The Emergence & Developmentof Perovskite Based Photovoltaics

Michael StringerRegistration Number, 140148240

PhD Supervisor, Prof. David Lidzey.

The University of Sheffield,Department of Physics & Astronomy.

EPSRC Centre for Doctoral Training,‘New and Sustainable Photovoltaics’

Abstract

In recent years perovskite based solar cells have become a promising photovoltaic technology, most notablefor their high power conversion efficiencies and potential for cheap solution-processable module produc-tion. This report identifies the ongoing research and development into the behaviour of the perovskitematerial as a semiconducting active area and the ever improving fabrication of perovskite solar cell devices.

The largest problems and unanswered questions reviewed in this report include the long term sta-bility of perovskite devices, and the challenge associated with up-scaling production of perovskites intohigh performance large area devices. There is no consensus on whether or not the stability of perovskitephotovoltaics will improve enough so that their cost and relative lifetime will be competitive with siliconor cadmium telluride modules. Continued research will determine if the perovskite solar cell will everbecome a commercially viable product.

Literature Review

June 2015

Contents

1 Emergence of the PerovskiteSolar Cell 11.1 Perovskite Crystal Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 PSSC Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 The Potential of PerovskiteSolar Cells 52.1 Up-scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2 Tandem Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3 Research Developments& Unanswered Questions 63.1 Excitonic or Non-Excitonic Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73.2 Hysteresis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73.3 Humidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.5 Material Replacements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.5.1 Perovskite Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113.5.2 Electron Transport Mediums

& Hole Transport Mediums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.6 Deposition Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.7 Encapsulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163.8 Modeling the Perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4 A solar cell to competewith silicon? 18

1 Emergence of the PerovskiteSolar Cell

With the race for renewable energy to meet the goals ofthe climate change act, there has been a surge in largescale photovoltaic (PV) deployment on rooftops and solarfarms in the past ten years. Total global PV installed ex-ceeded 100GWp (Giga-watt peak) by the close of 2012[1]

and it has been predicted that close to a third of all newlyinstalled electrical generation will be PV by 2030.[2] Govern-ment feed-in tariffs across Europe have stimulated growthand investment interest in solar energy, and the ever fallingprices of crystalline silicon (c-Si) is bringing c-Si solar en-ergy up to grid cost parity with other energy sources.[3] Thisdrive for solar energy means that there is a future for pho-tovoltaics even alongside the large scale research programsinto carbon capture technology and nuclear power.

Thin-film and organic PV active areas are the focal point ofa well established scientific community, aiming to providea semiconducting material to compete with popular solarmodule materials: c-Si, amorphous silicon (a-Si), cadmiumtelluride (CdTe) or copper indium gallium selenide (CIGS).Thin film active area materials most commonly have lowerefficiencies but rely on using less bulk material and cheaperprocessing techniques in order to produce a cost competi-tive device.[4]

A surprise development in this PV community came withthe fabrication of the perovskite sensitized solar cell (PSSC).In 2006 a dye sensitized solar cell (DSSC) adapted withperovskite within its nanoporous titanium oxide (titania orTiO2) layer achieved a power conversion efficiency (PCE)of 2.2%.[1,5] This report aims to collate the literature sur-rounding this development, and every stage of evolutionthat the PSSC has undergone since. In 2006 the perovskiteused was an organometal halide CH3NH3PbBr3, a crys-tal structure containing methylammonium (CH3NH3), lead(Pb) and bromide (Br). The most prominent perovskiteactive areas use iodine (I) as the halide. There have beenmany reiterations and attempts to change the perovskiteformula with the goal to improve efficiency, lengthen sta-bility and search for a metal less toxic than Pb, with nocarcinogenic manufacturing bi-product.[2,6]

The reason why many articles similar to this literature re-view are being published, and why the scientific communityis so excited by the advent of the PSSC is best explainedby observing the rapid increase in reported PCEs. By theend of 2014 a device with an active area made from a for-mamidinium (FA) and methylammonium (MA) perovskiteblend (FAPbI3)0.85(MAPbBr3)0.15 achieved a 20.3% PCEon a reverse bias J-V (current density-voltage) sweep, otherdevices with the same architecture managed an average of17.3% steady state PCE.[7]

Figure 1: The crystal structure of a perovskite in a cubic orpseudocubic -phase (α). The positions and common substancesfor the organic cations A, metal cations B and halide anions X inthe ABX3 formula are indicated. Whilst in the cubic phase, theperovskites unit cell may be taken as a repetition of A organiccations (as shown) or as a unit cell of B metal cations. Thehalides form octahedra which may tilt depending on the size ofthe surrounding and enclosed cations.

Perovskites have had the advantage of immediately inte-grating into the 20 years of engineering already completedon their predecessor DSSC. PSSCs utilize the same archi-tecture as DSSCs and lean on the same chemical and elec-tronic understanding of the supporting electron and holetransport mediums (ETMs & HTMs), the electrode con-tacts, the transparent conducting oxides (TCOs) and all ofthe interface physics involved with planar and mesoporousstructures between each layer.[2] There are still many prob-lems and unanswered questions surrounding the perovskite,but the scientific community has certainly proved this is anactive field that can rise to face these challenges.

1.1 Perovskite Crystal Structure

The perovskite chemical structure is of the form ABX3,[1,2,6,8]

consisting of A and B Cations with X Anions forming acrystal structure demonstrated in figure 1. In the organic-inorganic organometal halide hybrid perovskites implementedin typical perovskite solar cells: A is typically the large or-ganic cations MA (CH3NH3) or FA (HC(NH2)2), B hasremained as Pb for all high PCE devices but a few researchgroups are looking at the replacement tin (Sn), and thehalides iodine (I), chloride (Cl) and bromide (Br) are allimplemented individually or as halide blend perovskites.(APbX3 (A= MA, FA)(X = I, Br, Cl)). Successful per-ovskite blends include MAPbI3−xClx,[9,10] MAPb(I1−xBrx)3,[11]

FAPbIyBr3−y,[12] FAPbI3−xClx[13] and a recent blend with

record breaking PCE, (FAPbI3)0.85(MAPbBr3)0.15.[7]

The perovskite crystal structure is often represented as acage of organic A cations enclosing a octahedral structureof halide anions X with the metal cation B in the centre.There are several phases of the extended perovskite crystalstructure with differing unit cells.

1

Figure 2: Edited from Nature Photonics,[2] the commonly re-ported unit cell of a perovskite in a cubic (α, a=b=c) or pseu-docubic phase (a=b=c, θ=φ=ω 6=90o) taken across the metalcations. This α phase is temperature, pressure and compositiondependent, for a certain composition of molecules it will onlyremain in a α crystalline state within a range of temperatures.Other typical crystalline perovskite phases include tetragonal(β, a=b 6=c,), orthorhombic (γ, a 6= b 6=c) or trigonal. The mostcommon trigonal phase is the hexagonal phase (δ, a=b6=c, θ=90o

φ=90o ω=120o)

Phase changes are influenced by changes in temperature,pressure and the composition of the perovskite formula.A larger organic A molecule may cause the tilting of theinternal octrahedra and the shifting of the halide-Pb bondangles.[2,14,15] In order for the perovskite cubic crystal struc-ture to form, the relative size of the anions and cationsmust produce an ideal Goldschmidt tolerance factor (t) be-tween 0.8 → 1, lower values of t may cause tetragonal ororthorhombic structures. Ionic radii come with a range ofreported values, in the case of MAPbI3: RA ∼ 1.8 A forMA, RX ∼ 2.22 A for iodine and RB ∼ 1.19 A for Pb.MAPbI3 is reported to have t=0.99. Other important ionicradii include: FA ∼ 1.9→2.2 A , Br ∼ 1.96 A and Cl ∼1.81 A .[2,16] t is calculated using equation 1.

t =RA +RX√2(RB +RX)

[17] (1)

Figure 2 begins to demonstrate the unit cells of perovskitecrystals by showing a typical unit cell for a cubic and pseu-docubic phase, it also contains the geometry of the othercommon tetragonal and orthorhombic phases. In the case ofMAPbI3 the perovskite is in a orthorhombic phase (Pbnm)untill 162K at which point it becomes tetragonal (I4/mcm).It will remain tetragonal at room temperature and only en-ters a cubic phase (Pm3m) state in temperatures above327K.[18–21] It is notable that this expected phase statechange from tetragonal to cubic occurs within the expectedoperating temperature of a solar cell (up to ∼ 350K). Thedisorientation of the halides is better represented in fig-ure 3.[15] The implementation of different cations or cationblends, and halides or halide blends; leads to different phasestructures of perovskite around room temperature. Eachphase has different electronic and optical properties, forexample, the dielectric constant will change with phase.[14]

Figure 3: From Physical Chemistry Letters,[15] phases of thehalide perovskite. (a) α cubic phase, (b) β tetragonal phase, (c)γ orthorhombic phase. (d), (e) and (f) are the [001] views of(a), (b) and (c) respectively, indicating the halide disorientationassociated with each phase.

Figure 4: Edited from Nano Letters,[22] analysis ofMAPb(I1−xBrx)3 with varied compositions. (a) X-ray diffrac-tion (XRD) focussed on the identifiable peaks for the cubic(200)C and tetragonal phases (004)T and (220)T . (b) Top downview along the [001] direction of crystal structure demonstratingthe expanded unit cell of the tetragonal phase (β, I4/mcm) atthe top, with the typical cubic phase (α, Pm3m) below. (c) Lat-tice parameters as a function of x, Br composition. A differentlinear relation is found for the tetragonal phase (light blue) andcubic phase (beige).

2

Figure 5: From Physical Chemistry,[15] table of band gaps and lattice constants for phases of MAPbX3 (X = I, Br, Cl) calculatedfrom first principle projector augmented-waves with Monte Carlo simulations (Perdew-Burke-Ernzerhof, spin-orbital coupling andhybrid functional method) based on cluster expansion methods (CEM). Also contains some comparable experimental data.

A few groups have published work using density functiontheorem (DFT) which predicts that the band gap of per-ovskites is relatively insensitive to the size of the organiccation RA. Their results suggest that changing RA couldbe a way to tweak perovskite band structure.[8,23] A largeamount of effort has been put into the tuning of band gaps,tailoring of absorption ranges, and the optimization of car-rier diffusion lengths over many iterations of perovskite.For example, the larger ionic radius of the cation FA ∼ 2 Aover MAs ∼ 1.8 A obtains an FAPbI3 tolerance factor oft=1.01,[16] effectively doubles the electron diffusion lengthsLe from ∼0.1µm to ∼0.18µm, increases the hole diffusionlength Lh from ∼0.1µm to ∼0.81µm[19] and narrows andred-shifts the optical band gap (absorption edge) from 1.52eV to 1.47 eV,[15,16,19,24] moving the absorption closer to anoptimum for the Shockley-Queisser limit (∼ 1.4eV).[24] Thisis at the expense of a lower absorption coefficient α. αMA at550nm is ∼1.5x10−4cm−1[25] meanwhile it has been foundthat αFA at 550nm is ∼1.3x10−4cm−1.[24] The X halidesare regularly changed and mixed to heavily tune the bandgap. Analysis of phase states induced by changing the per-ovskite formula are often performed as function of mixedhalide content x. For example figure 4 covers a study onthe previously mentioned MAPb(I1−xBrx)3

[11] with X-raydiffraction (XRD) to identify the changes in key identifi-able peaks and determine the changing lattice parameter.Figure 4 (c) indicates a tetragonal to cubic phase transitionfor bromide content at x=0.2.[22]

Compendium tables containing the phase symmetry groupnames, lattice contants, bang gaps (optical and electrical),conduction band minima (CBM) and valence band minima(VBM) for none mixed halide MAPbX3 and FAPbX3 filmscan be found in figure 5 and 6.[15,19]

Figure 6: From APL Materials,[19] table summarizing per-ovskites band gaps (eV), structure at room temperature, carrierdiffusion lengths and conduction band minima (CBM) and va-lence band minima (VBM) with energy levels collated from wellcited literature. The common chemicals used in the perovskiteactive layer forumlas included are; Iodide (I), Bromide (Br),Chlorid bathe (Cl), with methylammonium (MA, CH3NH3) andformamidinium (FA, HC(NH2)2).

