long electron–hole diffusion length in ... - pld.nju.edu.cn

1706246 (1 of 6)Adv.Mater.2018,30, 1706246 © 2018WILEY-VCH Verlag GmbH& Co. KGaA, Weinheim

de

CommuniCationPerovskite Solar Cells

Long Electron–Hole Diffusion Length in High-QualityLead-Free Double Perovskite FilmsWeihua Ning, Feng Wang, Bo Wu, Jun Lu, Zhibo Yan, Xianjie Liu, Youtian Tao,Jun-Ming Liu, Wei Huang, Mats FAHLMAN, Lars HULTMAN, Tze Chien SUM,*and Feng Gao*

Solution-processed lead halide perovskites have shown supe-rior optoelectronic properties, including strong and tunablelight absorption/emission, long carrier diffusion lengths, andhigh carrier mobilities.[1,2] As a result, the power conversion

efficiencies of perovskite solar cells haveincreased from 3.8% to 22.1% withinonly a few years, making perovskitesthe fastest-advancing technology in thephotovoltaic history.[3,4] To ensure thesustainability of the perovskite photovol-taic technology, the number of studies toaddress the lead (Pb) toxicity and devicestability issues has increased.[5–7] The mostobvious option for lead-free perovskites isthe substitution of Pb2 with another diva-lent cation (e.g., germanium (Ge2) or tin(Sn2)).[8,9] Unfortunately, the resultingperovskites based on Sn2 or Ge2 areeasily oxidized by O2, limiting their prac-tical applications.[10] Bismuth (Bi)-basedorganic–inorganic metal halides have alsobeen studied as an alternative for solar cell

applications.[11] Different from the 3D lead-based perovskites,the 0D to 2D structures of Bi-based organic–inorganic halideslead to strongly bound excitons with low mobilities.[5]A new generation of perovskites, lead-free halide double per-

ovskites with a general formula of A2MM3X6, where bothDr.W.Ning, Dr.F.Wang,Dr. J. Lu, Dr.Z. Yan,Dr.X. Liu,Prof. M. Fahlman, Prof. L. Hultman, Prof. F. GaoDepartment of Physics, Chemistry, andBiology (IFM)Linköping UniversityLinköping SE-581 83, SwedenE-mail: [email protected]. W. Ning, Prof. Y. Tao, Prof. W. HuangKeyLabforFlexibleElectronics&InstituteofAdvancedMaterialsJiangsu National Synergistic Innovation Center for AdvancedMaterials (SICAM)Nanjing Tech University30SouthPuzhuRoad,Nanjing211816,P.R.ChinaDr. B. Wu, Prof. T. C. SumDivision of Physics and Applied PhysicsSchool ofPhysicalandMathematicalSciencesNanyang Technological University (NTU)21 Nanyang Link, Singapore 637371, SingaporeE-mail: [email protected]. Z. Yan, Prof. J.-M. LiuLaboratory of Solid State Microstructures and InnovationCenter of AdvancedMicrostructuresNanjing UniversityNanjing 210093, P. R. China

TheORCID identificationnumber(s) for theauthor(s)of thisarticlecanbefoundunderhttps://doi.org/10.1002/adma.201706246.

DOI: 10.1002/adma.201706246

A and M are monovalent cations, M3 is a trivalent cation, andX is a halide, provide rich substitutional chemistry and promisingoptoelectronic properties.[12] Several groups have successfully syn-thesized double perovskite powders and single crystals, and car-ried out crystal characterizations and fundamental studies.[13–15]Double perovskites show tunable band gaps spanning the visibleto near-infrared spectra and possess relatively low carrier effec-tive masses that are favorable for efficient charge transport andextraction, similar to 3D lead-based perovskites.[14,16] Moreover,these materials provide rich substitutional chemistry, whichcan dramatically change their photophysical properties.[17,18] Forexample, Tl-doped Cs2(Ag1aBi1b)TlxBr6 (x 0.075) results ina decrease in the bandgap of 0.5 eV.[18] Recent first-principlecalculations also indicate that halide double perovskites arepromising candidates for photovoltaic applications.[16,19,20] Fur-thermore, these double perovskites are much more stable thanGe or Sn perovskites in repelling the attacks by O2 and H2O.[15,21]However, since the precursors of double perovskites cannot dis-solve in common solvents (for example, dimethylformamide)which are frequently for lead-based perovskites, it is still a chal-lenge to fabricate double perovskite solar cells.[22]And also, mostof the fundamental questions concerning the photophysics ofdouble perovskite films remain unexplored and unknown due inpart to the lack of uniform and high-qualityfilms.

www.advmat.

