nphoton.2014.134

8/18/2019 nphoton.2014.134

1/9

506 NATURE PHOTONICS | VOL 8 | JULY 2014 | www.nature.com/naturephotonics

R ecent substantial reductions in the manu acturing costs omainstream silicon solar cell technology assure the uturelarge-scale use o photovoltaics, with a recent orecast antici-pating photovoltaics will contribute nearly a third o new electric-ity generation capacity worldwide between now and 20301. As inmicroelectronics, silicon has a combination o strengths that hasmade it difficult to displace as the avoured photovoltaic material.Opportunities exist or technologies that promise either signi-cantly higher energy conversion efficiencies or signicantly lowerprocessing costs. A new generation o mixed organic–inorganic hal-ide perovskites offers tantalizing prospects on both ronts2–6.

Some key attributes o these perovskites include ease o abrica-tion, strong solar absorption and low non-radiative carrier recom-bination rates or such simply prepared materials, plus the ability tocapitalize on over 20 years o development o related dye-sensitizedand organic photovoltaic cells. A reasonably high carrier mobil-ity is an important property or some cell architectures, as is the

range o properties accessible by orming mixed compounds withina compatible materials system. One negative aspect o perovskitesis the act that lead has been a major constituent o all highly per-

orming perovskite cells to date, raising toxicity issues during deviceabrication, deployment and disposal. Also, they generally undergo

degradation (sometimes quite rapid) on exposure to moisture andultraviolet radiation.

Perovskites are materials described by the ormula ABX3, whereX is an anion and A and B are cations o different sizes (A beinglarger than B). Te crystal structure o perovskites is depicted inFig. 1a. Teir crystallographic stability and probable structure canbe deduced by considering a tolerance actort and an octahedral

actorμ (re . 7); here,t is dened as the ratio o the distance A−X tothe distance B−X in an idealized solid-sphere model (t = (RA + RX)/

{√2(RB + RX)}, whereRA, RB andRX are the ionic radii o the corre-sponding ions) andμ is dened as the ratioRB/RX. For halide perovs-kites (X = F ,Cl ,Br , I)7, generally 0.81

8/18/2019 nphoton.2014.134

2/9

NATURE PHOTONICS | VOL 8 | JULY 2014 | www.nature.com/naturephotonics 507

but they dissolved in the electrolyte, resulting in a rapid degradationo per ormance21.Tis stimulated the replacement o problematic electrolytes by a

solid-state H M22,23. Park, Grätzel and colleagues22 introduced a spiro-MeO AD (2,2’,7,7’-tetrakis(N,N-di-p-methoxyphenylamine)-9,9’-spirobiuorene) H M, which was developed or organic LEDs24 but was also ound to be effective in solid-state dye cells25. Whendissolved in an organic solvent, spiro-MeO AD penetrates nanopo-rous iO2, leaving only solute molecules afer solvent evaporation.Spiro-MeO AD not only improved the stability, as expected, it alsoboosted the reported efficiency to 9.7% (re . 22). Te cell structureis encompassed by the more general device o Fig. 2a i the optionalcontinuous perovskite layer is removed, leaving only scaffoldinginltrated by perovskite (and subsequently H M).

Almost simultaneously (mid-2012), Snaith and co-workers23

also reported success with spiro-MeO AD along with our addi-tional developments that split the eld wide open. One o thesedevelopments was the use o the mixed-halide CH3NH3PbI3−x Clx ,which exhibited better stability and carrier transport than itspure iodide equivalent23,26. A second involved going beyond ear-lier nanoparticle structures by coating nanoporous iO2 sur aceswith a thin perovskite layer and thereby orming extremely thinabsorber (E A) cells. A third was replacing conducting nanopo-rous iO2 by a similar but non-conducting Al2O3 network. Tisimproved the open-circuit voltage (V oc), boosting the reportedefficiency to 10.9%; it also demonstrated that perovskites have abroader potential than just being used as sensitizers, as they areable to transport both electrons and holes between cell terminals.

Te ourth development exploited such ambipolar transport bydemonstrating simple planar cells with the scaffolding (Fig. 2a)completely eliminated.

A jump to a reported efficiency o 12.0% came rom the com-bined efforts o Seok, Grätzel and colleagues using both optionallayers shown in Fig. 2a, including a solid perovskite capping layeroverlying the scaffolding27 (nanoporous iO2 inltrated by per-ovskite). O the H Ms they investigated (which included spiro-MeO AD), poly-triarylamine proved to be the best. Seok’s group

urther improved the per ormance to realize a reported efficiencyo 12.3% using similar structures and mixed-halide CH3NH3PbI3−x Brx perovskites28. A low Br content (20%)provided a better high-humidity stability. Tis was correlated witha tetragonal to pseudo-cubic structural transition arising rom a

highert actor due to the smaller ionic radius o Br (molecules inposition A exclude ull cubic symmetry 29).Further progress was reported at the European Mate

Research Symposium in May 2013, with two groups repefficiencies above 15%. Grätzel’s group used iO2 scaffolding andtwo-step iodide deposition, which improved the morphol30.Tey also reported the rst independent measurements ociency by an independent accredited test centre, which conan efficiency o 14.1%. (Independent measurements wereto be an essential quality-control measure with other photovtechnologies to prevent inated results inltrating the litera31;inexperience or overenthusiasm ofen result in ‘in-house’ dataoverestimates.) Snaith’s group reported similar results usingdifferent planar cells that did not have scaffolding. Te sim

structure allowed CH3NH3PbI3−x Clx deposition by two-source thermal evaporation, again giving a better morphology 32 and a reportedefficiency o 15.4%.

No efficiency improvements were announced at the 201Materials Research Society (MRS) Meeting, although a specovskite session attracted considerable attention. At the end oSeok’s group achieved an independently conrmed efficie16.2% by using the mixed-halide CH3NH3PbI3−x Brx (10–15% Br)and a poly-triarylamine H M (S. I. Seok, personal commtion). Both optional layers shown in Fig. 2a were included, wthickness ratio o perovskite-inltrated iO2 scaffolding relative tothe continuous perovskite layer being the key to the improveciency (S. I. Seok, personal communication). Tis was increaa conrmed efficiency o 17.9% in early 2014 (S. I. Seok,

communication). An unconrmed efficiency o 19.3% was rat the 2014 Spring MRS Meeting6 (some uncertainty surrounds thireported efficiency due to the absence o supporting in orm

Signicantly, the latter three results were obtained using dstructures that span all the possibilities inherent in Fig. 2aor both optional layers, three different mixed-halide perovand two different H Ms were used (although some ambigremain30,33). Fabrication simplicity combined with similaritiesdye-sensitized and organic photovoltaics has resulted in aincrease in the number o researchers working in this eld.2013, only seven journal papers had been published that disphotovoltaic devices based on halide perovskites, whereas byo 2013, relevant publications were appearing at the rate o smonth. Tis spurt urther increased the diversity o approachgave creditable per ormances.

