Seid Yimer Abateabc,
Wen-Ti Wu*a,
Someshwar Pola*d and
Yu-Tai Tao*a
aInstitute of Chemistry, Academia Sinica, Nankang, Taipei 115, Taiwan. E-mail: ytt@gate.sinica.edu.tw
bTaiwan International Graduate Program (TIGP), Sustainable Chemical Science and Technology (SCST), Academia Sinica, Taipei, Taiwan
cDepartment of Applied Chemistry, National Chiao Tung University, Hsinchu, Taiwan
dDepartment of Chemistry, Nizam College, Osmania University, Hyderabad, India
First published on 19th February 2018
Fabrication of perovskite solar cells (PSCs) in a simple way with high efficiency and stability remains a challenge. In this study, silver nanoparticles (Ag NPs) were sandwiched between two compact TiO2 layers through a facile process of spin-coating an ethanolic AgNO3 solution, followed by thermal annealing. The presence of Ag NPs in the electron-transporting layer of TiO2 improved the light input to the device, the morphology of the perovskite film prepared on top, and eliminated leakage current. Photoluminescence and electron mobility studies revealed that the incorporation of Ag NPs in the ETL of the planar PSC device facilitated the electron–hole separation and promoted charge extraction and transport from perovskite to ETL. Hysteresis-free devices with incorporated Ag NPs gave a high average short-circuit current density (Jsc) of 22.91 ± 0.39 mA cm−2 and maximum power conversion efficiency of 17.25%. The devices also showed enhanced stability versus a control device without embedded Ag NPs. The possible reasons for the improvement are analyzed and discussed.
Recently, perovskite materials are catching much attention of researchers for its potential application in solar cell.2 Perovskite absorbs visible light, with a low and tuneable band gap of ∼1.5 eV, high electron and hole mobility, which make it a good candidate as the active component in solar cells. Since the breakthrough report of Miyasaka et al. using perovskite as a visible light sensitizer for photoelectrochemical cells, giving 3.8% power conversion efficiency (PCE),3 various advances have been made in device architecture and fabrication process, and most recently a high PCE of 21.1% has been reached for the perovskite-based solar cell.2,4–13
To further push for higher efficiency and ultimate commercialization of this technology, optimization of each component in the device has been tried, such as electron transporting layer (ETL), hole transporting layer (HTL), and electrode, besides the light-absorbing perovskite layer. In terms of ETL, a layer of mesoporous TiO2 (mp-TiO2) together with the compact TiO2 (c-TiO2) layer is necessary to reach the high efficiency reported.14 Other metal oxides or metal-doped TiO2 have been reported to modulate the electron-collecting or electron-transporting properties of the ETL.15–29 Core–shell metallic nanoparticles (NPs) were also introduced in the meso-superstructure layer, with some improvements reported.30,31 Metallic nanoparticles were explored for their unique absorption properties in the visible and near-IR region, as well as the plasmonic properties.32 In particular, silver nanoparticles, Ag NPs, were incorporated in various ways in different layers of perovskite solar cells (PSCs) and beneficiary effects have been demonstrated.33–39 Nevertheless, these devices usually involved more complicated preparation procedures, with little elaboration of the detailed mechanism.
In this work, we report the use of Ag NPs-embedded c-TiO2 as ETL to improve the opto-electrical characteristics, charge extraction and transportation of planar PSCs with a device configuration of FTO/c-TiO2/Ag NPs/c-TiO2/perovskite (CH3NH3PbI3)/Spiro-OMeTAD/Ag. The sandwiched Ag NPs were formed by a simple and robust spin-coating of ethanolic AgNO3 solution on the c-TiO2 layer and later annealed at high temperature. It was shown that the incorporation of Ag NPs in the centre of c-TiO2 ETL layer promoted the formation of larger perovskite crystals with reduced grain boundaries, so that the electron trap density and the charge recombination pathways in the film were reduced. Also, the Ag NPs, with a specific amount, act as an electron scavenger and therefore electron-transporting nature of the TiO2 layer was improved in the PSCs.
A solution of titanium diisopropoxide bis(acetylacetonate) was prepared by dissolving 55 μL stock solution of (75% in 2-propanol) in 1 mL 1-butanol to make 0.15 M solution. Spin-coating of this solution in a three-stage procedure (see ESI†) on the FTO substrate gave the single-layered c-TiO2, double layered c-TiO2 or Ag-embedded c-TiO2. Double-layered c-TiO2 was prepared by another spin-coating of c-TiO2 on the single-layered TiO2. To prepare Ag-sandwiched c-TiO2, different concentrations of ethanolic AgNO3 solutions (with drops of DI water added to dissolve AgNO3) was spin-coated on the single layered c-TiO2 and annealed at 130 °C for 10 min, followed by one additional spin-coating of the c-TiO2 precursor on the Ag NP-covered c-TiO2. Finally, all of the ETL layers were sintered at 500 °C for 30 min. Before coating perovskite layer, the ETL substrates were treated by UV–ozone for 30 min to clean the surface.