Developing environmentally friendly perovskites has become important insolving the toxicity issue of lead-based perovskite solar cells. Here, the firstdouble perovskite (Cs2AgBiBr6) solar cells using the planar structure aredemonstrated. The prepared Cs2AgBiBr6 films are composed of high-crystal-quality grains with diameters equal to the film thickness, thus minimizingthe grain boundary length and the carrier recombination. These high-qualitydouble perovskite films show long electron–hole diffusion lengths greaterthan 100 nm, enabling the fabrication of planar structure double perovskitesolar cells. The resulting solar cells based on planar TiO2 exhibit an averagepower conversion efficiency over 1%. This work represents an important stepforward toward the realization of environmentally friendly solar cells and alsohas important implications for the applications of double perovskites in otheroptoelectronic devices.

mailto:[email protected]

mailto:[email protected]


www.advancedsciencenews.com www.advmat.de

Figure 1. a) Low-magnification andb) high-magnification SEM images, c) the TEM image andSAEDpattern, and d) the XRDpattern of the preparedCs2AgBiBr6 films annealed at 250 C for 5 min.

In this work, we demonstrate the first double perovskitesolar cells using the planar structure. We prepare high-qualityfilms with single-layer Cs2AgBiBr6 crystals. Through photo-physical investigations, we find the coexistence of excitons andfree carriers in the material. These Cs2AgBiBr6 films show along photoexcited carrier diffusion length of 110 nm. Theresulting solar cells based on planar TiO2 exhibit an averagepower conversion efficiency (PCE) over 1%.The high quality Cs2AgBiBr6 films are prepared through a

one-step spin-coating process from single-crystal Cs2AgBiBr6solutions. Figure 1a,b shows typical scanning electron micro-scopy (SEM) images of the perovskite films from a 0.5 m solu-tion. The surface roughness (Rq) is only 24 nm (Figure S1,Supporting Information). The smooth film is essential forthe following photoluminescence (PL)-quenching measure-ments and photovoltaic performance. The films are composedof closely packed polycrystalline grains with diameters of 100–500 nm. To examine each individuate grains, transmis- sionelectron microscopy (TEM) and selected electron diffrac- tion(SAED) are performed. Impressively, both TEM and SAEDreveal that each grain is a single crystal (Figure 1c), althoughthe films are polycrystalline. This feature is beneficial for thephotovoltaic performance since there is no grain boundaryin between from top to bottom of the film. The limited grainboundary in the vertical direction would be important forefficient carrier transfer in devices. In addition, the X-ray dif-fraction (XRD) pattern confirms the pure phase in the Cs2Ag-BiBr6 films, matching well with the results of the simulation(Figure 1d).

Figure 2 shows the UV–vis absorption of Cs2AgBiBr6 thinfilm. There are three parts in the absorption spectrum: below400 nm, with a flat absorption feature; an excitonic absorp-tion band in the region from 400 to 500 nm; and a very weakindirect absorption band between 500 and 538 nm, similar tothat in single crystals.[14] The absorption coefficients at 439 nmreach up to 1 105 cm1. By using the Elliott formula, thedirectbandgap (Egd) is 3.26 eV (See Figure S2 and the SupportingInformation). The indirect band gap cannot be determined bythe thin films due to its weak absorption, but was extracted pre-viously to be 1.95 eV from single crystals.[14]We obtain PL spectra and time-resolved photolumines-

