A lot of work has been made in this area since 2012, when solid-state perovskite solar cells (PSCs) were first made. In 2009, the organic-inorganic lead halide perovskite was added as a sensitiser to a dye-sensitized solar cell structure. This led to a 25.2% power conversion efficiency (PCE) that was confirmed in 2019. MAPbI3 (MA = methylammonium) has been used in the early stages of PSCs because it is easy to make and stays stable in its structure at room temperature. But problems with temperature stability have come up because the organic MA ion breaks down and evaporates.
People think that FAPbI3 (FA = formamidinium) is more stable at high temperatures because the FA ion has a stable chemical resonance structure. It also works better as a solar material because its bandgap is smaller, at 1.47 eV. Smaller cations or anion are added to the FA site and/or I site to fix phase instability, which leads to higher PCEs.
However, strange feedback in the relationship between photocurrent density (J) and voltage (V) has been found. This hysteresis is caused by flaws in the perovskite itself, charge buildup at surfaces, ferroelectric behaviour, and the movement of ions. Defect engineering is important to lower or get rid of hysteresis because it is linked to flaws and can cause perovskite to break down. Hysteresis must be reduced or eliminated in order to improve both the performance and safety of solar systems.
KI doping in the perovskite layer is one way that has been tried to lower or get rid of hysteresis. In situ photoluminescence is talked about as a way to show that ions are moving.
Hysteresis in Current–Voltage Performance
The difference of J–V curves recorded in two different scan directions, called forward scan (FS) and reverse scan (RS), was first reported by Snaith et al. [11]. They saw a big change in fill factor (FF) because Jsc dropped a lot during FS. Because of this, the gap between the J and V shapes makes the average PCE go down. The J–V hysteresis also has an effect on stability as it is linked to flaws in the bulk and/or at the surfaces.
First, a mesoscopic PSC with a mesoporous TiO2 layer and a flat PSC without mesoporous TiO2 were used to study the hysteretic effect in more detail. To understand the J–V hysteresis, a device should be scanned at a slow enough rate to give temporary processes time to settle. The J–V hysteresis was caused by a number of things, such as the direction of the scan, the rate of the scan (measurement delay time is a more useful metric than scan rate [23]), light, voltage preconditioning, and others.
We measured the current and voltage step-by-step, as well as at different delay times and after light soaking at different bias voltages. We also looked at the FB-SC (SC-FB) current-voltage curves of perovskite solar cells at different scan directions, such as forward scan (FS) and reverse scan (RS). The steady-state Jsc and PCE (here, PCE is shown by 𝜂) were measured at the maximum power point (MPP).
The perovskite layer has ferroelectricity that is linked to a ferroelectric domain that includes a non-centrosymmetric lead halide octahedral structure like MAPbI3 [24] and the long-range ordering of dipoles of organic cations by an electric field [25]. Because VOC is proportional to Vin, polarisation (ΔP) caused by ferroelectric behaviour might lower built-in potential (Vin). This makes VOC go down, which leads to a J–V loop. But J–V hysteresis was seen in a device that used PbI2 (a non-ferroelectric material) as a light-absorbing layer. This clearly shows that the ferroelectric property is not the only reason for the J–V hysteresis. So, ion movement caused by flaws at the interface and gain boundary can be seen as either a major cause of the hysteresis or a competing cause.
Defects in crystallography can be point defects (like gaps, interstitial defects, and substitution defects), line defects (like screw dislocation and edge dislocation), and surface defects (like surface and grain border). Point defects in MAPbI3 are caused by iodine (I) interstitials (Ii), MA on Pb sites (MAPb), MA vacancies (VMA), Pb vacancies (VPb), MA interstitials (MAi), Pb2+ on MA sites (PbMA), I vacancies (VI), and MA on I sites (MAI). It was worked out that those point defects had low defect formation energies, which meant that they made shallow traps. When we think about how photoexcited carriers move through the trapping and de-trapping process, the carriers are drawn into the traps by light before they are released, because the de-trapping process moves more slowly.
Capacitive current (Jcap), which builds up charge carriers in the gadget, can be caused by the trap-filling process. This leads to J–V hysteresis because the photocurrent density changes based on the direction of the scan (J = J0 ±Jcap). Because carriers are lost through non-radiative recombination, J–V hysteresis caused by flaws can also make the device less stable.
There is also the idea that the J–V hysteresis can be caused by the movement of ions. With the heavy elements lead and iodine in lead halide perovskite, it may not be clear how it can act as an ion carrier. However, research into the hysteresis in the perovskite solar cell has shown that it can be changed by a number of things, such as the presence of flaws in the perovskite layer and the movement of ions.
