1

Planar inverted perovskite solar cells (PSCs) with high efficiency and ambient stability

are fabricated based on a reduced graphene oxide (rGO) doped PBCM electron transporting

layer (ETL). The performance improvement is attributed to enhanced electron extraction,

while the enhancement in the lifetime is due to the stabilization of the Perovskite/ETL

interface of the rGO doped device compared to the pristine.

Keywords: perovskite solar cell, reduced graphene oxide, graphene based electron

transport layer, improved electron extraction, enhanced ambient stability

George Kakavelakis,* Temur Maksudov, Dimitrios Konios, Ioannis Paradisanos, George

Kioseoglou, Emmanuel Stratakis, Emmanuel Kymakis*

Efficient and Highly Air Stable Planar Inverted Perovskite Solar Cells with Reduced

Graphene Oxide doped PCBM Electron Transporting Layer

2

Efficient and Highly Air Stable Planar Inverted Perovskite Solar Cells with Reduced

Graphene Oxide doped PCBM Electron Transporting Layer

George Kakavelakis,* Temur Maksudov, Dimitrios Konios, Ioannis Paradisanos, George

Kioseoglou, Emmanuel Stratakis, Emmanuel Kymakis*

G. Kakavelakis, T. Maksudov, Dr. Dimitrios Konios, Prof. E. Kymakis

Center of Materials Technology and Photonics &Electrical Engineering Department,

Technological Educational Institute (TEI) of Crete, Heraklion 71004 Crete, Greece

E-mail: kakavelakis@staff.teicrete.gr; kymakis@staff.teicrete.gr

G. Kakavelakis, T. Maksudov, Prof. G. Kioseoglou, Dr. E. Stratakis

Department of Materials Science and Technology, University of Crete, Heraklion, 710 03

Crete, Greece

G. Kakavelakis, I. Paradisanos, Prof. G. Kioseoglou, Dr. E. Stratakis

Institute of Electronic Structure and Laser, Foundation for Research and Technology - Hellas,

P. O. Box 1527, 711 10 Heraklion Crete, Greece

Keywords: perovskite solar cell, reduced graphene oxide, graphene based electron transport

layer, improved charge extraction, enhanced ambient stability

Abstract

Reduced graphene oxide (rGO) was added in the PCBM electron transport layer (ETL) of

planar inverted perovskite solar cells (PSCs), resulting in a power conversion efficiency

(PCE) improvement of ~ 12%, with a hysteresis-free PCE of 14.5%, compared to 12.9% for

the pristine PCBM based device. The universality of our method was demonstrated in PSCs

based on CH3NH3PbI3-xClx and CH3NH3PbI3 perovskites, deposited through one step and two

step spin coating process respectively. After a comprehensive spectroscopic characterization

of both devices, it is clear that the introduction of rGO in PCBM ETL resulted in an important

increase of the ETL conductivity, together with reduced series resistance and surface

roughness. As a result, a significant photoluminescence quenching of such perovskite/ETM

was observed, confirming the increased measured short circuit current density. Transient

absorption measurements revealed that in the rGO-based device, the relaxation process of the

excited electrons was significantly faster compared to the reference, which implies that the

charge injection rate was significantly faster for the first. Furthermore, the light soaking effect

3

was significantly reduced. Finally, ageing measurements revealed that the rGO stabilizes the

ELT/perovskite interface, which results in the stabilization of perovskite crystal structure after

prolonged illumination.

Introduction

Organic−inorganic lead halide perovskites have been recently demonstrated as one of the

most promising materials for next generation photovoltaic and light emitting applications.[1]

Owing to some key properties, such as the medium optical bandgap and strong absorption

coefficients,[ 2 ]

the long carrier diffusion lengths,[ 3 ]

the low recombination losses,[ 4 ]

and

bandgap tunability,[ 5 ]

hybrid perovskite-based absorbers have attracted great interest

presenting the potential to be established as an efficient, low cost and flexible, large-scale

photovoltaic technology. As a result, lead halide perovskite solar cells (PSCs) power

conversion efficiencies (PCEs) have rapidly grown from 3.9% to more than 22% in a few

years’ time,[6,7]

resulting in performances comparable with current state-of-the-art commercial

technologies.

Despite the high record PCEs, the commercialization of this technology requires the

development of low-cost and highly stable PSCs. In this context, open issues such as the

hysteresis in the current-voltage (I-V) curve, depended on the direction of the voltage scan,[8]

the light-soaking requirement,[9 , 10 ,11 ]

which is responsible for enhancing the PCE under

continuous illumination and the long-term operational instability of PSCs,[12,13]

need to be

effectively resolved. Recent experimental studies have demonstrated that the anomalous

hysteresis response of PSCs is probably due to the migration of the moving ion species,[14]

the

light soaking effect is related to the accumulation of space charges and the formation of light-

activated meta-stable trap states,[15]

while the observed lifetime degradation may be related to

various external issues such as moisture and oxygen presence, temperature increase, UV light,

as well as some intrinsic factors such as ion migration and interfacial reactions.[11]

Hysteresis

is currently one of the key issues in the normal planar structure PSC, while on the other hand

4

it has been successfully tackled through the adoption of the bulk heterojunction organic

photovoltaics structure in PSCs,[9]

the so-called planar inverted structure. However, the high

sensitivity of lead halide perovskites to ambient conditions (mainly humidity) and the

significant photo-charging effect during the first few minutes of the I-V measurement are still

significant bottlenecks towards the successful commercialization of the planar inverted PSCs.

