inso

, KHsin-Ming Cheng d, Chi-Chung Kei c, Zong-Liang Tseng a, Chun-Guey Wu a,e

inuences the photovoltaic performance in terms of open-circuit voltage (VOC), ll factor (FF), and short-circuit current density (JSC). In

absorption spectrum (DInnocenzo et al., 2014; Wolfet al., 2014). The exciton binding energy (Urbach energy)of PA is about 50 meV (15 meV), which is slightly largerthan that of GaAs. The small exciton binding energy(Urbach energy) and the low optical bandgap are the

Corresponding author. Tel.: +886 3 4227151x25360; fax: +886 34252897.

E-mail addresses: shchang@ncu.edu.tw (S.H. Chang), t610002@cc.ncu.edu.tw (C.-G. Wu).

Available online at www.sciencedirect.com

ScienceDirect

Solar Energy 120 (2015the case of the thicker ZnO lm, the photovoltaics have the better FF and Voc, as a result of the smaller electron recombination. Thismeans that a thicker ZnO lm can block the electron recombination from the Fermi level of the ITO to the valance band of the TPA. Onthe other hand, the thicker ZnO lm results in a higher JSC due to the better exciton dissociation at the interface between TPA and ZnO,which means that the electron mobility of the thicker ZnO is higher. Consequently, the photovoltaic performance can be expected to beimproved by using a transparent cathode electrode with high conductivity and electron mobility. 2015 Elsevier Ltd. All rights reserved.

Keywords: Time-resolved photoluminescence; Photo-induced absorption; Energy transfer; Photovoltaic

1. Introduction

Mixed halide (CH3NH3PbI3xClx) and tri-iodide(CH3NH3PbI3) perovskite absorbers (PAs) have been inten-sively investigated because with them a power conversioneciency (PCE) of 15% can be achieved by using a solution

process under low temperatures (Burschka et al., 2013; LiuandKelly, 2014). In order to absorbmore sun-light, the opti-cal bandgap of PA has been engineered to range from 2.2 eVto 1.1 eV (Eperon et al., 2014; Stoumpos et al., 2013). Theexciton binding energy and Urbach energy of PA can bedetermined by analyzing the temperature-dependentaResearch Center for New Generation Photovoltaics, National Central University, Taoyuan 32001, Taiwan, ROCbDeaprtment of Chemical and Material Engineering, National Central University, Taoyuan 32001, Taiwan, ROCcNational Applied Research Laboratories, Instrument Technology Research Center, Hsinchu 30013, Taiwan, ROC

dMaterial and Chemical Research Laboratories, Industrial Technology Research Institute, Hsinchu 31040, Taiwan, ROCeDepartment of Chemistry, National Central University, Taoyuan 32001, Taiwan, ROC

Received 11 May 2015; received in revised form 3 June 2015; accepted 16 July 2015

Communicated by: Associate Editor Takhir M. Razykov

Abstract

Comprehensive studies were carried out to understand the thickness eects of ZnO thin lms for tri-iodide perovskite absorber (TPA)based photovoltaics, including the absorption spectrum, photoluminescence, nanosecond time-resolved photoluminescence (NTRPL),and photo-induced absorption (PIA) of TPA/ZnO/ITO/glass. These were carried out in order to explore the Urbach energy of TPA lmsand the exciton dissociation at the interface between TPA and ZnO. The results show that the thickness of the ZnO thin lm signicantlyThickness eects of ZnO thof tri-iodide perovskite ab

Kun-Mu Lee a,b, Sheng Hsiung Chang a,http://dx.doi.org/10.1016/j.solener.2015.07.030

0038-092X/ 2015 Elsevier Ltd. All rights reserved.lm on the performancerber based photovoltaics

ai-Hung Wang b, Chun-Ming Chang c,

www.elsevier.com/locate/solener

) 117122

The thickness of ZnO thin lm was varied from 5 nm to40 nm by controlling the ALD cycle times.