1.2 PSSC Architecture

The success of the perovskite as a semi-conducting activearea, capable of achieving high PCE, is due to the remark-able properties of the perovskite. As previously discussedthe perovskites large absorption coefficients allow manyphotons to be collected with the perovskite layer being onlya few 100nm thick. A high charge mobility of 8cm2(Vs)−1

for MAPbI3 and 11.6cm2(Vs)−1 for MAPbI3−xClx and alow bi-molecular charge recombination encourage a highfill factor (FF).[26] JSC is high as charge generation occurson the order of picoseconds whilst recombination is on theorder of microseconds.[21] A low-non radiative recombina-tion rate means that there is a small difference of ∼450meVbetween VOC and Eg/q, better than most current PV semi-conductors.[21,27]

3

Figure 7: From Nature,[29] a Dye-Sensitized Solar Cell(DDSC). The well-established DDSC configuration, which wasadapted by replacing the dye with halide perovskite to make thefirst ever perovskite sensitized solar cell (PSSC).

The perovskite was originally implemented in a photovoltaicactive area as a hybrid of a standard DSSC in 2006;[5]

achieving 2.2% PCE. Figure 7 is the schematic for a typi-cal DSSC. The photoactive dye generates free charges fromincoming photons, the electrons transfer into the opticallytransparent TiO2 mesoporous scaffold and the holes passthrough the liquid electrolyte to the electrode and counterelectrode respectively. The perovskite directly replaced thedye to produce the first ever PSSC. In 2009 the same groupreplaced the bromide with iodide in a similar architectureand reached 3.8% PCE[28] and by 2011 another group hadreached 6.2% PCE by sensitizing the TiO2 with perovskitequantum dots (QDs).[25] It was at this time that perovskiteswere observed to be easily dissolved in the liquid electrolyte,causing damaging losses to the PCE.[8]

In 2012 the move away from PSSCs using the DSSC liq-uid redox electrolyte occurred with two major pieces ofpublished work, completed by H.Snaith’s and M.Gratzel’sgroups. They produced device with PCEs 10.9% and 9.7%respectively.[30,31] The liquid electrolyte was replaced by theHTM spiro-MeOTAD, doped with bis(trifluoromethane) sul-fonimide lithium (Li-TFSI) and 4-tert-butylpyridine (TBP).The device structure implemented is similar to schematic(a) in figure 8. These two groups have remained central tothe evolution of the perovskite solar cell and are involvedin many more of the papers covered in this review.

There has been a lot of attention focused on the use ofa scaffold of TiO2 as the ETM. In 2012 a Al2O3 scaffoldwas implemented to replace TiO2 and it was believed thatopen circuit voltage (VOC) was boosted by this change, pro-ducing a 10.9% device.[8,27,30–32] In addition, several articlessince have outed TiO2 as having a bad interface with typical

Figure 8: From Nature Materials,[6] the evolution of the per-ovskite solar cell away from DSSCs. (a) The liquid electrolytereplaced by a hole transport medium (HTM), the perovskitecoats the titanium dioxide (Ti02) scaffold or an alternative alu-minium Oxide (Al2O3 scaffold). This all remains upon a trans-parent conductive oxide layer (TCO) and Compact TiO2 layer.(b) The perovskite is a layer above the scaffold material and theHTM is a separate layer above this, due to the highly conduc-tive nature of the perovskite this structure has been proven tofunction effectively, even without the HTM. (c) Thin film planarstructure with HTM above and no mesoporous scaffold.

perovskite materials.[33,34] A study focusing on the origin ofJ-V sweep hysteresis (a common problem with perovskitedevice testing) compared several combinations of perovskiteinterlayers and came to the conclusion the interface betweenthe scaffold TiO2 and the perovskite could be a signifi-cant contributor to hysteresis.[34] Several studies on stabil-ity have also led to conclusion that the TiO2/perovskite in-terface is ultraviolet (UV) unstable as it is a photo-catalystfor oxidizing materials. Iodine may be extracted by the pro-cess described in equation 2. The solutions offered by oneof these studies recommends the replacement of Al2O3 orthe introduction of an antimony sulfide (Sb2S3) interlayerbetween the perovskite and the TiO2 layer.[21]

2I− ↔ I2 + 2e−

3CH3N+3 ↔ 3CH3NH2 + 3H+

I− + I2 + 3H+2e− ↔ 3HI [21] (2)

The next step for perovskite hybrids was the analysis ofMAPbI3−xClx which was found to have a electron and holediffusion length Ld,e of well over one micron, over 10 timesthat of MAPbI3 with Ld/e∼100nm. This allows charge tobe extracted over thicker perovskite films. In addition ex-perimental evidence revealed the perovskite itself could pro-duce working devices as a thicker perovskite active area, notjust as a sensitizing substance[30,31,35] (See Figure 8 (b)).

4

It was also found that the perovskite later didn’t have tobe fully intercalated with an ETM scaffold, in fact, devicescould function without an ETM scaffold at all[36](See Figure8 (c)). In 2013 a planar compact titania c-TiO2/MAPbI3−x

Clx/Spiro-OMeTAD device reached a PCE above 15%. Withlonger diffusion lengths these perovskite films no longerneeded the scaffold to sufficiently extract charges into theHTM.[21] The variation in charge diffusion lengths that oc-curs with different compositions of perovskite allows flexi-bility in the device architecture, and opened several avenuesof research for the ever growing perovskite community.[27]

Device architectures for perovskites from 2013 onwards typ-ically include the mesoporous structure already shown infigure 8 (b) or the planar structure in figure 8 (c). Devicesare not limited to being produced with the HTM abovethe perovskite (n-i-p devices), and have also been fabri-cated with alternative HTMs below the perovskite (p-i-ndevices).[37] Choices of material for ETMs and HTMs arelimited by use of appropriate electron affinity to create anideal energy landscape for strong charge extraction and alsoby solvent matching in the fabrication process.[38] Figure 9shows a device with ETM Phenyl-C61(or C71)-butyric acidmethyl ester (PCBM) layer above the perovskite, alongsidethe energy levels associated with each layer. In this casePCBM could not have been fabricated below the perovskiteas the typical solvent used for depositing the perovskite isdimethylformamide (DMF), which also dissolves PCBM.Figure 10 shows the typical energy landscape for the com-mon TiO2/perovskite/spiro-MeOTAD device.

The band gap, energy landscape and morphology of everylayer involved varies between different choices of: device ar-chitectures, perovskite formulas, transport layers and inter-layers. Optimization by fabrication or computational mod-els help refine ideal formulas, such as the optimum thick-ness’s of each layer for charge generation and extraction.Every interface within the perovskite is important, chargeextraction processes and degradation effects are often foundto depend heavily on the perovskite/transport layer inter-faces. Each material and fabrication technique used for alayer will effect the electronic structure and morphologyof that layer, which in turn also effects the structure andmorphology of the layer deposited above it.[3,14,39,40]

2 The Potential of PerovskiteSolar Cells

In 2013 it was predicted that a stabilized perovskite solarcell could potentially cost ∼ 0.30 $ per Wp (Watt peak),lower than the price of c-Si (1 $/Wp), CIGS (0.55 $/Wp)and CdTe (0.67 $/Wp) at the time.[3,41] However the priceof silicon has consistently dropped in recent years, by 2015some prices for c-Si solar cells were estimated at 0.36 $/Wp.[3]

Figure 9: From Scientific Reports,[37] planar device schematicand energy level diagram for a reverse structure devicewith HTM PEDOT:PSS (poly(3,4-ethylenedioxythiophene)polystyrene sulfonate) below the MAPbI3 perovskite and ETMPC60/70BM (Phenyl-C61/C71-butyric acid methyl ester).

The cost gap is closing between the thin film technologiesand c-Si, but the potential advantages of the perovskitesolar cells over conventional c-Si are still enough to encour-age progress. As well as all the remarkable electronic andoptical properties of the halide perovskites, they also offerthe best of both worlds between being a self-assembling so-lution processable layer and having remarkable crystallinestructures.[30,42,43]

2.1 Up-scaling

The dream for the perovskite solar cell is a high efficiency,stable, large area, flexible, transparent device capable of be-ing deposited by reproducible and scalable processes suchas: spray coating, slot-dye coating, gravure coating, doc-tor blade coating or inkjet printing. These up-scaling pro-cesses come with many challenges.[44–47] It is commonly ob-served that large scale print and spray techniques producefilms with less surface coverage. This is compounded withthe fact that these techniques are often performed in at-mospheric conditions where perovskite formulas have beenknown to produce devices with lower efficiencies and shorterlifetimes than those made in a nitrogen atmosphere or un-der vacuum.[45] Degradation processes linked with moistureduring fabrication have been identified,[48,49] and the com-monly used PCBM/electrode interface has been known tocause device failure when deposited in the presence of mois-ture.[49] Conversely there is also evidence to suggest thatslight humidity or oxygen during fabrication can improveperovskite morphology and reduce trapped states in trans-port layers respectively.[33,50]

Fast and low-temperature fabrication is another cost savinggoal for the perovskite. The typical 60→90 minute annealtime can be lowered by crystallization quenching throughsolvent washing and has also been demonstrated to dropfrom over an hour to ∼ 5 minutes by the introduction oflead acetate PbAc2 into the formula, replacing the halidePbX2 (X=I, Cl, Br) as the lead source component.[51,52]

5

Figure 10: From Energy & Environmental Science,[38] energy level diagram of interfaces between HTM spiro-MeOTAD (with bandbending) and perovskite MAPbX (X = I3, I3−xClx, Br3) on TiO2 with experimentally determined work function, ionization energy,electron affinity and Fermi level position.

The use of alternative material or deposition techniques forthe compact TiO2 layer was investigated as this layer oftenrequires annealing temperatures of 450oC for over an hour.The alternatives include: switching the device architectureto that indicated in figure 9 (for which PCBM and poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS)only require a short anneal at temperatures ∼ 100oC or noanneal at all), implementation of graphene into the TiO2, orthe use of solution processable TiO2 nanoparticles.[49,51,53–55]

Few high efficiency large area devices (bigger than the typi-cal ∼ 0.1 cm2) have been reported, due to the lower surfacecoverage and the increasing number of shunt inducing pinholes that appear as the device size is scaled up.