Material and Structure Design to Reduce Hysteresis
It’s still not clear where J-V hysteresis comes from, but many ways have been found to lessen or get rid of it. This chapter talks about different ways to get rid of or reduce hysteresis. These include engineering the grain size, engineering the makeup, post-treatments of perovskite films, selective contact materials, engineering the grain border with chemicals, engineering the interface, and engineering defects.
Grain Boundary Engineering
Grain borders are thought to be places where non-radiative recombination can happen because of ion flaws that are not aligned. Because ions move more along grain boundaries than inside grains, creating grain boundaries is a way to lower hysteresis. Another way to passivate grain borders is to make the grains bigger or add extra or added source material (non-stoichiometry).
How much J–V feedback there is depends on how big the perovskite crystals are. In a two-step spin-coating method, the percentage of the MAI solution can be changed to change the average grain size of MAPbI3 from 130 to 440 nm. The hysteresis index (HI) from Equation (7.1) shows the amount of J–V hysteresis.
A method that doesn’t use stoichiometry and a precursor solution that has too much MAI was shown to be an effective way to passivate the MAPbI3 grain border. The drawing method consistently gives a PCE of about 20%. Transmission electron microscopy (TEM) studies showed that extra MAI that had not been reacted with formed on the grain surface. A fourier transformation of a 0.31 nm lattice fringe shows the MAPbI3 phase when there is no extra MAI. On the other hand, a new 0.36 nm lattice fringe seen in the sample with extra MAI shows the (110) plane of MAI.
A long heating time causes grain boundary passivation without changing the grain size in grain boundary engineering. By making the heating time longer, a thin layer of PbI2 can form on the grain surface. This lowers J–V hysteresis by preventing non-radiative trapping.
Interfacial Engineering
In PSCs, the perovskite layer touches an electron-transporting layer (ETL) and a hole-transporting layer (HTL) directly. This creates the surfaces between the perovskite and ETL and the perovskite and HTL. This can cause J–V feedback by building up charge. To cut this down, interface engineering is needed. The J–V hysteresis was looked at using different contact materials and device structures. For example, PEDOT:PSS, NiO, spiro-MeOTAD, TiO2, and PCBM were used with either normal or upside-down p–i–n structures.
What the study found is that J–V hysteresis is strong in the normal device structure with a TiO2 ETL and a spiro-MeOTAD HL, but not at all in the reversed device structure with a PEDOT:PSS HTL and a PCBM ETL. It is seen that ¡Jsc/t changes a lot for the normal structure, but it stays the same over time for the reversed structure.
In order to lower hysteresis in the device with a TiO2 HTL and a CuSCN HL, a 2D stacked perovskite ((5-AVAI)2PbI4, 5-AVA = 5-ammoniumvaleric acid) was added to the point where the perovskite and CuSCN layers meet. When compared to when the gadget didn’t have the interface layer, it had a lot of hysteretic behaviour. The interface layer greatly lowers this.
It is known that TiO2 is more connected to hysteresis in a normal shape with a TiO2 layer and a spiro-MeOTAD layer. To lower this, the TiO2 film needs to be changed, and a strong connection between the perovskite layer and the TiO2 layer through a chlorine-capping method was found to help lower the hysteresis.
Defect Engineering
Ion migration is a key step in making flaws. Halide ions move to the next empty space and pass through unstable intermediate sites. To control ion movement and lower hysteresis, defect engineering is a must. It is possible to do this with chemical engineering or liquid engineering.
In the precursor solution, 1,8-Diiodooctane (DIO) was added to temporarily bind Pb2+ with DIO. This made a solid that didn’t have any flaws. Because there are fewer defects, this method improves carrier lifetime and charge separation. It was discovered that light can make ion movement faster in MAPbI3. This movement can be slowed down by switching the organic MA cation for an artificial Cs cation. The activation energy (Ea) for ion migration, also known as the energy barrier of ion migration, is a good way to measure how much ions are moving.
The slope of the 𝼏ionT vs. 1/T plot can be used to find the activation energy since 𝼏ionT = 𝼏0exp(Ea/kT). To get the ionic conductivity, take the total conductivity from the side device arrangement in Figure 7.7d and take the electronic conductivity away. CsPbI2Br’s Ea values stay almost the same, going from 0.45 eV at 0.1 mW/cm2 of light to 0.43 eV at 25 mW/cm2.
It is talked about how adding alkali cations to MAPbI3 changes it. The energy barrier of ion movement in CsPbI2Br is not affected by light, so there is no J–V hysteresis even after 1500 hours of MPP measurement. For MAPbI3, on the other hand, serious hysterical behaviour starts after only 50 hours. It was suggested that adding Br to the I site lowers the hysteresis in MAPbI3. The hysteresis decreased greatly as the Br concentration rose. This was because the higher Ea was caused by local structure warping, as proven by density functional theory (DFT) calculations.