3. Results and discussion

Fig. 1 presents the X-ray diraction (XRD) patterns ofZnO thin lm and TPA lm. The XRD characteristics ofZnO thin lm cannot be obtained due to the thin thickness.The four diraction peaks of 13.4, 19.5, 24.0 and 27.4indicate the formation of crystalline CH3NH3PbI3(Kojima et al., 2009). The diraction peak of 12.1 origi-nates from the unreacted PbI2, which is presented intwe-step depostion because of dense PbI2 layer for planarstructure. The substrate-induced changes in the energy

nergy 120 (2015) 117122reasons why the PCE of PA based photovoltaics can be lar-ger than 15%. The rst attempt of using CH3NH3PbX3(X = Br and I) as light absorber was made indye-sensitized liquid-type photovoltaic (Kojima et al.,2009), which was signicantly improved to 6.5% by optimiz-ing perovsikte fabrication process (Im et al., 2011).However, the perovskite absorber suers from liquid-typedye-sensitized photovoltaic because it tends to easily dis-solved in polar liquid electrolyte. Long-term durable allsolid state perovskite solar cell was rst developed andreported in literature (Kim et al., 2012). The rst high e-cient PA based photovoltaic device was realized using thefollowing structure: Ag/Sprio-OMeTAD/PA/TiO2/FTO/glass (Lee et al., 2012), where Spiro-OMeTAD and meso-porous TiO2 were used as the hole acceptor and electronacceptor, respectively. The main drawback is that the sinter-ing temperature for anatase TiO2 has to be higher than450 C (Schattauer et al., 2012) which can damage the plasticsubstrate. For low-temperature fabrication, ZnO thin lmshave been used to replace the mesoporous TiO2 as an elec-tron acceptor (Liu and Kelly, 2014; Kumar et al., 2013;Son et al., 2014). ZnO nanoparticle lm andSpiro-OMeTAD lm were used to extract the electronsand holes, respectively (Liu and Kelly, 2014). The optimizedthickness of a ZnO nanoparticle lm for PA based photo-voltaics is 25 nm, which results in a high PCE of 15.7%. Incontrast, in this study we investigate the thickness eectsof ZnO thin lm fabricated by atomic layer deposition(ALD) method (Wang et al., 2010).

We seek to understand the thickness eects of ZnO thinlm on the performance of PA based photovoltaics.Nanosecond time-resolved photoluminescence (NTRPL)and photo-induced absorption (PIA) were carried out toobserve the exciton dynamics of PA/ZnO/ITO/glass fordierent thicknesses of ZnO thin lm.

2. Fabrication

ZnO thin lm was deposited on top of ITO/glass with asheet resistance of 10 X/sq under a temperature of 80 Cusing the ALD method. The detailed fabrication processof the ZnO thin lm is described in our previous report(Wang et al., 2010). A two-step process was used to growthe TPA lm. This consisted of spin coating a lm ofPbI2 onto the ZnO/ITO/glass, followed by spin coating asolution of CH3NH3I in isopropyl alcohol. After thetwo-setp spin coating process, the lms were heated at100 C for 30 s. The detailed fabrication process of theTPA lm is described in our previous report (Chianget al., 2014). Spiro-OMeTAD was spin coated on top ofTPA/ZnO/ITO/glass as the hole acceptor. Then, MoO3and Ag were thermally evaporated onto the sample to actas the anode electrode. The nal photovoltaic structurewas comprised of Ag/MoO3/Sprio-OMeTAD/TPA/ZnO/ITO/glass. The thicknesses of ITO, TPA,