The perovskite precursor solution was prepared by following a one-step deposition procedure reported in the literature.40 To a solution of DMSO (1 M) containing PbI2 (1 M) and CH3NH3I (1 M), DMF (600 mg) is added and stirred at room temperature for 60 min. The precursor solution was passed through a 0.22 μm PVDF filter and then spin-coated on c-TiO2/FTO substrate at 4000 rpm for 25 s. Diethyl ether was used to wash DMF away to give CH3NH3I·PbI2·DMSO adduct. At the end of the spinning process, transparent perovskite film formed, which was converted to a thick black film after annealing for 2 min at 65 °C and 5 min at 100 °C.
For HTM, 40 mg Spiro-OMeTAD was dissolved in 0.5 mL chlorobenzene and 14.5 μL TBP, 7.5 μL Li-TFSI, and 14.5 μL FK209 were added successively. The solution was spin-coated on top of perovskite surface. Finally, 70–80 nm thick Ag metal was deposited as the terminal contact by thermal deposition at a pressure of 8 × 10−6 torr in a metal-masked area of 0.04 cm2.
The UV-Vis absorption spectra (Fig. S2a, ESI†) revealed the absorption edge for pristine c-TiO2 shifted from 387 nm for the single layer c-TiO2 toward around 413 nm for the Ag NPs-embedded c-TiO2 layers, presumably due to the incorporation of Ag NPs, which is known to give UV absorption above 400 nm.37,44–46 Importantly, the TiO2 ETL embedding Ag NPs showed stronger UV-Vis absorption compared to the pristine c-TiO2, which may be the result of Surface Plasmon Resonance (SPR) of Ag NPs. All the aforementioned characterizations confirmed that Ag NPs were successfully formed in between the c-TiO2 layers. The optical transmittance of ETL is an important criterion to prepare high efficiency PSCs. So that, the optical transmittances are recorded for 1× SC, 2× SC and 4 mM Ag NPs-embedded TiO2 layers (Fig. S2b, ESI†). About 80% transmittance was recorded for all of the samples, which indicates that the incorporation of adequate amount Ag NPs insignificantly change the transmittance of the TiO2 film. With the Ag NPs in place, planar perovskite solar cells were then fabricated by spin-coating perovskite layer on top of various ETLs prepared under the same condition, followed by spin-coating Spiro-OMeTAD and depositing Ag back electrode to complete the device structure. A preliminary survey of the effect of incorporating Ag NPs on device performance was carried out, with the Ag NPs prepared from aqueous AgNO3 solutions of 1 mM to 20 mM concentrations. As a reference cell, TiO2 layers by spin-coating once and twice were also used. Abbreviations of 1× SC, 2× SC, 2× SC AgNO3 (x mM) are used to represent devices with single spin-coating, double spin-coating, and double spin-coating with embedded Ag NPs obtained from x mM AgNO3 solution respectively. The results showed that better performances were obtained with low concentration AgNO3 solutions and peaked for that from 4 mM AgNO3 solution.
Thus the device with 2× SC Ag NPs (4 mM) as ETL showed a maximum PCE of 17.25%, an open-circuit voltage (Voc) of 1.01 V, short-circuit current density (Jsc) of 23.36 mA cm−2, and a fill factor (FF) of 73.11% (Fig. 1c, Table 1). However, when 2× SC Ag NPs (12 mM, and 20 mM) were used as ETL, the device PCE decreased to 15.92% and 15.26% respectively with decreased Jsc and FF. In contrast, the reference solar cells fabricated by using 1× SC and 2× SC TiO2 as ETL showed smaller PCE than the 2× SC Ag NPs (4 mM) device, at 12.35% and 15.17% respectively. Compared to the others, the 1× SC device also had very low PCE for the forward sweep (5.80%), suggesting this device had substantial hysteresis, an undesirable situation in device operation. This phenomenon was eliminated for devices with TiO2 embedding Ag NPs, particularly for the 2× SC Ag NPs (4 mM) device (Table S1, ESI†).