Some groups maintain that larger scale spray or slot dyetechniques could still be cost effective within vacuum evap-oration or in a filtered nitrogen glove box.[40]

2.2 Tandem Device

Henry Snaith suggested a major advantage that perovskitesmight be able to utilize is their absorption band, which ef-ficiently collects wavelengths of light which the major com-mercial solar cells such as c-Si and CIGS do not.[27] A tan-dem solar cell was proposed, where a layer of perovskite filmcould theoretically be deposited on top of one the standardbulk solar cells. Such a tandem cell would require signif-icant optimization on its architecture in order to achievegood lattice matching and charge transfer between any in-terlayers and ETMs/HTMs used. For feasible up-scalinga device with fewer layers would be preferable as it shouldincur fewer parasitic losses, however this would increase thelikelihood of bad interfaces within the device.[56]

It is estimated that such a tandem cell should reach 1.3times the maximum amount of photons absorbed than thatset for a single junction by the Shockley-Queisser limit.A tandem perovskite/c-Si cell that can limit its parasiticlosses and achieve good charge transfer could reach 29.6%PCE.[27] Further studies using an oscillator model on opti-cal n-k data predicted a similar 29% PCE for a perovskite/-CIGS tandem cell.[57,58]

3 Research Developments& Unanswered Questions

Despite the progress made on perovskites solar cells so far,our understanding of perovskites is still in its infancy, manygroups still have conflicting opinions on properties such asthe exact excitonic functionality of perovskites or the originof hysteresis for J-V sweeps at different speeds. In generalthe quality of fabrication techniques and optimization offilms should continue to reduce imperfections in perovskitedevices, this trend will reduce series and shunt resistances,increase maximum power points (Vmpp) and therefore in-crease device PCEs.[27] The ever evolving perovskite devicearchitecture may also provide answers to some of the cur-rently unsolved problems that are present in the perovskitecommunity. It is not encouraging that the majority of pub-lished research comes with a range of quoted values formany important properties, something as fundamental asthe bandgap commonly varies by 20% from group to group.This is one of the many indications that the repeatabilityof producing perovskite devices between laboratories is stilllimited. Many groups are careful to only postulate their so-lutions instead of providing certifying evidence.

6

3.1 Excitonic or Non-Excitonic Behaviour

After a photon has caused the excitation of a electron withina semi-conductor, the electron and hole generated are gen-erally understood to be extracted in one of two ways. Theelectron and hole pair are refereed to as a quasi-particleknown as an exciton and depending on the semi-conductormaterial this exciton can either be a large weakly boundWannier-Mott exciton with a binding energy on the orderof ∼10meV or a small tightly bound Frenkel exciton with abinding energy typically between 0.1 and 1 eV. c-Si activeareas generate Wannier-Mott like excitons, where the elec-tron and hole have diffusion lengths large enough to coverthe depth of active area and are extracted to the electrodesas free chargers. Typical organic cells have more tightlybound Frenkel excitons, this limits organic cells quantumefficiency as charge can only be extracted if the exciton ran-domly diffuses to an hole or electron transport interface.Many organic active areas are intercalated with PCBM sothat the excitons have to diffuse less distance before reach-ing an interface.[2,30,31,35]

The consensus is that most MA or FA perovskites with anyhalide or a mix of halides generate weakly bound Wannier-Mott excitons, resulting in free charge separation. Howeverthe experimentally and computational obtained binding en-ergies that form this consensus vary in value between EB

∼(1 → 50)meV.[18,26,39,43,57,59–65] If the excitons did indeedhave one of the higher quoted values of binding energy, likethose approaching 50meV, then there is some contentionthat the excitons may exhibit tightly bound effects, and re-duce the quantum efficiency of a perovskite device.

The impressive PCEs of perovskites devices fundamentallysuggest that their generated excitons are unlikely to betightly bound. After the realization that perovskites didnot need a mesoporous scaffold it became apparent that thecharge extraction properties of the perovskite were remark-able. This encouraged the investigation of hole and electrondiffusion lengths through several time-resolved photolumi-nescence (PL) studies.[24,26,62,66,67] MAPbI3 has Le ∼ 0.1µmwith a slightly larger Lh,[19] FAPbI3 has Le ∼ 0.18µm andLh ∼ 0.81µm,[7] MAPbI3−xClx has Le ∼ 1µm and Lh ∼1.2µm.[10,26,59] For all of these perovskite formulae the dif-fusion length is of the order of the typical width of a thinfilm active area ∼ 0.5µm, but the mixed halide and FAperovskites have the advantage of their diffusion lengthsgreatly exceeding film thickness.[2,30,31,35]

The calculation to obtain exciton binding energy valuesusually involves considering the dielectric constant of theperovskite. There is currently a conflict in how to interpretthe value of the dielectric constant through the use of n-kdata and the Kramers-Kronig relations.[68]

One recent study highlights the current discrepancy in theWannier-Mott model and poses an argument on how thetypical use of the optical frequency dielectric constant ε

′

may be inappropriate as the difference between the staticεSt and high-frequency ε

′′values is typically very large.[39]

Another common method of determining EB is by observ-ing the separation of Landau levels using magnetoasborp-tion measurements, calculated using equation 3 where Nis the Landau quantum number, wc = eB/m*, B is themagnetic field, m* is the exciton effective mass, geff is theeffective Zeeman splitting and µB is the Bohr magnetron.[60]

E(B) = Eg + (N + 1/2)~wc ± 1/2geffµBB[60] (3)

Studies using an ε′

for determining EB typically find a ε′

∼ 6.5 and hence a higher EB ∼ 50meV.[61,65] Alternativelystudies using εSt ∼ 70 obtain a lower EB ∼ 2meV.[39,69] An-other paper proposes that this choice of dielectric causes themagnetoasborption measurements using equation 3 to ob-tain questionable overestimates of EB ∼(37→ 50)meV.[59].The most recent unpublished (as of 06/2015) attempt toclarify the binding energy of perovskites uses the latestfabrication techniques of quality films and very large fastpulsing magnetic fields up to 150T.[60] This study finds ageneral EB=(16±2)meV and m*∼ 0.1me for the low tem-perature MAPbI3 orthorhombic phase, which collapses tolower EB for higher temperatures due to dielectric constantbeing frequency dependent. This study attributes the causeof previously inaccurate EB to: the limited range of 1s lowmagnetic field transitions, the approximation of dielectricconstants and the assumption of EB being a constant func-tion of magnetic field or temperature. Two reports reach aconsensus that a MAPbI3 film at room temperature, in thetetragonal phase, has a binding energy which recedes to ∼ 5meV under the influence of dielectric screening from rotat-ing MA cations. Mixed halide films were also found to havereduced excitonic screening due to collective reorientationsof the MA cations.[59,60]

3.2 Hysteresis

A common method of determining the PCE is by takingJ-V curves under solar illumination Air Mass 1.5 (AM 1.5,100mWcm−2) and obtaining the VOC , FF and current den-sity short-circuit voltage JSC . Typically the solar cell isheld under forward bias (FB), then the voltage is reduceddown to short circuit voltage (SC) at a sweep speed vs,to obtain a FB-SC sweep. The opposite can be done toproduce a SC-FB sweep. Perovskites have been found toexhibit a significant difference in their J-V curves depend-ing on the direction and speed of the sweep, an anomalycalled hysteresis. This process has been previously identi-fied in other DSSC or organic devices, and can be linked tolight soaking leading to doping and band bending and/orthe effective capacitance of slowly responding devices.

7

Figure 11: From Physical Chemistry,[34] J-V CurrentDensity-Voltage curves demonstrating hysterisis for compactTiO2/MAPbI3−xClx/spiro-OMeTAD perovskite devices takenin forward bias to short circuit (FB-SC) and vice versa (SC-FB), at several scan speeds. These scans were performed underAM1.5 100mW cm−2 after a stabilization time of 60 seconds atforward bias.

During high speed scans a build of charge may occur whena cell is under FB, which is observed as additional photo-generated charge on the backward sweep. Another effectcan occur with fast SC-FB scans partially charging the cellwhich is observed as a reduced photogenerated charge.[34]

A study on hysteresis published in 2014 took particular caretoo wait for several minutes to achieve steady state scans,and found that hysteresis in perovskites gets more extremewith slower scan rates. This is the opposite of what hasbeen observed in many other photovoltaics, including hys-teresis observed in a PSSCs architecture revealed with J-Vscans and impedance spectroscopy earlier that year.[34,70]

Figure 11 demonstrates a typical J-V curve and shows theincreasing hysteresis with scan speed observed in a planarMAPbI3−xClx device. J-V curves of other devices contain-ing mesoporous TiO2 yielded the same scan speed trend anddemonstrated that the magnitude of hysteresis is highly de-pendent on device architecture, including; material choices,deposition techniques and layer thicknesses.[34]

Another group used capacitance - voltage (C-V) and ca-pacitance - frequency (C-F) scans of perovskite films toidentify the potential changes occurring in perovskite filmsduring light soaking. The accumulation of charge at elec-trode surfaces was found to neutralise interfacial defectsand increases VOC , meanwhile PL suggested that photo-

generated electrons were more likely to reduce the densityof bulk defects, therefore reducing one of the inhibitors ofeffective charge transport and increasing FF. It was alsofound the light soaking reduces the bulk polarization of theperovskite film, increasing recombination rates inside thefilm and decreasing JSC .[71]

The hysteresis in perovskites is speculated to be causedby any combination of three processes. The first potentialcause of hysteresis is that there may be a large number ofsurface trapped states in the perovskite layer. These stateswould be filled under FB and create good perovskite in-terfaces with the ETM/HTM or contacts directly. Thesestates would then empty fast into the contacts under SC,whereby the performance of device would be reduced untilthe trapped states were once again filled. This effect wouldbe exaggerated in the architectures with no transport lay-ers. Secondly, hysteresis may be caused by the ferroelectricpolarization of the perovskite crystal structure. It has beenobserved that the organic cations may slowly polarize acrossthe film depending on voltage bias, which could potentiallyincrease charge separation and theoretically allow a VOC

higher than the band gap.[34,72] The final suspect for hys-teresis is ion interstitial migration, whereby excess ions maybe free to move within the film, screen extra space chargeand generate increased charge collection.[34,73]

By comparing multiple architectures the TiO2 layer hasbeen ruled out as the sole source of hysteresis, however itconsidered likely that it is the interfaces between the per-ovskite and ETM/HTMs like TiO2 that are causing hys-teresis. It has been shown that hysterysis can be reducedby only using a thicker TiO2 scaffold, due to the increase insurface area allowing a better p-type contact with the per-ovskite. Hysteresis free films have been reported, with thelack of the hysteresis anomaly attributed to: the high qual-ity morphology of the films,[39] a lower number of trappedstates and an increase in crystal grain size.[74] Recently ithas been reported that high quality n-i-p structured devicesdemonstrate negligible hysteresis but it is difficult to sig-nificantly reduce hysteresis in p-i-n structured devices.[75]

Groups working on understanding hysteresis have also cometo realise that in in order to properly analyse perovskite de-vices, a consistent method avoiding light soaking and hys-teresis must be followed. It is suggested that the steadystate J-V curves are recorded at several scan speeds andthat the stabilised maximum power point is followed forseveral minutes. This is important as fast scan speeds canmask hysteresis and trapped states by scanning faster thanthe emptying of traps. It is also important because thestabilised maximum power point often yields PCEs signifi-cantly lower than fast J-V scans. This effect is most promi-nent in FB-SC scans of planar devices and has been usedto question the validity of many reported PCEs.[34,73]

8

3.3 Humidity

It has been demonstrated that MAPbI3−xClx films depositedfrom a solution with 3:1 molar ratio of PbCl2:MAI in DMFshow significant improvement in morphology for those an-nealed in ambient air with humidity (35±5)%, over thoseannealed in nitrogen. It is assumed that grain boundarycreep caused by moisture between crystal grains encouragesgrain growth, however films annealed in over 80% humid-ity showed rapid decay into PbI2. Devices fabricated inthis study achieved an increase in PCE from 12% for thoseannealed in the glove box to 15.4% for those annealed inair, this is attributed to a decrease in non-radiative decaydemonstrated by the increase of PL lifetime from 33ns to99ns. This moisture induced device improvement is ex-pected to occur up to 60% humidity at which point FFbegins to fall as more PbI2 is found present in the film.[50]

