Guanidinium: A Route to Enhanced Carrier Lifetime and Open-CircuitVoltage in Hybrid Perovskite Solar CellsNicholas De Marco,†,‡ Huanping Zhou,†,‡ Qi Chen,†,‡ Pengyu Sun,† Zonghao Liu,†,‡ Lei Meng,†

En-Ping Yao,† Yongsheng Liu,†,‡ Andy Schiffer,‡ and Yang Yang*,†,‡

†Department of Materials Science and Engineering, ‡California NanoSystems Institute, University of California, Los Angeles,California 90095, United States

*S Supporting Information

ABSTRACT: Hybrid perovskites have shown astonishing power conversion efficienciesowed to their remarkable absorber characteristics including long carrier lifetimes, and arelatively substantial defect tolerance for solution-processed polycrystalline films.However, nonradiative charge carrier recombination at grain boundaries limits opencircuit voltages and consequent performance improvements of perovskite solar cells. Herewe address such recombination pathways and demonstrate a passivation effect throughguanidinium-based additives to achieve extraordinarily enhanced carrier lifetimes andhigher obtainable open circuit voltages. Time-resolved photoluminescence measurementsyield carrier lifetimes in guanidinium-based films an order of magnitude greater thanpure-methylammonium counterparts, giving rise to higher device open circuit voltagesand power conversion efficiencies exceeding 17%. A reduction in defect activation energyof over 30% calculated via admittance spectroscopy and confocal fluorescence intensitymapping indicates successful passivation of recombination/trap centers at grainboundaries. We speculate that guanidinium ions serve to suppress formation of iodidevacancies and passivate under-coordinated iodine species at grain boundaries and within the bulk through their hydrogenbonding capability. These results present a simple method for suppressing nonradiative carrier loss in hybrid perovskites tofurther improve performances toward highly efficient solar cells.

KEYWORDS: Perovskite, solar cell, guanidinium, passivation, open circuit voltage, carrier lifetime

1.0. INTRODUCTION

The hybrid organic−inorganic perovskite has emerged as astrong candidate for photovoltaic cells, achieving an unprece-dented rise in performance from 3.8% PCE1 as a liquid-basedsolar cell to an initial 9.7% in the solid state,2 eventually soaringrapidly to reach over 20% in just 5 years.3 This alluring materialpossesses several key attributes of an ideal solar cell absorber,such as a favorable band gap, high absorption coefficient, longambipolar carrier diffusion lengths, high carrier mobility, and arelatively high defect tolerance.1,4−9 Furthermore, its capabilityto be processed via low temperature solution techniquesrenders it substantially more cost-effective than the well-established silicon solar technology. As such, perovskite hasbecome highly attractive as an affordable and scalable next-generation photovoltaic technology with the potential to matchthe continuously increasing global energy demands.The ability of an absorber material to effectively generate and

extract charge carriers is of paramount importance to create ahighly efficient solar device. Theoretical studies have showndefect energy levels to lie relatively shallow,10,11 and it has beencommonly accepted that grain boundaries do not contributelargely to recombination losses in perovskite. However, suchrecombination centers in perovskite have recently been deemedless benign than previously believed.12,13 Hence, there stillremains considerable room to improve the performance of

perovskite solar cells. In order to effectively mitigate non-radiative carrier loss and enhance efficiency, it is necessary tosuppress recombination pathways within the perovskite filmitself. As a result of the solution-processed nature of perovskite,under-coordinated ions may exist at grain boundaries andsurfaces, as has been recently described, that can act as chargecarrier trap/recombination centers.14,15 Passivation of suchspecies has been demonstrated by Abate and co-workers, wherean iodopentafluorobenzene (IPFB) post-treatment was used tosuccessfully passivate under-coordinated iodine ions.14 In asimilar manner, Noel et al. used the Lewis bases, thiophene andpyridine, to passivate under-coordinated Pb ions and achievecarrier lifetimes of perovskite films up to 2 μs.15 Internalpassivation methods have also been reported. For instance,Chen and collaborators demonstrated a self-induced passivationeffect due to residual PbI2 to serve as recombination barriers.16

In addition, Br and Cl have been recently suggested topreferentially locate at grain boundaries where they serve toassist in suppressing recombination and decoupling electron−hole pairs.17,18 Thus, one may speculate that there is a highappeal to utilize the compositional flexibility of perovskite and

Received: October 5, 2015Revised: December 14, 2015

Letter

pubs.acs.org/NanoLett

© XXXX American Chemical Society A DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

explore new species for suppression of nonradiative recombi-nation pathways.In this light, the organic molecule CH6N3