118 K.-M. Lee et al. / Solar ESpiro-OMeTAD, MoO3, and Ag were controlled at ca.250 nm, 350 nm, 150 nm, 5 nm, and 100 nm, respectively.band position of TPA lm has been investigated (Milleret al., 2014). In order to include the substrate-inducedchanges in the optical and electrical properties of theTPA lms, the ZnO/ITO/galss substrate is used in the opti-cal characterizations of TPA lms. Fig. 2 presents theabsorbance and PL spectra of the TPA/ZnO/ITO/glass.In the linear absorbance spectrum, there are two prominentpeaks located at 480 nm and 760 nm, which is in agreementwith the perivously reported values (Xing et al., 2013). Inthe PL spectrum, there is an emission peak located at768 nm when pumping at k = 532 nm, which is the directbandgap emission from the rst conduction band to therst valence band (Xing et al., 2013). The exciton bindingenergy ranging from 30 meV to 50 meV (DInnocenzoet al., 2014; Koutselas et al., 1996; Ishihara, 1994), whichis slightly larger than the thermal energy (25 meV) atroom temperature. For an exciton binding erngy of50 meV (30 meV), the faction of the photo-generated exci-tons can be calculated from statistical physics to yield: 34%(45%) of the excitons dissociating spontaneously andremaining 66% (55%) of the excitons (Sum and Mathews,2014). Therefore, the PL originages mainly from the radia-tive recombination of electronhole pair (exciton). The PLintensity can be used to eavuate the exciton quenching(exciton dissociation) at the interface between TPA andZnO. The thickness of ZnO thin lm highly inuenscesFig. 1. X-ray diraction patterns of ZnO/glass and TPA/glass.

optical bandgap, PL peak, and Urbach energy of samplesA, B, C, D, and E are almost the same and are listed inTable 1, which indicates that the optical and electricalproperties of TPA lms do not change when the thicknessof ZnO thin lm is varied from 5 nm to 40 nm.

PL decay time constant can be viewed as the lifetime ofexcitons in TPA/ZnO/ITO/glass. PIA of TPA/ZnO/ITO/

K.-M. Lee et al. / Solar Energy 120 (2015) 117122 119Fig. 2. Absorbance and photoluminescence spectra of TPA/ZnO/ITO/glass. The measurements are carried out at room temperature.the PL intensity of TPA, which indicates that the excitonquenching eciency is related to the electrical propertiesof ZnO thin lm.

Fig. 3 presents the absorbance in the optical band edge.In this wavelength range of 767785 nm, the absorptionoriginates from the TPA lm. The optical bandgap andUrbach energy of TPA can be obtained by tting theabsorbance spectrum using the empirical equation(Urbach, 1953) which can be written as follows:

a a0 expEp Eg=Eu; 1where a0 is the absorption coecient, Ep is the photonenergy, Eg is the optical bandgap, and Eu is the Urbachenergy. This Urbach energy represents the thermal disorderor the occupancy level of phonon states in TPA. The

Fig. 3. Absorbance spectrum of TPA/ZnO/ITO/glass and its tting curve.

Table 1Optical bandgap, photoluminescence peak, and Urbach energy of TPA.

Sample number Thickness of ZnO (nm) Optical bandgap (e

A 5 1.578B 10 1.581C 20 1.579D 30 1.580E 40 1.581glass occurs when excitons stay in excited state.Therefore, the PIA strength is proportional to the lifetimeof excitons. It means that the lifetime of excitons (PIAstrength) is subject to the substrate due to the built-in driv-ing force at the pn junction interface. The shorter (longer)lifetime of excitons corresponds to the faster (slower) exci-ton dissociation at the pn junction interface. Therefore,the lifetime of excitons can be used to evaluate theeciency of the exciton dissociation at the pn junctioninterface. In order to understand the thickness eects ofthe ZnO thin lm, the NTRPL and PIA of theTPA/ZnO/ITO/glass were measured. Fig. 4(a) presentsthe normalized NTRPL intensity of the TPA lm at thewavelength of the PL peak. To obtain the relaxation time(lifetime of excitons), the data are tted in a constant expo-nentially decaying function. Fig. 4(b) presents the experi-mental and tting curves. The curve tting begins at1 ns in order to avoid the laser response. The lifetimesof excitons for samples A, B, C, D, and E are 15.84 ns,40.58 ns, 33.79 ns, 18.16 ns, and 22.1 ns respectively.Fig. 5 presents the PIA spectra of TPA lms. The largerPIA strength corresponds to the longer PL lifetime.There is a shorter PL lifetime and a stronger PIA strengthwhen the thickness of the ZnO lm is 5 nm. In such a case,the 5-nm thick ZnO cannot completely cover the rough-ened ITO surface (Jonda et al., 2000) as shown in Fig. 6,which results in a better exciton dissociation at the inter-face between the TPA and ITO due to the large built-indriving potential. The driving potential (1 V) is propor-tional to the dierence between the Fermi level of ITO(4.9 eV) and the conduction band of TPA (3.9 eV)(Chang et al., 2013). When the thickness of the ZnO lmis increased to 10 nm, the lifetime of excitons (PIAstrength) in the TPA lm is increased to 40.58 ns whichmeans that the exciton dissociation at the interface betweenTPA and ZnO is slower. In such a case, the driving poten-tial (0.5 V) is proportional to the dierence between theFermi level of ZnO (4.4 eV) and the conduction bandof TPA (3.9 eV). The lifetime of excitons (PIA strength)in the TPA lm is decreased when the thickness of ZnOlm is increased from 10 nm to 30 nm, which can be