cence (TRPL) studies to understand the photoexcited species.As shown in Figure 2a, the broad PL peak centered at 2.0 eV(620 nm) can be attributed to the indirect bandgap emission, asit corresponds well with the indirect absorption and emissionfound previously in single crystals.[14] The excitation intensitydependence of PL intensity just after photoexcitation is gene-rally a good indicator of the nature of the radiative recombi-nation processes.[23] Briefly, the initial PL intensity exhibitsa quadratic dependence on the photoexcitation density foremission by free-carrier band edge recombination: PLt0 n2,where n is the photoexcitation density (See the SupportingInformation for detailed explanation). For emission by radia-tive recombination of excitons or free carriers with doped car-riers, PLt0 n.[24–26] It is noted that the PL signal is too weakto be detected if the carrier density below 1016 cm3. When thecarrier density is between 1016 and 1017 cm3, the PL intensityshows power-dependence on the carrier density with a scaling

http://www.advancedsciencenews.com/

http://www.advmat.de/



Figure 2. a) UV–vis absorption and PL spectra of a 100 nmCs2AgBiBr6 film. b) PL intensity as a function of carrier density, the PL intensities were thevalues just after photoexcitation (PLt 0), rather than the integrated PL intensity. c) PL decay dynamics of the Cs2AgBiBr6, Cs2AgBiBr6/PC61BM, andCs2AgBiBr6/Spiro-OMeTAD films. d) The photocurrent of the Cs2AgBiBr6 devices as a function of solution concentration.

factor () of 1.34 (Figure 2b). This value suggests either the to 1/e of its initial intensity: A exp t A exp

t , i i

coexistence of radiative recombination of the free electrons–holes (recombination order 2) and the free carriers with dopedcarriers (recombination order 1), or the coexistence of exci-tons (recombination order 1) and free electrons–holes in themeasured excitation density range. However the doped carrierdensity measured by Hall Effect is on the order of 1013 cm3,which is not comparable to the photoexcitation density. Hence,it would be more reasonable to conclude that excitons and freecarriers coexist in the double perovskite films, and the excitonpossesses a higher ratio in the measured excitation range. Fur-ther increasing the carrier density to above 1017 cm3 causesthe linear scale factor () to decrease to 0.85. Meanwhile, theeffective PL lifetime (eff, the time at which the PL intensitydrops to 1/e of its maximum value) also decreases continu-ously with increasing carrier densities (Figure S3, SupportingInformation), implying strong high-order recombination suchas exciton–carrier and exciton–exciton Auger recombinationat high carrier densities. This unusual photophysical behaviorcontrasts from that of lead-based perovskites and further inves-tigations are warranted.To quantify the carrier diffusion length of the Cs2AgBiBr6

film, we carry out TRPL experiments. The thicknesses ofperovskite films are 100 nm, and the quenching samples areprepared by spin-coating layers of either a hole-transportingacceptor (Spiro-MeOTAD) or an electron-accepting fullerene(PC61BM) on top of the perovskite film. The results show thatthe PL decay dynamics of Cs2AgBiBr6 film can be fitted by abiexponential decay function at low excitation density, with timeconstants being 1 2.5 0.4 ns (52%), 2 35 1 ns (48%),which arises from the crystal size inhomogeneity. As a result,the effective excitation lifetime eff (the time for the PL decaying

i eff i i

where i is the ith fitted lifetime component of the decay curveand Ai is its weighted amplitude) in pristine film is 13.7 0.4 ns (Figure 2c). After the film is coated with PC61BM, thePL decays much faster, with eff being 2.4 0.4 ns. This indi-cates highly efficient electron transfer from the perovskite toPC61BM. A similar value of 2.6 0.2 ns was obtained for Spiro-MeOTAD coated perovskite films. Based on the PL quenchingmodel,[25,26] we estimate an average photoexcitation mobilityof 0.37 0.15 cm2 V1 s1, and the photoexcitation diffusionlength for electrons and holes 110 20 nm. The similarity ofdiffusion length is consistent with the dominance of excitonsas the primary photoexcitation species in the excitation densityrange. With photoexcitation diffusion lengths above 100 nm,the double perovskite polycrystalline film already shows excel-lent carrier diffusion properties comparable to those of lead-based perovskite films possessing typical carrier diffusionlengths of 100 nm–1 m.[25–27] It is noted that the diffusionlength measurements were performed at a carrier density of5 1016 cm3, which is expected to be higher than that ofdevices under solar illumination (1015–1016 cm3).[27] It is dif-ficult to make sure whether there is a trap-filling effect duringthe TRPL measurements, as we did not observe an increase ofcarrier lifetime with increasing the carrier density (Figure S4,Supporting Information), possibly due to low trap density inthe devices. Therefore, trap-filling as well as the tendency of thephotoexcitation species changing from excitons to free carrierswere not considered in the measurements.The long diffusion lengths and high mobility of the Cs2Ag-