Method | Voc (V) | Jsc (mA cm−2) | FF (%) | PCE (%) | Rsh (KΩ cm−2) | Rs (Ω cm−2) |
---|---|---|---|---|---|---|
1× SC | 0.95 ± 0.02 (0.98) | 16.76 ± 1.37 (18.47) | 68.90 ± 2.64 (68.20) | 11.02 ± 1.01 (12.35) | 0.35 | 6.27 |
2× SC | 0.99 ± 0.01 (0.98) | 21.79 ± 0.50 (21.95) | 69.33 ± 1.46 (70.51) | 14.91 ± 0.43 (15.17) | 1.67 | 4.76 |
Ag NPs (4 mM) | 1.00 ± 0.01 (1.01) | 22.91 ± 0.39 (23.36) | 71.40 ± 1.81 (73.11) | 16.37 ± 0.32 (17.25) | 6.02 | 2.87 |
Ag NPs (12 mM) | 0.99 ± 0.01 (0.99) | 22.40 ± 0.59 (22.52) | 67.82 ± 1.85 (71.41) | 15.03 ± 0.43 (15.92) | 3.12 | 3.34 |
Ag NPs (20 mM) | 0.99 ± 0.01 (1.00) | 21.09 ± 1.06 (21.62) | 68.35 ± 2.54 (67.44) | 14.23 ± 0.68 (15.26) | 0.57 | 4.43 |
The PCE for the 2× SC device also improved over that for 1× SC device, to a maximum of 15.17%. This means ca. 23% improvement over the pristine 1× SC-based device. The possible reason for the improved performance could be a better c-TiO2 coverage on FTO layer when a second c-TiO2 layer was employed. This would reduce the possible direct contacts between perovskites and FTO, which is expected to decrease the chance of charge separation and increase recombination for the excited perovskites. However, 2× SC devices showed higher series resistance (Rs) and much smaller shunting resistance (Rsh), compared to the Ag NPs-incorporated PSCs. As shown in Table 1, the Rs for 2× SC (ca. 4.76 Ω cm−2) device is about two folds higher than that with 2× SC Ag NPs (4 mM) device (ca. 2.87 Ω cm−2). A higher Rs would suggest lower electron mobility for the device without Ag NPs-embedded, as will be discussed later. The same phenomena were observed for Rsh measurement, which quantifies explicitly the degree of recombination at the interface and relates to photocurrent loss.47 The device with 2× SC Ag NPs (4 mM) showed a Rsh value of 6020 Ω cm−2, which is more than three-fold photocurrent loss resistance than that with 2× SC (1670 Ω cm−2). This suggests that in the device fabricated with Ag NPs (4 mM) ETL, the photogenerated charge carriers are successfully separated and transported by suppressing charge recombination at the interfaces and resulting in a higher PCE. The Rs and Rsh properties were improved for nearly all of the Ag NPs-embedding devices compared to 1× SC and 2× SC devices, although when the amount of Ag NPs increased by using 12 mM and 20 mM solution, the series resistance started to build up and Rsh decreased to 3120 Ω cm−2 and 570 Ω cm−2 respectively, resulting in a decrease in performances of the cells. The box charts in Fig. 2 summarize the photovoltaics performance characteristics for devices prepared on various TiO2 substrates. Apparently, the highest Jsc values were recorded for devices prepared from that with 2× SC Ag NPs (4 mM) ETL, so were the PCEs for these devices. The histogram in Fig. S3, ESI† also showed the average PCE and their corresponding Voc, Jsc, FF for at least 15 devices each for 1× SC, 2× SC and 2× SC Ag NPs (4 mM)-based ETL and confirmed the high reproducibility of the performance characteristics. Thus there is ∼39% and ∼14% PCE enhancement for devices prepared from the 2× SC Ag NPs (4 mM)-based device over the pristine 1× SC and 2× SC respectively.
