induced by the interaction between two M–O groups (I2 M) or nucleophilic attack by water (WNA). I2 M mechanisms are achieved by reductive coupling and reductive removal, or by radical coupling through inner/intramolecular pathways. In the WNA mechanism, when the M–O is sufficiently electrophilic, the water molecules attack the M–O to form O–O bonding
.
Figure 5.
Photoelectrochemical processes involved in (
a
) water reduction and (
b
) water oxidation. (
c
) Proposed mechanistic pathways for H
2
generation at a metal center of M
n+
. (
d
) Two mechanistic pathways to form an O–O bond for the molecular catalysts. (ref. [55]).
The perovskite photoelectrochemical cell consists of a perovskite light absorption layer, an electron transporting layer, and a hole transporting layer (HTL) to extract the generated electron and hole, and a passivation layer to prevent perovskite degradation in aqueous solution. A simplified working principle of the device may be described as follows: when the light falls on the device, the perovskite layer absorbs the light and generates excitons. The electron and hole pairs (EHPs) are created by the thermal energy, which is diffused and separated through the electron and hole transporting layer, respectively
[55]
. The performance of the perovskite-based device is influenced by the diffusion length, lifetime, and mobility of the generated carriers. The diffusion length of the perovskite depends on the qualities such as the crystallinity and grain size of the perovskite film, indicating that the diffusion length varies based on the perovskite preparation method. In the case of MAPbI
3
, as the diffusion length of the hole is longer than that of the electron, mesoporous TiO
2
is used to compensate for this short electron diffusion length. Furthermore, performance is improved when the injected charge mobility is fast. Therefore, designing the charge transporting materials and thickness in consideration of the diffusion length and mobility of the charge is one way to improve performance
.
The basic function of the electron transporting layer is to enhance the transport of photo-generated electrons through electron-selective contact with the perovskite layer and to prevent hole injection, so as to improve carrier separation and reduce recombination. Generally, TiO
2
, ZnO, [6,6]-phenyl-C
61
-butyric acid methyl ester (PCBM), etc. are used as electron transporting materials. Though TiO
2
is the most commonly used inorganic electron transporting layer (ETL) material and its electron injection rate is very fast, it has limitations, as recombination occurs due to low electron mobility
. While ZnO has higher electron mobility than TiO
2
(Bulk mobility: 205–300 cm
2
V
−1
s
−1
), its chemical instability is a drawback
. PCBM is a conductive polymer material with low photocurrent hysteresis and high short circuit current density. PCBM plays a critical role in improving the quality of the light-absorbing layer by filling the pinholes and vacancies between perovskite grains, resulting in a film with large grains and fewer grain boundaries
.
The main function of the hole transporting layer is to improve electron-hole pairs separation by collecting and transporting the generated hole from the perovskite layer. The material used for the hole transporting layer is spiro-OMeTAD, poly (3,4-ethylene dioxythiophene) polystyrene sulfonate (PEDOT:PSS), carbazole functionalized with a naphthalene (Cz-N) or pyrene (Cz-Pyr)
[67]
and CuI. Although Spiro-OMeTAD is one of the best performing HTL materials, its cost can be an impediment, and it is also a small molecule substance which diffuses well into perovskite
. PEDOT:PSS also exhibits good performance and is a conductive polymer material like PCBM. Alternatively, while CuI has about five times the hole mobility of spiro-OMeTAD and a relatively large particle size, signifying that it does not readily diffuse into the perovskite layer, its shortcomings are that the surface is rough and it is not well bonded with perovskite due to the large particles
.
Perovskite-based device configuration is derived from the dye-sensitized solar cell, and is classified into two types. One of them comprises the transparent conductive substrate coated with the electron transporting material (mesoporous/planar n-type material) under the perovskite layer and hole transporting layer (p-type material) deposited on the perovskite layer. This configuration, which is called n–i–p type, strives for an oxygen evolution reaction. In n–i–p type devices, electron-hole pairs (EHPs) generated in the perovskite layer are transported by the electron transporting material to the transparent conductive oxide (TCO) substrate, and through the hole transporting material, the holes are transported to the metal electrode (Figure 6a). Holes that reach the metal electrode through this mechanism react with oxygen ions to produce oxygen. The other common configuration is inverted compared to the former, and the transparent conductive substrate is coated with the hole transporting material, followed by the perovskite layer and electron transporting layer. This configuration, which is called p–i–n type, strives for a hydrogen evolution reaction. In the p–i–n type perovskite photoelectrochemical cell, a hydrogen evolution reaction occurs by moving electron-hole pairs (EHPs) in a converse mechanism to that of n–i–p type perovskite PEC devices. The EHPs generated in the perovskite layer are transported by the hole transporting material to the TCO substrate, and through the electron transporting material, the electrons are transported to the metal electrode (Figure 6b). The electrons are transported to the metal electrode to produce hydrogen through the reaction with the electrolyte.
Figure 6.
The operating mechanism of perovskite-based photoelectrochemical device (
a
) n–i–p configuration, (
b
) p–i–n configuration.