One way to improve the photovoltaic properties and deal with the above issues of PSCs is

by the incorporation of solution processable graphene-based materials in the device

architecture, mainly aiming to improve the interface between the perovskite absorber and the

contacts and simultaneously extend the device operational stability.[16]

In particular, reduced

graphene oxide (rGO) and other solution processable graphene derivatives have been mainly

investigated in the normal (mesostructured and planar) perovskite photovoltaic structure to

increase the electron extraction from the perovskite to the compact TiO2 and suppress the

hysteresis effect.[17,18]

On top of that, recent studies demonstrated the positive function of

graphene-based materials in a way to increase the grain size of the perovskite absorber[19]

and

improve the operational stability of PSCs by using it as the hole transport layer (HTL)[20,21]

and exploiting its exceptional electrical, optical, and physical properties as well as its low-cost

solution-phase production techniques.[22]

However, there is no direct evidence regarding the

graphene-based materials beneficial role in perovskite/electron transport layer (ETL) interface

of planar inverted PSCs.

Here, we demonstrate for the first time the utilization of rGO as an additive in the PCBM

ETL of planar inverted PSCs, using both the CH3NH3PbI3-xClx perovskite deposited through

one step method and the CH3NH3PbI3 using a two-step spin coating process. It is firstly

shown that, the addition of rGO in the PCBM layer increases its conductivity, leading to

higher short circuit current density (Jsc) and fill factor (FF) values, while it simultaneously

reduces the surface traps and passivates the perovskite surface, resulting in higher open circuit

voltage (Voc) and reduced light soaking effect. Secondly, rGO stabilizes the PCBM/perovskite

5

interface with the degradation rate significantly suppressed, making the rGO doped devices

highly stable under continuous solar illumination in ambient conditions. The resulting two

step solution processed CH3NH3PbI3 PSCs exhibit an average hysteresis-free PCE of 14.5%,

compared to 12.9% PCE of the reference cell, where all the device layer components, except

the back metal contact, were deposited at room temperature through solution process.

Results and Discussion

2.1 Characterization of the CH3NH3PbI3−xClx and rGO thin films

The as crystallized mixed halide perovskite, CH3NH3PbI3−xClx, on glass/ITO/PEDOT:PSS

substrate was fully characterized. To avoid any misconception, in the case of

CH3NH3PbI3−xClx perovskite absorber, Cl, that was the lead source for the crystal formation,

is not entered in the lattice of the perovskite, but acts as a crystallisation inhibiter.[23,24]

Figure

S1a demonstrates the typical X-Ray diffraction (XRD) pattern of a mixed halide perovskite

fabricated using the typical reported method,[25]

with the 14.15o and 28.47

o peaks ascribed to

(110) and (220) lattice plain, confirming the as expected cubic perovskite phase.[26]

The top

view morphology of the as-crystallized CH3NH3PbI3−xClx can be seen in the scanning electron

microscopy (SEM) images (Figure S1b). It is obvious that a compact perovskite layer with

only a few pinholes and with quite large grains was obtained. The perovskite grains

distribution (Figure S1c) was extracted by fitting the data (SEM), presenting an average size

of 325 nm. The UV-Vis absorption spectra is plotted together with the sharp

photoluminescence spectra at ~770 nm (Figure S1d) to further confirm the proper formation

of the mixed halide perovskite onto ITO/PEDOT:PSS substrate.[23]

The as-fabricated rGO was studied and compared to the pristine GO in terms of the

chemical bonding and the reduction degree, by X-ray photoelectron spectroscopy (XPS).

Figure S2 shows the XPS C1s spectra of GO and rGO. The peaks for GO clearly indicate a

substantial degree of oxidation with four components corresponding to carbon atoms in

different functional groups: the non-oxygenated C at ~285 eV (C=C/C-C), the

6

hydroxyl/epoxy groups at ~286.5 eV (C-OH/C-O), the carbonyl groups at ~288 eV (C=O),

and the carboxylate carbon at ~289 eV (O=C-OH).[ 27 ]

XPS analysis revealed that a

significantly larger proportion of oxygen in GO existed in the form of hydroxyl/epoxy groups

and fewer proportion of oxygen was associated with C=O and O=C-OH functionalities.

Compared to GO, the rGO C1s spectrum clearly exhibited much smaller related to oxygen

peaks, indicating an efficient restoration of the sp2 C-C bonds.

2.2 Optical and Electrical Characterization of the PSCs

To assess the viability of rGO in the planar inverted PSC, devices with architecture of

ITO/PEDOT:PSS/CH3NH3PbI3−xClx or CH3NH3PbI3/PCBM or rGO:PCBM/PFN/Ag were

fabricated (Figure 1a). Figure 1b displays the current density-voltage (J-V) curves of the

reference and the rGO doped PCBM PSCs using the mixed halide perovskite as the solar

absorber. An important enhancement in the short-circuit current density (Jsc) from 20.65 to

22.92 mA cm−2

was observed, accompanied by a slight enhancement in the open-circuit

voltage (Voc) from 836 to 850 mV and the fill factor (FF) from 65.3 to 65.8%. As a result, a

~14% enhancement in PCE was obtained (Table 1 and Figure 1c). After systematic

investigation of the effect of rGO concentration in PCBM on PSCs performance, we

concluded that a concentration of 5% v/v rGO was the optimum, in terms of PCE. While,

further spectroscopic characterization and analysis was conducted in the devices incorporating

the rGO:PCBM ETL with the optimum concentration of rGO (5% v/v), in direct comparison

with the pristine ones. Figure S3a,b present the J-V characteristic plotted for both the forward

and reverse scan, showing almost no discrepancy among the device parameters, indicating the

absence of the hysteresis effect for both devices tested. The standard deviation of the

photovoltaic parameters for the pristine and the rGO doped PSCs are shown in Figure S4. In

order to extract the standard deviations 5 identical photovoltaic devices for each type were

prepared consisting of 4 solar cells. The detailed statistics clearly display that all the

photovoltaic parameters were improved when the rGO was introduced in the PCBM ETL. It is

7

important to mention, that the solar cells tested were measured at a typical scan rate bias of 10

mV s−1

and without the application of any preconditioning procedure prior to the J-V

measurement.[28]

Figure 1 and Table 1

Figure 1. (a) Schematic device architecture of the fabricated planar inverted perovskite solar

cells, (b) The J-V curves of perovskite solar cells based on 5% rGO-doped PCBM and PCBM

ETL measured under AM 1.5G (100 mW cm-2) illumination, (c) PCE distribution of the

devices with and without rGO in the PCBM ETL extracted from 20 identical devices.