BiBr6 films implies that carriers can travel through the film;thus solar cells are expected to operate well on planar-structured





devices. The detailed information about the fabrication ofan indium tin oxide (ITO)/compact TiO2/Cs2AgBiBr6/spiro-MeOTAD/Au device is described in the Supporting Information.The thickness-dependent photocurrent is compared by varyingthe concentration of the solution as follows: 0.4, 0.45, 0.5, and0.55 m. The corresponding thicknesses are 145 12, 170 15,205 10, and 223 10 nm, respectively. J–V results show thatthe photocurrent initially increases with increasing thickness,with an optimised Jsc 1.7 mA cm2 at a 205 nm thickness(Figure 2d; Figures S5 and S6, Supporting Information). Fur-ther increasing the thickness to 223 nm decreases the photo-current to 1.1 mA cm2. The optimum thickness of 205 nmconfirms the long diffusion length of Cs2AgBiBr6 films.The annealing temperature also has a significant effect on

the device performance. The Voc values are all 1.0 V for dif-ferent annealing temperatures (Figure S7, Supporting Infor-mation). However, Jsc and fill factor (FF) values first graduallyincrease with increasing annealing temperatures, and thendrop down when the temperature is above 250 C. This resultcan be rationalised by considering the characteristics of crys-tallinity, grain size, and pinholes of films together (Figures S8and S9, Supporting Information). The grain size and crystal-linity gradually increase with increasing annealing tempera-ture. Unfortunately, a large number of pinholes appear whenthe annealing temperature increases above 300 C. Furtherincrease in the annealing temperature to 400 C results in thedegradation of the films (Figure S8, Supporting Information).The optimized devices exhibit an average PCE of 1.05%

(averaged from 40 devices from five different batches) andoutstanding PCE of 1.22% with a Voc of 1.06 V (Figure 3c).Figure 3d shows the corresponding stabilized power outputwith a bias of 0.82 V.The device exhibits a rapid response after

illumination, resulting in a stable PCE of 1.17% over 600 s ofillumination. Notably, there is almost no hysteresis behaviorin the J–V curves, implying less trapping/detrapping or ionmigration in Cs2AgBiBr6 compared with lead-based hybrid per-ovskites. Very recently, Scanlon and co-workers estimated thespectroscopic limited maximum efficiency (SLME) of Cs2Ag-BiBr6 (200 nm) to be 7.92%.[28] The SLME takes into account thestrength of optical absorption and the nature of the band gapin the overall theoretical efficiency of an absorber material.[19]As an indirect band gap semiconductor, the SLME of Cs2Ag-BiBr6 is significantly dependent on the thickness of the films,and hence future approaches are required either to enhancethe thickness of Cs2AgBiBr6 films or to make Cs2AgBiBr6 intodirect bandgap semiconductor, e.g., through doping.[18]We note that the efficiency we obtained (up to 1.22%) is

much lower than the SLME of Cs2AgBiBr6 at 200 nm, in spiteof high crystal quality. One of the reasons might be due to thefact that the charge extraction efficiency of TiO2 for the Cs2Ag-BiBr6 films is not as efficient as those of Spiro-MeOTAD andPC61BM. PL intensity and decay of TiO2/Cs2AgBiBr6 showno obvious difference compared with those of pure Cs2Ag-BiBr6 (Figure S10, Supporting Information). Since the bandenergy of Cs2AgBiBr6 matches with those of TiO2 and Spiro-OMeTAD (Figure S11, Supporting Information), the reason ispossibly due to the presence of an interfacial barrier from sur-face defects, similar to the MAPbI3:TiO2 heterojunction.[29] Wealso estimate the theoretical JSC value for Cs2AgBiBr6 devicesbased on the diffusion length values and the absorption coef-ficient.[30] The calculated theoretical Jsc value for Cs2AgBiBr6devices can be around 5.2 mA cm2, which is higher than1.7 mA cm2 in our results. This also implies poor chargeextraction of TiO2 in the devices. We thus further fabricate

