Spin-orbit coupling in metal halide perovskites

Metal halide perovskite present outstanding electronic properties. One hypothesis that explains those properties is the presence of a very strong bulk Rashba effect, reportedly the greatest among any other materials. This blog post introduces the Rashba effect in halide perovskites, explaining the origin of both the reported static and dynamic effects. Finally, I give some details about the theory from Byschkov & Rashba and about the photogalvanic effect.


Metal halide perovskites (structure ABX₃ where X is a halide) have been subject to an unparalleled interest in the materials science community for the past 6 years. One reason for such an interest is their incredible long carrier diffusion lengths of up to 1 micron in polycrystalline films and as high as tens of microns in single crystals, driving outstanding photovoltaic performance. It has also been observed a strong optical absorption, with incredibly low non-radiative recombination rates, leading long carrier lifetimes — up to tens of microseconds. These parameters are surprisingly high for the unavoidable presence of defects in solution-processed polycrystalline films. This differs from the image of conventional semiconductors in which defects and impurities are responsible for scattering of carriers, lowering their lifetime and diffusion length.

Therefore, many fundamental questions underlying these outstanding optoelectronic properties still remain unanswered. Particularly, the origin of carrier mobilities, of long lifetimes and very low rate of non-radiative recombination need further studies. In fact, different hypotheses have been proposed to explain both the origin of long-carrier lifetimes and low rate of non-radiative recombination:

  • The presence of a Rashba effect, from which an indirect band gap would originate;
  • A considerably low density of deep intrinsic defects, each one having high formation energies;
  • The existence of strong polaronic effects.

The Rashba effect would explain both the long carrier lifetimes and the intriguing low-rate of non-radiative recombination.

The Rashba effect in metal halide perovskites

Let’s start with the example of MAPI (perovskite phase where the A-site cation is methylammonium – CH₃NH₃, the metal, lead, and the halide is iodide), that is the most studied halide perovskite. In this structure, the presence of heavy atoms such as lead or iodine introduces spin-orbit coupling, that, along with inversion-symmetry breaking in the crystal, the electrons behave as if they were subject to an effective magnetic field which due to the spin-orbit coupling (SOC). This results in a momentum-dependent splitting of electron bands commonly referred to as the Rashba splitting. This effect yields a lifting of the degeneracy in k-space, inducing the shift of both the valence band (VB) maxima and conduction band (CB) minima, away from the high symmetry points of the Brillouin zone. This results in an indirect band gap, as shown in the figure below. Particularly, the splitting of the conduction band suppresses the rate of band-to-band recombination of charge carriers that has been proposed to be the origin of increased carrier lifetimes.

The spin-orbit coupling (SOC) results in a Rashba effect on the VB dispersion of MAPI perovskites. Adapted from Niesner et al (2016). (a,b) Without SOC, double spin-degenerate band with single maximum can be expected (b) is the VBM position represented in k-space. (d,e) SOC and inversion-symmetry breaking fields, the VB splits up in two spin-split bands. alpha is the Rashba parameter that quantifies the strength of the SOC.

The inversion-symmetry breaking can happen either in the bulk of non-centrosymmetric crystals, or at surfaces and interfaces. It has been originally theorised for two-dimensional electron gases (2DEG, see below). There are two main mechanisms that lead to the Rashba effect, breaking the inversion symmetry in the bulk of the MAPI crystals, one is static, the other dynamic.

The static Rashba effect

First, lead iodide sub-lattice can distort from a set of ideal corner-sharing octahedra (tetragonal structure). As recently suggested (Rakita et al, PNAS 2017), this forms a non-centrosymmetric phase that is the source of the static Rashba effect. Particularly, the room-temperature of MAPI is the tetragonal I4/mcm phase, which is not centrosymmetric, that may not be subject to spin-splitting. In other phases, when the orbitals with spin-orbit splitting are not subject to inversion symmetry, the spin-orbit coupling causes a spin-dependent shift of the electronic bands along the k-direction. The double spin-degenerate band reportedly splits into two bands, shifted in k-space by k₀. Thus, the bandgap becomes slightly indirect, as the optical transition become spin-dependent. Particularly, these effects are responsible for a high Rashba parameter, whose intensity has been reported to be ɑR = 2 E₀ / k₀ = 7± 1 eV Å in the orthorhombic perovskite and  11 ± 4 eV Å in the cubic phase . These values are among the highest one ever reported

This effect on the optical transition can be probed with a photogalvanic effect (see below), expected if coherent spin transport takes place on length scales large enough for spin-polarised currents to get through devices. That’s what Niesner and co-workers reported with their systematic study of the circular photogalvanic effect in MAPI single crystals.

The spin-splitting also causes a minimum of energy at the central high symmetry point, of depth E₀. Particularly, in the room temperature tetragonal phase, they observed (Niesner et al, 2016) a circular photogalvanic effect for excitations 110meV below the direct optical gap, indicating that the transitions between spin-polarised electronic bands happened below the direct gap.

Circular dichroism has also been observed in pump-probe spectroscopy, and spin dependence of the charge dissociation has been reported. This may allow for the creation of perovskite-based spintronic devices.

The dynamic Rashba effect

Recently, it has been proposed the existence of a dynamical Rashba effect, allowing for quantification of its magnitude. In fact, the cubic perovskite lattice allows high mobility of the organic dipolar cation that locally allow the breaking of inversion symmetry in the absence of any static distortion of the lead-iodide framework. Locally, this creates screening localisation domains that, combined with the presence of lead, provide MAPI with a giant SOC. The degrees of freedom of the molecular cation gives rise locally to a Rashba effect, fluctuating on the picosecond time scale, related to the dynamics of the methylammonium rotation. This shows that the Rashba effect exists in MAPI, regardless of whether the crystal lattice is centrosymmetric or not (i.e. is orthorhombic, tetragonal or cubic). Those dynamic structural fluctuations can occur as a result of the phonon modes or due to the interaction of the MA⁺ ions with the lead iodide framework.

Particularly, the soft nature of the lead-iodide bond allows it to be easily deformed under symmetry-breaking, due to the important fluctuation of the organic cations. At high temperature, this local structural disorder induces the dynamical Rashba effect.

The amplitude of this dynamical effect is very similar to the one that is observed for bulk Rashba systems (static Rashba). Hence, a change in the direction of the current, that is associated with the circular photogalvanic effect at the orthorhombic-tetragonal phase transition, demonstrates that this dynamical effect has two different physical origins in the two phases.

At room temperature, the perovskite is tetragonal, thus the energy splitting between spin-polarised transition and direct optical transition increase with the temperature, as well as the amplitude of the circular photogalvanic effect. Both effect are responsible for an increase of the Rashba parameter with the temperature, that has been measured experimentally (Niesner et al, 2018), giving support for the predicted dynamical Rashba effect.

The orthorhombic lattice is centrosymmetric. Hence, the bulk Rashba effect is impossible in this low-temperature phase. Thus the temperature-independent circular photogalvanic effect that has been measured is attributed to the reduced symmetry at the surface and interfaces. 

Finally, this Rashba effect explains the reported indirect-direct band gap and thermally activated radiative recombination in tetragonal MAPI (Hutter et al, Nature 2018). Particularly, this provides an evidence for the spin-splitting mechanism at elevated temperatures, that should be seen in materials with inversion symmetry that contains heavy elements like lead, and exhibit soft phonon modes.

Further applications

The Rashba mechanism described herein would potentially be very attractive for both optoelectronics and spintronics. Notably, the exploration of Rashba physics has now been at the heart of the growing research field of spin-orbitronics that focuses on the manipulation of non-equilibrium materials properties using SO coupling.

The reported Rashba effect in metal halide perovskites could pave the way towards new applications, thanks to the unique transport properties that emerge with such an effect. The tremendous field of spin manipulation using spin-orbit coupling could notably lead to a quantum spin Hall effect, to the possibility of creating spin-orbit Torque, spin-orbit Qubits or spin transistors or even study topological insulators and Majorana fermions…

I’m looking forward to reading papers involving halide perovskites in the fantastic field of spin manipulation. Given this, I now expect perovskite to have a pivotal role in modern physics with marvellous applications.