Table 1. Summary of the average photovoltaic characteristics of inverted CH3NH3PbI3-xClx

based planar perovskite solar cells with and without the addition of rGO in PCBM.

rGO

Concentration (%)

Jsc (mA/cm2) Integrated Jsc (mA/cm2)

(IPCE)

Voc (mV) FF (%) PCE (%)

(max.)

0 20.65 20.09 836 65.3 11.27 (12.30)

2.5 21.33 844 65.8 11.85 (12.76)

5 22.92 22.32 850 65.8 12.82 (13.50)

7.5 22.24 866 64.7 12.46 (13.15)

8

The interface between the semiconductror material and the metal electrode plays a crucial

role for the proper operation of solar cells, and in particular in PSCs in either the normal or

the inverted structure. Accordingly, the optimization of the perovskite/ETL/Metal Contact

interface is vital for the passivation of the pinholes formed during the crystallization process

of the perovskite absorber, and in turn for the passivation of the of anion vacancies at the

interface ETL/Perovskite. An ohmic contact is also required for the effective transport of the

photoexcited eletrons from the perovskite to the metal electrode. For this purpose, we

obtained AFM images (Figure S5) of the neat perovskite thin film as well as for the

perovskite film passivated with PCBM and rGO:PCBM ETL. It was observed that, following

the coating of the CH3NH3PbI3-xClx thin film with PCBM the root-mean-square (RMS) value

of surface roughness was significantly decreased from ~31 to ~10 nm. Moreover, following

the doping of PCBM with rGO the RMS was further decreased to ~9 nm, which is an

indication that the rough perovskite film became smoother when it is coated by rGO:PCBM

ETL compared to that coated by pure PCBM ETL. The graphene based ETL may also

passivate the grain-boundary of the perovskite and thus reduce the surface traps to mitigate

carrier recombination, since we have observed a slight increase in Voc of the rGO-based

devices.[ 29 ]

The improved morphology ensured efficient coverage of the perovskite film

together with a smaller series resistance measured for the graphene based device (4.92 Ω cm2)

compared to the reference device (5.47 Ω cm2), explaining also the slight improvement in the

FF value of the former.

To confirm the enhancement of the Jsc in the graphene-based device, the external quantum

efficiency (EQE) spectra of CH3NH3PbI3-xClx PSCs, with and without the addition of the

optimum concentration of rGO in PCBM (5% v/v) was measured (Figure 2a). It is clear, that

following the incorporation of rGO in the ETL the EQE exhibits a broad and almost uniform

increase in the entire active spectrum of the device. It is reported that the EQE is the product

of light harvesting efficiency (ηlh), charge injection/transfer efficiency (ηinj), and charge

9

collection efficiency (ηcc).[30]

In our case, the enhancement of the rGO based device quantum

efficiency can be attributed to the improved ηinj and ηcc; ηlh is the same for both cell types

studied, since the perovskite absorber is not affected by the addition of rGO in the PCBM

ETL. Another point to be noted, is that the integrated Jsc from EQE is less than 3% different

from the actual measured Jsc values (Table 1), indicating good accuracy of our electrical

measurements.

Τo get an insight into the charge extraction properties of the photogenerated carriers from

the hybrid perovskite to the two different ETLs, the samples’ steady state photoluminescence

(PL) spectra were measured and analysed as shown in Figure 2b (suggestion: I think Fig2b

will be better presented if the y-axis is normalized. In this way, someone can see the

presentage drop of the PL intensity). The PL spectra were collected from perovskite films

fabricated on glass/ETL substrates. It is evident that, the CH3NH3PbI3-xClx perovskite film

deposited on 5% rGO-doped PCBM shows significant PL quenching, compared to the pristine

PCBM films, proving that the rGO doping has successfully enhanced the rate of carrier

extraction at the ETL/perovskite interface.[31,3]

Nearly indistinguishable spectra were recorded

at different areas, suggesting reproducible and homogeneous PL properties. The PL results

further support our findings from the I-V and EQE spectra, suggesting that upon the addition

of rGO in the PCBM ETL, the Jsc was significantly enhanced.

To better understand the origin of the performance enhancement in the graphene-based

inverted PSC, the I-V characteristics were measured in sandwich cells composed by

ITO/ETL/Au, with PCBM or rGO:PCBM used as ETLs (Figure 2c). The DC (direct current)

conductivity (σ0) can be determined from the slope of I-V plot, using the equation I = σ0Ad-

1V,

[30,31] where A is the area of sample (0.04 cm

2) and d is the thickness of sample,

respectively. The thicknesses of PCBM and rGO:PCBM layers were ~100 nm. The

conductivity of pristine PCBM was 0.109±0.005 mS cm-1

, whereas the rGO doped PCBM

film presented ~5 fold higher conductivity (0.495 ± 0.001 mS cm-1

). As a result, it is expected

10

that the photogenerated charge carriers in CH3NH3PbI3-xClx based absorber are more

efficiently transported to the rGO:PCBM electron conductor than to the pristine PCBM. The

higher conductance of the graphene-based ETL justifies the observed PL quenching,

originating from the improved charge extraction from the perovskite-based absorber towards

the ETL.