Humidity has been identified as an important factor duringdevice fabrication, but little work has been done to isolatethe effect of humidity at each point of fabrication. Humid-ity could potentially change the crystal growth and mor-phology of the perovskite film as it is spun or spray coated,as it is drying or as it is being annealed. Humidity mayalter any interface reaction with the transport layers, oreven the deposition and interface reactions of the transportlayers with the contacts.[50]

3.4 Stability

Perovskite devices are as well known for their instabilityas they for their dramatic rise in efficiency. A typical pho-tovoltaic with encapsulation should aim to retain 80% ofits initial PCE after 25 years by being resistant to: con-tinued AM 1.5 illumination (including the UV part of thespectrum), maximum summer operating temperatures (upto ∼ 350K), precipitation and water vapour (up to ∼ 80%relative humidity). Each of these factors may effect the sto-ichiometry of any of the layers and interfaces in a solar celland cause degradation to the morphology or crystallinityof the active layer. In perovskites it is commonly observedthat the perovskite will degrade into hydrates or its originalconstituents PbX2 and MAX/FAX (X = I, Cl, Br) understress testing.[21,27,76,77]

Analysis of film constituents before and after stress test-ing devices are typically done using XRD, X-ray fluores-cence (XRF), X-ray photon spectroscopy (XPS), Energy-dispersive X-ray spectroscopy (EDX), thermogravimetricanalysis (TGA), differential scanning calorimetry (DSC)and fourier transform infrared spectroscopy (FITR). Manyof these techniques only probe the surface of the perovskite,however some conclusions can be drawn by comparing thespectral data from these techniques to the data obtainedby techniques than can probe the bulk of the perovskite,for example XRF. XRF detects more Cl in MAPbI3−xClx

films than that spotted with surface probing techniques,suggesting that the majority of Cl is in the bulk of theperovskite film, at the opposite interface, or within crys-tal grain boundaries. Unfortunately it is difficult to probesmall volumes with XRF as the spot size must be large.Despite the limitations of these probing techniques much asbeen learnt about the stability of perovskite devices.[21,70,78,79]

The effects of liquid water on unencapsulated perovskitefilms are expected to follow the degradation described inequation 4.[21,76], this has been shown to be a irreversibleprocess. The liquid water travels through the film alonggrain boundaries and quickly decomposes the crystal struc-ture, visibly changing the perovskite from brown to a yellow-transparent film.[77]

CH3NH3PbI3 ↔ PbI2 + CH3NH3I

CH3NH3I ↔ CH3NH2 +HI

4HI +O2 ↔ 2I2 + 2H2O

2HI ↔ H2 + I2[21] (4)

It has been suggested that tetramethylammonium mightbe a highly moisture resistant substitute for the organiccation, as it offers no protons to initiate the water degra-dation described in equation 4.[21] Another study foundthat MABr3 or hybrid MAPb(I1−xBrx)3 films are signif-icantly less sensitive to moisture with increasing bromidecontent, but also are bad absorbers due to Br’s larger bandgap. MAPb(I1−xBrx)3 devices were found to maintain theirPCE after 20 days in 55% humidity.[22]

It has also been observed that water vapour incident on per-ovskite films disperses along grain boundaries and forms hy-drated crystals CH3NH3PbI3H2O followed by (CH3NH3)4PbI62H2O. When saturated with water vapour these hy-drates reduce device PCE by an order of magnitude, butafter devices are dried in nitrogen for 6 hours PCEs returnto their original value, however; a greater level of hysteresisis observed. With insufficient encapsulation for humid en-vironments, the perovskite could fluctuate between an effi-cient and inefficient state following this reversible hydration(with an ever increasing number of defects). Devices withlarger grains could reduce the penetration of water vapourand decrease the number of trapped stated caused by highlevels of humidity.[77] This could be linked to reports of de-vices with large grain sizes and high quality morphologiesshowing negligible amounts of hysteresis.[39,74]

As previously stated, the maximum operating temperaturea perovskite solar cell should have to endure is ∼ 350K,this is good because the chemical degradation of the per-ovskites crystal structure starts occurring above ∼ 380K.[21]

Beyond this temperature many more halide deficiencies areobserved, which is the reason why perovskite annealing of-ten does not exceed this temperature.[52]

9

Figure 12: From Physical Chemistry,[80] wide angle XRD analysis of FAPbI3 stabilised with MAI. (a) yellow trigonal δ-phase and(b) stabilised black α-phase with 15% MAI.

However it has also been identified that there is a phasestate change between 327K and 330K from tetragonal tocubic, causing a small decrease in band gap and a switch toan acentric structure exhibiting ferroelectric effects.[18–21,80]

FA perovskites have been shown to be more thermally sta-ble showing no degradation or discolouration when exposedup to ∼ 400K, whilst MA based perovskites fully decom-pose after ∼ 30 minutes at such high temperatures. It isunfortunate that FA seems to offer no observable advan-tages in resistivity to degradation caused by moisture.[12].

FA has often been refereed to as a good potential sub-stitute for MA in perovskite solar cells, this may be bestdemonstrated by figure 13 showing its smaller band gap andgood energy landscape alignment with common transportlayers. There is some disagreement in the FAPbI3 crys-tal phase structure as it has been previously identified asTetragonal[19] and also as trigonal (P3m1, α) at room tem-perature.[7,80] It is agreed that for MAPbI3:FAPbI3 mixedfilms the phase state is tetragonal when the MA contentmakes up >20% of the total organic anions. It also seemswell established that despite the stability of the black αphase in high temperatures up to ∼ 430K, α is susceptibleto ambient humid environments. In a humid environmentthe α phase is no longer stable and at lower temperaturesit slowly converts to a yellow hexagonal δ phase (P63mc).The phase change to δ from α appears to be reversible, theα phase can be re-established by annealing again. The δphase contains linear chains of PbI6, has a larger opticalband gap and exhibits poor charge separation and transferproperties.[7,80] Figure 14 demonstrates the crystal struc-ture of both the α (a) and δ (b) reported phases.

The phase instability of perovskites has been combated bythe incredibly successfully implementation of (FAPbI3)0.85(MAPbBr3)0.15 which maintains the optimum absorptionrange of FAPbI3 over MAPbI3.

Figure 13: From Physical Chemistry C,[24] energy level dia-gram of mesoporous scaffold TiO2/FAPbI3/spiro-OMeTAD per-ovskite device.

Figure 14: From Physical Chemistry C,[24] diagrams of theblack α (a) and the yellow δ (b) phases of FAPbI3. The bluepolyhedral are the PbI6 octahedral, the Pb and I atoms are theyellow and orange circles respectively. In the δ phase the largerstructure consists of linear chains of face sharing octahedral.

10

This blend is observed to have no hexagonal δ yellow phasepresent during or after lifetime testing in an ambient hu-mid atmosphere.[7] Following this development a separatepublication has argued that the implementation of bromidecontent, which increases the band gap of the device, couldbe a significant limiting factor for its PCE. They suggestthat mixing MAPbI3 and FAPbI3 stabilises the trigonal αblack phase effectively and does not limit PCE through thepresence of bromide.[80] The stability of FAPbI3 with 15%MAI is demonstrated in figure 12 (b) where no phase statechange is observed in the crystal XRD spectra after anneal-ing and then re-cooling the samples. Figure 12 (a) showsthe un-stabilised perovskite is in a δ phase for temperaturesbelow 400K. Conversely, the original study clearly statesthat mixing MAPbI3 and FAPbI3 would require too muchMA content to phase stabilise FA based devices and hencethe performance would be dominated by MAPbI3, whichwould be counter-productive.[7]

The spectroscopic break down of a comparable FAPbIyBr(3−y)

device architecture reaching 14.2% PCE is given in figure16, which identifies the implementation of a high bromidecontent into FA perovskite causing cubic and tetragonalcrystal structures similar to MA only perovskite devices.

In terms of UV stability the potential instability of TiO2

has already been addressed in the first few iterations of per-ovskite solar cells. Further theoretical work has been doneto suggest that exposure to oxygen is needed on the TiO2

surface in order to passivate surface trapped states. Thisshould increase UV stability and reduce hysteresis originat-ing from the bad perovskite/TiO2 interface.[33]

3.5 Material Replacements

This report has already covered some iterations of per-ovskite device architectures used to study the functionalproperties of perovskite. Many more steps have been takenby research groups to increase stability and efficiencies,some of which are reviewed in the following sections. Theexperimentally determined energy levels for further alter-native device materials are presented in figure 15. Thisincludes the antimony sulfide (Sb2S3) and the photosensi-tive dye N719. The popular MA replacement FA has al-ready been discussed, however it is worth clarifying thatin general FA films need to be slightly thicker (∼500nm →∼600nm) than MAI:PbCl2 films (∼350nm → ∼500nm) orMAI:PbI2 (∼200nm)[67] due to FA’s weaker absorption.[75]

3.5.1 Perovskite Formula

The band gap of the common perovskite film MAPbI3 hasalready been established as ∼1.52eV. This is reduced by∼50meV and ∼150meV by the direct replacement of iodidefor bromide and chloride respectively.

Figure 15: From Materials Today,[63] energy levels for alterna-tive materials of HTM, ETM and Perovskite active layers.

The implementation of these halides causes contraction ofthe octahedra, decreases the unit cell size and causes strainin the crystal lattice that blue shifts the absorption range.[15,19]

The role of phases in mixed halide perovskites has beenaddressed several times but is still not fully understood.The introduction of bromide into FA is summarised in aspectroscopic breakdown in figure 16. The effect of bro-mide in MA can be summarised by considering figure 17where for MAPb(I1−xBrx)3 devices: the absorption shiftwith additional bromide is shown in (a), the change in op-tical band gap is fitted in (b) and the efficiency change asa function of x in (c). It is prompted that a Br content ofx=0.2 might be a sweet spot, maintaining high PCE butgaining the extended stability previously reported in per-ovskites containing bromide.[22] It is worth noting that thisx=0.2 bromide content is also where the shift from tetrag-onal to cubic phase in MAPb(I1−xBrx)3 films occurred.

The uniformity of halide distribution in mixed halide per-ovskite films is still uncertain. An anomalous PL changehas been observed in MAPb(BrxI1−x)3 films under lightsoaking. Figure 18 (a) shows how the expected PL signalfrom a fully mixed halide phase (x=0.4, PL peak ∼1.85eV) shrinks and is replaced by a PL with peak at 1.68 eV,similar to that observed in MAPbI3 (iodine only) films.This new PL peak returns to its original location and in-tensity after the film is left in the dark for 5 minutes, asdemonstrated in figure 18 (c) for a x=0.6 film. It has beensuggested that this effect can be attributed to a changein perovskite film morphology under illumination, wherebythe mixed halide film converts into bromide and iodide richphases. If this is the case, then any value of overall bromidecontent x would still cause a PL peak at ∼1.68 eV underillumination, where the photo-excited bromide rich statesare relaxing to iodide rich states before charge extractionor recombination occurs.