A

B

X

0.75

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

0.80 0.85 0.90 0.95 1.00 1.05

Tolerance factor, t

O c t a

h e

d r a

l f a c t o r ,

MAPbl 3MASnl 3EAPbl 3EASnl 3

MAPbBr 3MASnBr 3EAPbBr 3EASnBr 3

MAPbCl 3MASnCl 3EAPbCl 3EASnCl 3

a b

Figure 1 | Perovskite crystal structure and associated tolerance and octahedral factors. a , Cubic perovskite crystal structure. For photovoltaicallyinteresting perovskites, the large cation A is usually the methylammonium ion (CH 3NH3), the small cation B is Pb and the anion X is a halogen ion(usually I, but both Cl and Br are also of interest). For CH 3NH 3PbI3, the cubic phase forms only at temperatures above 330 K due to a low t factor (0.83).b, Calculated t and μ factors for 12 halide perovskites. The corresponding formamidinium (NH 2CH=NH 2) based halides are expected to have intermediatevalues between those of the methylammonium (MA) and ethylammonium (EA; CH 3CH2NH 3) compounds shown.

REVIEW ARTICLENATURE PHOTONICS DOI: 10.1038/NPHOTON.2014.134

8/18/2019 nphoton.2014.134

3/9


Key recent results include demonstrations o an additional depo-sition process involving PbI2 deposition rom solution within situ conversion to perovskite by a vapour-phase CH3NH3I reaction34.Promising results have also been reported with perovskites based onorganic cations other than CH3NH3+. Cations with larger ionic radii,specically ethylammonium (CH3CH2NH3+)10 and ormamidinium(NH2CH=NH2+)11–13, increase thet actor and push structures towardsthe symmetrical cubic phase. Different proportions o organic cati-ons, inorganic cations (Pb , Sn) and halide anions (I , Br , Cl) can beincorporated in mixed perovskites, allowing their properties to bene-tuned; however, the use o mixed halides has dominated.

Recent work has also demonstrated the use o new H Ms andelectron transport media (E Ms). Effective E Ms have been reported

in which the standard uorine-doped tin oxide (F O)/compactiO2 combination is replaced by indium tin oxide as a transparentconducting oxide combined with a thin (25 nm) ZnO-nanoparticlelayer35; this gave a reported efficiency o 15.7% or planar cells onglass. Low-temperature processing also gave a creditable per or-mance on exible polyethylene terephthalate. Inorganic H Ms suchas CuI36 and CuSCN37 also give reasonable results, as do organic-photovoltaic-derived organics or both E Ms and H Ms, speci-cally a (6,6)-phenyl C61-butyric acid methyl ester E M combinedwith a poly(2,3-dihydrothieno-1,4-dioxin)-poly(styrenesulphonate)H M38–40 (efficiencies up to 12%, re s 38,39).

Such diversication increases potential applications. Flexiblecells5 require low processing temperatures o less than 150 °C, ratherthan 500 °C, which is typical or compact iO2. Graphene nano-

akes in the normally compact iO2 layer allow such processing41

.An efficiency o 15.6% has been reported or a structure that includesall optional layers and Al2O3 scaffolding (0.6% graphene/ iO2 byweight)41, and an efficiency o 15.9% has been achieved by com-bining small iO2 nanoparticles with a titanium diisopropoxidebis(acetylacetonate) binder42. In other work, the non-uni ormityproduced by solution deposition is exploited to produce neutral-colour semi-transparent cells43. Instead o continuous perovskite,small invisible islands are grown, which should be less expensivethan laser approaches or abricating semi-transparent amorphous-silicon cells44.

Enabling attributesStrong optical absorption is the key to the outstanding per ormanceo these perovskite cells, reducing both the required thickness and

the challenges in collecting photogenerated carriers. Absomeasurements (Fig. 3a) are consistent with calculations ing direct-bandgap properties or perovskites o interest14,45. wostrong, spin–orbit split, excitonic absorption thresholds are aent, as in direct-bandgap III–V semiconductors46. However, reverseordering o band-edge states (specically a p-like condband)47 results in splitting in the conduction band, rather thathe valence band. Interestingly, reverse band-edge orderingives a bandgap that increases with increasing temperature given phase48,49, which is the opposite trend to that o tetrahedcoordinated semiconductors.

Te strong excitonic absorption edge also means there ibasis or the common practice o determining bandgaps usi

plots50

. Te absorption edge is determined by a broadened excimpulse response, as is the case or direct III–V semicondwith the unbroadened response described by Elliott’s theo46,51.Te relatively high exciton binding energy compared to thoIII–V semiconductors with a similar bandgap (37–50 meV hareported or iodide in the low-temperature phase52 and 35–75 meV

or the mixed chloride at room temperature49), not only lowers theabsorption threshold, but also increases the strength o the bandgap absorption that generates unbound electron–hole pCorrespondingly, above-bandgap absorption is comparablestronger than that in many direct-bandgap III–V semicondusuch as GaAs, although it is lower than that o some inorgancogenides (see Fig. 3a).

For devices with a continuous perovskite layer or an insu

scaffold (Fig. 2a), transport across the perovskite is importdevice operation. Te exciton binding energy is still sufficientor photogeneration o both excitons and unbound electron

pairs, with thermal dissociation o excitons into ree carriereassociation into excitons) expected. Te respective photocucontributions ideally depend on the coupling between these ptions, as characterized by the intrinsic exciton dissociation tithe time that would prevail i excitons were not mobile or cobe ionized at heterojunctions.

At the low-coupling extreme (dissociation times large comto recombination times), the two populations will be indepeand relative concentrations will be determined by the spectraposition o illumination. For sunlight, ree electron–hole peration is expected to be completely dominant, as primary egeneration will occur only at wavelengths near band edges (

Light

Scaffold/perovskite(optional)

Perovskite(optional)

Au or Ag

Electronenergy

Holeenergy

Glass

FTO

Compact TiO 2

Perovskite

HTM

Au

−8.1

−7.2

−5.5−5.2

−5.1

−2.1−3.9

−4.0

−4.4

Electrons

Holes

Figure 2 | Perovskite cell structure and associated vacuum energy levels. a, General organic–inorganic halide solar cell, which includes two optional layersthat are not essential for high performance; an energy conversion efficiency of over 15% has been reported for devices that have both optional layers, withonly the scaffold layer inltrated by the perovskite (and then by the HTM) and without scaffolding; the structure then corresponds to a simple planar thin-lm cell. b, Vacuum energy levels (in eV) for corresponding materials (CH 3NH3PbI3 perovskite, conducting TiO 2 scaffold).