About the Rashba spin-orbit coupling

In non-centrosymmetric crystals, the electronic energy bands are split by spin-orbit coupling. In those systems with structural inversion symmetry breaking, SOC becomes odd in momentum p, that, in two-dimensional electron gas (2DEG), reduces to a linear dependence of the Rashba effect in momentum.

For electrons moving in an electric field (even in the absence of external magnetic field), and effective magnetic field is felt by the electrons in their frame of motion. This is called the spin-orbit field, and couples to the electron’s magnetic moment. When inversion symmetry breaking is present, this SO field becomes odd in electron momentum. Let’s take the mathematical expression to better understand this:

Let’s chose an electron with momentum p moving across a magnetic field B. This electron experiences a Lorentz force in the direction perpendicular to its motion, F= -ep×B/m and possesses a Zeeman energy μB σ·B, where σ is the vector of Pauli spin matrices, m the mass of the electron, e its charge, and μB is the Bohr magneton.

If this electron moves across an electric field E, it experiences the effective magnetic field Beff ~ E×p/mc² in its rest-frame (c is the speed of light), and this field induces a momentum-dependent Zeeman energy called spin-orbit coupling,which Hamiltonian can be written ĤSOB (E×pσ/mc². In usual crystals, the electric field can be given by the gradient of the potential: E=-V. In quantum wells — that have structural inversion symmetry breaking, along the growth direction z — the spin subbands are split in energy. The band splitting was explained by Bychkov and Rashba considering an electric field E = Ez z resulting in an effective spin-orbit coupling of the form:

ĤR = ɑR/ℏ * (z × p) · σ

Here, ɑR is the Rashba parameter. This formula has been derived for 2D plane waves, and is only phenomenological, thus do not apply as such on real systems.

About the photo-galvanic effect

The spin of electrons and holes in solid state systems has been intensively studied in quantum mechanics, and originates an outstanding number of phenomena. The dominant method to generate and investigate the spin polarisation has been optical orientation. Indeed, light propagation within a semiconducting material and scattering by inhomogeneities or mobile carriers is able to generate either a DC current (for short-circuit conditions) or a voltage (for open-circuit), that is called the photogalvanic effect.

Under illumination with a circularly polarised light, it is possible to generate a transformation of the photon angular momentum into a translational motion of free charge carriers. To better understand this phenomenon, one can imagine it as the electronic analogue of a mechanic screw or wheel, that transform a rotation into a linear motion, respectively tangential or perpendicular to the rotation momentum). This effect has a strong signature due to this circular motion, and leads to the possibility to produce helicity-dependent currents, whose behaviour upon the variation of the radiation helicity, the cristallographic orientation on the sample geometry, can be probed.


The cover photo shows ARPES on beamline 5-4 at SSRL. Courtesy of SLAC National Accelerator Laboratory. ARPES is a technique allowing to observe the band dispersion.

The papers that made 2018

The perovskite field has been evolving so rapidly that it’s been hard to follow all its evolution. Here I give a list of the papers published last year that I think advanced perovskite research and marked the important research milestone. I also extended this list to other papers I’ve particularly enjoyed reading in 2018.

Halide perovskites

There are currently huge debates about whether metal halide perovskites present ferroelectric properties, as do their oxide counterparts. In a Science paper this summer, Heng-Yun Ye et al. proposed metal free halide perovskite compositions showing ferroelectricity properties. This could trigger new ferroelectric studies in this field.

Layered halide perovskites (usually called “2D-perovskites”) have been subject to a huge interest because the large organic cations used to separate different sheets of corner-sharing octahedra also provided the structure with enhanced stability against moisture and oxygen degradation. Mike Toney’s and Ted Sargent’s groups collaborated in a Nature materials paper published in September. They provide a kinetic model for the formation of those layered perovskites, providing with new insights allowing orientational and compositional control of the solar cells made out of this material.

The Rashba effect had been theorised for a long time to happen in halide perovskites, as it would explain both the high electron diffusion length and low recombination rates. In May, Kyle Frohna et al. reported in a Nature Communications paper a static Rashba effect, induced by the breaking of inversion symmetry in some phases. In their PNAS paper published in September, Daniel Niesner and co-workers provided experimental evidence of a dynamical Rashba effect, characterised by spin-splitting at elevated temperature.

In September, Roald Hoffman and Maarten Goesten published in JACS a theoretical study: “Mirrors of Bonding in Metal Halide Perovskites”. In this article they investigate the different interactions within CsPbBr₃, providing new insights both on the hydrogen bonding and the band structure of this incredible material.

Solar cell engineering

If I had to highlight a review about tandem solar cells this year, that would be undoubtedly Tomas Leijtens’ in Nature Energy, published this summer. He and his co-workers discuss the latest developments in perovskite tandem solar cells and give perspectives to move this kind of research forwald.

Condensed matter physics

Ming-Min Yang et al. reported in Science a flexo-photovoltaic effect in some non-centrosymmetric crystals like strontium titanate SrTiO₃ (a structural analogue of the photovoltaic halide perovskite CsPbBr₃). This effect is different from the typical photovoltaic effect originating from p-n junction. The presence of such an effect may allow boosting the power-conversion efficiency of solar devices. In my mind, this might provide further explanations of the already incredible solar conversion properties of halide perovskites, in the presence of the debated ferroelectric properties in these compositions.

On a completely different ground, Pablo-Jarillo-Herrero from MIT presented in two Nature papers the presence of superconductivity in twisted graphene bilayers. According to many observers, this is likely to be the breakthrough of the year, in a field that hasn’t seen such enthusiasm since the discovery of graphene in 2004.

These results took everyone in the community by surprise because although superconductivity had been previously observed in heterostructures made with graphene, they always involved another superconductive material. Here, they’ve shown that bilayers, rotated with an angle of 1.1° allowed the generation of a non-conducting state (Mott insulator) that can be turned into a superconducting state if charge carriers are added to the graphene system, under 1.7K. Interestingly, this paper caught attention because superconductivity appears in a much simpler system than what has been studied previously (cuprates), making graphene bilayers a possible Rosetta stone for the understanding of unconventional superconductivity.

In the semiconductors community, Eve Stenson from Max Plank Institute reported in PRL that a beam of positrons could boost semiconductor luminescence, more than 100 times that what an electron beam could provide. This suggests that the electron antiparticles may annihilate on collision with particles in the semiconductor, boosting the efficiency of luminescent devices.

Another paper that sparked a huge interest on Twitter has been Thapa & Pandey’s claim of the discovery of a superconducting material at -37°C. This claim has been largely criticized, notably by Brian Skinner, who discovered a weird noise pattern.


A large part of condensed matter chemistry is moving toward machine learning to be able to predict new molecules. In the perovskite field, this has been highlighted at MRS Fall by Marina Leite who proposed to build an open-access library, useful for all the perovskite community. With a much larger materials science background, Aron Walsh and collaborators published a Nature review detailing tools and principles for this emerging field.

Early this year, new kind of chemical reactions has been intended. Whereas typical chemical reactions are carried out with a number of molecules about the Avogadro number, a reaction on the atomic scale has been encountered by these scientits. They built 1 molecule by merging 2 atoms, trapped in laser-beams of different wavelenghts. This has been interesting because it came out just before the Nobel prize of physics, attributed to the laser tweezers, showing a great example of this technology. On a basic scale, this may be useful to understand chemistry in low pressure systems, far in the universe, and also gives perspective in the design of complicated molecules that could be engineered with such a precision. This paper is very important to design a whole new kind of chemistry.

Climate change

I started this blog by being deeply convinced of the need to tackle climate change, and that the latest research advances in solar energy could be of great help in this domain.

Without a surprise, climate science has continued to evolve this year, providing new proofs of the disaster that already started. In a Nature paper, Springmann and co-workers presented options for keeping the food system within environmental limits. This convinced me to turn vegetarian to preserve life on earth as we know it.


Finally, the mainstream press has been enthusiast about a paper from two of my friends published in Nature Physics this summer. With a groundbreaking experiment, they provide new insights on a system that has been known for centuries: Leidenfrost effect. The most notable result is that they understood intrinsic movements within small droplets floating above a hot surface. I’ve been quite happy seeing such popularity for their results.