Figure 2. (a) External quantum efficiency (EQE) spectra of the optimal rGO:PCBM and

PCBM based devices, (b) Photoluminescence (PL) spectra (excitation at 543 nm) of

CH3NH3PbI3-xClx/PCBM and rGO:PCBM/glass substrates , (c) I-V characteristics of

ITO/PCBM/Au (black squares) and ITO/rGO:PCBM/Au (red spheres) devices.

As already reported, the light soaking effect (LSE) taking place at the first few minutes of

solar illumination is another unsolved and thus critical issue in the PSCs technology. In our

case, and without any preconditioning procedure in our measurements, the Jsc, FF and PCE of

the reference device presented significant LSE (Figure S6a), before reaching the steady

state.[ 32 ]

In contrast, Voc reaches almost its maximum value from the beginning of the

measurement, attribured to the presence of the poly[(9,9-bis(3’-(N,N-dimethylamino)propyl)-

11

2,7-fluorene)-alt-2,7-(9,9-dioctylfluorene) (PFN) interlayer. This is a very important finding

since in the majority of past studies Voc was shown to increase significantly with time before it

reaches the maximum value,[32]

a phenomenon we also observed without the presence of PFN

interlayer (not shown). Moreover, following the incorporation of rGO into the PCBM ETL,

the LSE, showed by the respective J-V curves, is reduced by a factor of ~3, as shown in

Figure S6b. This result can be directly correlated with the higher conductivity of the rGO

doped PCBM ETL compared to the pristine one (Figure 2c), which implies that the

photoexcited charges could possibly be transferred faster from the perovskite to the back

contact and/or at the interface between the perovskite and the ETL.[32]

Our rGO doped device

exhibits one of the lowest reported photo-charging effects in PSCs, considering the absence of

preconditioning procedures, reaching its stabilizedd PCE value within a few I-V cycles. This

is a very important result towards the PSC commercialization, providing the potential for

improving the stability of such high efficient low cost solar cells.

To further shed light on the photoexcited carrier dynamics in our devices we have

performed femto-second transient absorption spectroscopy (TAS) of the glass/PCBM and

rGO:PCBM/CH3NH3PbI3-xClx systems. TAS has been recently demonstrated to be a useful

tool to detect charge separation and electron injection in PSCs.[33,34]

As it can be seen in

Figure 3a,b, where the respective TAS spectra are displayed, our samples presented two main

features: a photoinduced bleaching at ~ 736 nm and a photoinduced transient absorption at ~

800 nm.[33]

To study the relaxation dynamics, the decay kinetics of the related band edge

bleaching were measured (Figure 3c), and the kinetic fits obtained are summarized in Table

2. It is widely known from previous studies, that the photoinduced bleaching band dynamics

will follow the electron and hole population dynamics in the hybrid perovskite semiconductor

as it originates from the transparency induced at the onset of the optical absorption after

population of the bottom of the conduction band and top of the valence band by

photogenerated electrons and holes, respectively.[3]

Also in our case, the bleaching band at

12

736 nm is directly correlated with the onset in the ground state absorption spectra (Figure S

1d), located at practically the same spectral region. Following calculation of the average

lifetime of the photo-bleaching decay, we observe that a significantly shorter time (152 ps) is

required for the carriers in the rGO based sample to relax compared to the pristine sample

(548 ps). Specifically this average lifetime corresponds, to the time required for an excited

electron to be separated from a hole,[35]

and be inject to the ETL.[33]

Accordingly, upon the

addition of rGO in PCBM ETL, the photoexcited electrons exhibit the tendency to inject

significantly faster from the photo-excited perovskite to the graphene based ETL. It may be

postulated that the rGO additive functions as a bridge for charge injection from the perovskite

towards the ETL. The faster decay kinetics can also be correlated with the reduced LSE of the

rGO based device as well as with the increased conductivity of the rGO doped PCBM ETL.

Figure 3. fs-TAS spectra, relative optical density as a function of wavelength, at various time

delays after photoexcitation of (a) pristine and (b) rGO doped PCBM ETLs. The broad

negative band with minima at ~ 736 nm for both samples is associated to the photo-bleaching

(PB) of the material due to the presence of hot carriers. (c) Transient band edge of bleach

(symbols) and their decay fits (lines) for perovskite with (cyan) and without (green) rGO in

the PCBM respectively.

13

Table 2. Kinetic Fit Parameters of TA bleaching for samples with and without rGO in PCBM.

Sample λprobe (nm) τ1 (ps) τ2 (ps) <τ> (ps)

Without rGO in PCBM 736 39.84 796.08 548.44

With rGO in PCBM 736 43.65 303.13 152.23

2.3 Ambient Stability Tests of the PSCs

Apart from the improved PCE, the incorporation of rGO in PCBM ETL affects the device

stability as well (Figure 4). It has been recently demonstrated that the PCBM layer has very

low fracture energy and thus planar inverted PSCs based on PCBM ETL can easily

degrade.[36]

Indeed, compared to the reference device, the rGO based devices preserved an

almost 5-fold higher PCE after ~50 h of continuous solar illumination at high levels of

relative humidity (RH) (>50%). In Figure 4, the Jsc, Voc and FF were plotted independently to

provide a better understanding on the characteristics of this remarkable difference in stability

between the two cell types. The most prominent difference is observed in the plots of Voc and

FF against illumination time. In the reference device, a constant decay of the Voc, which is the

fingerprint of the perovskite degradation, was observed, while on the other hand the rGO

doped device exhibited an almost unchanged Voc after more than 100 h of continuous light

soaking under ambient conditions. The above results suggest a perfect and very stable

ETL/perovsktite interface that efficiently retards the CH3NH3PbI3-xClx degradation. Indeed,

focusing on the FF degradation rate, which is determined by the changes of the interface

between the perovskite and the hole/electron transport layer, one can clearly observe a fast FF

decay in the reference device, contrary to the rGO doped device, that exhibited an extremely

stable FF. One way to understand this important improvement in stability could be that the

addition of rGO enhances the fracture resistance of the PCBM layer. Indeed, rGO has already

shown the tendency to enhance the fracture resistance when combined with other materials.[37]

It can be concluded that doping of PCBM with rGO gives rise to the passivation of the

14

perovskite top surface and in turn, to improved stabilization of the ETL/perovskite interface

against photodegradation.