+, more commonlyknown as guanidinium (GA), was investigated as an additive inMAPbI3 to observe influences on structure, film quality, andperformance. GA possesses an approximate zero dipolemoment that has been hypothesized to have influence onbias-induced ionic motion and hysteric effects.19,20 However,GA is substantially larger in size (278 pm) than the commonlyemployed methylammonium (MA; CH3NH3

+) cation (217pm), and ought to not form a 3D perovskite structure as thesole A-cation.21,22 Interestingly, GA has been well investigatedas an additive for dye-sensitized solar cells (DSSCs) to improveperformance.23 In such devices, many claim that GA serves topassivate TiO2 surfaces. More recently its contributions havebeen shown to lie within the interaction of the dye/liquidelectrolyte interface.23 In this manuscript we demonstrate apassivation effect through partial GA incorporation in perov-skite films, producing significant mitigation of nonradiative

decay enabling an order of magnitude enhanced carrier lifetimeover that of pure MAPbI3, open circuit voltages as high as 1.112V, and improved device performances over 17% PCE. Areduced defect activation energy and enhanced carrier lifetimewithin a full device indicate passivation effects as a result of theGA inclusion. We believe that the hydrogen bonding capabilityof GA provides enhanced grain size and continuity and moresignificantly serves to effectively passivate under-coordinatediodine species between adjacent crystalline grains.

2.0. RESULTS AND DISCUSSIONMixed solutions of MA and GA precursors (MAI, GACl, GAI)with molar ratios ranging from 1:0 to 0:1 (MA:GA) werestudied. Perovskite films were fabricated via a sequential two-step deposition technique using spin-coating similarly toprevious reports24 (detailed in the Supporting Information).To directly observe the effects of GA on device performance,current−voltage (J−V) measurements were employed forvarying GA content with a planar device architecture of ITO/

Figure 1. (a) Current−voltage characteristics of the champion GA and MA reference devices. The optimized GA device shows an improved Voc andpreserved FF and Jsc yielding an overall efficiency improvement, as depicted in the inset. (b) Device averages demonstrating the consistency ofenhanced performance characteristics for GA addition.

Figure 2. (a) Molar and weight ratios ranging from pure MA (ref) to pure GA (Film 7) used in this study; (b) corresponding film colors forpreannealed (top) and postannealed (bottom) GA films; (c) absorption spectra for varying GA content with a log scale plot in the inset.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

B

TiO2/perovskite/spiro-OMeTAD/Au. Intriguingly, we ob-served a prominent enhancement of Voc through GAincorporation. It was found that for only a small quantity ofGA incorporation, ranging between a 30:1 and 6:1 molar ratioof MA:GA, the Voc could be improved while preserving Jsc. Anoptimal molar ratio of 6:1 (MA:GA), hereinafter denoted asGA when directly compared to the reference MA device,provided the best performance characteristics. Figure 1a showsthe J−V curve of the champion device in comparison with thereference (MA) sample, demonstrating an enhancement of Vocfrom 1.025 V in the standard MA-based device to 1.071 V forGA. Furthermore, a slight increase in fill factor (FF) from 75.00to 75.31 for MA and GA was observed, respectively. The Jscremained relatively unaffected with GA inclusion, achieving21.24 mA/cm2 compared to 21.27 mA/cm2 for the pure MAdevice. The much improved Voc and slightly increased FF led toan overall power conversion efficiency (PCE) gain of 17.13%compared to 16.35% for the MA reference device. The averagedevice characteristics (Figure 1b) show the consistency of theobserved performance enhancement for GA addition. We cansee that GA provides an average increase in Voc ofapproximately 50 mV, while preserving the Jsc and improvingthe FF from 74.87% to 76.04%. These improvements lead to adevice performance average of 16.27% over the 15.75% PCE ofits MA-based counterpart. As the GA content was increasedover a molar ratio of 6:1 (MA:GA), a steady decrease in Jsc wasobserved. For molar ratios above 6:1 up to 3:1 (MA:GA), a Vocbetween 1.07 and 1.09 V was consistently achievable, where ahigh of 1.112 V was achieved, as depicted in Figure S1. It can beobserved that this increased Voc comes at the cost of Jsc, and didnot lead to the highest performing device. Despite inferioroverall efficiencies of the higher Voc devices to that of the

champion device (due a large sacrifice in the Jsc and FF), thisresult shows that even higher open circuit voltages may beattainable in the future.A prominent reddish color change of the preannealed