The cover photo has been taken from this Youtube video.

New year, new milestones

Happy new year to all my readers! I’m sure 2019 will bring new exciting advancement in the field. At a minimum, there’ll be new stability developments and increased efficiency of halide perovskite solar cells. There also will be more and more fundamental studies. I’ll try to keep pushing with more new posts this year.

I’ve been carrying research in two different labs in 2018 so it’s been a bit hard to manage writing blog posts, engage with the community and curate a twitter feed. Now that I’m carrying out my own research at the University of Manchester for the next 3 years, things will go smoothly.

This year I’d like to talk here about Rashba-Dresselhaus effects and spin-physics within perovskites, the use of machine learning to predict perovskite properties and materials, and the all expanding field of so-called “2D perovskites” (even though I prefer the terminology layered perovskites to describe those systems). More to come pretty soon.

Finally, here is a video on Scientific Video Protocols about the history of perovskite photovoltaics, featuring Prof Henry Snaith:

A Few Notes on Photo-induced Phase Segregation

Tandem solar cells are a very promising technology that would allow to overpass the Shockley-Queisser efficiency limit for photovoltaic devices. They can be manufactured by stacking two solar devices, one of low band-gap (to absorb low-energy photons), on top of which is placed a high band-gap absorber (that would absorb the high-energy photons, therefore reducing the heat losses inside the solar cell). Mixed halide perovskite are particularly interesting for the assembly of those tandem solar cells, because compositional arrangements can allow acute band-gap tuning.

In perovskite of the form ABX₃, where X is a mixture of halide ions, generally iodide and bromide, researchers have observed a phase segregation under illumination. In fact, the halide ions form clusters inside the perovskite, leading to a hysteresis cycle in the J-V characteristic curves and other instabilities. This issue is currently one of the most important to address before perovskite can hit the market.

Here is a collection of a few papers on the subject, that I have found particularly interesting:

  • Eric T. Hoke et al. reported first the photo-induced phase segregation effect. (October 2014)
  • Christopher Eames et al. proposed a mechanism explaining the ion migration in perovskites. (February 2015)
  • Dan Slotcavage, from the same group, continued research on effect. (September 2016)
  • Sergiu Draguta et al. propose a model suggesting that the phase segregation is driven by the band-gap reduction of the iodide-rich phase. (September 2016)
  • Connor G. Bischak et al. proposed that the effect originates in a polaron. (October 2016)
  • Alex J. Barker et al. reported that the role of defects is of utter importance. They hypothesise that the phase segregation is driven by the generation of charge carrier gradients through the thickness of the film. (March 2017)
  • Felix Lang et al. proposed that radiation introduce phase-segregation. (May 2017)
  • Ute B. Cappel et al. measured enrichment of bromide ions at the surface of thin films upon illumination. (July 2017)
  • Gergely F. Samu et al. reported that light-soaking influenced the segregation. (July 2017)
  • Carolin M. Sutter-Fella et al.: review about phase segregation (February 2018)
  • John M. Howard et al. probed a water-induced phase segregation (April 2018)
  • Connor G. Bischak et al. proposed that the photo-induced phase segregation is controlled by tunable polaron distortion, suggesting that phase segregation may be an intrinsic effect of mixed-halide perovskites. (May 2018)

The cover photo is taken from Bischak et al. in ACS Energy Letters. It shows the evolution of perovskite films emission wavelengths (blue = homogeneous film, yellow = iodine clusters).

Band-gap Tuning for Tandem Perovskites

Tuning the active materials for the engineering of tandem solar cells is an arising field of research among the perovskite community. In this post, I present the main targets of tandem solar cells engineering: band-gap tuning. Then, I showcase two solutions that have been reported by students I am currently working with.

Tandem Solar Cells or the Necessity of Wide Band-gaps

Metal halide perovskites, of the form ABX₃ (where A can be methyl ammonium, formamidinium, caesium or a mixture of both, B lead or tin and X a halide, typically iodide, bromide, or mix of both), are gaining much attention for thin film photovoltaics, thanks to their long carrier diffusion lengths, their defect tolerance enabling versatile solution (or vapour) processing and a very strong optical absorption. The efficiencies of so-called perovskite solar cells are comparable to the one achieved with the incumbent silicon technologies.

Two market solutions can be proposed for perovskites: either challenging the incumbent silicon or use perovskite as a layer that can be stacked on top of the silicon cell, boosting its efficiency. For this later solution, a wide band-gap needs to be achieved with perovskites, in order to maximise the energy collection. In fact, the silicon solar cells have a very low band-gap. It means that every incident photon having an energy higher than the silicon band-gap will be collected, but the energy difference (the energy higher than the band-gap), will be dissipated as heat loss. Thus the need of a layer that would extract a portion of higher voltage photo-generated carriers will be valuable in order to overpass the fundamental efficiency limit for single junctions solar cells, commonly known as the Shockley-Queisser limit.

Because the perovskite has a tunable band-gap, that can be controlled with halide substitution (for example), they tend to be considered as the best solution for more efficient solar cells. Their ability to be solution-processed or vapour-deposited would allow benefiting high quality scalable inexpensive integration on existing devices. State of the art perovskite-silicon tandem solar cells achieved record efficiencies of 23.6% for monolithically integrated cells, and 26.4% for mechanically stacked configuration. Attempts to make perovskite-perovskite tandem solar cells have also been tried with record efficiencies of 18.5%, but these rely largely on the development of more efficient and more stable low-end-gap perovskite absorbers (1.1 to 1.3eV). The PCE figures are expected to grow significantly with a deeper understanding of the materials.

All these aspects make the development of perovskites for tandem devices an active field of research.

Band-gap Tuning Induced via Compositional Engineering

In order to achieve higher band-gap solar devices, compositional tuning of the solar cells has been tried by the Materials Science & Engineering group led by Mike McGehee. His students, Bush et al. describe their result in their paper: “Compositional Engineering for Efficient Wide Band-Gap Perovskites with Improved Stability to Photo-induced Phase Segregation”.

They explore various compositions of hybrid perovskites of the form FA1-xCsxPb(I1-yBry)3. They investigate the performance of solar cells as well as the photo-stability by varying the composition between formamidinium (FA) and caesium (Cs) in the perovskite A-site and between iodide and bromide on the X-site. With that, they show that increasing the concentration in caesium into the solid-solution thin film results in higher VOC and greater photo-stability, as it raises the bandgap. This can be compared with the previously known raise of bromide, that yields higher band-gap, with a fewer photo-stability and VOC.

That article might be particularly important in the development of high-efficiency perovskite tandem solar cells, as the authors identify stable compositions with high band-gap (1.68 to 1.75eV), that demonstrate high device efficiencies.

Band-gap Tuning Induced with Change of the Lattice Constants

Variation of those A-site cation changes the band-gap. Prasanna et al. proposed an interpretation in their JACS paper: “Band-gap Tuning via Lattice Contraction and Octahedral Tilting in Perovskite Materials for Photovoltaics”. By exploring both lead and tin on the B-site, they allow different octahedral sizes, as the length of the metal-halide bond changes. In this paper, they show that the use of a smaller A-site cation can distort the perovskite lattice in two different ways: either by tilting the BX₆ octahedra (pictured on the right) or by contracting the lattice isotropically (depicted on the left). The first effect yields to a change of the metal-halide orbital overlap, that varies the band-gap. The second results in a relative contraction that increases the orbital overlap, thus cutting down the band-gap.

scheme: two strategies
Two strategies implemented by Prasanna et al: either octahedral tilting enlarging the band-gap, or lattice contraction yielding to a smaller one.

Those two strategies can be achieved with partial substitution of the large FA cation with the smaller Cs. With lead halide perovskites, this results in octahedral tilting but with tin-based materials to lattice contraction, due to the smaller size of the tin.

Hence the authors provide a framework to tune the band-gap as well as the valence and conduction band positions by controlling the A-site cation composition.