Figure 3. a) Planar Cs2AgBiBr6 solar cell structure, ITO/compact TiO2/Cs2AgBiBr6/Spiro-MeOTAD/Au. b) Cross-sectional SEM image of Cs2AgBiBr6solar cells. c) J–V curve of the optimized device. d) The stabilized power output of the Cs2AgBiBr6 solar cells with a bias of 0.82 V.





an ITO/perovskite/Spiro-MeOTAD/Au device without TiO2 tounderstand the charge extraction process. The resulting Jsc is1.7 mA cm2 (Figure S12, Supporting Information), which isalmost the same as that of devices with TiO2. This result con-firms less effective electron extraction by TiO2. However, theVoc of the devices without TiO2 is only 0.5 V,much lower thanthe 1.0 V shown with TiO2. The high Voc suggests that TiO2

can prevent the carrier recombination at the ITO side.In summary, we fabricate a uniform Cs2AgBiBr6 thin solid

film of high crystal quality through a one-step spin-coatingprocess from single-crystal Cs AgBiBr solution. Upon excita-

G2 TF20 UT with a field emission gun operated at 200 kV and a pointresolution of 0.19 nm. Sample thicknesses were measured using anAlpha step 500 Surface profilometer. The current density–voltage (J–V)curves were measured (Keithley Instruments, 2400 Series SourceMeter)under simulated AM 1.5 Solar Simulator. The effective area of thecell was defined as 0.075 cm2. The external quantum efficiency (EQE)data were obtained using a solar cell spectral response measurementsystem (QE-R3011, Enli Technology Co. Ltd), and the light intensityat each wavelength was calibrated with a standard single-crystal Siphotovoltaic cell. PL and TRPL measurements were performed using400 nm femtosecond excitation pulses (50 fs). The laser pulses weregenerated by passing the strong 800 nm femtosecond laser beam

2 6 (Coherent Libra, 50 fs) through a beta barium borate (BBO) crystaltion, excitons and free carriers coexist in double perovskitefilms, with a long diffusion length of 110 nm. We achieve anaverage PCE over 1% based on a planar device structure witha maximum value of 1.22%. The photovoltaic performanceis expected to be further boosted by replacing the TiO2 (usedpresently in this study) with more suitable ETL materialsand by increasing the film thickness while maintaining thefilm quality. In addition, it will be favorable to develop directbandgap double perovskites, e.g., through doping. The longcarrier diffusion length of high-quality double perovskite filmsopens a route toward developing environmentally friendly per-ovskite-based solar cells.

Experimental SectionCs2AgBiBr6 Single Crystals Synthesis: CsBr (213 mg, 1.00 mmol), BiBr3

(225 mg, 0.5 mmol), and AgBr (94 mg, 0.5 mmol) were dissolved in3mLof 47%HBr.The solutionwas transferred to aTeflon-lined reactor.After reacting at 120 C for 24 h, and cooling to room temperatureslowly, redCs2AgBiBr6octahedral single crystalswith the sizeof 2–5mmcan be collected (Figure S13, Supporting Information). The yield is85% calculated from Ag.