perovskite film, and a corresponding increase in transparencypostannealing, was observed for incremental amounts of GAaddition, as shown in Figure 2a,b. It was found that a molarratio of 2:1 (MA:GA) was the upper limit of GA inclusion forsuccessful 3D perovskite formation indicated by a distinctyellow resultant film color. UV−visible absorption spectroscopywas utilized to investigate the optical effects of GA inclusion.Figure 2c shows the absorption spectra for increasing GAcontent, from which it is readily apparent that an increase inGA content results in a decrease in light harvesting capability,as evidenced by the reduction in Jsc. The band gap (Eg) remainsrelatively unaffected with GA incorporation, varying between1.53 and 1.55 eV for different GA amounts corresponding toabsorption tails spanning 790−800 nm. The calculated bandgap for MA and GA were 1.56 and 1.55 eV, respectively. Wenote that this small fluctuation within this range is commonlyobserved for reference samples prepared with identicalconditions, and is based on morphological and film qualitydifferences between samples. Tauc plots are provided in FigureS2 to show variation in optical band gap. Interestingly, thereduction in Jsc from GA incorporation does not appear toresult from a shift in band gap, but rather, a reduction inabsorption intensity across the spectrum for increasing GAcontent. This indicates that GA does not directly substitute forMA, as if this were the case the band gap ought to shift inaccordance with the large size of the GA cation. This result is incontradiction with previous theoretical studies that suggestsubstitution of MA with the larger GA cation would increase

Figure 3. (Top) Scanning electron microscopy (SEM) and (Bottom) atomic force microscopy (AFM) of (a,c) MA and (b,d) GA films, respectively.Scale bars are 1 μm for the SEM images, and the width of AFM images is 5.0 μm. An increase in film continuity and slightly improved surfaceroughness is observed via GA inclusion, showing an improved morphology.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

C

the band gap due to a reduction in antibonding interactionsbetween Pb and I.19 This would indeed be the case if GA wereto successfully substitute for MA in the perovskite crystalstructure. However, as there is no experimental evidenceregarding the crystal structure for GAPbI3 to show how itincorporates into the perovskite crystal lattice, calculationsbased on the GA ion in perovskite must make such anassumption. We believe that an excess amount of the oversizedGA within the perovskite film likely disrupts the crystallinity ofthe 3D structure that constitutes the lower absorption,consequent reduction in Jsc, and unsuccessful perovskiteformation over the upper molar ratio limit. Evidence for thisclaim is provided in the following characterization results.Ultraviolet photoelectron spectroscopy (UPS) was further

used to observe Fermi level and valence band edge positions ofthe MA and GA champion device (Figure S3). There is anegligible Fermi level (Ef) shift, with values located atapproximately 4.59 and 4.54 eV for MA and GA, respectively,indicating an n-type nature of the perovskite film. A slightchange in the valence band energy level (EVB) is observed. Thepure-MA reference device yields an EV value of 5.6 eV incomparison to a slight shift to 5.72 eV for GA. Interestingly,this downward shift does not seem to affect the hole transferbetween perovskite and Spiro-OMeTAD. In order to observestructural impacts of GA on the perovskite crystal, X-raydiffraction (XRD) analysis was conducted (Figure S4a). Both inthe case of MA and GA, strong characteristic perovskite 110and 220 peaks were observed, located at approximately 14.15°and 28.5° (2θ), respectively, in accordance with previousresults.25 Along with the absorption spectra and observed filmcolor change, these results show that a small amount of GA

does not affect the perovskite structure, while a larger amountwill ultimately disrupt the crystallinity. To further explore thenature of GA incorporation within the perovskite crystal, weconducted XRD measurements for pure GA (no MA) content,as provided in Figure S4b. From the spectra we can see that thecharacteristic perovskite peaks diminish entirely as a result ofthe GA cation inclusion. While this confirms that GA does notincorporate into the lattice, this spectra cannot be wellinterpreted as there appear to be other factors as a result ofthe thin film fabrication process that affect the spectra.Therefore, we cannot ascribe this XRD spectra to be anaccurate representation of a pure GAPbI3 phase. A thoroughstudy on the structural characteristics of the pure GAPbI3 singlecrystal phase is thus necessary and is currently underinvestigation for future work. Nevertheless, these resultsdemonstrate that GA cannot serve as a direct substitution forMA within the perovskite crystal lattice, suggesting that its rolein enhanced performance characteristics lie elsewhere.One plausible explanation for the enhanced Voc is facilitation

of improved film morphology resulting from GA that mayreduce recombination at interfaces. It is believed thatintroducing organic ammonium cations could help facilitatecrystal growth to improve film uniformity and compactness.26

Indeed, this effect is observed through scanning electronmicroscopy (SEM) images provided in Figure 3a,b that depictmorphological impacts of the resulting films. A slightimprovement in film continuity is observed for GA-basedfilms as there appears to be higher grain continuity with fewersmall grain protrusions and less prominent grain boundariespresent. To investigate such features, atomic force microscopy(AFM) was used to measure the surface profiles of the MA and