In order to achieve high efficiency perovskite-based tandem solar cells, it is necessary to adjust the band-gap of the top cells. As shown in this post, this can be improved through control of the cation composition, resulting in higher band-gaps. These results will, for sure, be needed for future high-efficiency solar panels.

The cover picture is from Michigan Engineering on Flickr. Scheme from Prasanna et al.

Proposal: Graphene Sheets with Halide Perovskite Layers

Some months ago, I had conversations with some researchers to propose them a graduate project proposal that could be done within the scope of a PhD. I have been admitted at the University of Manchester to carry out this project under the supervision of Prof Sir Andre K. Geim and Prof Irina V. Grigorieva. I have decided to publish this proposal here, as it may give ideas to other students.

Controlling photons and charge carriers interactions in graphene sheets with halide perovskite layers

Julien Barrier

ABSTRACT. In the pursuit of highly efficient photovoltaic absorbers, halide perovskites are real prospects, but they lack stability and recombinations limit their efficiency. To fix current shortcomings, I propose to widen the range of investigations. In particular, I am seeking to advance the optoelectronic properties of graphene sheets with metal halide perovskite layers. Lowering recombination rates is a challenge for perovskites that may be addressed by transferring charge carriers from the perovskite to the graphene layers. Various synthetic strategies would be methodically elucidated in the aim of enhancing the efficiency of both perovskite solar cells, transistors and light emitters. This proposal may have remarkable outcomes for energy efficient technologies.


In the past five years, metal halide perovskites have drawn huge excitement from the photovoltaics research community because of their high power conversion efficiency. The 2D-structured perovskite demonstrate promising stability properties [Wang 2017], large carrier mobility [Brenner 2016], strong light absorption, superb photo- and electro-luminescence [Stranks 2017, Dou 2015] and strong quantum confinement effects. These properties have enabled LEDs [Yuan 2016, Friend 2017], lasers and photodetectors [Li 2017] powered with perovskite materials.

In the realm of 2D materials, the use of stacked layers, forming so-called van der Waals heterostructures, has empowered in-band structure engineering [Geim 2013] to create tunnel junctions with unparalleled performances. To date, the best materials for building such heterostructures have been graphene, frequently achieving superior performances for surface science, endowing it with favourable electronic, optical, thermal and mechanical properties [Novoselov 2016].

This proposal aims to create heterostructures based on layered 2D perovskite with graphene sheets. This challenge has been addressed with thin film perovskites and has shown excellent FET performance but poor stability in ambient conditions [Cheng 2015]. Here, two dimensional perovskite may be a promising alternative for enhanced stability and attractive electronic and optical properties [Yang 2016]. Moreover, 2D perovskite can be bound to graphene sheets by various techniques as I present in this research proposal.

Project Details

Scope of the project

I propose to investigate various bonding strategies between 2D perovskite layers and graphene sheets to create stacks and assess their properties. Various synthetic methods as well various bond types may be explored: van der Waals, ionic, covalent and hydrogen bonds [Liu 2012].

The question to be addressed is: to what extend does the nature of the bonding between graphene sheets and 2D perovskite layers influence the optoelectronic properties of the stack?


Graphene and perovskite space groups are different therefore stacking the two materials in a heterostructure will induce lattice strain. However, interface modelling studies [Guo 2017] showed that this kind of heterostructure can be stable.
Van der Waals heterostructures will be manufactured via physical deposition under vacuum. Ionic heterostructures will be developed by doping the graphene structure to make it more or less electroattractive. Hydrogen and covalent bonding will involve structure modification, enabled by the 2D perovskite where the crystallographic structure is not influenced by the cation [Weidman 2016]. Thus both the excitonic energy and the bandgap will stay unchanged. However, it may require the modification of graphene via grafting organic cations molecules on its surface.

Van der Waals heterostructures of graphene and perovskite have been theorised [Guo 2017]. It has been predicted that the electronic structure of both 2D halide perovskite and graphene will be preserved after stacking. To my knowledge, there has never been any attempt to carry out the practical experiment.

With grafting methods, I anticipate that recombination rates may be reduced by transferring charge carriers from the perovskite to the graphene layers. This may lead to higher efficiency solar cells. Finally, I anticipate that the different binding approaches will tune the position of the Fermi level by modifying the interface between the two layers. The orbital overlapping may be altered, as well as the electronic configuration at the interface.


Characterisation of the manufactured materials would be carried out with X-Ray Diffraction techniques, as well as observations with Scanning and Transmission Electron Microscopy. The optoelectronic properties would be explored via photoluminescence spectroscopy; charge transport with microwave conductivity measurements; surface states with transient absorption spectroscopy; and the band structure with Compton and Raman scattering.


The first year of the PhD would be dedicated to exploring the property range through methodical testing, with the aim of identifying the most efficient process for keeping the heterojunction stable. During the two following years, I would focus on the optoelectronic properties of the manufactured materials. The various bonding investigations would be carried out simultaneously in order to be able to compare results.

Objectives and Final Outcomes

I believe that combining two outstanding materials may lead to unparalleled results. Covalent bonding will affect the graphene conjugation system, therefore compromising some of its properties. I suppose that non-covalent interactions may preserve all of its electronic properties. The afforded assembly with two-dimensional perovskite structures may certainly give rise to a new class of layered semiconductors, as the assembly with the bulk perovskite counterparts already gave astonishing performances [Cheng 2015].

I am convinced that the perovskite is a class of materials that has the potential to impact the future of modern civilisation. My investigations may lead to meaningful industrial opportunities with great impact on energy saving and energy collection.



You may have realised that there were two shortcomings in this proposal. How would you manage to fix it?

The cover picture is from Charles van Meter on Flickr.

Investigating Carbon Nanotubes

When we think of the long term future for perovskite solar cells, there are basically two strategies. The first is to enhance their efficiency by making tandem solar cells. The second is to create flexible devices with roll to roll technologies, as industrials currently do with organic solar cells. This latter strategy has an enormous potential, but shows an important issue: to be scalable, we absolutely need enhanced stability, especially at the interface between the electrodes and the active layer. One way to enhance this stablity is to use carbon nanotubes (CNTs).

To present CNTs for solar cell applications, let me introduce Dr Severin N. Habisreutinger (@sevnhabis) who, during and after his Ph.D. at the University of Oxford, worked extensively with carbon nanotubes for perovskite applications. He has now ben awarded the Director’s fellowship at the National Renewable Energy Laboratory (NREL) in Golden, Colorado, for investigating the nature of CNT-perovskite interface.

About CNTs

To imagine a (single-wall) carbon nanotube think of a sheet of paper being rolled up such that it forms a tube. If you replace the sheet of paper with a monoatomic sheet of carbon atoms (known as graphene) you get a carbon nanotube. Commonly the diameter is around one nanometer (4-5 orders of magnitude smaller than the diameter of a single strand of hair). CNTs exhibit several very interesting properties. On one hand, their mechanical strength make it one of the strongest materials we know. In addition they are chemically quite resilient, which means they do not readily form strong chemical bonds with other materials. This is important, in particular, at interfaces with very reactive materials. And finally, the property which makes them particularly interesting for photovoltaic application is their extraordinary charge-transport characteristics, above all the high charge carrier mobility. This means that a charge carrier can travel a long distance through a CNT before it is lost when it recombines with a charge carrier of the opposite polarity. The basic idea how to exploit this particular property is to use CNTs at the interface with an absorber material in which light generates charge carriers. These photogenerated carriers can then be rapidly transferred away from the interface through the CNTs which act like high-speed channels for those charges moving them away from a region, namely the illuminated absorber, in which they can readily recombine and would thus be lost.

A challenge for using CNTs is the fact that they have the tendency to form bundles, which makes it more difficult to process them out of solution and it negates some of their excellent electronic properties. A strategy which successfully overcomes this challenge is wrapping individual CNTs with conjugated polymers, known from organic electronics.

With this approach, CNT thin films can be used in photovoltaic applications, such as perovskite solar cells, for example as hole-selective layers. As opposed to conventional solar cells, charge separation in a perovskite device does not rely on an internal electric field, instead charge selective contacts introduce a driving force for electrons on one side, and holes on the other side.