Cs2AgBiBr6 Solar Cell Fabrication: The TiO2 compact film precursorsolution in ethanol consists of 0.3 m titanium isopropoxide (Sigma-Aldrich, 99.999%) and 0.01 m HCl. A 35 nm dense TiO2 film wascoated onto an ITO substrate by spinning a titanium precursor at5000 rpm, followed by annealing at 200 C for 2 h. The synthesizedCs2AgBiBr6 single crystals were dissolved in dimethyl sulfoxide (DMSO)with a temperature of 100—130 C. After the crystals were completelydissolved, the solution was cooled to room temperature, and thendeposited onto the TiO2/ITO substrate by spin-coating at 3000 rpmfor 60 s. The films were annealed at 250 C for 5 min in order toobtain better crystallization. The thickness of the Cs2AgBiBr6 filmswas controlled by varying the concentration of the precursor solutionfrom 0.4–0.55 m. The highest PCE value of Cs2AgBiBr6 solar cells wasachieved from the 0.5 m solution. The spiro-MeOTAD based hole-transfer layerwas prepared bydissolving 60mgspiro-MeOTAD, 17.5Llithium-bis(trifluoromethanesulfonyl)imide (Li-TFSI) solution (520 mgLi-TFSI in 1 mL acetonitrile), and 28.8 L 4-tert-butylpyridine in 1 mLchlorobenzene. The devices were put into a dry cabinet for 15 h for theoxidization of Spiro-MeOTAD. The hole-transfer layer was depositedby spin-coating at 5000 rpm for 30 s. Finally, a 100 nm gold layer wasdeposited by thermal evaporation at a pressure of 1 104mbar. Alldevice fabrication stepswere carried out in an N2-purged glovebox.

MEASUREMENT and Characterization: The XRD patterns of theproducts were recorded with a X’Pert PRO X-ray diffractometer usingCu K1 irradiation ( 1.5406 Å). The Ultraviolet–visible absorptionspectra were measured on a Shimadzu spectrophotometer (UV-2450).The general morphologies of the films were characterized by SEM

(frequencydoubler). Theemitted lightwas collectedat abackscatteringangle by a spectrometer (Acton, Spectra Pro 2500i) and CCD (PrincetonInstruments, Pixis 400B) in PL measurements and by an OptronisOptoscope streak camera system which has an ultimate temporalresolution of 6 ps in TRPL measurements.

Supporting InformationSupporting Information is available from the Wiley Online Library orfrom the author.

AcknowledgementsW.N. and F.W. contributed equally to this work. The authors thankLijun Zhang (Jilin University, China) for insightful discussions. Thework was financially supported by the Joint NTU-LiU PhD programmeon Materials- and Nanoscience, the Swedish Government StrategicResearch Area inMaterials Science on Functional Materials at LinköpingUniversity (Faculty Grant SFO Mat LiU No. 200900971), the EuropeanCommission Marie Skłodowska-Curie actions (No. 691210); theEuropean Commission SOLAR-ERA.NET, the Swedish Energy Agency(Energimyndigheten), and the Swedish Research Council (FORMAS).F.G. is a Wallenberg Academy Fellow. Both F.W. and Z.Y. are VINNMERMarie Skłodowska-Curie Fellows. W.N. is supported by the ChinaScholarship Council. T.C.S. acknowledges the financial support fromthe Singapore Ministry of Education Academic Research Fund Tier 1grantsRG101/15andRG173/16,andTier 2grantsMOE2014-T2-1-044,MOE2015-T2-2-015, andMOE2016-T2-1-034; and from the SingaporeNational Research Foundation through the Competitive ResearchProgram NRF-CRP14-2014. L.H. acknowledges support from the KnutandAliceWallenberg (KAW) Foundation for a Scholar Grant 2016.0358,and for support to the Linköping Ultra ElectronMicroscopy Laboratory.

Conflict of InterestThe authors declare no conflict of interest.

Keywordsdouble perovskite, lead free, long diffusion length, planar solar cell

Received: October 27, 2017Revised: February 1, 2018

Publishedonline:March30,2018

(LEO 1550). The Atomic force microscope measurement was carriedout using a Dimension 3100/NanoScope IV system equipped with aC-AFM module (Veeco, Bruker). TEM was performed in the FEI Tecnai

[1] M. A. Green, A. Ho-Baillie, H. J. Snaith, Nat. Photonics 2014, 8, 506.[2] M. Gratzel, Nat. Mater. 2014, 13, 838.



long electron–hole diffusion length in ... - pld.nju.edu.cn

Documents