Highly Efficient Perovskite Solar Cells with Gradient Bilayer Electron

Letter Cite This: Nano Lett. 2018, 18, 3969−3977

pubs.acs.org/NanoLett

Highly Efficient Perovskite Solar Cells with Gradient Bilayer Electron Transport Materials Xiu Gong,† Qiang Sun,† Shuangshuang Liu,† Peizhe Liao,† Yan Shen,† Carole Graẗ zel,‡ Shaik M. Zakeeruddin,‡ Michael Graẗ zel,‡ and Mingkui Wang*,† †

Downloaded via IOWA STATE UNIV on January 26, 2019 at 21:20:59 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Wuhan National Laboratory for Optoelectronics, School of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan, Hubei 430074, P. R. China ‡ Laboratoire de Photonique et interfaces (LPI), Ecole Polytechnique Federale de Lausanne, CH-1015 Lausanne, Switzerland S Supporting Information *

ABSTRACT: Electron transport layers (ETLs) with suitable energy level alignment for facilitating charge carrier transport as well as electron extraction are essential for planar heterojunction perovskite solar cells (PSCs) to achieve high open-circuit voltage (VOC) and short-circuit current. Herein we systematically investigate band offset between ETL and perovskite absorber by tuning F doping level in SnO2 nanocrystal. We demonstrate that gradual substitution of F− into the SnO2 ETL can effectively reduce the band offset and result in a substantial increase in device VOC. Consequently, a power conversion efficiency of 20.2% with VOC of 1.13 V can be achieved under AM 1.5 G illumination for planar heterojunction PSCs using F-doped SnO2 bilayer ETL. Our finding provides a simple pathway to tailor ETL/perovskite band offset to increase built-in electric field of planar heterojunction PSCs for maximizing VOC and charge collection simultaneously. KEYWORDS: Doping, electronic material, Fermi level, perovskite, photovoltaic, charge transfer

1. INTRODUCTION The power conversion efficiency (PCE) of emerging organic− inorganic halide perovskite solar cells (PSCs) has rapidly risen from an initial 3.8% efficiency in 2009 to a recently certified 22.7% due to their intriguing properties1,2 including high ambipolar carrier transport,3 large light absorption coefficients,4 and long charge diffusion lengths.5 The photovoltaic performance of planar heterojunction (PHJ) PSCs, using recently developed techniques such as interface, composition engineering, or latest incoming nanoscale perovskites, has been largely augmented to a level of being close to their mesoscopic counterparts.6−11 Indeed, PSCs hold potentials for producing high efficiency at low cost. Furthermore, the characteristics of semitransparent, flexible, thermal stability, and lightweight of hybrid perovskites have opened up paths to diverse applications for solar cells.11,12 The high PCE achieved for PHJ PSCs could be attributed to a large short circuit current (JSC, 22−23 mA cm−2 under standard testing conditions by considering light scatering for instance),13,14 which is approaching the theoretical value with deducting reflective and parasitic absorption losses (23.8 mA cm−2 for CH3NH3PbI3).15 However, we have observed that the open-circuit voltage (VOC, ∼1.1 V) remains lower than the predicated value of 1.32 V from the Shockley Queisser limit for perovskite compounds with band gap of 1.59 to 1.63 eV.16 Several approaches have been proposed to further increase the PCE of PHJ PSCs by boosting their VOC. The open-circuit © 2018 American Chemical Society

voltage of perovskite devices depends on several factors like the optical band gap of perovskite compounds, the quasi-Fermi level difference between electron transport layers (ETLs) and hole transport layers (HTLs), relative conduction/valence band energy levels, the generation rate of bound electron−hole pairs, and the dissociation probability of bound electron−hole pairs into free charge carriers and interfacial recombination as well as nonradiative recombination.16−18 Among these factors, the energy band difference (i.e., energy band offset) between the ETL/perovskite and HTL/perovskite interfaces offers the driving force (about 150 mV for each side) for efficient charge transfer, reducing interfacial and nonradiative recombination, and thus has a significant influence on the VOC. To achieve an ideal VOC, therefore, it is crucial to build up a functional interface with well-matched energy bands and superior electron mobility and conductivity to avoid excessive band offset while accelerating carrier extraction and transport. To date, the TiO2 nanocrystal (NC) thin film is one of the most often used ETLs in PSCs. However, the optoelectronic characteristics of TiO2 nanocrystal still exhibit some shortfalls such as low electron mobility (0.1−1 cm2 V−1 s−1) and conductivity (∼1.1 × 10−5 S cm−1). The PSCs using TiO2 as ETL are likewise sensitive to UV illumination, which has Received: April 11, 2018 Revised: May 9, 2018 Published: May 21, 2018 3969