For CNTs, the mechanism leading to the collection of holes is a good alignment of the valence band of the CNTs and the HOMO of the perovskite active layer. Another aspect is their ability to block the electrons. The combination of these characteristics makes them a good hole-transport layer. Severin explains:

Actually, the exact underlying mechanism is still a bit contested, but we know that when CNTs are in contact with ambient oxygen, they become more p-type therefore there is a higher concentration of holes available in the carbon nanotubes. That makes them an interesting hole-transporting material which is the main aspect most recent publications are currently looking at.

Transport properties

However, in some conditions, he explains, PCBM can be used to dope the material and make it more n-type:

I think there is a lot of potential in using CNTs both as anode and cathode with the active layer sandwiched in between. Looking forward, I would imagine that quite a few researchers will look into using CNTs in such a capacity, but for now we still need to answer a range of questions.

Specifically for the use in perovskite devices, we would like to know: what exactly happens at the interface in terms of charge transfer kinetics? Do the metallic-type CNTs have a positive or a negative effect? What about hybrid layers comprising CNTs? We expect that the answers will lead to interesting new concepts and ideally to more stable and more efficient contacts.

Large-scale use of CNTs in this capacity still needs a lot of work to be done to better understand the interface interactions. Guess what? That is exactly what Severin is looking at. Therefore we expect a deeper understanding leading to higher quality charge selective layers and more stable solar cells.

Sci-Fi ideas

Our discussion evolved to more futuristic ideas, such as direct inclusion of CNTs into the perovskite absorber. The basic idea is to extract photogenerated charges even more directly thus further reducing recombination losses. At the moment, it seems hard to tell whether this would be beneficial since the charge transport in perovskite is already remarkably good. But we imagined grafting carbon nanotubes on perovskite crystals to avoid surface recombination. In fact, this concept of a heterojunction is essential for organic solar cells because the electron diffusion length is limited, and one needs to provide a direct conducting path between the active layer and the CNT electrode. As mentioned above, this is less of a concern for the perovskite bulk. However, it could also be an approach to avoid recombination at grain boundaries, if CNTs were to penetrate a perovskite film. There is one interesting study, Severin mentioned, by Prof Horváth and co-workers from EPFL who tried some interesting in that direction. In fact, they grew a perovskite single crystal around carbon nanotubes, achieving a vertical protogenic inclusion of CNTs into a MAPbBr₃ single-crystal.

We also talked about using graphene layers:

In general, if you have a pristine graphene layer, it will have better charge transport characteristics than a CNT film. There is a common issue with the charge transfer between individual carbon nanotubes within a conductive film. The inter-tube junctions create obstacles for the charge-carriers often due to the formation of Schottky barriers, limiting the transport of charges on a macro-scale. That is not present in pristine graphene. However, there is still a big challenge for graphene: fabricating it large scale. One interesting way to implement graphene nonetheless is to make a hybrid layer of graphene flakes and CNTs. This way we could use the excellent electrode properties of graphene in combination with the charge selectivity of CNTs.

Thus, we might expect synergistic effects from the combination of both graphene and CNTs. That may be the kind of idea to explore in the future, especially combined with other nanostructured materials.

 The cover photo is from NASA Goddard Space Flight Center on Flickr.

Understanding degradation of perovskites

Degradation of perovskite has been a well-developed topic on this blog. Actually, it represents one of the greatest challenges to address. Scientists try to tackle the degradation issues from many approach angles. Here I showcase two articles I’ve read in the past months about how the lattice evolves and how it leads to cell degradation: “Effects of Small Polar Molecules on Degradation Processes of Perovskite Solar Cells”, from Prof. Chun-Sing Lee et al. and “Study on degradation mechanism of perovskite solar cell and their recovering effect by introducing CH₃NH₃I layers” from Prof Masanori Ozaki et al. I find important links between those two articles, that is actually what I wanted to present.

As we all know, degradation of perovskite materials can be caused by humidity, oxygen, light, or heat. After being exposed to humidity, the perovskite’s colour changes, making it a perfect moisture sensitive material. In this very interesting article, Prof. Eugene A. Katz and co-workers found that the decomposition of perovskite is induced by illumination in the presence of water. In fact, concentrated sunlight experiments have shown that light triggers degradation and the mechanism depends highly on the temperature and the composition of the materials.

Water has been considered as the main cause of the instability of the perovskite, causing hydrolysis reactions that degrade the solar cells. Prof. Aron Walsh and co-workers have shown in 2014 that upon illumination and heating, hydrated MAPI (methylammonium lead triiodide) perovskite phases, such as MAPbI₃.H2O and MA₄PbI₆.2H₂O are formed and are followed by the release of a gas phase, composed of MAI. Those hydrated phases present no optoelectronic properties, therefore, this effect is considered responsible for the degradation along which the cell proceeds.

The release of methylammonium ignites the degradation

Prof. Nam-Gyu Park and co-workers reported last year that mobile MA⁺ ions appear in the perovskite film during degradation. Here Prof. Lee’s article is of particular importance. It appears that polar molecules, such as MA⁺ or H₂O are closely related to the degradation processes. Knowing that there are mobile MA⁺ ions in the perovskite film during degradation, the authors investigated the degradation processes of MAPI perovskite solar cells with different PbI₂/MAI ratios. They show with X-ray diffraction that the perovskite lattice is expanded by incorporation of water molecules. This mechanism is followed by a crystal breakage, in which MA⁺ and I⁻ ions are released. Those ions diffuse to form PbI₂ crystals.

The formed lead iodide crystals do not present electronic properties thus they are blocking the conducting channels. Hence the ion diffusion coefficient is lowered with the formation of those crystals. Later on, the scientists have demonstrated that the adsorption of water leads to a decrease in the Jsc along with a bulk degradation that accompanies an ion diffusion process. It becomes important when putting in relation with Prof. Ozaki’s article.

Effects of MAI in recovering the performances of degraded structures

In this interesting article, MAPI perovskite materials are evaluated before and after degradation due to moisture and sunlight. It has been reported that during degradation, the crystalline structure changed with peaks assigned to planes of hexagonal PbI₂ and orthorhombic I₂. Simultaneously, the colour of the perovskite changed, indicating its degradation. Because they observed such a phenomenon, Ozaki et al. tried to add MAI layers on top of the perovskite. They have shown that with this layer addition, the MAPI solar-cell performance recovered. The hypothesis developed is that PbI₂ formed with the degradation allows recrystallization of perovskite after introducing MAI layers.

Thus, with the assumption that evaporated MAI cannot be recycled, the authors tried to add it directly to the surface. That allows the cell to be “refurbished”. This article shows an example of what can be done with MAI to reverse the degradation process. This could have many derived applications in the future.

Those two examples show how scientists have been able to understand, in a very basic way, how degradation happens. These understandings have led them to ideas of better, more stable and more efficient cells. That was to say that the perovskite field is rapidly evolving and we will rapidly have a better understanding of the degradation processes that will allow us to develop more stable and more efficient solar cells.

The cover photo has been taken from Swansea University on Flickr

Defect tolerance within halide perovskite

The properties of crystalline semiconductors, and particularly those giving rise to the photoelectric effect, are mainly due to the presence of defects inside a crystalline lattice. Two weeks ago, at the annual Perovskite Solar Cells and Optoelectronics Conference (#PSCO17) in Oxford, one of the main topic developed was the exceptional defect tolerance found in the perovskites and the doping allowed by it. I contacted Prof. Aron Walsh, from the Department of Materials at Imperial College London (United Kingdom) that published recently a an interesting commentary in Nature Materials: ‘Instilling defect tolerance in new compounds’ and presented new results at the conference.

The role of defects in crystalline semiconductors

There are different type of defects in crystalline semiconductors: interruptions (crystallographic defects) or foreign atoms in the lattice (impurities). Those interruptions can be present in the form of point defects, including atomic vacancies, interstitials or anti-site substitutions, and higher dimensional defects like dislocations, grain boundaries or precipitates. Defects are formed in the lattice because there are thermodynamically favourable: their equilibrium concentration depends highly on their formation energy.