Effect of Alkali Cation Doping
Reduction in Hysteresis by KI Doping: A Universal Approach
By adding K+ ions, the PCE of MAPbI3-based PSCs was raised because the longer carrier lifetime and better crystallinity made the materials better. The methylammonium cation was supposed to become vacant when the potassium ion occupied the intermediate space. This could shorten the c axis and move the (040) plane to a lower angle. But not as much research was done on the J–V feedback when KI was added. It was reported that adding the micromole of KI in the perovskite precursor solution was a general way to get rid of the hysteresis.
It was looked into how adding KI changed the J–V hysteresis for various perovskite material types. The big hysteric J–V curves in the pure perovskite films are greatly reduced or erased when 10 μmol KI is added, and this is true no matter what kind of perovskite is used. The J–V hysteresis was found to be affected by the concentration of KI. A concentration lower than 10 μmol greatly lowered the hysteresis. The hysteresis and lower performance were caused by a higher concentration of KI (more than 20 μmol). When the efficiency and the inertia were looked at together, 10 μmol was found to be the best value.
It was also looked into how alkali iodides affected things, but only KI worked best. The J–V hysteresis is still very clear for smaller cations like Li+ and Na+, but it tends to get weaker as the ionic radius gets bigger. For bigger cations like Rb+ and Cs+, the J–V hysteresis tends to get worse as the ionic radius gets bigger. This shows that the ionic radius in alkali iodides is linked to the J–V hysteresis.
In addition, adding KI might make flaws less noticeable and stop ions from moving around. DFT was done on MAPbI3 as a model molecule, and the I_Frenkel flaw was found to be energetically favourable in MAPbI3.
We look at what happens to the J–V hysteresis in perovskite (PSC)-based PSCs when alkali iodides are added, both before and after KI doping. It is thought that the intermediate Oh sites filled by alkali metal ions will guard the Frenkel flaw, but this isn’t clear. The cation/anion radius ratio between K+ and I− is 0.616, which shows that the K+ ion is stable in the Oh site, where it lies between 0.414 and 0.732. Na+ (0.449) and Rb+ (0.681) are also in the range, but they are not as useful as K+ because they are on the edge, with four coordination (below 0.4141) and eight coordination (above 0.732). The gaps between the OH sites are too small for Li+ (0.315) and too big for Cs+ (0.782).
The EIS analysis in Figure 7.9c,d shows that the J–V hysteresis is mostly caused by the bulk defect rather than the interface. This is because the capacitance frequency in the 10−1–∼104 Hz range behaves the same before and after KI doping, even though different interfaces were used, such as FTO/m-TiO2/perovskite/spiro-MeOTAD/Au and FTO/perovskite/Au. The space-charge-limited current (SCLC) equation in Equation (7.2) was used to measure the trap density (nt) of pure perovskite with and without the K+ ion. The relation 𝼀 = Cd/A𝼀0 can be used to figure out the trap density. In this equation, A is the area and C is the physical capacitance taken from the EIS high frequency region (about 104 Hz).
The dark current–voltage (I–V) curves in Figure 7.9e and f show that nt goes from 1.367 × 1016 cm−3 for the pure perovskite to 0.846 × 1016 cm−3 after 10 μmol KI doping. This is because VTFL goes from 0.788 V to 0.617 V after KI doping. This shows that the surfaces are not as good at protecting the Frenkel flaw as the intermediate Oh sites that are filled by alkali metal ions.
Passivation Effect of Excess KI
The research looks into how potassium ions (KI) can fix surface flaws in a perovskite material made of (Cs0.06FA0.79 MA0.15)Pb(I0.85Br0.15)3. The amount of KI changed based on the number of monovalent cations in the starting solution. The pure perovskite grains were thought to be lacking in halide anions, so KI was used to fill them up with potassium halide. The best quantity was 0.4x. Time-resolved photoluminescence (TRPL) showed that KI greatly slowed down the fast non-radiative monomolecular recombination in the pure perovskite, which caused a large rise in radiative bimolecular recombination. When the light was on for a longer time, the potassium passivated film (x =0.4) had a higher photoluminescence quantum efficiency (PLQE) and stayed that way longer than the pure film (x =0). The PLQE slowly rises over time for the perovskite film that has not been passivated. This may be because of photoinduced halide movement. The steady-state PL doesn’t change its PL peak position after being passivated with KI, but the pristine perovskite (x = 0) shows large redshifts over time. This suggests that the mixed composition is unstable under continuous illumination because of phase segregation and/or photoinduced trap formation.