DOI: 10.1021/acs.nanolett.8b01440 Nano Lett. 2018, 18, 3969−3977

Letter

Nano Letters

Figure 1. (a, b) TEM and HR-TEM images of F:SnO2 nanocrystals. The inset shows the selected-area electron diffraction. (c) Bandgap of (F:SnO2)180 and (F:SnO2)380 deposited on quartz substrates versus F content. Solid squares and solid lines indicate the raw data and fitted curves with Gauss function, respectively. (d) Electronic Fermi levels of (F:SnO2)180 and (F:SnO2)380 deposited on quartz substrate versus F content.

become one critical issue for practical application in terms of long-term stability.19−21 Therefore, several alternatives have been proposed as effective and promising ETLs to the conventional TiO2.22,23 For example, the semiconducting SnO2 with a wide optical band gap (3.6−4.0 eV) has attracted great attention because of its excellent chemical stability, high electron mobility, and conductivity that can be tailored easily through low cost processes such as deposition approach, doping, and composition engineering. Fang et al. first reported solution-processed SnO2 ETL for efficient planar PSCs showing an overall PCE over 17%, which was attributed to good antireflection and high electron mobility.24 Nonetheless, the PCE under forward voltage scans was reduced to 14.82% due to a large density of surface traps on the as-synthesized SnO2 and inefficient electron transfer at the SnO2/perovskite interface. Without further surface/interface modification, these traps eventually result in photocurrent hysteresis phenomenon and poor fill factor (FF).25−27 Lately, doping metal ions (Li+, Mg2+, Nb5+, Y3+, Sb5+, etc.) into the SnO2 ETL has led to improvement of the device photovoltaic performance.28−30 For instance, p-dopant such as Li+ or Mg2+ induces a downward shift of the conduction band (CB) minimum of SnO2 and thus enlarges the band offset to accelerate the injection and transfer of electrons from perovskite absorber to the ETL. To avoid VOC losses by the ETL, it must be assured that at open circuit the quasi-Fermi level of electrons set up under illumination in the perovskite film is equal or higher than that attained in the absence of the ETL. This implies that the rate of interfacial recombination of electrons injected into the ETL with the holes in the perovskite must not exceed their intrinsic rate of carrier recombination in the perovskite. Thus, it is imperative to develop new ETLs with low interfacial carrier recombination velocity. By tailoring the Fermi energy, some ETLs seem to meet this requirement by forming well-matched energy levels

with perovskite active layer while reducing interfacial carrier recombination (Table S1). Herein we reported a new ETL via fabrication of F-doped SnO2 (F:SnO2) nanocrystals using a facile low-temperature solution-process method for efficient n-i-p planar PSCs. We further demonstrated that the device VOC could be tailored by gradually moderation of the band offset at the interface of perovskite active layer and F-doped bilayer SnO2 ETL via tuning Fermi level of the latter.31,32 Grain boundary barriers and defects within the bilayer ETL can be largely restrained due to a matched lattice constant. Consequently, efficient PHJ PSC devices using bilayer ETL can be achieved with a PCE of 20.2%, a VOC of 1.13 V, a short-circuit current density (JSC) of 22.92 mA cm−2, and FF of 78%. The advantages of bilayer ETL synthesized with low-temperature processing can be easily foreseen. This represents an important proof of concept that paves the way to further realization of high-performance and large-scale production of PSCs based on bilayer transparent substrates with different carrier concentration.