11

Figure 16: From Energy & Environmental Science,[12] analysis of the FAPbIyBr(3−y) perovskite films with varying y. (a) Ab-sorbance. (b) Steady-state Photoluminescence (PL). (c) Images of films, y increases from 0 to 1 from left to right. (d) XRDindicating that as y decreases, the (100) cubic peak changes to the (110) tetragonal peak. (e) Variation of band gap as a functionof pseudocubic lattice parameter.

12

Figure 17: From Nano Letters,[22] analysis of MAPb(BrxI1−x)3 films and devices. (a) Absorbance of films with varying bromidecontent x, with image of each film. (b) Band gap variation with increasing x. (c) PCE of the devices with structure mesoporousTiO2/perovskite/polytriarylamine (PTAA).

Figure 18: From Chemical Science,[81] PL spectra under 457nm at 300K light for MAPb(BrxI1−x)3. (a) PL spectra of x=0.4film over 45 seconds in 5 seconds increments showing growth of peak at 1.68ev. (b) Normalised PL spectra after films have beenilluminated for 5-10 minutes. (c) PL spectra of x=0.6 film before illumination and after illumination for 2 minutes, where films areleft in the dark for 5 minutes before the next set of measurements.

13

The exact composition and halide uniformity in MAPbI3−xClxfilms has been a cause of disagreement since the adventof the first mixed halide devices. Despite the commonMAI:PbCl 3:1 molar ratio implemented by most researchgroups, the films appear to only possess a Cl content be-tween 2% and 4% after annealing. One study measured anabsolute upper boundary of 300 ppm (parts per million) forCl in their MAPbI3−xClx devices.[10,52]

PbX2 + 3CH3NH3I → CH3NH3PbI3 + 2CH3NH3X

(X = I, Cl, Br) [52] (5)

Consider the perovskite formation equation 5 which demon-strates the generation of the by-product 2CH3NH3X. Theenergetic requirements of the decomposition of these by-products can be calculated as 226.7oC for MACl and 245oCfor MAI, which would suggest that it is unlikely that themajority of Cl could decompose and evaporate off the sur-face of the film, and so does not explain the apparent lack ofCl commonly measured. The possibility for the majority ofCl to be located in the perovskite/transport layer interfacesor amongst grain boundaries has already been considered,however; it is suggested that a large amount of Cl muststill be subliming and out-gassing off of the films given thequantity of Cl unaccounted for.[78] Some publication sug-gest that the high surface area:volume ratio and efficientheat dissipation through perovskite films allow MACl tosublime at only 100oC instead of 245oC, and suggest thatthe out-gassing of MA and Hydrochloric Acid (HCl) occursinstead of MACl. This is strongly backed up by in situ ev-idence of slowly decreasing Cl content during annealing. Itis also mentioned that whilst MACl is not the major out-gassing component, it has been measured on petri dishesused to contain old films.[52,78] Other separate studies con-clude that no MAI is given off during deposition or anneal-ing, but that HCl can be seen coming off old films heatedto 150oC.[70,79] Conversely, there is published evidence thatshows only solvent out-gassing can be observed during theannealing process.[78] These disagreements do not help tosolve what actually occurs during deposition and annealingto produce the resultant Cl deprived MAPbI3−xClx films.

Despite some unanswered questions, the perovskite com-munity’s understanding of decomposition temperatures andenergetically favourable reactions has improved and has in-spired one of the biggest developments in perovskite for-mula to date. The by-product produced by using lead ac-etate (PbAc2) in MAI:PbAc2 perovskite films is MAAc,which decomposes at 97.4oC, making it energetically favour-able compared to other potential by-products MACl andMAI. Implementation of lead acetate as a lead source in-stead of the typical halide lead sources PbX2 is demon-strated in scanning electron microscopy (SEM) images offilms in figure 19.

Figure 19: From Nature Communications,[52] scanning elec-tron microscopy (SEM) images showing the morphology ofFTO/TiO2/MAI:PbX with differing lead sources. (a) PbCl2,(c) PbI2 and (e) PbAc2, are top down SEM images of the per-ovskite films. Also included are cross-sections of full devicesusing the (b) PbCl2, (d) PbI2 and (f) PbAc2 lead sources.

An example of a smooth and even film obtained from MAI:PbAc2 with a 3:1 molar ratio is shown in figure 19 (e),the quality of the film is described by the group who madeit as comparable to a vapour deposited sample. The av-erage PCE of devices using PbAc2, PbI2 and PbCl2 were14.0%, 9.3% and 12% respectively; demonstrating the im-provement from using this alternative lead source.[52]

Another notable addition to the perovskite formula is theuse of iodopentaflurobenzene (IPFB), this material is de-posited before the HTM in n-i-p structured devices and hasbeen shown to passivate trapped states in the perovskite.[82]

The idea of producing lead-free non-toxic perovskites so-lar cells has received a lot of attention in research level andpopular science publications. The best potential candidateso far is tin (Sn). To demonstrate the functionality of Sn,one study integrated Sn into a lead halide perovskite hy-brid cell with the formula MASn1xPbxI3. This hybrid cellwas found to have a bang gap tunable between 1.17→ 1.55eV with the highest device PCE only reaching 3%. Withconfirmation that Sn could work another group went on toproduce the first ever MASnI3 lead free device, and reacheda champion PCE of 5%.[8] No doubt research into lead-freeperovskites will continue to remain a popular topic, howeverHenry Snaith’s group maintain that the restrictions withinthe European Union will not stop perovskites containinglead from reaching a commercial market.

14

3.5.2 Electron Transport Mediums& Hole Transport Mediums

Despite TiO2 being one of the most common ETMs em-ployed in perovskite n-i-p architectures, common TiO2 lay-ers require a 500oC calcination, and have also been linkedto interfacial UV degradation when in contact with per-ovskite materials. A recent publication has attempted toovercome the first of these problems by developing a solu-tion processable TiO2 layer which could eventually replacethe high temperature TiO2 deposition for n-i-p devices thatdo not need a precisely engineered TiO2 scaffold. In addi-tion, a solution processable TiO2 could be used as an inter-layer or transport layer in p-i-n architectures to replace orlie below PCBM. Since TiO2 is often deposited in air thismay provide a route to fabricate a p-i-n device architecturethat is fully air-processable; using PEDOT:PSS as an HTMand TiO2 as an ETM. Unfortunately the champion devicesusing this new solution processable TiO2 currently haveless that half the champion PCE of those without TiO2 orusing standard calcinated TiO2.[49] Another alternative tothe traditional TiO2 layer is the implementation of inter-calated graphene flakes into the TiO2 layer, which recentlyhave produced devices yielding a maximum PCE of 15.6%with fabrication temperatures not exceeding 150oC.[2,44,53]

Spiro-MeOTAD is a common HTM used in n-i-p devicearchitectures which requires doping with li-TSFI in orderto increase hole mobility, however it also makes the Spiro-MeOTAD more susceptible to moisture. Another dopantused in Spiro-MeOTAD is TBP which has been shown toreact with perovskite and also increase measurable par-asitic resistance when used in conjuncture with li-TSFI.This resistance may be caused due to it being energeticallyfavourable for li-TSFI to migrate to the TiO2 surface inthe presence of TBP, decreasing VOC and FF.[70] It hasbeen suggested that TBP could be replaced with mont-morillonite in order to improve the effectiveness of Spiro-MeOTAD without causing these additional resistances anddegradation effects.[70]

Another low temperature solution processable ETM is zincoxide (ZnO) which has been demonsatrated to work as aninterlayer alongside PCBM in a p-i-n architecture produc-ing a device with 16.8% PCE.[83]

In a traditional p-i-n architecture PEDOT:PSS should bedeposited onto hot ITO held at 120oC in order to create asmooth surface on which to deposit the perovskite.[84] Analternative method for perovskite solar cells is to removethe ETM entirely, one group demonstrated a FTO/MAPbI3−xClx/Spiro/Au device that achieved 14.4% PCE.[85] ETMfree devices may not ever reach higher PCE’s given thathole and electron transport rates should be balanced in or-der to achieve the maximum theoretical PCE for a partic-ular solar cell.[7]

Figure 20: From Advanced Functional Materials,[46] depen-dence of perovskite morphology on annealing temperature andthickness. (a) SEM of film perovskite MAPbI3−xClx morpholo-gys. Top row: annealing temperatures 90 170oC with constantfilm thickness 650 ± 50 nm. Bottom row: film thickness 50-700nm with constant annealing temperature 95oC. (b) Perovskitefilm coverage as a function of annealing temperature. (c) Per-ovskite film coverage as a function of film thickness.

3.6 Deposition Optimization

It is clear that the formation of the perovskite film willdepend heavily on the choice of contacts, transport layersand perovskite formula used, for example the inclusion ofCl has shown evidence of increasing the MA out-gassingduring annealing.[10] There are other additional fabricationtechniques that have been reported to improve the forma-tion of perovskite films in order to increase the stability andperformance of perovskite devices.

The choice of solvent (or solvents) used to dissolve the per-ovskite precursor can change the morphology of the film.Common solvents used are DMF, γ-butyrolactone and Dimethylsulfoxide (DMSO). For the current record efficiency per-ovskite devices an improvement in film coverage and smooth-ness has been partially attributed to the use of γ-butyrolactone:DMSO solvent blend (7:3, volume ratio).[7]

Two obvious factors affecting the morphology of perovskitedevices are the thickness of the layer being deposited andthe annealing temperature chosen. Figure 20 give an ex-ample of how a MAPbI3−xClx film is affected by these twoproperties and demonstrates how detrimental over-annealingcan occur at temperatures above 110oC.[46] Annealing causescrystallisation only after the excess organic component isdriven out, therefore a volatile organic component is needed.

15

Crystallisation should be driven only by temperatures be-low 110oC for MA based devices, as it is calculated that alarge increase in halide deficiencies occur over 110oC.[52]

The deposition of the perovskite can be completed using co-evaporation to produce a very uniform and smooth film,[84]

however most research groups try to avoid this time con-suming and expensive technique and use spin coating in-stead. Spin coating has achieved resounding success whencombined with several additional techniques. A toluenewash is commonly deployed in a second spin cycle quicklyfollowing the deposition of the perovskite, the toluene washesaway excess unreacted organic material and encourages anearlier initiation of the crystallisation process. This as beenassociated with large crystal grains and a overall smoother,pin-hole free film morphology.[7,51,86] A similar improvementin perovskite devices has been observed by the implementa-tion of vapour or solvent annealing, where the film crystalgrowth is made slower in the presence of excess solvent, en-couraging much larger crystal grain sizes.[83] Hot casting theperovskite precursor at a high temperature (∼70oC) onto ahot substrate (∼180oC) has reported increased grain sizesof up to millimetres in length achieving PCE’s of up to18%.[74] Various groups use either hot solution or hot sub-strates at some point in their fabrication routines but thereseems to be no universal standard technique to improvedevice morphology. Several groups have reported that anadditional amount (∼ 6% volume) of stabilised HI can beadded to the precursor ink to form much smoother films.[12]

Common 2-step fabrication processes include the depositionof the lead source PbX2 in DMF and organic halide com-ponent MAX in IPA one after the other.[84] These are oftenspin coated sequentially or, in some cases, a 2-step solutiondump has been used where a PbI2 film has been dipped ina bath of MAI in IPA. Devices fabricated using a 2-step-dipprocess kept 4/5ths of their original PCE after 500 hours ofageing in ambient conditions.[87] FA has recently proved it-self a capable MA replacement again as FAPbI3 films withover 20% PCE have been made by using intramolacularexchange. These impressive devices utilize the reactionPbI2-DMSO + FAI ↔ PbI2-FAI + DMSO, whereby it isenergetically favourable for a PbI2(DMSO) precipitate ob-tained from PbI2(DMSO)2 to replace DMSO with the FAIdeposited on top through intermolecular exchange. A sim-ilar process worked for producing the formidable FAPbI3:MAPbBr3 devices.[75]

Other additional factors that may also improve the devicequality include the implementation of a drying step beforeannealing, or the removal of airflow by the switching offglovebox circulation and/or lamina flow hoods during per-ovskite and transport layer deposition and annealing. Areduction in airflow could be a contributing factor to thesuccess of sealed solvent annealed perovskite films.[83]

Figure 21: From Nano Letters,[88] schematic of a scaffold meso-porous Al2O3/Perovskite with a novel HTM-encapsulate hybridof P3HT/SWNTs capped with PMMA.