Figure 4. Confocal fluorescence intensity images for (a) MA and (b) GA (scale bar is 5 μm). Darker low-intensity regions are prominent in the caseof MA, which we attribute to grain boundary regions. (c) Corresponding histogram for the intensity range of fluorescence images of MA and GA,respectively. A magnified view of (d) MA and (e) GA films with 4 μm linear intensity profiles shown in (f). The MA film shows a prominent dropacross the darker region, whereas the GA remains relatively constant across.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

D

GA films (Figure 3c,d). GA shows a slight reduction in surfaceroughness of 12.7 nm in comparison to 13.2 nm of pure MA.To further observe any morphological trends, we fabricatedfilms with higher GA content (Figure S5). An even furtherimproved film continuity and larger grain regions are apparentin the case of the SEM image for increasing GA content,suggesting an improved morphology and reduced surfaceroughness. Interestingly, it would appear that the higher GA(Film 2) content provides a smoother and more continuousmorphology than the optimized GA content (Film 1) accordingto the SEM images. However, AFM measurements yield asurface roughness of 16.3 nm for Film 2, which is larger thanthat of Film 1. This indicates that though there is a larger graincontinuity for increased amounts of GA, the larger surfaceroughness may reduce interfacial contact that would affectcharge extraction across the perovskite transport layerjunctions. Consequentially, the very slight decrease in surfaceroughness for GA (Film 1) compared to the reference (MA)does not provide conclusive evidence that the observedperformance improvements may be attributed to an improvedfilm roughness and interfacial contact effects. Such a smallchange in surface roughness is commonly observed within agiven sample due to local topographical variations acrossdifferent regions of the surface, resulting from the imperfectnature of the spin-casting technique.It has recently been suggested that the microstructure can

greatly affect local carrier lifetime in perovskite.12 In particular,it was shown through PL microscopy that between differentgrains the local carrier lifetime varies considerably, and that thedark low-intensity regions of the PL image, corresponding tograin boundaries, represent a region of reduced carrier lifetime.In this light, we conducted confocal fluorescence microscopy tocorrelate observed morphological features to the carrierdynamic effects of GA films, where a low intensity (darker)region corresponds to a reduced lifetime. The fluorescenceintensity mapping of the MA and GA films are depicted inFigure 4. The samples were prepared on a glass substrateidentically to the two-step procedure, as specified in theSupporting Information. Darker regions are clearly distinguish-able in the case of the MA reference (Figure 4a). In the case ofthe GA film (Figure 4b), we observe a reduced quantity of theprominent dark regions. In correlation to our SEM and AFMimages, we presume that the larger regions representmorphological differences observed between the film surfacesof MA and GA. Individual grains are not highly distinguishable

due to the limited resolution of this technique, however, we canobserve what appears to be small grains with sizes between 500nm to 1 μm, which is in accordance with grain sizes observedunder SEM. The darker regions surrounding these grains arebelieved to be grain boundaries. Accordingly, we can observeless prominent darker grain boundaries surrounding individualgrains in the case of GA, indicating that recombination withinthese grain boundary regions is successfully suppressed.Histogram plots (Figure 4c) of average intensity show a higheraverage intensity in the case of the GA sample. The meanintensity values, within a relative intensity range of 0−255, were140.47 in the case of GA and 131.54 for MA. We can alsoobserve over an order of magnitude increase in the number ofcounts for GA, indicating that it fluoresces more strongly.Furthermore, 4 μm linear intensity profiles were taken for boththe MA and GA samples, as shown in Figure 4d,e. Plottingthese linear intensity scans as a function of distance (Figure 4f)yields a consistent profile for GA, whereas there is a strong dipin the profile in the case of MA. These results indicate that thedarker regions in MA indeed correspond to a reduced intensity,and corresponding lifetime, suggesting that GA serves topassivate grain boundaries.Photoluminescence spectroscopy is a useful tool to extract

information regarding charge carrier dynamics in semi-conductor materials. Specifically, the quality of the materialcan be analyzed by determining the degree of nonradiativerecombination loss. The carrier dynamics of the GA perovskitefilms were carefully examined through photoluminescence (PL)and time-resolved photoluminescence (TRPL) measurements(Figure 5). The perovskite films for this measurement wereprepared on glass substrates with an identical sequential two-step deposition technique as used for device fabrication. To oursurprise, GA shows a large rise in PL intensity over four timesthat of the pure MA devices, as shown in Figure 5a. Even moreintriguingly, the TRPL spectra shows a substantial enhance-ment in carrier lifetime (τ), superseding that of the reference byan order of magnitude. The curves display biexponentialcharacter and were fitted accordingly with both fast and longdecay components, where the long decay component (τ) canbe described as free carrier recombination.16 Figure 5b portraysan enhanced τGA of 800 ns for the GA film compared to τMA =80 ns in the case of the MA reference. It is rather intriguing thatsuch an exceptional improvement in carrier lifetime does notyield an as significantly improved performance. In principle, animproved carrier lifetime ought to result in a larger open circuit