Figure 4. Evolution of normalized (a) PCE, (b) Jsc, (c) Voc and (d) FF of PCBM (black) and

rGO:PCBM (red) ETLs based PSCs under continuous solar illumination in ambient

conditions (~50% RH).

2.4 Universality of the method

To prove the universality of our approach, solar cells based also on the CH3NH3PbI3 were

fabricated using the sequential deposition technique.[38]

The rest of layers comprising the

sandwich like architecture were kept identical to the mixed halide device. Figure 5 shows, the

J-V characteristics of planar inverted PSCs using the two different ETLs, one consisting of

pristine PCBM and the other of rGO doped (5% v/v) PCBM. It is clear that the PCE and in

general all the photovoltaic parameters, were significantly enhanced in the latter case and

following the same trend as in the mixed halide perovskite case. In particular, the main reason

for the PCE enhancement is again the Jsc improvement, which can be attributed to the higher

15

conductivity of the rGO:PCBM ETL compared to the pristine PCBM one; moreover both Voc

and FF follow again the same trend, i.e. show a slight increase. Following the systematic

optimization for the sequentially deposited CH3NH3PbI3 fabrication, a higher average PCE

was obtained for the reference device (12.9% for CH3NH3PbI3 compared to 11.3% for the

CH3NH3PbI3-xClx). This is mainly attributed to the significantly high Voc of 926 mV,

suggesting the high quality of the fabricated perovskite absorber. On the other hand, the rGO

doped device, exhibited an average PCE of 14.5%, which is one of the highest PCEs ever

obtained for low temperature, fully solution processable planar inverted perovskite solar cells,

exhibiting no hysteresis.

Figure 5. The J-V curves of two step processed CH3NH3PbI3 based PSCs with and without

rGO in the PCBM ETL measured under AM 1.5G (100 mW cm-2

) illumination

3. Conclusion

In summary, rGO was successfully added in the PCBM ETL of both fully solution

processable CH3NH3PbI3−xClx and CH3NH3PbI3 based planar inverted PSCs. The introduction

of rGO in PCBM resulted in a hysteresis-free and high efficient (14.5%) PSC, with all the

device component layers deposited at room temperature and treated below 120 oC. On top of

that, significantly extended stability in prolonged illumination under ambient conditions and

reduced light soaking effect compared to the reference device was also identified. The PCE

16

enhancement was mainly due to the significantly increased Jsc of the graphene-based device,

attributed to the improved conductivity of the rGO:PCBM ETL and the more effective

extraction of photoexcited electrons from the perovskite to the back metal contact, which also

explains and the reduced light soaking effect. On the other hand, the improved stability of the

rGO doped devices was due to the stabilization of ELT/perovskite interface, which in turn

results in the stabilization of perovskite crystal. This is also demonstrated by the stable values

in Voc and FF of the graphene based device after intense ageing stress. Our work opens new

pathways for efficiency and stability optimization of PSCs, perovskite light emitting diodes,

perovskite photodetectors and other novel perovskite based devices using low cost, solution

processable graphene-based materials, deposited and treated at low temperature, giving the

opportunity for the development of such high efficient devices in flexible, ultra-light and

flexible substrates.

Experimental Section

Preparation of lead halide Perovskite Precursor Solutions: The mixed halide perovskite

(CH3NH3PbI3-xClx) precursor solution was prepared by mixing methylammonium iodide

(MAI, Dyesol) with lead(II) chloride (PbCl2, 99.999% Sigma Aldrich), molar ratio of 1:3, in

anhydrous N,N-dimethylformamide (DMF). The final concentration was 40 wt% and the

prepared solution was stirred overnight at 70oC. For the fabrication of CH3NH3PbI3

perovskite, 450 mg lead(II) iodide (PbI2, 99.999% Sigma Aldrich) were dissolved in 1 ml

anhydrous DMF and 45 mg of MAI were dispersed in 1 ml anhydrous 2-propanol. Both

solutions were kept at 70 oC and stirred overnight in separate vials.

Preparation of Graphene Oxide (GO):[ 39 ]

GO was prepared from graphite powder (Alfa

Aesar, ~200 mesh) according to a modified Hummers’ method. In more detail, graphite

powder (0.5 g) was placed into a mixture of sulfuric acid, H2SO4 (40 mL, 98%) and sodium

nitrate, NaNO3 (0.375 g). The mixture was then stirred and cooled in an ice bath. While

maintaining vigorous stirring, potassium permanganate, KMnO4 (3.0 g) was then added in

17

portions over a period of 2 h. The reaction mixture was left for 4 h in order to reach room

temperature before being heated at 35 °C for 30 min. It was then poured into a flask

containing deionized water (50 mL) and further heated at 70 °C for 15 min. The mixture was

then decanted into 250 mL of deionized water and the unreacted KMnO4 was removed by

adding 3% hydrogen peroxide, H2O2. The reaction mixture was then allowed to settle and

decanted. The obtained graphite oxide was purified by repeated centrifugation and redispersed

in deionized water until a neutralized pH was achieved. Finally, the resulting GO was dried at

60 °C in a vacuum oven for 48 h before use.

Preparation of reduced graphene oxide (rGO):[40]

The reduction of GO was performed using

a mixture of hydriodic acid (55%)/acetic acid (HI/AcOH). In detail, the as prepared GO

powder (0.1 g) was sonicated in AcOH (37 mL) for 2 h. HI (2 ml) was then added and the

mixture was stirred at 40 °C for 40 h. After being isolated by filtration, the product was

washed through a three step procedure using saturated sodium bicarbonate, NaHCO3 (3×2.5

mL), distilled water (3×2.5 mL) and acetone (2×2.5 mL). Finally, the resulting rGO was dried

at 60 °C in a vacuum oven overnight. A solution of rGO in chlorobenzene with a resulting

concentration of 0.055 mgmL-1

was prepared through sonication for 2 h.