3.7 Encapsulation

Encapsulation can stop many of the degradation effectswhich destabilise perovskite solar cells. Typical laboratoryUV activated encapsulation is relatively ineffective com-pared to some novel encapsulates that have been developedspecifically to target degradation effects and increase thelifetime of perovskite and organic solar cells.

The nano-phosphor layer europium doped yttrium vana-date (YVO4:Eu3+) has been implemented as an extra en-capsulation layer to protect the perovskite device beneathfrom UV illumination. The YVO4:Eu3+ red shifts UV pho-tons into the absorption region of the perovskite which pro-tects the UV sensitive interfaces in the perovskite whilstsimultaneously increasing the PCE of the devices. It ispredicted that an extra 8.5% PCE could be achieved froma well engineered down shifting layer. Sample devices im-plementing YVO4:Eu3+ retained half of their original PCEinstead of a third for those without YVO4:Eu3+ after pro-longed UV illumination.[21]

A very successful encapsulating layer that has been de-veloped is a HTM P3HT/Single walled nanotube (SWNT)layer.[89] This layer fits into a typical n-i-p perovskite archi-tecture as demonstrated in figure 21. It has remarkable holeextraction properties which allow it to entirely replace thetypical doped spiro-OMeTAD layer. P3HT/SWNT cappedwith Polymethyl methacrylate (PMMA) protects the de-vice from moisture whilst removing the degradation of per-ovskite associated with li-TSFI and TBP. Figure 23 showsthe images taken of perovskite/HTM samples over 98 hoursheld at 80oC in ambient atmospheric conditions, it is clearthat P3HT/SWNT-PMMA layers stabilise the perovskite.[89]

16

Figure 22: Edited from Nature Communications,[14] density function theorem (DFT) calculations on tuning of A cations steric sizeinfluencing the X-B halide-metal bond angle as the octahedral tilt and band gap of the perovskite APbI3 changes. (a) Octehedra01, 02, 03, 04 of different tilts demonstrating the apical and equatorial bond angle (αa and αe). (b) Map of the DFT calculatedband gaps as a function of αa and αe. The dashed line shows the crystalline symmetry where αa = αe. Dark circles are DFTcalculated perovskites and white circles are previously synthesised perovskites. The size of circles indicates the Pb-I bond lengthvarying away from the MA-PbI3 (3.18 A). (c) Assuming αa ∼ αe this graph shows the correlation between bond angle and the sizeof the replacement A cation.

Figure 23: From Nano Letters,[88] pictures showingthe degradation of MAPbI3−xClx/HTMs over 98 hours at80oC in ambient atmospheric conditions. The HTMsused are: Li-spiro-OMeTAD (spiro-OMeTAD doped withLithium bis(trifluoromethane sulfonyl) imide), P3HT (Poly(3-hexylthiophene-2,5-diyl)), PTAA (Poly(triaryl amine)), PMMA(Poly(triaryl amine)) and P3HT/PMMA-SWNT (Single-walledcarbon nanotubes).

The P3HT/SWNT-PMMA samples were also found to re-sist running water.[89]

3.8 Modeling the Perovskite

Computational models can sometimes save time in the lab-oratory by removing the need for rigorous trial and errorexperimentation to reach a device architecture that works.For example the band gaps, VBMs and CBMs found infigure 6 were determined using computational spin orbitalcoupling SOC-GW and QSGW quasiparticle self-consistentGW DFT models (Using the GW approximations). Someof the predicted band gaps of MAPbX3 and FAPbX3 (X =I,Br,Cl) are within ∼ 0.1eV of experimental obtained val-ues.[19] We can now be fairly sure that a DFT model canproduce an decent estimate of a semiconductor’s band gap.This means the functionality of any new potential itera-tions of perovskite (with replacement organic cations) canbe tested computationally before they are implemented inthe laboratory.[52]

An example of a piece of published computational engineer-ing is shown in figure 22, where changes in apical and equa-torial (X halide)-(B Metal) bond angle (αa and αe shownin (a)) caused by the various steric sizes (95% Density ofionic radii) of possible A cation replacements are mappedagainst their DFT calculated band gap (shown in (b)).[14]

17

n-k data of the perovskite and transport layers can be usedto model the device as an optical cavity and determine theoptimum architecture of a perovskite device. One groupused such a model to produce ideal thicknesses for a devicewith architecture: ITO(850nm)/PEDOT:PSS (15nm)/PCD-TBT(5nm)/perovskite(350nm)/PC60BM/Ag. The deviceachieved 16.5% PCE with minimal laboratory optimiza-tion.[39]

4 A solar cell to competewith silicon?

Ultimately; the perovskite community needs to produce astress tested perovskite device that emulates enough yearsof operational life time to be cost competitive with c-Si andCIGS. Oxford-PV and Snaith’s research group believe thatthis goal is soon to be reached with a first generation ofcommercial glove-box or vacuum processed devices. Thesegroups have identified that perovskites can pass many of thegovernmental and European Union restrictions pertainingto lead and have applied for several patents surroundingthe perovskite solar cell.

There is still promise for air processed p-i-n devices suchas ITO/PEDOT:PSS/Perovskite/PCBM/Ca/Al. Ambienthumidity has been proved advantageous for some perovskiteformulae, but a new deposition technique or replacementmaterial for PCBM is needed to get ETMs integrated intofully air-processable p-i-n devices.

Device stability and efficiencies have continued to improvealongside a reduction in hysteresis by using new formu-las and deposition techniques. Smooth spin cast films ofFA/MA perovskite blends have produced supposedly hys-teresis free n-i-p devices by implementing either intramolec-ular exchange or toluene washes.

Publications using the FAI:MAX:PbI2:PbX2 blends haveso far only used devices architectures that implement theexpensive 500oC calcination step for the TiO2 layer. It alsoappears that the lead acetate source has not been used incombination with any FAI:MAX blend or in any fabricationroutine in air. This may be a potential avenue for futureresearch but it is likely that lead acetate perovskite fabrica-tion routines will not work will in air. A new formula thatcould prove successful is a FAI:MA(Br/Cl):PbAc2 blend ina ITO/PEDOT:PSS/Perovskite/PC70BM/Ca/Al device, us-ing an acetate lead source to obtain short annealing times.An air-processable version may need a solution processableoxide ETM layer instead of PC70BM and an alternativeconventional PbX2 halide lead source since PbAc2 may notimprove film morphology in air.

A toluene wash to quench excess solvent could be imple-mented instead of PbAc2 to maintain high quality mor-phology.

If you extrapolate the amazing steps taken by the per-ovskite community so far then the future of perovskite solarcells looks promising... and if you are feeling cynical aboutthe devices’ undesirable muddy-brown appearance holdingthem back, there are even groups working specifically onmaking them rose tinted.

18

References

[1] Gary Hodes. “Perovskite-Based Solar Cells”. In: 342.October (2013), pp. 317–319.

[2] M.a. Green, a. Ho-Baillie, and H.J. Snaith. “The emergence of perovskite solar cells”. In: Nature Photonics 8.7 (June2014), pp. 506–514. issn: 1749-4885. doi: 10.1038/nphoton.2014.134. url: http://www.nature.com/doifinder/10.1038/nphoton.2014.134.

[3] Pablo Docampo et al. “Efficient organometal trihalide perovskite planar-heterojunction solar cells on flexible polymersubstrates.” In: Nature communications 4 (Jan. 2013), p. 2761. issn: 2041-1723. doi: 10.1038/ncomms3761. url:http://www.ncbi.nlm.nih.gov/pubmed/24217714.

[4] Robert F Service. “Turning up the light.” In: Science (New York, N.Y.) 342.6160 (2013), pp. 794–5, 797. issn:1095-9203. doi: 10.1126/science.342.6160.794. url: http://www.ncbi.nlm.nih.gov/pubmed/24233703.

[5] Akihiro Kojima et al. “Novel Photoelectrochemical Cell with Mesoscopic Electrodes Sensitized by Lead-halide Com-pounds (11)”. In: Meeting Abstracts MA2008-02.1 (2008), p. 27. issn: 2151-2043. url: http://ma.ecsdl.org/content/MA2008-02/1/27.abstract.

[6] Michael Gratzel. “The light and shade of perovskite solar cells”. In: Nature Materials 13.9 (Aug. 2014), pp. 838–842.issn: 1476-1122. doi: 10.1038/nmat4065. url: http://www.nature.com/doifinder/10.1038/nmat4065.

[7] Nam Joong Jeon et al. “Compositional engineering of perovskite materials for high-performance solar cells”. In:Nature 517.7535 (2015), pp. 476–480. issn: 0028-0836. doi: 10.1038/nature14133. url: http://dx.doi.org/10.1038/nature14133.

[8] Baohua Wang, Xudong Xiao, and Tao Chen. “Perovskite photovoltaics: a high-efficiency newcomer to the solar cellfamily.” In: Nanoscale (Sept. 2014). issn: 2040-3372. doi: 10.1039/c4nr04144e. url: http://www.ncbi.nlm.nih.gov/pubmed/25257655.

[9] Silvia Colella et al. “MAPbI3-xClx mixed halide perovskite for hybrid solar cells : the role of chloride as dopant on thetransport and structural properties MAPbI 3-x Cl x mixed halide perovskite for hybrid solar cells : the role of chlorideas dopant on the transport and stru”. In: Chemistry of Materials 25.22 (2013), p. 4613. doi: 10.1021/cm402919x.

[10] Hui Yu et al. “The Role of Chlorine in the Formation Process of CH 3 NH 3 PbI 3-x Cl x Perovskite”. In:Advanced Functional Materials (2014), n/a–n/a. issn: 1616301X. doi: 10.1002/adfm.201401872. url: http:

//doi.wiley.com/10.1002/adfm.201401872.

[11] Sneha a. Kulkarni et al. “Band-gap tuning of lead halide perovskites using a sequential deposition process”. In:Journal of Materials Chemistry A 2.24 (2014), p. 9221. issn: 2050-7488. doi: 10.1039/c4ta00435c. url: http://pubs.rsc.org/en/content/articlehtml/2014/ta/c4ta00435c.

[12] Giles E Eperon et al. “Environmental Science Formamidinium lead trihalide : a broadly tunable perovskite for e fficient planar heterojunction solar”. In: (2014), pp. 982–988. doi: 10.1039/c3ee43822h.

[13] Siliu Lv et al. “One-step, solution-processed formamidinium lead trihalide (FAPbI(3-x)Clx) for mesoscopic perovskite-polymer solar cells”. In: Phys. Chem. Chem. Phys. 16 (2014), Ahead of Print. issn: 1463-9076. doi: 10.1039/C4CP02113D. url: http://pubs.rsc.org/en/content/articlepdf/2014/cp/c4cp02113d.