Figure 5. (a) Comparison of relative PL intensities and (b) TRPL spectra of pure MA and GA PVSK films deposited on glass substrates. GA showsan increased PL intensity over 4 times MA and an enhanced carrier lifetime an order of magnitude over that of MA.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

E

voltage as a reduction in minority carrier Shockley−Read−Halltype recombination (due to a reduced defect density) wouldyield larger Fermi level splitting. However, the complexity of afully assembled device under operating conditions is muchgreater than that of a bare perovskite film during opticalmeasurements. It is probable that there are other factors withinthe device, such as the selective contacts, that create additionalbottlenecks that pose additional limitations on device perform-ance regardless of the quality of the perovskite film itself. Withthis concept in mind, we may still correlate the enhancedcarrier lifetime to the enhanced device performance character-istics, mainly, the open circuit voltage, for which a lessprominent improvement is observed. Interestingly, it wasobserved that additional GA species reduces the PL intensityand carrier lifetime, indicating that exceeding the 6:1 molarratio will have a reverse effect. The PL and TRPL spectra foradditional GA content is provided in Figure S6. This result is inaccordance with the observed morphological trends, however, itcontradicts our observed performance trends, as a higher Voccan still be obtained through use of Film 2, but a reduced Jscand FF led to reduced performances.Transient photovoltage decay was conducted to further

explore carrier dynamics through investigation of carrierdynamics within a fully assembled device. In this technique,illuminated devices are probed with a light pulse to generate aphotovoltage perturbation. The voltage decay rate thereafter,which represents the change in carrier density versus time, isobserved to extract information regarding carrier lifetime withinthe device. Figure 6 shows the photovoltage decay spectra forthe reference MA and GA samples. The carrier lifetimes weretaken as the time for which the photovoltage decayed to 1/e ofits peak intensity.27 The MA reference device shows a chargecarrier lifetime of 3.15 μs, in agreement with previous reports(Figure 6a).25 The GA sample shows an improved chargecarrier lifetime of 7.95 μs, over twice that of the MA device,verifying suppression of carrier recombination. Figure 6bdepicts the extracted carrier lifetimes as a function of lightintensity. GA consistently yields carrier lifetimes approximatelytwice that of MA-based devices, providing further validation ofthis result. We note here that use of both GAI and GAClyielded comparable lifetime enhancements, signifying that theobserved effects are indeed a result of incorporation of theguanidinium ion.Carrier recombination dynamics in solar devices are largely

influenced by defects within the absorber material. Admittance

spectroscopy provides a useful route to extract the energy levelsof defects residing within semiconductor materials. Thecapacitance is directly associated with the charging anddischarging of trapped charge carriers within the material, ashas been described in further detail in our previous report.28

Occupancy of these trap sites depends on the location of theFermi level, where traps below the Fermi level are assumedfilled. Accordingly, by varying the AC voltage frequency and thetemperature of the system we can probe these trap levels perFermi level shift and extract information regarding the defectactivation energy of the system. A comparison of defectactivation energies determined from admittance spectroscopymeasurements is provided in Figure S7. The activation energyof the MA-based device was calculated as 21.11 meV, which isin accordance with our previous report.29 We can observe thatthe defect activation energy of the GA-based device issignificantly lower (14.86 meV), indicating that recombinationwithin the bulk of the film has been successfully suppressed.Among the native point defects within perovskite, iodinevacancies have shown to be most detrimental for chargetrapping and nonradiative recombination due to their deeplying energy levels.30 It was suggested that chloride inclusioncan alter the lattice constant of perovskite and prevent theformation of iodide vacancies. Because GA does not directlysubstitute for MA into the perovskite crystal lattice, itsinteractions between adjacent crystalline domains of theperovskite film may similarly affect the lattice constant, therebysuppressing the formation of deep iodide vacancies, asevidenced by the reduced defect activation energy.On a similar note, passivation of under-coordinated iodine

species has recently been demonstrated via postfilm formationtreatments using the IPFB Lewis acid.14 As a free cation, GA isa Lewis acid owed to the difference in covalent characterbetween the central C atom and surrounding nitrogen. The C−N partial charge difference for GA within the solid stateperovskite film has been shown to be greater than that for thefree GA cation according to recent theoretical studies.19 In fact,the unique hydrogen bonding capability of GA has beenpreviously demonstrated in metal−organic frameworks(MOFs).31 The symmetry of the amine groups allows theGA molecule to form six hydrogen bonds with neighboringoctahedra of the MOF, which enhanced its mechanical stability.Hence, we expect an increased quantity of hydrogen bondingcapability between the partial negative (δ−) iodine ions ofexposed PbI6