2. RESULTS AND DISCUSSION The F:SnO2 films with different doping levels were prepared by spin-coating SnCl2·2H2O and NH4F in ethanol solution onto precleaned FTO substrates. Then the films were thermally annealed at 180 °C, 380 °C, and 500 °C for 1 h in air and were represented as (F:SnO2)180, (F:SnO2)380, and (F:SnO2)500, respectively. The (F:SnO2)180 films with 0.1 molar ratio of F/ Sn doping are denoted as (F:SnO2)180-0.1 for clarity and brevity serving as an example. Figure S1 presents XRD patterns of the as-synthesized pristine SnO2 and F:SnO2 films. Three major peaks located at 26.62°, 33.91°, and 51.76° correspond to (110), (101), and (211) diffraction planes, respectively, which are well-matched with tetragonal rutile SnO2.33 Moreover, no additional diffraction peaks from any other impurity phases were detected in the F:SnO2 thin films. This indicates the SnO2 3970

DOI: 10.1021/acs.nanolett.8b01440 Nano Lett. 2018, 18, 3969−3977

Letter

Nano Letters

Figure 2. (a) Flowchart of fabrication process. (b) Cross-sectional SEM image of a typical perovskite device based on F:SnO2..

concentration according to the Bursteine-Moss effect.38 However, further augmentation in F doping could narrow the band gap due to the many-body interaction effect between ionized impurities and free carriers.39 Figure S6a and b compare the transmission of the SnO2 and F:SnO2 film deposited on FTO substrates under different temperatures. The F:SnO2 film with 0.2 F doping showed a slightly higher transmittance than films without doping. This can be beneficial for device light harvesting. The thin films electrical parameters including carrier concentration (n), mobility (μ), and electrical conductivity (σ) were investigated as a function of F doping ratio (Figure S7 and Tables S3). As clearly shown in Figure S7b, the carrier concentration and electrical conductivity increases with the annealing temperature at the same F content. All these parameter values first increase and then decrease along with F doping under a constant temperature, and achieving maximum at 0.2 doping, except for electron mobility. This is consistent with the change of optical band gap. Clearly, an appropriate F doping can improve electrical properties of SnO2 films. This once again proves that replacing O ions by F ions leads to higher free electron concentration in films. The mobility depends heavily on ionized impurities and grain boundaries according to the scattering mechanism.40,41 The density of electrons (nc) in conduction band for semiconducting NCs is given by

lattice structure has not been disrupted by introducing F atoms. We found that the sample crystalline quality was improved with temperature but decreased with F-doping concentration (Figure S1). This could be caused by lower melting point of SnOF compound as well as by thermally promoted ion diffusion during crystal growth associated with incorporation of F into SnO2 NC lattices.34 Figure 1a and b present transmission electron microscopy (TEM) images of F:SnO2 NCs of 6−9 nm size that is evenly distributed. A clear lattice spacing of 0.332 nm, corresponding to (110) facet of rutile SnO2, can be observed in the high resolution TEM image of F:SnO2 NCs, which further demonstrates their high crystallinity. Moreover, the selected area electron diffraction pattern of F:SnO2 NCs also confirms tetragonal structure. Figure S2 shows mapping of Sn, O and F in the scanning electron microscope-energy dispersive spectrometer, further indicating the homogeneously distribution of F. Figures S3−S5 display top-view scanning electron micrographs (SEM) images of SnO2 and F:SnO2 films annealed at different temperatures. The (F:SnO2)180 films with low doping ratio exhibited conformal and pinhole-free film coverage (Figure S3). As for the high doping ratios, at 0.7 and 1, for example, the films presented obvious small cracks (marked with red circle in Figure S3g,h). The samples annealed at 380 °C displayed slightly bigger particle size with tiny agglomeration (Figure S4), indicating the importance of annealing temperature for crystallization of F:SnO2 NCs. However, significant pinholes and cracks were easily observed for the samples annealed at 500 °C (Figure S5). This situation becomes even more significant for samples of high doping ratio.35 Therefore, we selected (F:SnO2)180 and (F:SnO2)380 as ETL for PSC device fabrication during the following experimental studies in avoid of high leakage current due to low coverage or high pinholes. To accurately obtain the band-gaps of films and to avoid the shielding effect from FTO substrates,36 we measured the transmission spectra of SnO2 and F:SnO2 films coated on quartz substrates (Figure S6). Optical band-gaps were calculated from linear line portion of the plot of (αhv)2 versus (hv) based on the equation of αhv = A(hv − Eg)2 (Table S2).37 As depicted in Figure 1c, the optical bandgap slightly broadens with not only increasing temperature but also F doping level (less than 0.2 F:Sn molar ratio). These observations can be explained with the filling of states in the conduction band due to lifting of the Fermi level as a function of increased carrier