Location of Potassium Ion in Perovskite
A study was done to find out where in the perovskite film the K+ ion is located. For the K+ ion to be in the lattice, it can either substitute at the A site or move between the atoms at the intermediate site. It was found that smaller cations, like K+, were thermodynamically stable at the intermediate spot. The iodide diffusion routes were used to figure out the iodide movement hurdles. The interstitial K+ made it harder for iodide to move along all routes. This suggests that the presence of K+ in the interstitial site of the perovskite lattice can stop iodide from moving. It was also suggested that when K+ was added to the lattice, FA ions changed direction. This made the iodide diffusion barrier higher because the dipole moment changed to point more towards K+. By adding 1% KI to Cs0.11MA0.15FA0.74Pb(Br0.17I0.83)3, the J–V hysteresis was greatly reduced. This was because the interstitial K+ made it harder for iodide to move around. Theoretical studies showed that the foreign K+ ion would fit well in the intermediate space, but there is still disagreement about where alkali metal ions are most likely to be found. However, it is clear that KI doping greatly lowers the J–V hysteresis.
In situ Photoluminescence (PL) as a Tool to Measure Ion
Migration Kinetics
An in situ PL imaging device was used to see ion movement in perovskite film. We used a home-built microscope and a charge-coupled device (CCD) camera to take a picture of the PL image of a big area that is about 60 μm across. A PL image can be seen with an exposure time of as little as 50 ms per picture. Inset in Figure 7.12g shows a diagram of the in situ PL imaging setup. To take in situ PL pictures, a lateral structure with two Au electrodes on the perovskite film spaced 200 μm apart was used.
The bright picture is caused by PL emission, and the dark area is made up of photogenerated charge carriers. The dark area gets lighter over time when the electric field stays the same because the dark layer on the positive side grows. The dark layer grows from the positive electrode to the negative electrode. This is because negative ions, like iodide anion, move and build up. Figure 7.12g shows the movement speed as a function of electric field (E). This helped us figure out that the iodide ion’s mobility () was about 5 1012 m2/V s, which is equal to 5 108 cm2/V s. It was also calculated that the diffusion coefficient (D) for iodide is about 10−13 m2/s at room temperature, which is one order of magnitude less than the measured and calculated value of 10−12 m2/s.
The KI doping almost got rid of the J–V hysteresis, so it is interesting to compare the in situ PL images of the perovskite films before and after the KI doping. We looked at how adding KI changed in situ PL images with a perovskite makeup of (FAPbI3)0.875(CsPbBr3)0.125 to figure out how the J–V hysteresis was taken away.
Because iodide moves, as expected, the PL dark area grows at the positive electrode as the bias voltage goes up. But even at a high bias voltage of 0.5 V, it is not possible to clearly see the PL dark layer getting thicker on the positive electrode. This is a clear sign that KI cheating is stopping the movement of ions.
To do this, the electrochemical impedance spectroscopy (EIS) method was used to find ionic diffusion. This was done by drawing a 45° line for the Warburg diffusion element, a vertical line for ion accumulation, and an ankle between the two areas with the typical frequency of diffusion (d). For the pure perovskite that hasn’t been doped with KI, the characteristic frequency of d is seen to be 0.17 Hz. For the KI-doped perovskite, on the other hand, d is seen to be much lower at 0.01 Hz, and there is no noticeable ankle. This study also shows that it’s not impossible for K+ to be present at the perovskite surface.
Summary
Photocurrent density-voltage (J-V) hysteresis happens when charges build up at perovskite/selective contacts. This can happen because of ion movement and/or the ferroelectric effect. Several ideas have been put forward to lower hysteresis, including grain boundary engineering, interface engineering, and flaw engineering. Grain boundary engineering includes making grains bigger to cut down on the number of grain borders and manage how fast they grow. Post-treatment of selected contacts and/or perovskite films is part of interface engineering. Compositional engineering or liquid engineering are used to make perovskite films, which are the centre of defect engineering.
A substance called KI has been shown to greatly reduce or get rid of J-V hysteresis under certain testing settings, as shown in most studies. How much J-V hysteresis there is may depend on how much KI there is and what kind of perovskite it is made of. On the other hand, the micromole size of KI can greatly decrease J-V hysteresis, no matter what kind of perovskite is used.
In situ PL imaging showed that J-V hysteresis is closely linked to iodide movement. The hysteresis-free perovskite with KI stopped ions from moving and stopped anions from building up on the positive electrode. The most likely reason is that K+ is occupied in the spaces between the perovskite lattice atoms, but it’s also possible that K+ is found close to the interface if J-V hysteresis is connected to interfacial capacitance.
Leave a Reply