4− of neighboring perovskite crystals and the

Figure 6. (a) Normalized log scale transient photovoltage decay measurements for MA and GA samples. GA shows an improved carrier lifetime of7.95 μs compared to 3.15 μs for MA. (b) Carrier lifetimes for varying light intensities of both MA and GA samples. GA consistently achievesapproximately twice the lifetime of MA-based devices at different light intensities.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

F

partial positive (δ+) ammonium ions of the organic GA species.Accordingly, we propose that GA preferentially resides at grainboundaries and forms hydrogen bonds with under-coordinatediodine species to effectively suppress these charge trapping/recombination regions. Furthermore, the hydrogen bondingcapability between neighboring grains may aid in grain growthduring film formation, facilitating the higher film continuity andlarger grain regions observed through SEM images. It is worthnoting that incorporation of the passivant GA molecule directlyinto the precursor solution, as opposed to postsurfacetreatments, may allow for the passivant species to addressdeeper internal defect regions that are otherwise inaccessiblethrough postfilm formation surface treatments. This may beone advantage in using additive-based passivation methods infuture works. However, the majority of the GA species ought toreside at grain boundaries and interfaces between the perovskitefilm and selective contacts.Despite the tremendous growth in performances of perov-

skite solar cells, there still remain two critical issues unresolvedthat prevent its commercialization in (1) the current−voltagehysteresis, and (2) the inherent instability of the perovskitefilm. Hysteresis effects are commonly observed for forward andreverse current−voltage scans of perovskite solar cells. GA hasbeen hypothesized to potentially solve hysteresis in perovskitefilms owed to its zero dipole moment.19,20 To compare thehysteric effects between the GA and MA-based devices, weconducted forward and reverse current−voltage scans for thechampion devices, provided in Figure S8. As can be observed,hysteresis is slightly more prominent for GA addition incomparison to MA alone. At first we may be surprised at thisresult since it is in contradiction with what we would expectaccording to the zero dipole moment of GA and the work fromGiacomo et al.19 However, since this GA ion is not confinedwithin the perovskite crystal, the ionic molecule would thus beresponsive to an external electrical influence. Tress et al. haverecently conducted thorough studies on hysteresis, for whichthey attributed such effects to migration of ionic species tointerfaces that can effectively screen the applied electric field.32

Since the GA species do not directly substitute for MA into thecrystal structure and are rather weakly hydrogen-bonded to theunder-coordinated ionic species mainly at grain boundaries(due to its approximate zero dipole moment19), the GA ionsought to be more responsive to an external electrical influence.As previously mentioned, the amount of GA species located atgrain boundaries will be significantly more than those residinginternally at defect regions within the bulk. Consequentially,the more mobile GA species between interfaces and at grainboundaries would readily migrate to the junction under theinfluence of an applied bias, leading to further internal electricfield cancellation and enhanced hysteric effects.Now turning our attention toward the stability of perovskites,

it is inarguably the most pressing issue preventing perovskitefrom commercialization as perovskite films exposed to ambientenvironment will degrade within a matter of hours to daysdepending on the humidity level.25,33 The stability for the GA-based device is compared to the standard MA-based device(Figure S9). The devices were stored under a dry oxygenenvironment and only exposed to ambient environment duringmeasurement. We can see that the both devices are relativelystable for approximately 7.5 days (180 h), where afterward bothdevices undergo a rapid decrease of the initial PCE.Interestingly, the rate of PCE drop for the pure MA-baseddevice is much higher than that of the device with GA

inclusion, where the GA device maintains over 80% of its initialPCE compared to approximately 60% for the pure MA device.While there are several factors involved in device stability, andconclusive evidence regarding the role of GA in enhancedstability requires further investigation, we speculate that thefollowing as a potential explanation of the observed results: Asgrain boundaries are highly susceptible to invasion by incidentwater molecules, the less prominent, and therefore vulnerable,grain boundaries as observed from the SEM images couldmitigate the effects of attacking water molecules. Moreover,since water molecules will primarily attack the more susceptiblegrain boundary regions, the weakly bound GA ions located inthese regions will be the first to interact with the incomingwater molecules. If this were the case, it would preserve thecrystalline perovskite domains for longer than films without GApresent. Nevertheless, a thorough study on the role of GA indevice stability is required and is currently under investigation.