REVIEW ARTICLE NATURE PHOTONICS DOI: 10.1038/NPHOTON.2014.134

8/18/2019 nphoton.2014.134

4/9


or continuous perovskite layers). o contribute to the photocurrentin this low-coupling limit, excitons would need to dissociate at aninter ace (that with the E M or the H M or both), as in organicphotovoltaic devices.

At the opposite extreme o high coupling (rapid dissociation),ree electron–hole and exciton concentrations will equilibrate with

relative concentrations theoretically determined by a mass actionlaw rather than by the illumination (this law will take the ormdeduced by Combescot53,54, when corrected to account or excitedexciton states and scattering states associated with electron–holeattraction55). Excitons are expected to ow in the same direction aselectrons or holes (Fig. 2b), depending on which has the larger elec-trochemical gradient56. Because electron and hole currents increasein opposite directions (Fig. 2b), these gradients will probably beequal at one point in the device, where the total ux is ree carriers.

No additional exciton dissociation eatures will be required in thiscase. ransport could then be treated as ree carrier with transportand recombination parameters determined by a weighted combi-nation o ree carrier and excitonic values56. For a relatively narrowcoupling range lying between the low and high extremes, interme-diate behaviour is expected54. As well as strengthening absorption,excitons provide an additional pool o carriers that may assist carriertransport. Benets are expected i the li etime o exciton diffusion orrecombination exceeds that o ree carriers54. However, despite rela-tively high exciton binding energies and uncertainties arising romattempting to separate correlated electron–hole pairs into bound andscattering states55, it seems as i exciton concentrations will be lowcompared to both ree-carrier concentrations except in extreme cir-cumstances, a conclusion supported by a recently published work 49.

Another striking attribute o these perovskites is their low non-radiative recombination rates compared to other thin-lm poly-crystalline semiconductors. Tis property mani ests itsel in therelatively small difference betweenV oc o experimental cells andtheir effective bandgap potential3 (Eg/q) or alternatively by theirhigh external radiative efficiency, a parameter deducible romV oc and the spectral response57. Te best perovskite cells have relativelylow values or the differenceEg/q − V oc (about 450 meV, re . 3) andrelatively high calculated external radiative efficiencies (0.058%;I. Al Mansouri, personal communication). Tis makes perovskitesparticularly interesting or highEg cells in tandem cell stacks3, wherethe highV oc values o such cells give rise to substantial efficiencyadvantages. Te diverse range o abrication methods that have beenused, which include low-temperature approaches, urther enhancesthe prospects o these perovskites as a tandem cell component.

In conventional polycrystalline semiconductors, low radiative recombination generally requires large grain sizesgrain-boundary activity and a low density o intragranular dTe present perovskites give narrow X-ray diffraction peakare consistent with both near micrometre grain sizes and a reably low intergranular de ect density 58. Large grain sizes might bexpected rom the high ‘effective’ homologous temperaturesponding to relatively low processing temperatures due to thimputed perovskite melting points (most decompose be oreing29). For metals, a 0.2 increase in the homologous temperincreases the grain size by a actor o ten59. o reduce detrimentalgrain boundary activity, specic processing steps are requirpolycrystalline Si , Cd e and copper–indium–gallium diselen(CIGS) cells. Tese perovskites do not seem to require such pring steps60. Tis may be because o their high effective homolo

temperatures, the tendency or the inorganic planes to align pto substrates61 (which reduces one degree o misorientations),tallographic exibility (allowing more grace ul accommodmisorientations than in more rigid materials), or a combinatiall these actors.

A recent study 62 investigated the activity o intrinsic intergrande ects in CH3NH3PbI3 using density unctional theory. wo typeintrinsic de ects were studied: neutral Schottky de ects (equber o positive and negative vacancies) and Frenkel de ectnumber o vacancies and interstitials o the same ion). Sde ects, such as PbI2 and CH3NH3I vacancies, were ound not generate de ect states having energies within the perovskitegap; this was attributed to the ionic bonding o the perovskiimplies Schottky de ects are unlikely to be effective as non-

recombination centres. Elemental de ects, such as Pb, I and C3NH3 vacancies associated with Frenkel de ects, were ound to low levels near band edges, again reducing the effectivennon-radiative recombination centres. However, these conclumay need to be moderated by the well-known limitations osity unctional theory studies14,45. Another recent study 63 that used asimilar approach reached similar conclusions. Dominant dethe iodide are deduced to be p-type Pb vacancies and n-type methylammonium interstitials, with growth conditions determininnal doping polarity. Tis study also considered anti-site subtions, which, along with Pb interstitials, ormed states near thdle o the bandgap. However, ormation energies were high, that these de ects are expected to orm in low concentrationrelatively low material deposition temperature involved. Altexperimental conrmation is required, these results may help e

R e

l a t i v e p e r m

i t t i v

i t y

Log (frequency, Hz)

05 6 7 8 9 10 11 12 13 14 15

10

20

30

40

50

60

70Real

Imaginary A b s o r p t i o n c o e

ffi c

i e n t

( c m −

1 )

Wavelength (nm)

0200 400 600 800 1,000 1,200 1,400

101

102

103

104

105

106

107GaAs

Si

InP

CIGS

CIS

CdTe

CdS

CH3NH3Pbl3

CH3NH3Pbl3− x Cl x

a b

Figure 3 | Absorption coefficients and relative permittivity. a, Absorption coefficient of CH 3NH 3PbI3 (ref. 40) (these values are about 50% higherthose obtained by Xing et al. 74) and CH 3NH 3PBI3− xCl x (ref. 96) compared to other solar cell materials (various sources). b, Real and imaginary parts of theCH3NH 3PbI3 dielectric constant at 300 K as a function of frequency. Dipolar and ionic components successively disappear as the frequency increases.Low-frequency values (circles) are from ref. 64, mid-frequency (90 GHz) values (squares) are from ref. 65 and optical-frequency value (triangle) isfrom ref. 66. The solid lines are Debye relaxation ts to mid-range values (ref. 65). The dashed lines show possible transitional values in the far infrared(qualitative only).


8/18/2019 nphoton.2014.134

5/9


why high-per ormance perovskite cells can be produced by a diverserange o deposition approaches and a wide variety o cell structures.

Although the dielectric properties o perovskites have not yetbeen shown to be important or good per ormance, they are soextreme compared to those o conventional semiconductors thatit is a possibility. Figure 3b shows a composite o reported dataobtained at 300 K or CH3NH3PbI3. At low requencies, the die-lectric constant is large; it has been reported to be 60.9 over the20 Hz – 1 MHz range64, which is consistent with the low- requency value o 60.2 obtained rom ts at higher requencies65. Tis low-

requency value is appropriate in determining steady-state prop-erties, with static elds consequently varying slowly with position— ve times slower than in silicon under similar electrostatic dis-

turbances. Te high value o the dielectric constant results rom acombination o dipolar, ionic and electronic contributions. As theexcitation requency increases, the permanent dipole associatedwith the organic cation can no longer respond and a new plateau isreached with a dielectric constant o 29.7 (re . 65). Finally, at in ra-red requencies, the ionic component drops out, leaving only theelectronic response. Te dielectric constant drops to 6.5 at optical

requencies66, lower than that o inorganic semiconductors with asimilar bandgap. Te dashed lines in Fig. 3b indicate possible tran-sition values o the constants.