⎛ E − EC ⎞ nc = NC exp⎜ F ⎟ ⎝ KBT ⎠

(1)

where NC is the accessible density of electronic states in the conduction band, kB is the Boltzmann constant, T is the temperature, Ec is the position of the lower edge of the CB, and EF is the position of the Fermi level of the semiconductor. Therefore, assuming an identical density of electron states NC, a variation of electron concentration would induce a shift in the Fermi energy level according to eq 2: ⎛n ⎞ E F1 = E F2 + KBT ln⎜ c1 ⎟ ⎝ nc2 ⎠

(2)

where nc1 and nc2 are the carrier concentrations for different SnO2 films, and EF1 and EF2 are the corresponding Fermi levels, respectively. Accordingly, Figure 1d compares the Fermi levels from Hall characterization on various F:SnO2 films at different annealed temperatures, and the corresponding values are 3971

DOI: 10.1021/acs.nanolett.8b01440 Nano Lett. 2018, 18, 3969−3977

Letter

Nano Letters

Figure 3. (a) J−V curves of PSCs based on doped SnO2 and bilayer ETL measured under simulated AM 1.5 G sunlight of 100 mW/cm2. (b) Histogram of PCEs measured from 50 bilayer ETL-based PSCs. Inset is the histogram of PCEs measured from 45 doped SnO2-based control devices. (c) EQE spectra of PSCs based on undoped SnO2 and bilayer ETL and the integrated current densities of two cells from the IPCE spectra. (d) Absorption spectra of perovskite films deposited on undoped SnO2 and bilayer ETL. (e) Energy band diagram of the bilayer ETLs based devices in the dark. (f) Quasi-Fermi level splitting of the bilayer ETLs based devices at open circuit under illumination. Ec, conduction band; Ev, valence band; Ef, Fermi energy; X, electron affinity; ΔEc, conduction-band offset; ΔEv, valence-band offset; Vbi, build-in potential; qVbi‑1, band bend of interface D1/D2; qVbi‑2, band bend of interface D2/perovskite; qVbi‑3, band bend of interface perovskite/spiro-OMeTAD; Enf, Fermi energy of ETL; Epf, Fermi energy of HTL.

summarized in Table S3. The Fermi levels of F:SnO2 films at a constant temperature follow a similar tendency with carrier concentration by increasing with the F doping ratio, reaching the highest value at 0.2 F doping. In this case, the carrier concentration increases from 2.04 × 1014 and 3.49 × 1015 to 1.90 × 1015 and 9.07 × 1016 cm−3 for the (F:SnO2)180 and (F:SnO2)380 samples, respectively. Thus, the up-shifted of Fermi level by doping can be evaluated to be around 68 and 95 meV (Table S3), respectively. This shift was confirmed with Kelvin probe microscopy measurements on the work function of F:SnO2 based on FTO (Table S4). A reduced work function of FTO cathode with F:SnO2 can be of benefit to free electron extraction and transport. These findings were expected to augment VOC and FF of PSCs, which have indeed been observed in this study. We fabricated PSCs with a planar heterojunction architecture of FTO/ETL (50 nm)/(FAPbI3)0.85(MAPbBr3)0.15 (500 nm)/ spiro-OMeTAD (200 nm)/Au (100 nm) by varying the ETL while maintaining the rest constant. Figure 2 shows a process flow for device fabrication and a clear cross-sectional SEM image of PSC device. Figure S8 compares the photocurrent