3.0. CONCLUSION

In summary we have demonstrated a simple route tosignificantly enhance carrier lifetimes and open circuit voltagesin hybrid perovskite solar cells via GA-based additives. Time-resolved photoluminescence and photovoltage decay techni-ques were used to extract charge carrier dynamics, demonstrat-ing an exceptional improvement in carrier lifetime for GA-basedfilms 1 order of magnitude larger than the pure MA films thatled to open circuit voltages exceeding 1.1 V. Fluorescencespectroscopy results showing a reduced PL intensity quenchingat grain boundary regions in correlation with a reduced defectactivation energy measured by admittance spectroscopyindicate successful passivation of nonradiative recombination/trap centers within the perovskite film. We propose that thelarge hydrogen bonding capability of the GA molecule allowsfor effective passivation of under-coordinated iodine specieslocated at grain boundaries both at the surface and internallywithin the bulk. SEM and AFM images show a slightimprovement in film quality with a lower surface roughness,higher degree of film continuity, and less distinct grainboundaries that may correlate to the observed stability of theGA-based device. The introduction of the GA species as aperovskite precursor overcomes the limitations of surfacetreatments by allowing for infiltration of the GA passivant intointernal regions within the perovskite bulk, where under-coordinated species may reside. This study demonstrates asimple and effective route to improve carrier lifetime and filmquality through precursor additives, bypassing the need foradditional processing steps and material usage to improveperformances of perovskite solar cells.

■ ASSOCIATED CONTENT

*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.nano-lett.5b04060.

Details regarding device fabrication and characterizationprocedures along with supporting figures (PDF)

■ AUTHOR INFORMATION

Corresponding Author*E-mail: yangy@ucla.edu.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

G

Author ContributionsN.D., H.Z., and Y.Y. generated the idea. N.D., H.Z., and Q.C.discussed the direction and experimental details. N.D.conducted device fabrication and all characterization with theassistance of L.M. for UPS measurements, Z.L. for photovoltagedecay, P.S. and E.Y. for admittance spectroscopy, and A.S. fordevice fabrication. N.D. wrote the manuscript with input fromH.Z., Q.C., and Y.Y.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was financially supported by UC Solar MRPI (GrantNo. 442551-YY-69090) as well as grants from the NationalScience Foundation (Grant Numbers ECCS-1202231, ProgramDirector: Dr Radhakisan S. Baheti; and ECCS-1509955,Program Director: Dr. Nadia El-Masry), Air Force Office ofScientific Research (Grant Number FA9550-12-1-0074, Pro-gram Manager Dr. Charles Lee) and UCLA Internal Funds.The authors graciously acknowledge the Advanced LightMicroscopy/Spectroscopy Lab at California NanoSystemsInstitute at UCLA for their assistance with fluorescenceimaging.

■ REFERENCES(1) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am. Chem. Soc.2009, 131, 6050−6051.(2) Kim, H. S.; Lee, C. R.; Im, J. H.; Lee, K. B.; Moehl, T.;Marchioro, A.; Moon, S. J.; Humphry-Baker, R.; Yum, J. H.; Moser, J.E.; Gratzel, M.; Park, N. G. Sci. Rep. 2012, 2, 591.(3) Research Cell Efficiency Records, NREL http://www.nrel.gov/ncpv/images/efficiencychart.jpg. Accessed: 12 December, 2015.(4) Burschka, J.; Pellet, N.; Moon, S.-J.; Humphry-Baker, R.; Gao, P.;Nazeeruddin, M. K.; Gratzel, M. Nature 2013, 499, 316−319.(5) Etgar, L.; Gao, P.; Xue, Z.; Peng, Q.; Chandiran, A. K.; Liu, B.;Nazeeruddin, M. K.; Gratzel, M. J. Am. Chem. Soc. 2012, 134, 17396−17399.(6) Lotsch, B. V. Angew. Chem., Int. Ed. 2014, 53, 635−637.(7) Park, N.-G. J. Phys. Chem. Lett. 2013, 4, 2423−2429.(8) Snaith, H. J. J. Phys. Chem. Lett. 2013, 4, 3623−3630.(9) Chen, Q.; De Marco, N.; Yang, Y.; Song, T.-B.; Chen, C.-C.;Zhao, H.; Hong, Z.; Zhou, H.; Yang, Y. Nano Today 2015, 10, 355−396.(10) Yin, W.-J.; Shi, T.; Yan, Y. Appl. Phys. Lett. 2014, 104, 063903−063903−4.(11) Kim, J.; Lee, S.-H.; Lee, J. H.; Hong, K.-H. J. Phys. Chem. Lett.2014, 5, 1312−1317.(12) deQuilettes, D. W.; Vorpahl, S. M.; Stranks, S. D.; Nagaoka, H.;Eperon, G. E.; Ziffer, M. E.; Snaith, H. J.; Ginger, D. S. Science 2015,348, 683−686.(13) Agiorgousis, M. L.; Sun, Y.-Y.; Zeng, H.; Zhang, S. J. Am. Chem.Soc. 2014, 136, 14570−14575.(14) Abate, A.; Saliba, M.; Hollman, D. J.; Stranks, S. D.;Wojciechowski, K.; Avolio, R.; Grancini, G.; Petrozza, A.; Snaith, H.J. Nano Lett. 2014, 14, 3247−3254.(15) Noel, N. K.; Abate, A.; Stranks, S. D.; Parrott, E. S.; Burlakov, V.M.; Goriely, A.; Snaith, H. J. ACS Nano 2014, 8, 9815−9821.(16) Chen, Q.; Zhou, H.; Song, T.-B.; Luo, S.; Hong, Z.; Duan, H.-S.;Dou, L.; Liu, Y.; Yang, Y. Nano Lett. 2014, 14, 4158−4163.(17) Yin, W.-J.; Chen, H.; Shi, T.; Wei, S.-H.; Yan, Y. Adv. Electron.Mater. 2015, DOI: 10.1002/aelm.201500044.(18) Kim, G. Y.; Oh, S. H.; Nguyen, B. P.; Jo, W.; Kim, B. J.; Lee, D.G.; Jung, H. S. J. Phys. Chem. Lett. 2015, 6, 2355−2362.(19) Giorgi, G.; Fujisawa, J.-I.; Segawa, H.; Yamashita, K. J. Phys.Chem. C 2015, 119, 4694−4701.(20) Giorgi, G.; Yamashita, K. Nanotechnology 2015, 26, 442001.