Te other possibility is that CH3NH3PbI3 may possess paraelec-tric or even erroelectric properties at room temperature and above,impacting device per ormance67,68. Te main evidences or this isthe hysteresis observed in resistivity measurements, which is not

seen in related compounds prepared similarly, and the remnantpolarization, which is apparent as a nite voltage output at zero cur-rent29. Te crystallographic point group (4mm) o both the room-temperature and high-temperature iodide phases is consistent with

erroelectric behaviour29. Hysteresis has also been reported orhigh-efficiency devices69,70, although many other explanations arealso possible or this besides erroelectric effects69. Best per ormancehas been realized when hysteresis is minimized (S. I. Seok, personalcommunication). Carrier collection along the boundaries o micro-scopic erroelectric domains has also been suggested68, althoughthis presently lacks experimental support.

Device operationHigh efficiencies have been realized using a diverse range o oper-ating modes, spanning rom generation and collection at sparse

points on the sur aces o titania nanoparticles26 to conventional pla-nar thin-lm operation32.

In the dye-sensitized and E A congurations, perovskitnot required to have good carrier transport, as the inter aciaerties mainly determine the per ormance (Fig. 4a depicts tprocesses involved71). Te desirable processes involve photoecitation in perovskite (1), electron trans er to titania (2) antrans er to the H M (3) (or, equivalently, electron trans er H M to the perovskite). Undesirable processes are recombio photogenerated species (4), back charge trans er at the ino iO2 and the H M with the perovskite (5,6) and between 2 and the H M (7) (this may occur i perovskite is absent inareas — or example, when nanoparticles or voids are prese

high per ormance, processes (4)–(7) must operate on much timescales than charge generation and extraction (1)–(3). resolved transient techniques have proved use ul or studyrelated kinetics72.

For thicker perovskite layers, photogenerated carrier tranbecomes important. Initial data suggested that the mixed hCH3NH3PbI3−x Clx has a distinct advantage over the pure iodideits carrier diffusion length (>1 μm; re . 73) is about ten timesthan that (about 100 nm) o the pure iodide73,74. Subsequent workwith the iodide prepared by the vapour-phase CH3NH3I conversiono PbI2 resulted in good carrier collection in 350-nm-thick iolms, suggesting that the diffusion lengths exceed this thick34;this is supported by recent electron-beam measurements60.

Recently, collection across a lm cross-section has been d

probed using the electron beam induced current (EBIC) tech(Fig. 4b)60,72. In this technique, the electron beam generates a co excited carriers (these are expected to be largely ree elecpairs, some o which may combine to orm excitons i they hassociation times). Because the resolution is limited by the sizcloud, approximately 1,500-nm-thick perovskite layers were which are appreciably thicker than those in optimal planar cel

Te typical line scan shown in Fig. 4b has two interesting eTe rst is the double peak in the EBIC signal within the pekite. Tis indicates the cell’s ambipolar response with carriers ated both near the spiro-MeO AD H M and near the E M 2 plus F O) being collected. Te second interesting eature is tbetween the two peaks, which arises because an appreciable o the generated carriers recombine on their way to their prcontact, due to the sample thickness exceeding the optimal thi

AuHTM Glass

Scan position

1 µm

Electron beam

PerovskiteTiO 2

FTO

E B I C

s i g n a

l

A

HTM

ФHTM

χ P

χ P

χ T χ F

Perovskite

TiO 2

FTO

FTO

a b c

5

6

3

2

41

hf

7

Perovskite

Perovskite

HTM

TiO2

Figure 4 | Electron-transfer processes in nanoparticle and bulk cells together with a bulk energy-band diagram. a, Schematic of electron-transferprocesses in a perovskite nanoparticle or ETA device 71. The thick green and thin red arrows respectively indicate the processes desirable for energyconversion and those associated with losses. hf , photon energy. b, Schematic of EBIC experiment 72. A scanned electron beam generates a cloud ofcarriers, creating a position-dependent current in a short-circuiting load. c, Energy-band diagram deduced from the vacuum energy levels shown in Fig. 2b. χ P, χ T and χ F represent the electron affinities of the perovskite, TiO 2 and FTO layers, respectively, and ΦHTM represents the work function of the HTM layer.(The barrier at the TiO 2 /FTO interface has been a source of discussion in the dye-cell literature 97,98 , because it impedes carrier collection in the directionshown. Some work suggests states generated at the interface under white-light illumination cause a pinning effect, reducing its signicance 98 . Other worksuggests a failure of the Anderson rule, which is not uncommon, and that the barrier has the opposite direction at thermal equilibrium 99 .)


8/18/2019 nphoton.2014.134

6/9


Tese eatures can be understood by constructing the expectedthermal-equilibrium energy-band diagram o the device rom the vacuum energy levels shown in Fig. 2b. At thermal equilibrium, acommon Fermi level prevails throughout the device. A rst-orderestimate o how the energies within the different materials alignwhen brought together is possible by the combined use o theAnderson rule or semiconductor heterojunctions75 and the relatedSchottky–Mott rule or metal–semiconductor inter aces. Both areknown to have severe limitations, but they have the advantage overmore sophisticated approaches76 o being universal in application.In both rules, the vacuum re erence level is assumed to be continu-ous across material inter aces. For semiconductors, this means thatthe relative alignment o conduction band edges across inter acesdepends on the difference o the electron affinities o the two mate-rials (the energy rom the conduction band edge to the vacuumlevel). For metal–semiconductor inter aces, the conduction bandedge will lie above the metal Fermi level by an energy equal to themetal work unction minus the semiconductor electron affinity.