Device Fabrication: The PSC reported were fabricated on pre-patterned 20 x 15 mm indium-

tin-oxide (ITO) coated glass substrates with a sheet resistance of ~20 Ω sq-1

. The impurities

were removed from the ITO glass through a 3-step cleaning process (detergent deionized

water, acetone, isopropanol). Before the deposition of the hole transport layer (HTL), the

substrates were placed inside an ultraviolet ozone cleaner in order to remove the organic

contamination and increase the surface hydrophilicity of ITO coated substrates. Afterwards,

poly(ethylene-dioxythiophene) doped with poly(4-styrenesulfonate) (PEDOT:PSS, Heraeus)

was spin-casted from an aqueous solution on the ITO substrate (4000 rpm, 60 s). The average

thickness of the layer was 30 nm, followed by baking for 15 min in ambient conditions and

for 30 min inside a nitrogen-filled glove box at 120 °C. The substrates were cooled down to

18

room temperature, and either the MAI:PbCl2 precursor solution was spin-coated at 4000 rpm

for 30 s followed by a thermal annealing at 100 °C for 75 min to form the CH3NH3PbI3-xClx

perovskite or through the two-step spin coating deposition method, PbI2 solution was spin

coated on the PEDOT:PSS substrates at 6000 rpm for 35 s and were dried at 100 oC for 10

min. Subsequently, MAI was spin coated at 6000 rpm for 35 s and annealed at 100 oC for 40

min to form the CH3NH3PbI3 perovskite based absorber. The procedures for the fabrication of

both perovskites were conducted inside a nitrogen filled glove box. Before the ETL

deposition, the substrates were again cooled down to room temperature. Different

concentrations of rGO (0, 2.5, 5 and 7.5%) were added to the PCBM (Solenne B.V), having

an initial concentration of 20 mg ml-1

in chlorobenzene and stirred for at least 5 h. Afterwards

the different ETL solutions were spin-coated at 1000 rpm for 45 s and the films were left to

dry for ~30 min in a closed petri dish under inert atmosphere. Subsequently, an as-prepared

solution, consisting of 0.4 mg ml-1

PFN in methanol with a small amount of acetic acid, was

spin-coated at 2000 rpm for 45 s to form an ultra-thin interlayer. Lastly, 100 nm of silver was

deposited through a shadow mask by thermal evaporation to define an active area of 4 mm2

for each device.

Characterization and measurements: The performances of the devices were measured under

inert atmosphere with an Air Mass 1.5 Global (A.M. 1.5 G) solar simulator at an intensity of

100 mW cm-2

using an Agilent B1500A Semiconductor Device Analyzer. A reference

monocrystalline silicon solar cell from Newport Corp. was used to calibrate the light intensity.

For the degradation study, the as-fabricated cells were exposed in continuous solar irradiation,

using an A.M. 1.5 G solar simulator, under ambient conditions with relative humidity (RH)

constantly above 50%. To obtain the plot data, the devices were placed again in a nitrogen

filled glove box and measured until their stabilization. Afterwards, the cells were exposed in

A.M. 1.5 G under ambient conditions to continue to the degradation study. This process was

repeated until the end of the measurements. The external quantum efficiency measurements

19

were conducted immediately after device fabrication using an integrated system (Enlitech,

Taiwan) and a lock-in amplifier with a current preamplifier under short-circuit conditions.

The light spectrum was calibrated using a monocrystalline photodetector of known spectral

response. The PSCs were measured using a Xe lamp passing through a monochromator and

an optical chopper at low frequencies (~200 Hz) in order to maximize the signal/noise (S/N)

ratio. Micro-photoluminescence (μPL) studies at 295K were performed using a setup in

backscattering geometry, with a He-Ne 543 nm continuous wave laser as an excitation source.

With a microscope objective lens (Mitutoyo 50X) the laser beam was focused down to 1 μm

on the sample, placed on an XYZ translation stage, at normal incidence. A spatial filter

system was used to obtain the central part of the beam and acquire a uniform energy

distribution. In a typical μPL experiment, different excitation positions of the samples were

checked with low laser power (controlled by a neutral density filter). UV-vis absorption

spectra were recorded using a Shimadzu UV-2401 PC spectrophotometer over the wavelength

range of 300-800 nm. The size distribution and the morphology of the as-fabricated

CH3NH3PbI3-xClx was characterized by scanning electron microscope (SEM JEOL JSM-

7000F). The crystallographic properties of CH3NH3PbI3-xClx film on the ITO/PEDOT:PSS

substrate was investigated using a D/MAX-2000 X-ray diffractometer under monochromated

Cu Kα irradiation (λ=1.5418 Å) at a scan rate of 4° min−1

. The surface analysis studies on GO

and rGO were performed in a ultra-high vacuum chamber (P<10-9 mbar) equipped with a

SPECS LHS-10 hemispherical electron analyzer. The XPS measurements were carried out at

room temperature using un-monochromatized AlKa radiation under conditions optimized for

maximum signal (constant ΔΕ mode with pass energy of 36 eV giving a full width at half

maximum (FWHM) of 0.9 eV for the Au 4f7/2 peak). The analyzed area was an ellipsoid with

dimensions 2.5 x 4.5 mm2. The XPS core level spectra were analyzed using a fitting routine,

which allows the decomposition of each spectrum into individual mixed Gaussian-Lorentzian

components after a Shirley background subtraction. The samples were coated on a glass