All of these defects will have different impact on the electronic and optoelectronic structures. Defects with energy levels inside the bandgap will induce electron or hole trapping. In other cases, defects can imply electron or hole emission: one hole and electron can be captured by a recombination centre. To sum it up, defects in a crystalline lattice produce recombination or emission centres that are at the heart of the optoelectronic properties of a lattice.

Due to these defects, semiconductors create unintentional n- or p-type doping, leading to the formation of p-n junctions. Also, from Fermi level pinning by defect states, unfavourable band alignment can be created that limit the voltage of solar cells.

Representation of possible implication of defect states within the bandgap (a-e), scattering of a charge array (f) and possible band diagram deformation (h,i) in comparison with the ideal p-n junction band diagram (g)

Defect tolerance properties within perovskites

Perovskites are also crystalline semiconductors. For this reason, defects exist and are active within the lattice. Their implications are not negligible and contribute significantly to many properties like carrier recombination. Also, in some cases, the presence of defects seems to have no major implication on recombination that leads to astonishing performance of perovskite solar cells. Prof. Aron Walsh explains this defect tolerance:

Practically speaking, in halide perovskite we expect many point defects to be present, but they do not destroy the performance of solar cells. In optoelectronic devices, it is essential to avoid non-radiative processes that convert electronic to thermal energy via defects. These events lower efficiency and can trigger device breakdown. It turns out that many of the low energy point defects in halide perovskite are “inert”, which is one of the reasons for the materials success. In other technologies such as kesterite (Cu₂ZnSnS₄) solar cells, non-radiative recombination is high, limiting the solar cell efficiencies to below 15%. Such defect intolerance is common in many materials that have been studied for solar cells (e.g. SnS, Cu₂O, FeS₂), which is why the unusual behaviour of halide perovskites is so important to understand.

I think that understanding the reason leading to such properties will lead to ways to enhance not only the properties of perovskites, but also the properties of other semiconductors. That is one of the reasons why research with perovskite is of high importance.

There are other properties that are in relation with the presence of defects. The mobile ions issue, which was previously discussed on this blog, is said to be due to the defect properties within perovskite structures. The presence of halogen interstitial or vacancy within the structure plays a major role: it has been observed that under light exposure, iodide and bromide species migrate along the cell, creating iodide- and bromide-rich domains, that act as recombination centres. Many scientist think that improving the crystal quality would lower the separation in halogen-rich domains. Personally, I think that a better understanding of the defect tolerance will allow us to develop better materials, that will not present hysteresis cycles lowering the performance of the cell.

Doping the perovskites

In typical crystalline semiconductors, defects are very important for the reason they allow the doping of the structure. Doping means that a given material can be tuned to be a better electron- or hole-acceptor, due to the presence of charged defects. Therefore, the question arises as to whether the perovskite materials can be tuned or not. That is a real question. Aron has an enticing opinion on it:

In principle, doping of perovskite is simple: partial substitution of Pb²⁺ by a metal with a higher charge could generate excess electrons, while replacing it by a singly charged metal could generate excess holes. The challenge is achieving this without forming other defects that compensate the charge.

He explains the process in a clear and precise manner in his latest publication, even which there are now limitations. Working with these limitations, and trying to overcome these are a challenge most of the scientists in the perovskite community are working on.

Going further with p-n junctions

If one manages to dope perovskite with tangible results, these could lead to the creation of p-n junction and therefore to field-effect transistors made with perovskite. The perspective of Prof. Walsh is very interesting at this point:

It will be possible once we would have understood how to control defect concentrations and transport. The current generation of perovskite solar cells are based on a p-i-n architecture, where the perovskite layer is intrinsic (i) with low carrier concentrations. Alternative p-n junctions would enable lower cost perovskite devices with fewer layers of materials, and potentially higher efficiency comparable to the best silicon solar cells. If realised, this would be a major breakthrough in the field.

Thus, the challenge to address is of great importance. I am sure we will understand in the year ahead the main reasons why perovskite are defect-tolerant, and will find solutions to create p-n junctions that will improve both the quality, the performance of solar cells, but also will allow us to create new electronic devices based on perovskites that will revolutionise the semiconductor industry.

The cover photo and the figure have been drawn by James M. Ball and Prof. Annamaria Petrozza in Nature.

Are DSSC still growing ?

First reported in 1991 by Prof. Michael Graetzel from EPFL, Lausanne (Switzerland), Dye sensitised Solar Cells (DSSC) have known important developments in the late 1990s ans early 2000s. Then the field began to falter, the community having massively preferred turning to organic and then Perovskite solar cells. Here I present how DSSCs are working and with the example of an article from Anders Hagfeldt and co-workers published in Nature last June, show that the topic is still evolving.

How do DSSCs work?

The behavior of DSSC is mostly an electrochemical behaviour. The current generation is based on the absorption of light by a dye molecule. Just after the absorption of photons, charge carriers move to a mesoporous semiconductor, generally TiO2 nanoparticles, where electrons and holes are separated and collected at hole-conducting or electron-conducting electrodes. A good engineering of the energy levels is necessary, to confine recombinations and enhance the photoelectric effect.

scheme of DSSC
Architecture of a DSSC

In original DSSCs, the system needs an electrolyte solution, typically Iodide/iodine, to enhance the conductivity to the cathode. Nowadays, most of the DSSCs are composed of a solid electrolyte.

energy diagram
Energy diagram of a DSSC.

Perovskite solar cells have been, at first, in 2009 used in dye sensitzed solar cells, as the dye molecule. But rapidly it has been shown that the dye molecule could be used alone, thanks to its exceptional semiconducting properties. For this reason, all the research that had been carried out on solid state DSSC (ssDSC) had been transferred to the field of perovskite solar cells, allowing its unprecedented growth.

The latest evolutions of DSSCs have been focused on the stability and the efficiency of solar cells, particularly in indoor conditions. In 2005, Fukuzumi et al. reported that copper-based electrolytes worked well in indoor (low-light) conditions. Later on, in 2012, Cao and co-workers have shown that ruthenium-based dyes used with iodide-based electrolytes perform well in ambient conditions. That is what I want to share.

What efforts are done to improve their efficiency?

Nowadays, only a few articles each year are dealing with the field of DSSCs, because it is admitted in the community that the field is mature enough. Most of the research is carried out by Prof. Graetzel or Prof. Nazeeruddin teams at EPFL in Switzerland. That is where Marina Freitag carried out research  reported in his Nature Article: “Dye-sensitised solar cells for efficient power generation under ambient lighting”.

The authors show that a proper design of DSSCs can result in outstanding results in term of power onversion efficiency, and particularly in indoors applications. They introduce a DSSC design that uses a copper organic complex as a redox shuttle, with D35 and XY1 sensitisers. These materials are based on the one previously known to perform well in indoor applications.

In the article, many reasons are discussed to explain the good performance of the produced cell. The most important, to me, is that the good engineering of the cell is responsible of its behaviour. The binding of the dye to the TiO₂ has been well controlled, as well as the position of the arylamine donor group, tho that the energy loss is minimized during the electronic process. The great matching of the potentials are of utter importance for the application with DSSCs, as well as the good choice of sensitizers.


This example shows that DSSCs developments are still going towards more stable and more efficient cells, particularly in indoor applications. I think that the good engineering of materials must be done in other solar cells applications, from Organic Solar Cells to Perovskite, that also present tremendous performance in indoor conditions.

The cover photo is from QITS2012 on Flickr. The energy diagram is from MD McGehee et al., the DSSC architecture figure from N Anscombe.

Exploring Excitonic Binding Energy in Perovskites

As usually described on this blog, perovskite solar cells represent a topic that constantly needs optimization. Many studies try to bring new elements for improving their stability or their efficiency. Some month ago, Dr Arman M. Soufiani from UNSW (Sydney) published an interesting article: ‘Impact of microstructure on the electron-hole interaction in lead halide perovskites’ under the supervision of Dr Samuel D. Stranks. I find it appealing for two reasons: first, it shows that parameters previously thought to influence the performance of solar cells are not an optimization factor, leading to fewer constraints on the device’s design. Secondly, the work has been carried out involving many teams all around the world, showing the field is in constant evolution.