Applying these two rules and treating the H M as a metal givesthe energy-band diagram shown in Fig. 4c or the case when theperovskite layer is reasonably thick. Te associated potential varia-tions across the perovskite are due to the low work unction o thecompact iO2 layer (doped at about 1018–1019 cm−3, re . 77) com-pared to the H M layer. Background perovskite doping levels arelow (they are commonly estimated to be ~1014–1016 cm−3; resistivityand Seebeck coefficient measurements29 suggest even lower values,but they are probably dependent on preparation conditions). Tisresults in the build up o holes near the H M and electrons near theE M (their concentrations equal the product o the effective densityo states in the respective band and the negative exponential o thedifference between the Fermi level and this band edge normalizedby the thermal energy,kT , wherek is the Boltzmann constant andT is the absolute temperature). Te slope o either band edge givesthe local electric eld, which is strongest near the H M and theE M. For thick devices, these high-eld regions may be independ-ent o each other (as shown in Fig. 4c), whereas they merge or thindevices, eventually creating an essentially uni orm eld across the

device. With increasing device voltage, the high-eld regions tendto decouple.Te EBIC outputs are low short-circuit currents, which is con-

sistent with the use o small-signal calculations in which a thermal-equilibrium equivalent circuit is used to model carrier drif anddiffusion78. Te low response when the beam is centred at the H M/perovskite inter ace is probably because o a high electron recombi-nation velocity at this inter ace, corresponding to the back charge-trans er process (6) in Fig. 4a. Tis velocity needs to be lower thanthe drif velocity o electrons to prevent the ormation o a dead layernear this inter ace78. Te response at the iO2–perovskite inter aceis much higher, suggesting that there is no dead layer near this inter-

ace and that carrier generation in the iO2 itsel might contributeto the EBIC signal. Te higher peak value on the H M side o the

device may correspond to a more extended high-eld region on thisside as a result o the band alignment imposed at the H M inter ace.Recent work 79,80 has allowed the accuracy o the Anderson and

Schottky–Mott rules used or deducing Fig. 4c to be assessed.Lindbladet al.79 ound that the alignment between the occupied valence bands o iodide perovskite and nanoporous iO2 was 2.1 eVusing hard-X-ray photoelectron spectroscopy. Using ultravioletphotoemission spectroscopy and inverse photoemission spectros-copy to respectively interrogate occupied and unoccupied states,Schulzet al.80 sel -consistently deduced the work unctions, elec-tron affinities and valence-band maxima. Moreover, all measure-ments were conducted in the same ultrahigh vacuum or compactuncoated iO2 samples, iO2 samples coated with iodide, bro-mide and mixed chloride perovskites, and or the latter samplesoverlaid with different thicknesses o the H M. Te nal picture

obtained is roughly consistent with that shown in Fig. 4c.2 and the perovskite were deduced to have similar electron ties, as shown. Undoped spiro-MeO AD was used in the s80 with the alignments deduced consistent with those in Fig. the doped material.

Commercialization challengesA series o unsuccess ul attempts to commercialize new stechnology in recent years vividly demonstrates the non-trivialenges o commercialization. A key prerequisite or commetion is a compelling market advantage over incumbent technoFor perovskites, such an advantage might be low processinghigh conversion efficiencies through tandem device structuunique products, such as the above-mentioned exible or patransparent modules. Te present reliance on Pb as a key perovcomponent may militate against the adoption o such produconsumer or building integrated applications. Te present rellack o robustness may present an additional barrier or maibulk power applications.

In terms o cost, the closest commercial technology that pkites must compete with is Cd e, the photovoltaic thin-lmnology with the lowest production cost. Fabrication is by vapour-phase deposition onto F O-coated glass with a ‘glas‘module out’ time o 2.5 h (re . 81). Semiconductor costs with e currently costing

8/18/2019 nphoton.2014.134

7/9


are better, as the RoHS limit on Pb in a homogeneous layer is tentimes higher than that or Cd, but they also appear to have no pros-pects or compliance i they contain the optional continuous layershown in Fig. 2a. I the scaffold layer were accepted as a homogenouslayer, the Pb content would be diluted in both nanoparticle-sensi-tized and E A devices. (I 18 2.5-nm-diameter hemispherical iodideperovskite nanoparticles coat a 20-nm-diameter iO2nanosphere21,the scaffolding layer has a porosity o 60% and the pores are lled byspiro-MeO AD, the Pb content will be 0.4–0.5% by weight, whichexceeds the RoHS limit, although not appreciably). Cd and Pb com-pounds have different solubilities5, which does not affect compli-ance with RoHS criteria, but it may affect end-o -li e disposal.

Te low robustness o present perovskite technology to moistair and water vapour means that out o all the present commer-cial photovoltaic technologies it is most closely related to CIGS.Unencapsulated CIGS cells generally degrade during damp heattesting, although this can be controlled to acceptable levels byemploying suitable encapsulation. A recent study investigating the

easibility o encapsulating CIGS cells in exible modules concludedthat ‘breathable’ designs are not viable87,88. Layers that effectivelyrestrict water ingress are required together with internal encapsulantlayers with a high moisture solubility to keep their relative moisturesaturation low. Double glass layers are used in present commercialCIGS modules together with edge sealants, both o which effectivelyprevent moisture penetration89,90. Alternative approaches, such asintegrating moisture barriers like Al2O3 into the device structure,might eventually eliminate the need to use such restricted designs

or perovskite91 and CIGS92 cells. Degradation under ultravioletexposure, which is reportedly more severe or devices with iO2 scaffolding, may similarly be addressed by protective encapsulation

eatures (in this case, ultraviolet ltering) or device design93.

Summary and future prospectsTe next ew years promise to be exciting ones or research anddevelopment o organic–inorganic halide perovskite solar cells.On-going efficiency improvements are expected, as well as a rap-idly growing understanding o their material properties and optimal

cell designs.Advantages over existing photovoltaic technologies includematerial properties that simpli y the manu acture o high-per or-mance devices. Te diversity in demonstrated approaches may giverise to low processing costs and simple implementation o attractiveproducts, such as exible, transparent or all-perovskite tandem cellmodules. Tis diversity may also allow perovskite cells to be directlyintegrated with other cell technologies to orm high-per ormancetandem cells; Si and CIGS modules appear particularly promisingin this respect3,6.

In the present market, the toxicity o Pb is not a major impedi-ment to large-scale, pro essional applications, as is evidenced by the

act that Cd e cells have already gained a reasonable market share.Cd or Pb is also present in some CIGS and silicon modules at the

same general level as that likely in perovskite modules94

. Te dangeris that technology relying on toxic materials may be increasinglymarginalized as legislation becomes increasingly more pervasiveand restrictive. Elimination o Pb seems the only sure solution,with replacement by Sn a possibility 95. Alternatively, research intothe present perovskites might allow more precise determination othe eatures that have resulted in such rapid progress, encouragingidentication and investigation o non-toxic material systems withsimilar properties.

Te recent surge in interest makes it likely there will be multi-ple attempts to commercialize perovskite photovoltaic products incoming years. Perovskites have the advantage that they provide mul-tiple paths to commercialization. As well as the traditional approacho challenging established manu acturers, there is also the opportu-nity to work with these to develop high-per ormance tandem cell

technology that uses both perovskite and existing technologimay allow market introduction as a new premium product.