[14] Marina R Filip et al. “Steric engineering of metal-halide perovskites with tunable optical band gaps Marina”. In:Nature Communications 5 (2014), pp. 1–9. doi: 10.1038/ncomms6757. url: http://dx.doi.org/10.1038/

ncomms6757.

[15] Wan-jian Yin, Yanfa Yan, and Su-huai Wei. “Anomalous Alloy Properties in Mixed Halide Perovskites”. In: (2014).

[16] Anna Amat et al. “Cation-induced band-gap tuning in organohalide perovskites: Interplay of spin-orbit couplingand octahedra tilting”. In: Nano Letters 14.6 (2014), pp. 3608–3616. issn: 15306992. doi: 10.1021/nl5012992.

[17] John B. Goodenough and J.-S. Zhou. “Orbital ordering in orthorhombic perovskites”. In: Journal of MaterialsChemistry 17.23 (2007), p. 2394. issn: 0959-9428. doi: 10.1039/b701805c.

[18] Kewei Wu et al. “Temperature-dependent excitonic photoluminescence of hybrid organometal halide perovskitefilms”. In: Phys. Chem. Chem. Phys. 16.41 (2014), pp. 22476–22481. issn: 1463-9076. doi: 10.1039/C4CP03573A.url: http://dx.doi.org/10.1039/C4CP03573Ahttp://pubs.rsc.org/en/content/articlehtml/2014/cp/c4cp03573a.

[19] Simon a. Bretschneider et al. “Research Update: Physical and electrical characteristics of lead halide perovskites forsolar cell applications”. In: APL Materials 2.4 (2014), p. 040701. issn: 2166-532X. doi: 10.1063/1.4871795. url:http://scitation.aip.org/content/aip/journal/aplmater/2/4/10.1063/1.4871795.

19

[20] a. Poglitsch and D. Weber. “Dynamic disorder in methylammoniumtrihalogenoplumbates (II) observed by millimeter-wave spectroscopy”. In: The Journal of Chemical Physics 87.11 (1987), p. 6373. issn: 00219606. doi: 10.1063/1.453467. url: http://scitation.aip.org/content/aip/journal/jcp/87/11/10.1063/1.453467.

[21] Guangda Niu, Xudong Guo, and Liduo Wang. “Review of recent progress in chemical stability of perovskite solarcells”. In: J. Mater. Chem. A 3 (2015), pp. 8970–8980. issn: 2050-7488. doi: 10.1039/C4TA04994B. url: http://xlink.rsc.org/?DOI=C4TA04994B.

[22] Jun Hong Noh et al. “Chemical Management for Colorful, Efficient, and Stable Inorganic Organic Hybrid Nanos-tructured Solar Cells”. In: (2013).

[23] Tom Baikie et al. “Synthesis and crystal chemistry of the hybrid perovskite (CH3NH3)PbI3 for solid-state sensitisedsolar cell applications”. In: Journal of Materials Chemistry A 1.18 (2013), p. 5628. issn: 2050-7488. doi: 10.1039/c3ta10518k. url: http://xlink.rsc.org/?DOI=c3ta10518k.

[24] Teck Ming Koh et al. “Formamidinium-Containing Metal-Halide: An Alternative Material for Near-IR AbsorptionPerovskite Solar Cells”. In: (2014). issn: 1932-7447. doi: 10.1021/jp411112k.

[25] Jeong-Hyeok Im et al. “6.5% Efficient Perovskite Quantum-Dot-Sensitized Solar Cell”. In: Nanoscale 3.10 (2011),p. 4088. issn: 2040-3364. doi: 10.1039/c1nr10867k.

[26] Christian Wehrenfennig et al. “High charge carrier mobilities and lifetimes in organolead trihalide perovskites”. In:Advanced Materials 26.10 (2014), pp. 1584–1589. issn: 15214095. doi: 10.1002/adma.201305172.

[27] Henry J Snaith. “Perovskites : The Emergence of a New Era for Low-Cost , High-”. In: J. Phys. Chem. Lett 4 (2013),36233630. issn: 1948-7185. doi: 10.1021/jz4020162.

[28] a Kojima et al. “Organo Metal Halide Perovskites as Visible-Light Sensitizer for Photovoltaic Cells”. In: PrivateCommunication 1 (2009), p. 1. issn: 0002-7863. doi: Doi10.1021/Ja809598r.

[29] Brian O’Regan and Michael Gratzel. A low-cost, high-efficiency solar cell based on dye-sensitized colloidal TiO2films. 1991. doi: 10.1038/353737a0.

[30] Michael M Lee et al. “Efficient hybrid solar cells based on meso-superstructured organometal halide perovskites.”In: Science (New York, N.Y.) 338.6107 (Nov. 2012), pp. 643–7. issn: 1095-9203. doi: 10.1126/science.1228604.url: http://www.ncbi.nlm.nih.gov/pubmed/23042296.

[31] Hui-Seon Kim et al. “Lead Iodide Perovskite Sensitized All-Solid-State Submicron Thin Film Mesoscopic Solar Cellwith Efficiency Exceeding 9%”. In: Scientific Reports 2 (2012), pp. 1–7. issn: 2045-2322. doi: 10.1038/srep00591.

[32] Olga Malinkiewicz et al. “Perovskite solar cells employing organic charge-transport layers”. In: Nature Photonics8.2 (2014), pp. 128–132. issn: 1749-4885. doi: 10.1038/nphoton.2013.341. url: http://www.nature.com/doifinder/10.1038/nphoton.2013.341$\backslash$nhttp://dx.doi.org/10.1038/nphoton.2013.341$

\backslash$n10.1038/nphoton.2013.341$\backslash$nhttp://www.nature.com/nphoton/journal/v8/n2/

abs/nphoton.2013.341.html\#supplementary-information.

[33] Tomas Leijtens et al. “Overcoming ultraviolet light instability of sensitized TiO with meso-superstructured organometaltri-halide perovskite solar cells.” In: Nature communications 4 (2013), p. 2885. issn: 2041-1723. doi: 10.1038/ncomms3885. url: http://www.ncbi.nlm.nih.gov/pubmed/24301460.

[34] Henry J. Snaith et al. “Anomalous Hysteresis in Perovskite Solar Cells”. In: The Journal of Physical ChemistryLetters iii (2014), p. 140324182322004. issn: 1948-7185. doi: 10.1021/jz500113x. url: http://pubs.acs.org/doi/abs/10.1021/jz500113x.

[35] Jin Hyuck Heo et al. “Efficient inorganicorganic hybrid heterojunction solar cells containing perovskite compoundand polymeric hole conductors”. In: Nature Photonics 7.6 (2013), pp. 486–491. issn: 1749-4885. doi: 10.1038/nphoton.2013.80. url: http://www.nature.com/doifinder/10.1038/nphoton.2013.80.

[36] Mingzhen Liu, Michael B Johnston, and Henry J Snaith. “Efficient planar heterojunction perovskite solar cells byvapour deposition.” In: Nature 501.7467 (2013), pp. 395–8. issn: 1476-4687. doi: 10.1038/nature12509. url:http://www.ncbi.nlm.nih.gov/pubmed/24025775.

[37] Ye-Jin Jeon et al. “Planar heterojunction perovskite solar cells with superior reproducibility”. In: Scientific Reports(2014), pp. 26–28. issn: 2045-2322. doi: 10.1038/srep06953.

[38] View Article Online et al. “Environmental Science”. In: (2014), pp. 1377–1381. doi: 10.1039/c4ee00168k.

[39] Qianqian Lin et al. “Electro-optics of perovskite solar cells”. In: Nature Photonics 9.2 (2014), pp. 106–112. issn:1749-4885. doi: 10.1038/nphoton.2014.284. url: http://dx.doi.org/10.1038/nphoton.2014.284.

20

[40] Qi Chen et al. “Planar heterojunction perovskite solar cells via vapor assisted solution process.” In: Journal ofthe American Chemical Society Scheme 1 (2013), pp. 3–6. issn: 1520-5126. doi: 10.1021/ja411509g. url: http://www.ncbi.nlm.nih.gov/pubmed/24359486.

[41] Qi Wang et al. “Large fill-factor bilayer iodine perovskite solar cells fabricated by a low-temperature solution-process”. In: Energy & Environmental Science 7.7 (2014), p. 2359. issn: 1754-5692. doi: 10.1039/c4ee00233d. url:http://xlink.rsc.org/?DOI=c4ee00233d.

[42] Maria Antonietta Loi and Jan C Hummelen. “Hybrid solar cells: Perovskites under the Sun.” In: Nature materials12.12 (Dec. 2013), pp. 1087–9. issn: 1476-1122. doi: 10.1038/nmat3815. url: http://www.ncbi.nlm.nih.gov/pubmed/24257133.

[43] Yu-Che Hsiao et al. “Fundamental Physics behind High-Efficiency Organo-Metal Halide Perovskite Solar Cells”. In:J. Mater. Chem. A 00 (2015), pp. 1–14. issn: 2050-7488. doi: 10.1039/C5TA01376C. url: http://pubs.rsc.org/en/Content/ArticleLanding/2015/TA/C5TA01376C.

[44] Tze Chien Sum and Nripan Mathews. “Advancements in Perovskite Solar Cells: Photophysics behind the Photo-voltaics”. In: Energy & Environmental Science (2014), pp. 2518–2534. issn: 1754-5692. doi: 10.1039/c4ee00673a.url: http://pubs.rsc.org/en/Content/ArticleLanding/2014/EE/c4ee00673a.

[45] Alex Barrows et al. “Efficient planar heterojunction mixed-halide perovskite solar cells deposited via spray-deposition”.In: Energy & Environmental Science 7 (2014), pp. 1–7. issn: 1754-5692. doi: 10.1039/C4EE01546K. url: http://pubs.rsc.org/en/Content/ArticleLanding/2014/EE/C4EE01546K.

[46] G E Eperon et al. “Morphological control for high performance, solution-processed planar heterojunction perovskitesolar cells”. In: Advanced Functional Materials 24.1 (Jan. 2014), pp. 151–157. issn: 1616301X. doi: 10.1002/adfm.201302090. url: http://www.scopus.com/inward/record.url?eid=2-s2.0-84891558874\&partnerID=40\&md5=479f23e34febc3dfafe8c02258a9fcc9.

[47] Zhengguo Xiao et al. “Efficient, high yield perovskite photovoltaic devices grown by interdiffusion of solution-processed precursor stacking layers”. In: Energy Environ. Sci. 7.8 (June 2014), Ahead of Print. issn: 1754-5692. doi:10.1039/C4EE01138D. url: http://pubs.rsc.org/en/content/articlepdf/2014/ee/c4ee01138d.

[48] Jinli Yang et al. “Investigation of CH 3 NH 3 PbI 3 Degradation Rates and Mechanisms in Controlled HumidityEnvironments Using in Situ Techniques”. In: 2 (2015), pp. 1955–1963.

[49] Yang-Yen Yu et al. “Perovskite photovoltaics featuring solution-processable TiO 2 as an interfacial electron-transportinglayer display to improve performance and stability”. In: Nanoscale 6.19 (2014), pp. 11403–11410. issn: 2040-3364.doi: 10.1039/C4NR03366C. url: http://xlink.rsc.org/?DOI=C4NR03366C.

[50] Jingbi You et al. “Moisture assisted perovskite film growth for high performance solar cells”. In: Applied PhysicsLetters 105.18 (Nov. 2014), p. 183902. issn: 0003-6951. doi: 10.1063/1.4901510. url: http://scitation.aip.org/content/aip/journal/apl/105/18/10.1063/1.4901510.