20

substrate with dimensions 1x1cm2. Wide Scans were recorded for both samples, while the C1s

and O1s core level peaks were recorded in detail. Errors in our quantitative data are found in

the range of ~10%, (peak areas). The femtosecond transient absorption spectroscopy (TAS)

measurements were performed using the Newport transient absorption spectrometer. The

incident to the instrument laser pulse is generated from a Yb:KGW-based laser system

(PHAROS, Light Conversion). The pulse duration used was 170 fs at 1026 nm wavelength

and repetition rate of 1 KHz. The initial laser beam splits into two beams, the pump (90% of

the incident), which excites the sample and the probe (10% of the incident), which generates a

white light supercontinuum probe, routed with reflective optics to minimize chirp. In a typical

pump-probe experiment, the sample was excited by the pump pulse and the dynamics of the

sample’s relative optical density is monitored as a function of wavelength at various time

delays after photoexcitation. The probe light is sensed as a function of wavelength by the

combination of a fiber coupled multichannel detector and imaging spectrograph. A

biexponential equation of the form 𝑦=𝑦0+𝐴1exp(−𝑥/𝜏1)+𝐴2exp(−𝑥/𝜏2) was used to fit the

normalized excited state decay kinetics of both samples. The coefficients 𝐴1 and 𝐴2 represent

the weighted contribution of each exponential component to the overall kinetics, and 𝜏1 and 𝜏2

are the time constants of the two exponential decay elements. The average lifetime at the

specified spectral position for film samples was calculated using the equation <𝜏>

=𝐴1𝜏1+𝐴2𝜏2.

Supporting Information

Supporting Information is available from the author.

Acknowledgements

This project has received funding from the European Union’s Horizon 2020 research and

innovation programme under grant agreement No. 696656 – GrapheneCore1.

21

[1] S. D. Stranks, H. J. Snaith, Nat. Nanotechnol. 2015, 10, 391

[2] N. K. Kumawat, M. N. Tripathi, U. Waghmare, D.Kabra, J. Phys. Chem. A 2016, 120,

3917.

[3] S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou, M. J. P. Alcocer, T. Leijtens, L.

M. Herz, A. Petrozza, H. J. Snaith, Science 2013, 342, 341.

[4] Y. H. Lee, J. Luo, R. Humphry-Baker, P. Gao, M. Grätzel, M. K. Nazeeruddin, Adv.

Funct. Mater. 2015, 25, 3925.

[5] T.-B. Song, Q. Chen, H. Zhou, C. Jiang, H.-H. Wang, Y. Yang, Y. Liu, J. You, Y. Yang,

J. Mater. Chem. A 2015, 3, 9032.

[6] A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, J. Am. Chem. Soc. 2009, 131,

6050.

[7] NREL Best Research-Cell Photovoltaic Efficiency Chart

[8] H. J. Snaith, A. Abate, J. M. Ball, G. E. Eperon, T. Leijtens, N. K. Noel, S. D. Stranks, J.

Tse-Wei Wang, K. Wojciechowski, W. Zhang, J. Phys. Chem. Lett. 2014, 5, 1511.

[9] P. Docampo , J. M. Ball , M. Darwich , G. E. Eperon , H. J. Snaith , Nat. Commun.

2013, 4, 2761 .

[10] A. T. Barrows, A. J. Pearson, C. K. Kwak, A. D. F. Dunbar, A. R. Buckley, D. G.

Lidzey, Energy Environ. Sci. 2014, 7, 2944.

[11] C. Zhao, B. Chen, X. Qiao, L. Luan, K. Lu, B. Hu, Adv. Energy Mater. 2015, 5, 1500279

[12] A. Mei, X. Li1, L. Liu, Z. Ku, T. Liu, Y. Rong, M. Xu, M. Hu, J. Chen, Y. Yang, M.

Grätzel, H. Han, Science 2014, 345, 295.

[13] T. A. Berhe, W.-N. Su, C.-H. Chen, C.-J. Pan, J.-H. Cheng, H.-M. Chen, M.-C. Tsai, L.-

Y. Chen, A. A. Dubaleb, B. –J. Hwang, Energy Environ. Sci. 2016, 9, 323.

[14] S. Meloni, T. Moehl, W. Tress, M. Franckevičius, M. Saliba, Y. H. Lee, P. Gao, M. K.

Nazeeruddin, S. M. Zakeeruddin, U. Rothlisberger, M. Graetzel, Nat. Commun. 2016, 7,

10334.

22

[15] Wanyi Nie, Jean-Christophe Blancon, Amanda J. Neukirch, Kannatassen Appavoo,

Hsinhan Tsai, Manish Chhowalla, Muhammad A. Alam, Matthew Y. Sfeir, Claudine Katan,

Jacky Even, Sergei Tretiak, Jared J. Crochet, Gautam Gupta, Aditya D. Mohite. Nat.

Commun. 2016, 7, 11574

[ 16 ] N. Balis, E. Stratakis, E. Kymakis, Mater. Today 2016, DOI:

10.1016/j.mattod.2016.03.018

[17] G. S. Han, Y. H. Song, Y. U. Jin, J.-W. Lee, N.-G. Park, B. K. Kang, J.-K. Lee, I. S.

Cho, D. H. Yoon, H. S. Jung, ACS Appl. Mater. Interfaces 2015, 7, 23521.

[18] A. Agresti , S. Pescetelli , L. Cinà , D. Konios , G. Kakavelakis , E. Kymakis , A. Di

Carlo, Adv. Funct. Mater. 2016, 26, 2686.

[19] M. Hadadian, J.-P. Correa-Baena, E. K. Goharshadi, A. Ummadisingu, J.-Y. Seo, J. Luo,

S. Gholipour, S. M. Zakeeruddin, M. Saliba, A. Abate, M. Grätzel, A. Hagfeldt, Adv. Mater.

2016, DOI: 10.1002/adma.201602785

[20] W. Li, H. Dong, X. Guo, N. Li, J. Li, G. Niu, L. Wang, J. Mater. Chem. A 2014, 2,

20105.