Received 10 February 2014; accepted 12 May 2014; publishedonline 27 June 2014

References1. urner, G. Global Renewable Energy Market Outlook 2013.Bloomberg New

Energy Finance https://www.bne .com/insightdownload/7526/pd(11 April 2014).

2. Park, N.-G. Organometal perovskite light absorbers toward a 20% effilow-cost solid-state mesoscopic solar cell. J. Phys. Chem. Lett. 4, 2423–2429 (2013).

3. Snaith, H. J. Perovskites: the emergence o a new era or low-cost, hiefficiency solar cells. J. Phys. Chem. Lett. 4, 3623–3630 (2013).

4. Kim, H.-S., Im, S. H. & Park, N.-G. Organolead halide perovskite: newhorizons in solar cell research. J. Phys. Chem. C118, 5615–5625 (2014).

5. Hodes, G. & Cahen, D. Photovoltaics: perovskite cells roll orward.NaturePhoton. 8, 87–88 (2014).

6. Service, R. F. Perovskite solar cells keep on surging.Science 344, 458 (2014).7. Li, C.et al. Formability o ABX 3 ( X = F, Cl, Br, I) halide perovskites. Acta

Crystallogr.B64, 702–707 (2008).8. McKinnon, N. K., Reeves, D. C. & Akabas, M. H. 5-H 3 receptor ion

selectivity is a property o the transmembrane channel, not the cytopla vestibule portals. J. Gen. Physiol . 138, 453–466 (2011).

9. Cohen, B. N., Labarca, C., Davidson, N. & Lester, H. A. Mutations in the selectivity o the mouse nicotinic acetylcholine receptor or organalkali metal cations. J. Gen. Physiol . 100, 373–400 (1992).

10. Im, J.-H., Chung, J., Kim, S.-J. & Park, N.-G. Synthesis, structure, andphotovoltaic property o a nanocrystalline 2H perovskite-type novel se(CH3CH2NH3)Pbl3. Nanoscale Res. Lett. 7, 353 (2012).

11. Koh, . M.et al. Formamidinium-containing metal-halide: an alternativematerial or near-IR absorption perovskite solar cells. J. Phys. Chem. C http://dx.doi.org/10.1021/jp411112k (13 December 2013).

12. Eperon, G. E.et al. Formamidinium lead trihalide: a broadly tunableperovskite or efficient planar heterojunction solar cells.Energy Environ. Sci.7, 982–988 (2014).

13. Pang, S.et al. NH2CH=NH2Pbl3: An alternative organolead iodide perovskitsensitizer or mesoscopic solar cells.Chem. Mater . 26, 1485–1491 (2014).

14. Umari, P., Mosconi, E. & De Angelis, F. Relativistic GW calculations CH3NH3PbI3 and CH3NH3SnI3 perovskites or solar cell applications.Sci. Rep. 4, 4467 (2014).

15. opsöe, H. Krystallographisch-chemische untersuchungen homologer verbindungen.Zeitschrif ür Kristallographie 8, 246–296 (1884).

16. Mitzi, D. B., Wang, S., Feild, C. A., Chess, C. A. & Guloy, A. M. Conlayered organic–inorganic halides containing -oriented perovskitsheets.Science267, 1473–1476 (1995).

17. Mitzi, D. B., Chondroudis, K. & Kagan, C. R. Organic-inorganic electIBM J. Res. Dev. 45, 29–45 (2001).

18. Kojima, A., eshima, K., Miyasaka, . & Shirai, Y. Novel photoelectrocell with mesoscopic electrodes sensitized by lead-halide compounds (Proc. 210th ECS Meeting (ECS , 2006).

19. Kojima, A., eshima, K., Shirai, Y. & Miyasaka, . Organometal halidperovskites as visible-light sensitizers or photovoltaic cells. J. Am. Chem. Soc. 131, 6050–6051 (2009).

20. Kojima, A., eshima, K., Shirai, Y. & Miyasaka, . Novel photoelectrocell with mesoscopic electrodes sensitized by lead-halide compounds (Proc. 214th ECS Meeting (ECS, 2014).

21. Im, J.-H., Lee, C.-R., Lee, J.-W., Park, S.-W. & Park, N.-G. 6.5% efficiperovskite quantum-dot-sensitized solar cell.Nanoscale 3, 4088–4093 (2011).22. Kim, H.-S.et al. Lead iodide perovskite sensitized all-solid-state submicrothin lm mesoscopic solar cell with efficiency exceeding 9%.Sci. Rep. 2, 591 (2012).

23. Lee, M. M., euscher, J., Miyasaka, ., Murakami, . N. & Snaith, H. Efficient hybrid solar cells based on meso-superstructured organometalperovskites.Science338, 643–647 (2012).

24. Salbeck, J., Yu, N., Bauer, J., Weissörtel, F. & Bestgen, H. Low molecuorganic glasses or blue electroluminescence.Synthetic Metals 91, 209–215 (1997).

25. Bach, U.et al. Solid-state dye-sensitized mesoporous iO2 solar cells withhigh photon-to-electron conversion efficiencies.Nature 395, 583–585 (1998).

26. Stranks, S. D.et al. Electron–hole diffusion lengths exceeding 1 micrometan organometal trihalide perovskite absorber.Science 342, 341–344 (2013).

27. Heo, J. H.et al. Efficient inorganic–organic hybrid heterojunction solar cecontaining perovskite compound and polymeric hole conductors.NaturePhoton. 7, 486–491 (2013).


8/18/2019 nphoton.2014.134

8/9


28. Noh, J. H, Im, S. H., Heo, J. H., Mandal, . N. & Seok, S. I. Chemicalmanagement or color ul, efficient, and stable inorganic-organic hybridnanostructured solar cells.Nano Lett. 13, 1764–1769 (2013).

29. Stoumpos, C. C., Malliakas, C. D. & Kanatzidis, M. G. Semiconducting tinand lead iodide perovskites with organic cations: phase transitions, highmobilities, and near-in rared photoluminescent properties.Inorg. Chem. 52, 9019–9038 (2013).

30. Burschka, J.et al. Sequential deposition as a route to high-per ormanceperovskite-sensitized solar cells.Nature 499, 316–319 (2013).

31. Green, M. A., Emery, K., Hishikawa, Y., Warta, W. & Dunlop, E. D. Solar cellefficiency tables (version 43).Prog. Photovolt. 22, 1–9 (2014).

32. Liu, M., Johnston, M. B. & Snaith, H. J. Efficient planar heterojunctionperovskite solar cells by vapour deposition.Nature501, 395–398 (2013).

33. Burschka, J. High per ormance solid-state mesoscopic solar cells. PhD thesis,École Polytechnique Fédérale de Lausanne 107 (2013).

34. Chen, Q.et al. Planar heterojunction perovskite solar cells via vapor-assistedsolution process. J. Am. Chem. Soc. 136, 622–625 (2014).