The genesis of the study

In the 1990s, many scientists used to study the electron-hole interaction in perovskite structure crystals. Since the advent of lead halide perovskites for optoelectronic applications, some scientists have focused on this. In 2015, Prof. Annamaria Petrozza reported in Nature that fabrication procedures and morphology had an important influence on the electronic behaviour of perovskites. The free carriers and excitonic regimes were said to be influenced by the microstructure.

The morphology and size of the crystals (usually called microstructure) are, in the case of perovskites, highly influenced by the deposition techniques. Polycrystalline thin films grown from solution processes do not produce the same crystals as of single crystal growth or nanocrystals. However, the octahedra (e.g. [PbI₆]⁴⁻ in CH₃NH₃PbI₃) in the crystalline structure is expected not to change significantly with the processing method. This idea was the starting point of Arman’s study:

Actually my PhD was composed of two parts. The first was looking at the fundamental properties of perovskites and in particular their excitonic characteristics whereas, the second was dedicated to characterisation of devices. All of my previous study lead me to the idea that the different processing methods might not influence the excitonic binding energy in these semiconductors. We proved it thanks to an international collaboration. We prepared several samples  both at UNSW and University of Cambridge and performed few preliminary characterizations before we sent the samples to colleagues at Laboratoire National des Champs Magnétiques Intenses (Toulouse) who are specialized in high magnetic field optical measurements.

Before starting the research, I had this confidence that the binding energy would not change noticeablywith the microstructure, because inorganic octahedra remains unchanged. Of course, we expect the grain size to influence the order and the orientation of the organic cation, but I supposed this impact not to be significant.

Observation of the exciton binding energy

In order to properly understand the optoelectronic properties of these materials, the strength of the Coulomb interaction between photo-generated electron in the conduction band and hole in the valence band is of high importance. It is called the exciton binding energy and refers to the fact that once separated the different carriers can recombine or collected in such different ways that can affect the optoelectronic properties. In perovskite semiconductors, the exciton binding energy is small and that contributes to their good device performances; after light absorption, mostly free-carriers are spontaneously generated.

The observation of the exciton binding energy gives insights not only on solar cells performances but also on luminescence applications. Because the binding energy affects the bandgap and the recombination behaviour of charge carriers, light emission following radiative recombination is a topic that deserves to be understood. Dr Samuel D. Stranks works on this in the Cavendish Laboratory (Cambridge, UK):

We are exploring luminescence as a tool not just for solar cells applications but also for LED and lasing applications, so I am looking at developing new optoelectronic materials. At this point in time, hybrid perovskite is one of our key focuses. We are looking at fundamental recombination processes and fundamental sciences in these materials, and using that to learn about how we can improve the devices and the materials, and how we can take them to their limits.

The research Arman achieved shows that the impact of microstructures of perovskite is not significant in the exciton binding energy and charge carrier effective masses

What else could be done?

This result might leave one puzzled. In the research of more and more performant and stable solar cells, knowing that changing a parameter does not affect an electronic property of a material might seem unsettling. With this it lets scientists with no clear idea of what to optimize. In fact it is quite a good result: the processing methods are usually different for manufacturing with the goal of commercialisation and in glovebox fabrication. Whatever the method employed, the described property will not change. The optimization is thus less constrained.

Following this study, there are many improvement paths. If the octahedral (i.e. the inorganic cage which is the main contributor to the electronic states at the band-edge) are not modified, it does not mean the bandgap does not change either, because it depends on a larger scale. As Arman told me:

The bandgap changes when the unit cell parameters changes. When we move from samples with larger grains to smaller crystals, strains are imposed into the crystal structure which changes the unit cell parameters and thus, affects the bandgap. Nevertheless, the binding energy and excitonic reduced mass of the charge-carriers seems to remain unchanged. Relatedly, replacing methylammonium by formamidinum, and keeping the lead iodide cage, the binding energy remains almost similar because you haven’t changed the inorganic octahedra, contrary to the bandgap.

This can be seen by observing the colour of the material. A change in the crystal size due to the cation size is reflected in changes in the bandgap of the perovskite. That was reported in my previous blog post.

Sam Stranks had a slightly different approach with the future studies that could be carried out from this:

I am looking forward to seeing industrial applications of perovskites. Of course we will see products soon, and certainly from OxfordPV. I am particularly excited about how we can re-imagine what PV can do. We started with silicon, but it will soon be possible to process coloured, lightweight, flexible photovoltaics. All of these would reduce the cost.

However, I think there is still quite a feature to adress to push the device at its limits. Long term stability and ion migration are absolutely essential to adress to push them into the market. We work with very good candidates with very good charge transport or good charge properties. We eare looking for a winner, that is part of the job we are trying to do.

It leads us to learn more about the recombination processes in solar cells. No wonder that it will be a topic we will hear about in the coming months…

Contact information:

Dr Arman M. Soufiani, University of New South Wales, Sydney: a.mahboubisoufiani@unsw.edu.au

Dr Samuel D. Stranks, Optoelectronics Group, Cavendish Laboratory, Cambridge University: sds65@cam.ac.uk

Perspective: Is Perovksite Fever Over?

I started this blog two months ago to share my opinion and enthusiasm about third generation photovoltaic technologies. In fact last year I used to ask questions to some of my teachers about perovskite solar cells, and each time, they were very sceptic about it and told me that the topic exploded some years ago, but that it could rapidly fall back. In the field, we call it the Perovskite Fever. I have never agreed with the fact perovskites could easily fall back, telling this was the original aim of this blog. In this perspective, I give my opinion about the past, current and future developments of perovskite solar cells and my vision on how to push it further.

Historical background: the Perovskite Fever

The perovskite technology has received tremendous interest in photovoltaic applications for five years. At the very beginning, this material was tin-based and its semiconducting properties were reported in 1994 by Mitzi et al. Actually, in 2009, Miyazaki et. al were the first to use perovskite materials into solar cells. At the time, they used lead-based perovskites in DSSCs to enhance their properties. Later on, in 2012, Kim et al. employed an all-solid configuration, reporting a 9% efficiency.

Since then, the topic received substantial attention from researchers all around the globe. Hence the performance of perovskite solar cells has been dramatically improved to more than 20% efficiency, boosted by material and interfaces engineering. Today, I consider perovskites as one of the most promising semiconducting material, that will boost the realm of optoelectronics in a very near future. Its low processing cost, abundance and excellent optical properties including a high absorption coefficient, strong photoluminescence, low trap-state density and long carrier diffusion length make it a perfect candidate for applications in solar cells, lasers, light emitting diodes, etc.

The crystalline structure

Before trying to talk about future developments, I think it is very important to understand the crystalline structure of perovskite. The term perovskite originated from the similarity with the CaTiO3 mineral, called Perovskite. In this structure, usually referred as ABX3 perovskite, a divalent metal B (lead in optoelectronics applications) is surrounded by six halogen atoms (iodide or bromide) in an octahedral network, and the cation A (methylammonium, formamidinum or caesium) is either located in the centre of an eight BX6 octahedral network (3D perovskite) or sandwiched between corner-sharing BX6 octahedral layers (2D perovskite). In fact, each layer of BX6 octahedra is only connected to each other by weak van der Waals forces. This structure is very important to understand because all the physical properties are induced by the crystallographic lattice.

picture: different perovskite structures
Different ABX₃ structures. Perovskite cubic(a), tetragonal (b) and orthorhombic (c) structure have PV properties. (d) has no PV property

It has recently been demonstrated that this crystalline structure evolves with the temperature. One must understand the phase dynamics leading to different lattice structure. Below 150K, lead halide perovskites are stable in an orthorhombic phase (Pnma). Behind 150K, the tetragonal phase (I4/mcm) is more stable and at room temperature or higher, the cubic phase is stable (Pm3̅m). Only the cubic phase presents optoelectronic properties. It means that at low temperature, the perovskite cannot absorb or emit light. However, there is no clear barrier: some scientists predicted metal halide perovskites to be metastable: particularly between 200 and 300K, both the tetragonal and the cubic phases can coexist. It is a problem to address: if this material cannot be represented on an equilibrium phase diagram, it means it can only be synthesised with a finite lifetime.