[21] A. L. Palma, L. Cinà, S. Pescetelli, A. Agresti, M. Raggio, R. Paolesse, F. Bonaccorso,

A. Di Carlo, Nano Energy 2016, 22, 349.

[22] K. S. Novoselov, V. I. Falko, L. Colombo, P. R. Gellert, M. G. Schwab, K. Kim, Nature

2012, 490, 192.

[23] W. Zhang, M. Saliba, D. T. Moore, S. K. Pathak, M. T. Hörantner, T. Stergiopoulos, S.

D. Stranks, G. E. Eperon, J. A. Alexander-Webber, A. Abate, A. Sadhanala, S. Yao, Y. Chen,

R. H. Friend, L. A. Estroff, U. Wiesner, H. J. Snaith, Nat. Commun. 2014, 6, 6142.

[24] M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami, H. J. Snaith, Science 2012, 338,

643

[25] S. Colella, E. Mosconi, P. Fedeli, A. Listorti, F. Gazza, F. Orlandi, P. Ferro, T. Besagni,

A. Rizzo, G. Calestani, G. Gigli, F. De Angelis, R. Mosca, Chem. Mater. 2013, 25, 4613.

23

[26] Y. Li, W. Sun, W. Yan, S. Ye, H. Peng, Z. Liu, Z. Bian, C. Huang, Adv. Funct. Mater.

2015, 25, 4867.

[27] D. R. Dreyer, S. Park, C. W. Bielawski, R. S. Ruoff, Chem. Soc. Rev. 2010, 39, 228.

[28] M. De Bastiani, G. Dell'Erba, M. Gandini, V. D'Innocenzo, S. Neutzner, A. R. S.

Kandada, G. Grancini, M. Binda, M. Prato, J. M. Ball, M. Caironi, A. Petrozza, Adv. Energy

Mater. 2016, 6, 1501453

[29] Y. Shao, Z. Xiao, C. Bi, Y. Yuan, J. Huang, Nat. Commun. 2014, 5, 5784.

[30] J. H. Heo, H. J. Han, D. Kim, T. K. Ahn, S. H. Im, Energy Environ. Sci. 2015, 8, 1602.

[31] B.-X. Chen, H.-S. Rao, W.-G. Li, Y.-F. Xu, H.-Y. Chen, D.-B. Kuang, C.-Y. Su, J.

Mater. Chem. A 2016, 4, 5647.

[32] W. Nie1, J.-C. Blancon, A. J. Neukirch, K. Appavoo, H. Tsai, M. Chhowalla, M. A.

Alam, M. Y. Sfeir, C. Katan, J. Even, S. Tretiak, J. J. Crochet, G. Gupta, A. D. Mohite, Nat.

Commun. 2016, 7, 11574

[33] L. Wang, C. McCleese, A. Kovalsky, Y. Zhao, C. Burda, J. Am. Chem. Soc. 2014, 136,

12205.

[34] K. G. Stamplecoskie, J. S. Manser, P. V. Kamat, Energy Environ. Sci. 2015, 8, 208.

[35] J. S. Manser, P. V. Kamat, Nat. Photonics 2014, 8, 737.

[36] B. L. Watson, N. Rolston, K. A. Bush, T. Leijtens, M. D. McGehee, R. H. Dauskardt,

ACS Appl. Mater. Interfaces, 2016, DOI: 10.1021/acsami.6b06164.

[37] N. Domun, H. Hadavinia, T. Zhang, T. Sainsbury, G. H. Liaghat S. Vahid, Nanoscale,

2015, 7, 10294.

[38] J. Burschka, N. Pellet, S.-J. Moon, R. Humphry-Baker, P. Gao, M. K. Nazeeruddin, M.

Grätzel, Nature 2013, 499, 316.

[39] D. Li, M. B. Muller, S. Gilje, R. B. Kaner, G. G. Wallace, Nat. Nanotechnol. 2008, 3,

101.

[40] I. K. Moon, J. Lee, R. S. Ruoff, H. Lee, Nat. Commun. 2010, 1, 73.

24

Supporting Information

Efficient and Highly Air Stable Planar Inverted Perovskite Solar Cells with Reduced

Graphene Oxide doped PCBM Electron Transporting Layer

George Kakavelakis,* Temur Maksudov, Dimitrios Konios, Ioannis Paradisanos, George

Kioseoglou, Emmanuel Stratakis, Emmanuel Kymakis*

Figure S1. (a) XRD pattern of CH3NH3PbI3-xClx/PEDOT:PSS/ITO substrate, (b) Top view

SEM images of ITO/PEDOT:PSS/perovskite thin film. Inset is a magnified image of the same

sample, (c) Size distribution of the as fabricated Perovskite on top of ITO/PEDOT:PSS

substrate extracted from the several SEM images, (d) Absorption (left axis) and PL (right

axis) of the mixed halide perovskite.

25

Figure S2. High-resolution XPS C1s spectra for (a) GO and (b) rGO.

Figure S3. J-V curves of the (a) reference and of the (b) rGO doped PSCs with both forward

(black) and reverse (red) scan of the applied voltage.

26

Figure S4. (a) PCE, (b) Jsc, (c) Voc, and (d) FF parameters based on devices with PCBM and

rGO:PCBM ETL. The data were extracted from 20 identical cells of each case.

27

Figure S5. AFM images of (a) neat Perovskite, (b) Perovskite passivated with PCBM and (c)

Perovskite passivated with rGO:PCBM, showing that roughness is decrease after the

incorporation of rGO in PCBM ETL.

Figure S6. J-V characteristics of the (a) reference and the (b) rGO based device before

(black) and after (red) the light soaking (stabilization).