Fig. 2 Box chart comparison of photovoltaics parameters for 1× SC, 2× SC and Ag NPs (4 mM, 12 mM and 20 mM) ETL-based planar PSCs. |
That the incorporation of Ag NPs in the TiO2 layer has a beneficiary effect on the device performance is further confirmed by additional evidences as described below. The incident photon-to-current efficiency (IPCE) mainly depends on the average size of the perovskite crystals, grain boundaries, and perovskite thickness, which influence the light harvesting efficiency (ηlhe), exciton diffusion efficiency (ηed), charge separation efficiency (ηcs), and charge collection efficiency (ηcc).48 The IPCE of the control and Ag NPs-embedded cells are presented in Fig. 1d. It can be seen that Ag NPs (4 mM)-embedded device having the highest integrated Jsc gave a maximum IPCE% of ∼88% at ca. 495 nm and the highest IPCE% in the broad wavelength range from 450 nm to 750 nm compared to the other ETLs. However, the Ag NPs embedded ETL showed a little absorbance enhancement at about 450–600 nm, the IPCE improvement is all over the 400–800 nm range. As a result of incorporation of Ag NPs, the light absorption of perovskite increased not only in a specific region but also in all panchromatic wavelength ranges so that the IPCE is influenced accordingly. The same phenomena are observed when Ag and Au core shall metallic nanoparticles are incorporated in ETL of perovskite solar cell.30,31 The rather flat spectral profile and high IPCE% led to outstanding light harvesting efficiency. Whereas, the control device with 1× SC and 2× SC TiO2 showed ∼78% and 86% IPCE respectively at the same wavelength. When the amount of Ag NPs increased, the IPCE% decreased. Thus similar trend as the PCE results is followed for respective devices. The integrated Jsc value is presented in double Y axis figure for IPCE in Fig. 1d, and integrated Jsc values of 18.1 mA cm−2, 20.8 mA cm−2, 21.1 mA cm−2, 20.5 mA cm−2 and 18.1 mA cm−2 are extracted for 1× SC, 2× SC, 4 mM Ag NPs-, 12 mM Ag NPs- and 20 mM Ag NPs-incorporated TiO2 films respectively. A small discrepancy (∼2–9.6%) was observed between the Jsc measured from J–V and Jsc integrated from IPCE because the latter measurement was carried out in a single wavelength with much lower intensity than the one sun radiation.49,50
In order to understand the root cause for suppressed PCE when increasing Ag NPs concentration in the ETL, the top view SEM images for the perovskite films grown on respective ETLs were taken. As shown in Fig. 3a–c, the top-view SEM images of perovskite films on 1× SC, 2× SC, and Ag NPs (4 mM) ETLs are different in terms of surface morphology. Grains in the perovskite film grown on 2× SC Ag NPs (4 mM) TiO2 were bigger compared to that on 1× SC and 2× SC TiO2. The grains in films on 2× SC Ag NP (12 mM and 20 mM) had similar crystal size as for Ag NP (4 mM), but more cracks appeared, particularly for Ag NP (20 mM) (Fig. S4a and b, ESI†). It is intriguing that the Ag NPs buried within c-TiO2 layer would affect the morphology of perovskite deposited on top. Considering the surface energy of a substrate would affect the morphology of a film formed on top, we measured the water contact angle on various c-TiO2 surfaces (Fig. 3e–g). The same treatment of 30 min UV–ozone exposure was applied to all of the ETLs before measuring contact angles. The water contact angle value on 1× SC and 2× SC TiO2 had a value of <5°, which suggests a highly wettable surface. However, the water contact angle on Ag NPs (4 mM), Ag NPs (12 mM) and Ag NPs (20 mM) TiO2 increased to ∼10.5°, ∼15° and ∼22° respectively. These may suggest that a delicate dependence of grain size on the wettability, smaller and continuous grains for highly wettable surfaces and yet cracks appearing for less wettable surfaces. Previous reports on wettability of the solid surface showed that the roughness of a solid surface will have an effect on wetting property in that a rougher surface tend to be more wettable.51 Our AFM micrographs show that 1× SC has the highest roughness (26.1 nm) and it showed high wettability with a contact angle of less than 5°. The surface roughness decreased to 23.0 nm for 2× SC, and a lower surface roughness was observed for all Ag NPs-incorporated surfaces, with a least roughness of 20.7 nm for 4 mM Ag NPs-embedded film. The perovskite film coated on the rougher surface (thus more wettable surface) gave smaller grain size because of more nucleation sites present on the rougher surface. A smoother ETL will induce the growth of larger grains of perovskite layer.
The thickness of 1× SC, 2× SC and Ag NPs (4 mM) TiO2 were measured by cross-sectional SEM and the results are presented in (Fig. S4c and d, ESI† and Fig. 3d), with an ETL thickness of ∼50 nm, ∼90 nm, and ∼95 nm respectively. The perovskite layer thickness grown on different underneath layers were slightly different, at ∼340 nm, ∼350 nm, and ∼395 nm on 1× SC, 2× SC, and 2× SC Ag NPs (4 mM) TiO2 respectively. The top view and cross-sectional SEM micrographs suggested a certain level of roughness increased the thickness and improved the morphology of the perovskite film formed on top and contributed to the performance of the device recorded.52
UV-Vis and XRD were recorded for perovskite films grown on 1× SC, 2× SC and 2× SC Ag NPs (4 mM) ETL layers (Fig. S5a, ESI†). The active perovskite layer prepared in our process absorbed light in UV-Vis range with an absorption band edge ca. 770 nm, which is in agreement with the previous report.53 UV-Vis absorbing for Ag NPs (4 mM) showed a little improvement in the whole range of UV-Vis compared to 1× SC and 2× SC. The powder XRD spectra (Fig. S5b, ESI†) presented dominant peaks at 2θ of 14.2° (110) and 28.5° (220) for perovskite, with no traces for either PbI2 or CH3NH3I, indicating the formation of high quality and pure perovskite crystal over the ETLs. There is nevertheless little difference in the UV-Vis and PXRD results for perovskite on these c-TiO2 ETLs, probably because the amount of Ag NPs embedded is small.