V) Photoluminescence peak (eV) Urbach energy (meV)

1.617 211.617 251.617 24

1.617 211.617 23

ner120 K.-M. Lee et al. / Solar Eexplained as beeing due to the thickness-dependent electronmobility of ZnO lms (Look et al., 2014). The lifetime ofexcitons (PIA strengh) in the TPA lm is not changed

Fig. 4. (a) Nanosecond time-resolved photoluminescence of TPA/ZnO/ITO/glass at k = 768 nm under 405-nm excitation. (b) Normalized PLintensity with the tting curve. The measurements are carried out at roomtemperature.

Fig. 5. Photo-induced absorption spectra of TPA/ZnO/ITO/glass under500-nm excitation.gy 120 (2015) 117122signicantly when the thickness of ZnO lm is increasedfrom 30 nm to 40 nm. In other words, the exciton dissoci-ation rate at the interface between the TPA and ZnOincreases with the electron mobility of the ZnO lm, as pre-dicted by the OnsagerBraun theory (Deibel et al., 2009).The electron mobility of ZnO lms was examined by hallmeasurements using van der Pauw geometry on aHL55WIN hall system. The measurement results show thatthe electron mobility of 30-nm (40-nm) thick ZnO is 17.5(18.3) cm2/V s, which is quite close to the value reportedin literature (Look et al., 2014). However, the electronmobility of ZnO lms cannot be obtained when the thick-ness is thinner than 30 nm, which is possibly due to anon-contiguous structure of the ZnO thin lms(Demaurex et al., 2014).

Fig. 6. The topography of ITO glass is imaged with tapping-mode AFM.The maximum peak-to-valley roughness (RPV) is 13 nm.

Fig. 7. JV curves of TPA based photovoltaics under 1 sun illumination.

is listed in Table 2. The series resistance (RS) and shuntresistance (RSh) are dened as the inverse of the slop of

, 1

Js

1012121413

nerthe JV curve of the best cell at V = 0 and J = 0, respec-tively. The thickness of the ZnO thin lm signicantlyinuences the photovoltaic performance in terms ofopen-circuit voltage (VOC), ll factor (FF), andshort-circuit current density (JSC). The dierence betweenthe VOC of sample A and the VOC of sample D equals0.4 V, which is almost the same as the dierence betweenthe Fermi levels of the ITO lm and ZnO lm (Yip andJen, 2012). Compared with sample D, the smaller FF andshunt resistance of sample A can be explained by the higherelectron recombination from the Fermi level of ITO to thevalence band of TPA (Liu and Kelly, 2014). Therefore, theFF is improved when the thickness of ZnO lm is increasedfrom 5 nm to 30 nm. However, the VOC, JSC, FF, and PCEare slightly decreased when the thickness of ZnO lm ischanged from 30 nm to 40 nm due to the increased seriesresistance.

4. Conclusions

In summary, we have successfully fabricated tri-iodideperovskite absorber (TPA) based photovoltaic devicesunder low-temperatures (

Koutselas, I.B., Ducasse, L., Papavassiliou, G.C., 1996. Electronicproperties of three- and low-dimensional semiconducting materialswith Pb halide and Sn halide units. J. Phys.: Condens. Matter 8, 1217.