(21) Kieslich, G.; Sun, S.; Cheetham, A. K. Chem. Sci. 2014, 5, 4712−4715.(22) Szafranski, M.; Jarek, M. CrystEngComm 2013, 15, 4617−4623.(23) Jeanbourquin, X. A.; Li, X.; Law, C.; Barnes, P. R. F.; Humphry-Baker, R.; Lund, P.; Asghar, M. I.; O’Regan, B. C. J. Am. Chem. Soc.2014, 136, 7286−7294.(24) Im, J.-H.; Jang, I.-H.; Pellet, N.; Gratzel, M.; Park, N.-G. Nat.Nanotechnol. 2014, 9, 927−932.(25) Zhou, H.; Chen, Q.; Li, G.; Luo, S.; Song, T.-b.; Duan, H.-S.;Hong, Z.; You, J.; Liu, Y.; Yang, Y. Science 2014, 345, 542−546.(26) Li, X.; Ibrahim Dar, M.; Yi, C.; Luo, J.; Tschumi, M.;Zakeeruddin, S. M.; Nazeeruddin, M. K.; Han, H.; Gratzel, M. Nat.Chem. 2015, 7, 703−711.(27) Stranks, S. D.; Eperon, G. E.; Grancini, G.; Menelaou, C.;Alcocer, M. J. P.; Leijtens, T.; Herz, L. M.; Petrozza, A.; Snaith, H. J.Science 2013, 342, 341−344.(28) Duan, H.-S.; Zhou, H.; Chen, Q.; Sun, P.; Luo, S.; Song, T.-B.;Bob, B.; Yang, Y. Phys. Chem. Chem. Phys. 2015, 17, 112−116.(29) Chen, Q.; Zhou, H.; Fang, Y.; Stieg, A. Z.; Song, T.-B.; Wang,H.-H.; Xu, X.; Liu, Y.; Lu, S.; You, J.; Sun, P.; McKay, J.; Goorsky, M.S.; Yang, Y. Nat. Commun. 2015, 6, 7269.(30) Du, M. H. J. Mater. Chem. A 2014, 2, 9091−9098.(31) Li, W.; Thirumurugan, A.; Barton, P. T.; Lin, Z.; Henke, S.;Yeung, H. H. M.; Wharmby, M. T.; Bithell, E. G.; Howard, C. J.;Cheetham, A. K. J. Am. Chem. Soc. 2014, 136, 7801−7804.(32) Tress, W.; Marinova, N.; Moehl, T.; Zakeeruddin, S. M.;Nazeeruddin, M. K.; Gratzel, M. Energy Environ. Sci. 2015, 8, 995−1004.(33) Yang, J.; Siempelkamp, B. D.; Liu, D.; Kelly, T. L. ACS Nano2015, 9, 1955−1963.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b04060Nano Lett. XXXX, XXX, XXX−XXX

H