35. Liu, D. & Kelly, . L. Perovskite solar cells with a planar heterojunctionstructure prepared using room-temperature solution processing techniques.Nature Photon.8, 133–138 (2014).

36. Christians, J. A., Fung, R. C. M. & Kamat, P. V. An inorganic hole conductoror organo-lead halide perovskite solar cells. Improved hole conductivity with

copper iodide. J. Am. Chem. Soc. 136, 758–764 (2014).37. Ito, S. Pb perovskite solar cells using inorganic hole conductor o CuSCN.

Paper Y-RS-14 in2013 MRS Fall Meeting & Exhibit (MRS, 2013).38. Docampo, P., Ball, J. M., Darwich, M., Eperon, G. E. & Snaith, H. J. Efficient

organometal trihalide perovskite planar-heterojunction solar cells on exiblepolymer substrates.Nature Commun.4, 2761 (2013).39. Malinkiewicz, O.et al. Perovskite solar cells employing organic charge-transport layers.Nature Photon.8, 128–132 (2014).

40. Sun, S.et al. Te origin o high efficiency in low-temperature solution-processable bilayer organometal halide hybrid solar cells.Energy Environ. Sci. 7, 399–407 (2014).

41. Wang, J. .-W.et al. Low-temperature processed electron collection layers ographene/ iO2 nanocomposites in thin lm perovskite solar cells.Nano Lett. 14, 724–730 (2014).

42. Wojciechowski, K., Saliba, M., Leijtens, ., Abate, A. & Snaith, H. J. Sub-150 °C processed meso-superstructured perovskite solar cells with enhancedefficiency.Energy Environ. Sci. 7, 1142–1147 (2014).

43. Eperon, G. E., Burlakov, V. M., Goriely, A. & Snaith, H. J. Neutral colorsemitransparent microstructured perovskite solar cells. ACS Nano8, 591–598 (2014).

44. Wang, J.et al. Per ormance improvement o amorphous silicon see-throughsolar modules with high transparency by the multi-line ns-laser scribingtechnique.Opt. Las. Eng. 51, 1206–1212 (2013).

45. Even, J., Pedesseau, L., Jancu, J.-M. & Katan, C. Importance o spin-orbitcoupling in hybrid organic/inorganic perovskites or photovoltaic applications. J. Phys. Chem. Lett . 4, 2999–3005 (2013).

46. Sell, D. D. & Lawaetz, P. New analysis o direct exciton transitions: applicationto GaP.Phys. Rev. Lett. 26, 311–314 (1971).

47. Even, J., Pedesseau, L., Dupertuis, M.-A., Jancu, J.-M. & Katan, C. Electronicmodel or sel -assembled hybrid organic/perovskite semiconductors: reverseband edge electronic states ordering and spin-orbit coupling.Phys. Rev. B 86, 205301 (2012).

48. Ishihara, . Optical properties o PbI-based perovskite structures. J. Luminescence 60, 269–274 (1994).

49. D’Innocenzo, V.et al. Excitons versus ree charges in organo-lead tri-halideperovskites.Nature Commun. 5, 3586 (2014).

50. auc, J. Optical properties and electronic structure o amorphous Ge and Si. Mater. Res. Bull. 3, 37–46 (1968).

51. Elliott, R. J. Intensity o optical absorption by excitons.Phys. Rev. 108, 1384–1389 (1957).52. anaka, K.et al.Comparative study on the excitons in lead-halide-based

perovskite-type crystals CH3NH3PbBr3 CH3NH3PbI3. Solid State Commun. 127, 619–623 (2003).

53. Combescot, M. Termodynamics o an electron-hole system insemiconductors.Phys. Status Solidi B 86, 349–358 (1978).

54. Corkish, R., Chan, D. S.-P. & Green, M. A. Excitons in silicon diodes and solarcells: a three-particle theory. J. Appl. Phys. 79, 195–203 (1996).

55. Geelhaar, F.Coulomb Correlation Effects in Silicon Devices (Series inMicroelectronics 147 Hartung-Gorre, 2004).

56. Green, M. A. Many-body theory applied to solar cells: excitonic and related carriercorrelation effects inProc.26th IEEE Photovoltaic Specialists Con erence (1997).

57. Green, M. A. Radiative efficiency o state-o -the-art photovoltaic cells.Prog.Photovolt. 20, 472–476 (2012).

58. Ungár, . Te meaning o size obtained rom broadened X-ray diffractionpeaks. Adv. Eng. Mater. 5, 323–329 (2003).

59. Edalati, K. & Horita, Z. Signicance o homologous temperature in sobehavior and grain size o pure metals processed by high-pressure tors Mater. Sci. Eng . A 528, 7514–7523 (2011).

60. Edri, E.et al. Why lead methylammonium tri-iodide perovskite-based solcells require a mesoporous electron transporting scaffold (but not neceshole conductor).Nano Lett . 14, 1000–1004 (2014).

61. Liang, K., Mitzi D. B. & Prikas, M. . Synthesis and characterization organic–inorganic perovskite thin lms prepared using a versatile two-dipping technique.Chem. Mater . 10, 403–411 (1998).

62. Kim, J., Lee, S.-H., Lee, J. H. & Hong, K.-H. Te role o intrinsic de emethylammonium lead iodide perovskite. J. Phys. Chem. Lett. 5, 1312–1317 (2014).

63. Yin, W.-J., Shi, . & Yan, Y. Unusual de ect physics in CH3NH3PbI3 perovskitesolar cell absorber. Appl. Phys. Lett. 104, 063903 (2014).

64. Onoda-Yamamuro, N., Matsuo, . & Suga, H. Dielectric study o CH3NH3PbX3 (X = Cl, Br, I). J. Phys. Chem. Solids 53, 935–939 (1992).

65. Poglitsch, A. & Weber, D. Dynamic disorder inmethylammoniumtrihalogenoplumbates (II) observed by millimeter-waspectroscopy. J. Chem. Phys. 87, 6373 (1987).

66. Hirasawa, M., Ishihara, ., Goto, ., Uchida, K. & Miura, N.Magnetoabsorption o the lowest exciton in perovskite-type compound(CH3NH3)PbI3. Physica B 201, 427–430 (1994).

67. Yuan, Y., Xiao, Z., Yang, B. & Huang, J. Arising applications o erromaterials in photovoltaic devices. J. Mater. Chem. A 2, 6027–6041 (2014).

68. Frost, J. M.et al. Atomistic origins o high-per ormance in hybrid halideperovskite solar cells.Nano Lett. 14, 2584–2590 (2014).

69. Snaith, H. J.et al. Anomalous hysteresis in perovskite solar cells. J. Phys. Chem.Lett. 5, 1511–1515 (2014).70. Hoke, E. ., Unger, E. L., Vandewal, K. & McGehee, M. D. Chargerecombination and transport in hybrid perovskite solar cells: why doperovskite solar cells have large Voc? inProc. MRS Fall Meeting andExhibit (2013).