Kumar, M.M., Yantara, N., Dharani, S., Graetzel, M., Mhaisalkar, S.,Boix, P.P., Mathews, N., 2013. Flexible, low-temperature, solutionprocessed ZnO-based perovskite solid state solar cells. Chem.Commun. 49, 1108911091.

Lee, M.M., Teuscher, J., Miyasaka, T., Murakami, T.N., Snaith, H.H.,2012. Ecient hybrid solar cells based on meso-superstructuredorganometal halide perovskite. Science 338, 643647.

Liu, D., Kelly, T.L., 2014. Perovskite solar cells with a planarheterojunction structure prepared using room-temperature solutionprocessing techniques. Nat. Photon. 8, 133138.

Look, D.C., Leedy, K.D., Thomson, D.B., Wang, B., 2014. Highconductance in ultrathin lms of ZnO. Jpn. J. Appl. Phys. 53,05FJ01-3.

Miller, E.M., Zhao, Y., Mercado, C.C., Saha, S.K., Luther, J.M., Zhu, K.,Stevanovic, V., Perkins, C.L., van de Lagemaat, J., 2014. Substrate-controlled band positions in CH3NH3PbI3 perovskite lms. Phys.Chem. Chem. Phys. 16, 2212222130.

Schattauer, S., Reinhold, B., Albrecht, S., Fahrenson, C., Schubert, M.,Janietz, S., Neher, D., 2012. Inuence of sintering on the structuraland electronic properties of TiO2 nanoporous layers prepared via anon-sol-gel approach. Colloid Polym. Sci. 290, 18431854.

Son, D.-Y., Im, J.-H., Kim, H.-S., Park, N.-G., 2014. 11% ecientperovskite solar cell based on ZnO nanorods: an eective chargecollection system. J. Phys. Chem. C 118, 1656716573.

Stoumpos, C.C., Maliakas, C.D., Kanatzidis, M.G., 2013.Semiconducting tin and lead iodide perovskite with organic cations:phase transitions, high mobilities, and near-infrared photoluminescentproperties. Inorg. Chem. 52, 90199038.

Sum, T.C., Mathews, N., 2014. Advancements in perovskite solar cells:photophysics behind the photovoltaics. Energy Environ. Sci. 7, 25182534.

Urbach, F., 1953. The long-wavelength edge of photographicsensitivity of the electronic absorption of solids. Phys. Rev. 92,1324.

Wang, J.-C., Weng, W.-T., Tsai, M.-Y., Lee, M.-K., Horng, S.-F., Perng,T.-P., Kei, C.-C., Yu, C.-C., Meng, H.-F., 2010. Highly ecientexible inverted organic solar cells using atomic layer deposited ZnOas electron selective layer. J. Mater. Chem. 20, 862866.

Wolf, S.D., Holovsky, J., Moon, S.-J., Loper, P., Niesen, B., Ledinsky,M., Haug, F.-J., Yum, J.-H., Ballif, C., 2014. Organometalic halideperovskite: sharp optical absorption edge and its relation to photo-voltaic performance. J. Phys. Chem. Lett. 5, 10351039.

Xing, G., Mathews, N., Sun, S., Lim, S.S., Lam, Y.M., Gratzel, M.,Mhaisalkar, S., Sum, T.C., 2013. Long-range balanced electron- andhole-transport lengths in organic-inorganic CH3NI3PbI3. Science 342,344347.

Yip, H.-L., Jen, K.-Y., 2012. Recent advances in solution-processedinterfacial materials for ecient and stable polymer solar cells. EnergyEnviron. Sci. 5, 59946011.

122 K.-M. Lee et al. / Solar Energy 120 (2015) 117122

Thickness effects of ZnO thin film on the performance of tri-iodide perovskite absorber based photovoltaics1 Introduction2 Fabrication3 Results and discussion4 ConclusionsAcknowledgmentsReferences