Photoluminescence and electron mobility studies were carried out to understand the characteristics of charge separation and transportation in these devices. Thus the steady-state photoluminescence (PL) was measured on half devices of perovskite/ETLs/FTO with 1× SC, 2× SC and 2× SC Ag NPs (4 mM) TiO2 as ETL. As shown in Fig. 4, the PL spectrum of perovskite on a bare glass substrate showed a strong emission at ∼774 nm. Much reduced PL intensity was recorded for perovskite grown on 2× SC Ag NPs (4 mM) ETL. This presumably could indicate a suppressed charge recombination in the Ag NP-embedded film as compared to that on 1× SC and 2× SC ETL films. This suggests a higher chance of charge separation and transportation to the electrodes when the 2× SC Ag NP (4 mM) TiO2 is used as ETL. This is in agreement with the Rs and Rsh measurements of the prepared devices.
Fig. 4 Normalized PL spectra of perovskite thin films prepared on glass, 1× SC ETL, 2× SC ETL and c-TiO2/Ag NPs (4 mM)/c-TiO2 ETLs. |
In order to confirm the influence of Ag NPs on charge carrier transport, electron-only device with a device configuration of FTO/c-TiO2/Ag NPs/c-TiO2/perovskite/Al (65 nm) was prepared, and the J–V characteristics were measured in the dark to determine electron mobility of the device. Mott–Gurney law is used to define Space-Charge-Limited Current (SCLC) region, with J = 9/8εoεrE2/L, where J is the current density, E is electric field, εo is permittivity of the vacuum (8.854 × 10−12 F m−1), εr is the relative dielectric constant of CH3NH3PbI3 (εr = 32),54 and L is perovskite film thickness (1× ST = ∼330 nm, 2× SC = ∼350 nm and 2× SC Ag NP (4 mM) = ∼395 nm).14,45 The mobility is dependent on the electric field and can be expressed by a Poole–Frenkel equation μ(E) = μoexp(βE1/2), where μo is the zero-field mobility and β is Poole–Frenkel factor. The slope (β) and intercept zero-field mobility (μo) were obtained from the plot of ln(JE−2) versus E1/2.55 The SCLC electron mobilities calculated are presented in Fig. 5. The device fabricated with Ag NPs (4 mM)-embedded TiO2 showed an electron mobility of 2.57 × 10−4 cm2 V−1 s−1, which is significantly higher than that with 1× SC (1.46 × 10−4 cm2 V−1 s−1) and 2× SC (2.03 × 10−4 cm2 V−1 s−1). This suggested the Ag NPs-incorporated ETL facilitated the charge extraction and prevent electron and hole recombination more than that with 1× SC and 2× SC ETLs. This also matched with the PL and charge resistance studies.
Fig. 5 Measurement of electron mobility by SCLC method at dark J–V characteristics for devices with and without the presence of Ag NPs. |
Finally, the long-term stability, which is one of the essential criteria for PSCs commercialization, was studied for devices with 1× SC, 2× SC and 2× SC Ag NPs (4 mM) as ETLs. Devices were stored in a dry box and only taken out for measurements in the atmosphere without encapsulation. The results are presented in Fig. 6, which revealed that PSC with 2× SC Ag NPs (4 mM) ETL retained ca. 89% of its initial PCE after 28 days, whereas that with 1× SC and 2× SC ETL degraded and retained ∼32% and ∼62% of their initial PCEs respectively after the same period. Also shown are the stability of other parameters like Voc, Jsc, and FF. The possible reason for better stability for Ag NPs (4 mM) may be the formation of bigger crystal with less grain boundaries so that the intrusion of moisture in the cell can be decreased.
Footnote |
† Electronic supplementary information (ESI) available: Preparation of ETL, ETL full XPS spectra, UV-Vis absorption of ETL and perovskite, perovskite XRD, top view and cross-section SEM monograph, histogram J–V comparison and set of tables shows J–V characterization at least for 15 devices each. See DOI: 10.1039/c7ra13744c |
This journal is © The Royal Society of Chemistry 2018 |