71. Marchioro, A.et al. Unravelling the mechanism o photoinduced chargetrans er processes in lead iodide perovskite solar cells.Nature Photon. 8, 250–255 (2014).

72. Edri, E.et al. Elucidating the charge carrier separation mechanism inCH3NH3PbI3-x Clx perovskite solar cells.Nature Commun. 5, 3461 (2014).

73. Stranks, S. D.et al. Electron–hole diffusion lengths exceeding1 micrometer in an organometal trihalide perovskite absorber.Science 342, 341–344 (2013).

74. Xing, G.et al. Long-range balanced electron- and hole-transport lengths iorganic–inorganic CH3NH3Pbl3. Science 342, 344–347 (2013).

75. Anderson, R. L. Germanium-gallium arsenide heterojunction.IBM J. Res. Dev .4, 283–287 (1960).

76. Peressi, M., Baldereschi, A. & Baroni, S. inCharacterization o SemiconductorHeterostructures and Nanostructures 2nd Edn (eds Agostini G. & Lamberti, CCh. 2 (Elsevier, 2008).

77. Kavan, L. & Grätzel, M. Highly efficient semiconducting iO2 photoelectrodes prepared by aerosol pyrolysis. Electrochimica Acta 40, 643–652 (1995).

78. Green, M. A. Te depletion layer collection efficiency or p-n junction,Schottky diode, and sur ace insulator solar cells. J. Appl. Phys. 47, 547–554 (1976).

79. Lindblad, R.et al. Electronic structure o iO2/CH3NH3PbI3 perovskite solarcell inter aces. J. Phys. Chem. Lett. 5, 648–653 (2014).

80. Schulz, P.et al. Inter ace energetics in organo-metal halide perovskite-baphotovoltaic cells.Energy Environ. Sci. 7, 1377–1381 (2014).

81. First Solar Sets New World Record or CdTe Solar Cell Efficiency . http://investor.rstsolar.com/releasedetail.c m?releaseid=743398 (26 February 2013)

82. Widmar, M.First Solar Q4’13 Earnings Call . http://les.shareholder.com/downloads/ slr/1347979521x0x728649/bdd d430-a025-43e4-8b91-c1915066b274/q413_earnings_call_presentation_nal1.pd(25 February 2014).

83. De Jong, .First Solar Manu acturing Update. http://les.shareholder.com/downloads/ slr/3084163747x0x652323/53 7a04 - c 5-4729-8b0ab 6033c046/4.%20 sanalystday_manu acturing.pd (11 April 201

84. Garabedian, R. echnology Update. 2013 Analyst Meeting,First Solar .http://les.shareholder.com/downloads/FSLR/3084163747x0x652328/d0a 6554-e193-47e4-9dd0-59b02968272b/ sanalystday_technologyu(11 April 2014).

85. Rinaldi, N.Solar PV Module Costs to Fall to 36 Cents per Watt by 2017 http://www.greentechmedia.com/articles/read/solar-pv-module-costs-to- all-cents-per-watt (18 June 2013).

86. Directive 2011/65/EU o the European Parliament and o the Counci2011 on the restriction o the use o certain hazardous substances in eand electronic equipment (recast). http://eur-lex.europa.eu/legal-conten

X /?uri=celex:32011L0065 (8 June 2011).


8/18/2019 nphoton.2014.134

9/9


87. Coyle, D. J. Li e prediction or CIGS solar modules part 1:modeling moisture ingress and degradation.Prog. Photovolt.21, 156–172 (2013).

88. Coyle, D. J.et al. Li e prediction or CIGS solar modules part 2: degradationkinetics, accelerated testing, and encapsulant effects. Prog. Photovolt. 21, 173–186 (2013).

89. Kempe, M. D., Dameron, A. A. & Reese, M. O. Evaluation o moisture ingressrom the perimeter o photovoltaic modules.Prog. Photovolt. http://dx.doi.

org/10.1002/pip.2374 (2013).90. Kempe, M. D., Panchagade, D., Reese, M. O. & Dameron, A. A. Modeling

moisture ingress through polyisobutylene-based edge-seals.Prog. Photovolt.http://dx.doi.org/10.1002/pip.2465 (2014).

91. Niu, G.et al. Study on the stability o CH3NH3PbI3 lms and the effect o post-modication by aluminum oxide in all-solid-state hybrid solar cells. J. Mater.Chem. A 2, 705–710 (2014).

92. Carcia, P. F., McLean, R. S. & Hegedus, S. ALD Moisture barrier or Cu(InGa)Se2 solar cells. inProc. 218th ECS Meeting (ECS, 2010).

93. Leijtens, .et al. Overcoming ultraviolet light instability o sensitized iO2 with meso-superstructured organometal tri-halide perovskite solar cells.Nature Commun. 4, 2885 (2013).

94. Werner, J. H., Zap -Gottwick, R., Koch, M. & Fischer, K. oxicsubstances in photovoltaic modules. inProc. 21st Int. Photovoltaic Sci. Eng.Con . (2011).

95. Noel, N. K.et al. Lead- ree organic-inorganic tin halide perovskites orphotovoltaic applications.Energy Environ. Sci. http://dx.doi.org/10.1039/c4ee01076K (2014).

96. Wehren ennig, C., Liu, M., Snaith, H. J., Johnston, M. B. & Herz, L. MHomogeneous emission line broadening in the organo lead halide perovCH3NH3PbI3-x Clx . J. Phys. Chem. Lett.5, 1300–1306 (2014).

97. Rühle, S. & Dittrich, . Investigation o the electric eld in iO2/F O junctionsused in dye-sensitized solar cells by photocurrent transients. J. Phys. Chem. B 109, 9522–9526 (2005).

98. Snaith, H. J. & Grätzel, M. Te role o a “Schottky barrier” at an electrcollection electrode in solid-state dye-sensitized solar cells. Adv. Mater. 18, 1910–1914 (2006).

99. Kron, G., Rau, U. & Werner, J. H. Inuence o the built-in voltage on actor o dye-sensitized solar cells. J. Phys. Chem. B 107, 13258–13261 (2003).

AcknowledgementsTe Australian Centre or Advanced Photonics (ACAP) is supported by the AustGovernment through the Australian Renewable Energy Agency (ARENA). Resp

or the views, in ormation or advice expressed herein is not accepted by the AuGovernment. H.J. S. is supported by the Engineering and Physical Sciences ReseCouncil UK and the European Research Council.

Additional informationReprints and permissions in ormation is available at www.nature.com/reprints. R

or materials and correspondence should be addressed to M.A.G.

Competing nancial interestsTe authors declare no nancial interests.


Documents

nphoton.2014.134