[51] Jae Woong Jung, Spencer T. Williams, and Alex K.-Y. Jen. “Low-temperature processed high-performance flexibleperovskite solar cells via rationally optimized solvent washing treatments”. In: RSC Adv. 4.108 (2014), pp. 62971–62977. issn: 2046-2069. doi: 10.1039/C4RA13212B. url: http://xlink.rsc.org/?DOI=C4RA13212B.

[52] Wei Zhang et al. “Ultra-smooth organic-inorganic perovskite thin-film formation and crystallization for EfficientPlanar Heterojunction Solar Cells”. In: Nature Communications (2014). doi: 10.1038/ncomms7142. url: file:///C:/downloads/GoogleDrive/JournalArticles/Mendeley/2014/Zhangetal.-2014-Ultra-smoothorganic-

inorganicperovskitethin-filmformationandcrystallizationforEfficientPlanarHeterojunct.pdf.

[53] Jacob Tse Wei Wang et al. “Low-temperature processed electron collection layers of graphene/TiO 2 nanocompositesin thin film perovskite solar cells”. In: Nano Letters 14.2 (2014), pp. 724–730. issn: 15306984. doi: 10.1021/

nl403997a.

[54] Guichuan Xing et al. “Low-temperature solution-processed wavelength-tunable perovskites for lasing.” In: Naturematerials 13.5 (May 2014), pp. 476–80. issn: 1476-1122. doi: 10.1038/nmat3911. url: http://www.ncbi.nlm.nih.gov/pubmed/24633346.

[55] Jingbi You et al. “Low-Temperature Solution-Processed Perovskite Solar Cells with High Efficiency and Flexibility”.In: acsNANO 2 (2014), pp. 1674–1680.

[56] Michael D Mcgehee. “Polymers and Perovskites for Hybrid Tandem Photovoltaics Tandem Photovoltaics Goal :Inexpensive cells with efficiency 25 %”. In: (), pp. 1–23.

21

[57] Chang-Wen Chen et al. “Optical properties of organometal halide perovskite thin films and general device structuredesign rules for perovskite single and tandem solar cells”. In: J. Mater. Chem. A 3.17 (2015), pp. 9152–9159. issn:2050-7488. doi: 10.1039/C4TA05237D. url: http://xlink.rsc.org/?DOI=C4TA05237D.

[58] Chun-chao Chen et al. “Materials Horizons”. In: Materials Horizons 00 (2015), pp. 1–9. issn: 2051-6347. doi:10.1039/C4MH00237G. url: http://dx.doi.org/10.1039/C4MH00237G.

[59] Jacky Even, Laurent Pedesseau, and Claudine Katan. “Analysis of multivalley and multibandgap absorption andenhancement of free carriers related to exciton screening in hybrid perovskites”. In: Journal of Physical ChemistryC 118.22 (2014), pp. 11566–11572. issn: 19327455. doi: 10.1021/jp503337a.

[60] Atsuhiko Miyata et al. “Direct measurement of the exciton binding energy and effective masses for charge carriersin organicinorganic tri-halide perovskites”. In: 11.July (2015). doi: 10.1038/NPHYS3357.

[61] Valerio D’Innocenzo et al. “Excitons versus free charges in organo-lead tri-halide perovskites.” In: Nature commu-nications 5 (2014), p. 3586. issn: 2041-1723. doi: 10.1038/ncomms4586. url: http://www.ncbi.nlm.nih.gov/pubmed/24710005.

[62] Guichuan Xing Nripan Mathews et al. “Long-Range Balanced Electron- and Hole-Transport Lengths in Organic-Inorganic CH3NH3PbI3 Guichuan”. In: Science 342 (2013), pp. 344–346.

[63] Pablo P. Boix et al. “Current progress and future perspectives for organic/inorganic perovskite solar cells”. In:Materials Today 17.1 (2014), pp. 16–23. issn: 13697021. doi: 10.1016/j.mattod.2013.12.002. url: http:

//dx.doi.org/10.1016/j.mattod.2013.12.002.

[64] James M. Ball et al. “Optical properties and limiting photocurrent of thin-film perovskite solar cells”. In: EnergyEnviron. Sci. 8.2 (2015), pp. 602–609. issn: 1754-5692. doi: 10.1039/C4EE03224A. url: http://xlink.rsc.org/?DOI=C4EE03224A.

[65] M. Hirasawa et al. “Magnetoabsorption of the lowest exciton in perovskite-type compound (CH3NH3)PbI3”. In:Physica B: Condensed Matter 201 (1994), pp. 427–430. issn: 09214526. doi: 10.1016/0921-4526(94)91130-4.

[66] Dong Shi et al. “Low trap-state density and long carrier diffusion in organolead trihalide perovskite single crystals”.In: ().

[67] Samuel D Stranks et al. “Electron-Hole Diffusion Lengths Exceeding 1Micrometer in an Organometal TrihalidePerovskite Absorber”. In: Science 342.October (2013), pp. 341–345.

[68] Qianqian Lin et al. “SUPPLEMENTARY INFORMATION Electro-optics of perovskite solar cells”. In: (2014). doi:10.1038/nphoton.2014.284.

[69] Kenichiro Tanaka et al. “Comparative study on the excitons in lead-halide-based perovskite-type crystals CH3NH3PbBr3CH3NH 3PbI3”. In: Solid State Communications 127.9-10 (2003), pp. 619–623. issn: 00381098. doi: 10.1016/S0038-1098(03)00566-0.

[70] Amalie Dualeh et al. “Thermal Behavior of Methylammonium Lead-trihalide Perovskite Photovoltaic Light Har-vesters”. In: Chem. Mater. (2014), Ahead of Print. issn: 0897-4756. doi: 10.1021/cm502468k. url: http://pubs.acs.org/doi/pdfplus/10.1021/cm502468k.

[71] Chen Zhao et al. “Revealing Underlying Processes Involved in Light Soaking Effects and Hysteresis Phenomena inPerovskite Solar Cells”. In: (2015), pp. 1–6. doi: 10.1002/aenm.201500279.

[72] Jarvist M Frost et al. “Atomistic origins of high-performance in hybrid halide perovskite solar cells.” In: Nano letters14.5 (May 2014), pp. 2584–90. issn: 1530-6992. doi: 10.1021/nl500390f. url: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4022647\&tool=pmcentrez\&rendertype=abstract.

[73] Jeffrey a. Christians, Joseph S. Manser, and Prashant V. Kamat. “Best Practices in Perovskite Solar Cell EfficiencyMeasurements. Avoiding the Error of Making Bad Cells Look Good”. In: The Journal of Physical Chemistry Letters6.5 (2015), pp. 852–857. issn: 1948-7185. doi: 10.1021/acs.jpclett.5b00289. url: http://pubs.acs.org/doi/abs/10.1021/acs.jpclett.5b00289.

[74] Wanyi Nie et al. “High-efficiency solution-processed perovskite solar cells with millimeter-scale grains”. In: Science347.6221 (2015), pp. 522–525. issn: 0036-8075. doi: 10.1126/science.aaa0472.

[75] Chan Kim et al. “High-performance photovoltaic perovskite layers fabricated through intramolecular exchange”. In:May (2015), pp. 1–8.

[76] Kenjiro Miyano et al. “Simple characterization of electronic processes in perovskite photovoltaic cells Simple char-acterization of electronic processes in perovskite photovoltaic cells”. In: 093903 (2015). doi: 10.1063/1.4914086.

22

[77] Aurelien Leguy et al. “The reversible hydration of CH 3 NH 3 PbI 3 in films, single crystals and solar cells”. In:Chemistry of Materials (2015), p. 150405140636003. issn: 0897-4756. doi: 10.1021/acs.chemmater.5b00660. url:http://pubs.acs.org/doi/abs/10.1021/acs.chemmater.5b00660.

[78] Eva L Unger et al. “Chloride in Lead Chloride-Derived Organo-Metal Halides for Perovskite-Absorber Solar Cells”.In: (2014).

[79] Alice E. Williams et al. “Perovskite processing for photovoltaics: a spectro-thermal evaluation”. In: J. Mater. Chem.A 2.45 (2014), pp. 19338–19346. issn: 2050-7488. doi: 10.1039/C4TA04725G. url: http://xlink.rsc.org/?DOI=C4TA04725G.

[80] Andreas Binek et al. “Stabilization of the Trigonal High-Temperature Phase of Formamidinium Lead Iodide”. In: TheJournal of Physical Chemistry Letters (2015), pp. 1249–1253. issn: 1948-7185. doi: 10.1021/acs.jpclett.5b00380.url: http://pubs.acs.org/doi/abs/10.1021/acs.jpclett.5b00380.

[81] Eric T Hoke et al. “Reversible photo-induced trap formation in mixed-halide hybrid perovskites for photovoltaics”.In: Chem. Sci. 00 (Nov. 2014), pp. 1–5. issn: 2041-6520. doi: 10.1039/C4SC03141E. url: http://pubs.rsc.org/en/Content/ArticleLanding/2014/SC/C4SC03141E.

[82] Nakita K Noel et al. “Enhanced Photoluminescence and Solar Cell Performance via Lewis Base Passivation ofOrganic A Inorganic Lead Halide Perovskites”. In: 10 (2014), pp. 9815–9821.

[83] Liu Qi Zhang et al. “Highly efficient and stable planar heterojunction perovskite solar cells via a low temperaturesolution process”. In: J. Mater. Chem. A 3 (2015), pp. 12133–12138. issn: 2050-7488. doi: 10.1039/C5TA01898F.url: http://xlink.rsc.org/?DOI=C5TA01898F.

[84] Chien-Hung Chiang, Zong-Liang Tseng, and Chun-Guey Wu. “Planar heterojunction perovskite/PC71BM solar cellswith enhanced open-circuit voltage via a (2/1)-step spin-coating process”. In: J. Mater. Chem. A 2 (2014), Aheadof Print. issn: 2050-7488. doi: 10.1039/C4TA03674C. url: http://pubs.rsc.org/en/content/articlepdf/2014/ta/c4ta03674c.

[85] Weijun Ke et al. “Efficient hole-blocking layer-free planar halide perovskite thin-film solar cells”. In: Nature Com-munications 6 (2015), pp. 1–7. doi: 10.1038/ncomms7700. url: http://dx.doi.org/10.1038/ncomms7700.

[86] Nam Joong Jeon et al. “Solvent engineering for high-performance inorganic-organic hybrid perovskite solar cells.”In: Nature materials 13.July (2014), pp. 1–7. issn: 1476-1122. doi: 10.1038/nmat4014. url: http://www.ncbi.nlm.nih.gov/pubmed/24997740.

[87] Julian Burschka et al. “Sequential deposition as a route to high-performance perovskite-sensitized solar cells.” In:Nature 499.7458 (July 2013), pp. 316–9. issn: 1476-4687. doi: 10.1038/nature12340. url: http://www.ncbi.nlm.nih.gov/pubmed/23842493.

[88] Severin N Habisreutinger et al. “Carbon Nanotube/Polymer Composites as a Highly Stable Hole Collection Layerin Perovskite Solar Cells”. In: (2014). issn: 1530-6992. doi: 10.1021/nl501982b.

[89] Severin N Habisreutinger et al. “Enhanced Hole Extraction in Perovskite Solar Cells Through Carbon Nanotubes”.In: (2014). issn: 1948-7185. doi: 10.1021/jz5021795.

23