density−voltage (J−V) characteristics of PSCs measured under AM1.5G irradiation (100 mW cm−2). The relevant performance parameters are summarized in Tables S5 and S6. Under the same F content, the (F:SnO2)180-devices exhibited better photovoltaic performance than that the (F:SnO2)380-devices. This is mainly contributed to the enhanced JSC and FF, stemming from the slightly high electrical conduction in the ETL as verified in Figure S8c,d. For the low FF of (F:SnO2)380device, however, it can be ascribed to poor contact correlated with nanoparticles agglomeration in this ETL film. Importantly, the photovoltaic parameters were increased with F doping under constant temperature and reached the maximum at 0.2 doping level, especially the VOC (Figures S8c). Specifically, as the doping level increased, the VOC increased from 1.03 and 1.04 to 1.1 and 1.14 V for the (F:SnO2)180-devices and the (F:SnO2)380-devices, respectively. This result clearly illustrates the correlation between VOC and electronic Fermi energy level of ETL. As expected, the upward-shifting of the Fermi level by moderation F doping can effectively decrease the band offset between ETL and perovskite absorber (Figure S9). We noted that at higher F doping above 0.2 the increased value of VOC by 3972

DOI: 10.1021/acs.nanolett.8b01440 Nano Lett. 2018, 18, 3969−3977

Letter

Nano Letters

Table 1. Photovoltaic Parameters of Perovskit Solar Cells Device A Based on Bilayer ETL Substrates and Device B Based on Undoped SnO2 Substratea samples

scan direction

Jsc (mA/cm2)

VOC (V)

FF (%)

PCE (%)

Rs (Ω cm2)

Rsh (KΩ cm2)

J0 (mA/cm2)

device A

RS FS RS FS

22.92 22.46 21.74 20.95

1.13 1.12 1.03 1.01

78.05 77.58 72.28 68.72

20.20 19.65 16.25 14.54

0.75

9.5

5.38 × 10−7

1.23

4.2

3.25 × 10−6

device B a

Rs, Rsh, and J0 are obtained from the fitting of dark J−V curves. RS(FS), scan from reverse (forward) direction.

Figure 4. SEM images of (a) FTO/(SnO2)180, (b) (F:SnO2)380-0.2/(F:SnO2)180-0.2. SEM images of (FAPbI3)0.85(MAPbBr3)0.15 perovskite films deposited on (c) FTO/(SnO2)180, (d) (F:SnO2)380-0.2/(F:SnO2)180-0.2. The insets in panels a and b are the corresponding SEM images with high amplification.

SnO2 nanocrystal for producing highly efficient PSCs. In this architecture, the configuration of bilayer ETL comprises an inner layer (noted as D1) employed on top of FTO substrate and outer layer (noted as D2) deposited on top of the inner layer. The D1 and D2 layers should have large differences in carrier concentrations of D1 > D2 aiming to imitating a spacecharge region and generating a built-in electric field (Ebi), while both layers should possess an appropriate band alignment (i.e., CBD1 < CBD2).42 Though the band alignment for both layers with CBD1 > CBD2 might be valid when considering the D1 layer could play a role of energy barrier to inhibit recombination of carriers,32,42 it is significant that these conditions can be closely satisfied by specially tuning F doping SnO2 ETL in this study. Therefore, a thin (F:SnO2)380-0.2 film (∼20 nm, n = 9.07 × 1016 cm−3) was applied onto the FTO substrate as D1 layer with the property of modifying surface work function and accelerating electron transport. A (F:SnO2)180-0.2 layer (∼40 nm, n = 1.90 × 1015 cm−3) was deposited as D2 layer, aimed at facilitating charge carrier extraction from the light absorber. The D1 and D2 layers have large difference in carrier concentration but small difference in

70 and 100 mV for both devices was obviously larger than the up-shifted Ef of 58 and 84 meV. This fact can be explained by a slight widening of bandgap (