The lattice evolves

The metastability is often associated with a very small energy barrier for the cation rotation. It has been predicted that molecular sub-lattice is dynamically disordered due to the ability for the cation to rotate, but scientists found that Bragg scattering does not probe such a local disorder, even if these vibrations are observed with the vibrational IR and Raman spectra. That lets a bunch of physical and chemical properties to understand.

Ion migration processes are also of utter importance. In fact, there are charged point defects in the bulk that allow the transport of ions, thus ensuring spatial fluctuations of the electrostatic potential of the solar cells. Schottky defects are prevalent in the crystallographic structure. Therefore it has been observed with X-ray diffraction that the ions segregate into crystalline phases under illumination: one iodide rich and one bromide-rich. The accumulation of charge carriers in different zones increases the lattice strain and is a path toward more and more segregation. That “doping” leads to a hysteresis cycle of the IV-curve, that reduces the photovoltaic performance. However, this is a reversible process. That means that in the dark, the iodide- and bromide-rich phases blend one each other. To date, the mobile ion issue is the one to address in priority before commercialisation. A way to find solutions is to focus on the crystallisation process, that is still unknown. Understanding this will surely lead to more reproducible, better performance solar cells, and will mitigate the hysteresis issue, providing greater assets to address the problem of long-term stability. During the crystallisation process, the role of chloride ions remains still unknown. It could be the source of a doping effect that would explain the large carrier diffusion length of perovskites because it is challenging to find chloride ions in the final perovskite. The reduction of ion diffusion is also important: the understanding of doping processes and exciton confinement will allow scientists to understand the dopant states and the impact of halide ion exchange.

The crystalline structure and the lattice vibrations drive directly the degradation of the materials. Even if there have never been any rigorous attempts to model the decomposition path of hybrid perovskites, including chemical reactions, anharmonic lattice vibration and thermal conductivity, many scientists are on their way. I think these are important things to be understood for the sustainability of the perovskites.

The electronic properties

All of the lattice behaviour changes the electronic properties, and particularly the electron-hole recombination. There are three types of possible recombination: non-radiative, radiative, and Auger recombination. Only the non-radiative recombination limits the efficiency of the solar cells, according to the Shockley-Queisser law. Hence the efficiency of perovskite solar cells is important due to the low number of non-radiative processes.

Moreover, a lot of work to understand the electronic properties have been carried out by simulations and theoretical calculation, often based on first principle calculations. These simulations allow scientists to compare different models and therefore to identify the crystalline structure based on measurable deductions. It has been shown that the valence and conduction band structure are quite sensitive to changes in the crystal structure. The band are also perturbed by the vibrations of the lattice and electron-phonon coupling, that increase the rate of non-radiative recombinations, thus decreasing the efficiency of the solar cells.

Furthermore, there are defect levels in the bandgap, and working further on finding and reducing these levels can allow scientists to understand the low rate of non-radiative recombination. I am sure that this will have a broad impact in solid state physics.

Another field of deep interest lie at the surfaces, the grain boundaries and the interfaces of perovskites. A lot of work has currently been achieved by many scientists to understand the influence of these parameters in the device performance and to push its limit further in terms of long-term stability.

Moving towards luminescence

Applications for perovskites have started in the solar cells field, but the research realm is moving to other optoelectronic properties.
The first is light emission allowed by metal halide perovskites. Working in this field is very important because these properties are due to radiative recombination. Enhancing the radiative recombination would enhance both the absorption and emission properties of the cells. That may seem counter-intuitive, but the perovskite solar cells that would present the best performances will also emit light when working.

Thus the optical properties, including tunable emission are important within perovskite solar cells. As we just saw, the material presents excellent exciton and carrier properties, leading to a high performance in the solar cell research area. The same properties make it an outstanding light emitter that can be used in LEDs and lasers. If we compare it with conventional III-V and II-VI semiconductors, an attractive feature of perovskite is the ability one can have to tune the emission wavelength, throughout the whole visible range. This can be achieved by controlling the stoichiometry, by substituting halide elements, for example, chloride or iodide. Replacing lead by other metal ions also enable a bandgap tuning, that can be achieved in the near infrared or ultraviolet.

image: tunable emission
Tunable emission of perovskite: (a) PL emission bands, (b) bandgap VS halide composition of the perovskite, (c) images of perovskite under UV light

Electronic properties of luminescence

Perovskites are excellent light emitters. Their high quantum efficiency is the result of a clear bandgap with a low rate of charge-trapping states, that promotes the exciton radiative recombination efficiency. Some scientists have discussed how the unique electronic band structure of these perovskites contribute to the formation of defect-tolerant CsPbX3 nanocrystals. One of the future challenges in modifying optical properties is by doping the crystals with Mn2+ or Bi3+, creating heterostructures with metal nanoparticles.

Luminescence has been mostly observed in perovskite nanocrystals. In fact, in that crystal the quantum confinement effect — described by a crystal size that is too small to be comparable to the Bohr radius of exciton — allows the energy levels to be discrete (instead of continuous as in solids). Thus, the optical absorption and emission properties can easily be tuned by changing the size of the semiconductor, i.e. the nanocrystal. Hence, when reducing the size of the nanocrystal, scientists have observed a blue shift in the bandgap. If most of these effects have been observed in all-inorganic CsPbBr3 perovskite nanocrystals, the improvement of this structure and use in organo-metal halide perovskites will surely lead to enhancement of the absorption properties of perovskite solar cells.

Main applications in light emission

There are two main applications for luminescence with perovskite: optical lasing and light emitting diodes. Due to the high absorption coefficient and strong photoluminescence properties, optical lasing is possible. In fact, when a high gain material is put in an optical cavity, one can expect lasing. Exploration of amplified spontaneous emission in perovskite allowed scientists to design some perovskite-based lasers. We understood in this perspective that the high crystallinity of perovskites was the base of many properties. Here, nanocrystals can act as a Fabry-Pérot cavity, allowing lasing in the materials. This will lead to applications in nanoscale optoelectronics, with the ability to tune the emission wavelength to achieve lasing in the full visible spectrum.

Light emitting diodes applications make perovskite one of the most promising material. The fact it is low cost, easy to prepare, with high performance and a high colour tunability, with very precise emission rays make it a technology we will hear about in the coming years. The main issue is overcoming the stability issue. I am sure that a combined work from two different communities (the one that works on solar cells and the one working with LEDs) will be the key towards these applications.

Forthcoming challenges

I do not think that the compound design should be application-driven. Perovskite is truly a multifunctional semiconductor, and the understanding of its properties for particular applications should drive the design for the whole applications. The understanding of the functions of each component, such as organic cations, metal halide octahedra network, and effect of each on the entire crystal structure in order to formulate better compound for each application is of paramount importance.

One of the biggest challenges for metal halide perovskite is their stability. Their vulnerability surrounding environments, such as moisture and oxygen or polar molecules, affect dramatically the performance of the devices. This casts a shadow towards their practical applications and commercialisation. Thus we need to develop a method to stabilise the material, both by chemical approaches, (capping agents, for example, stabilizers) or by physical methods such as device encapsulation.

Perovskite may have reached their high development peak, but we saw in this perspective there is still many years of research for chemist and physicist to improve the optical and electronic properties of perovskite by tailoring the crystallographic sciences and designing new devices. I think we are at the beginning of a new Perovskite era when scientists will not try to improve their devices based on existing knowledge of organic and inorganic semiconductors but will try to deeper understanding. And there is plenty of room for this!

I am deeply convinced that the world rapidly needs stable, efficient and low-cost photovoltaic solutions. Tackling climate change is a matter of days. A better understanding of Perovskite emission properties and their degradation mechanism will lead to a better understanding of this incredible material, that makes the Perovskite Fever not over yet. I am sure that we will understand all of their physical properties so that Perovskites will power what will become the major source of energy within 10 years, and I have a glimmer of hope that Perovskite will lead us to a sustainable future.

The cover photo is taken from NREL on flickr. The Perovskite structure illustration has been drawn by Walsh et al., the luminescence illustration by Pathak et al.