Solid state sensitized heterojunction photovoltaic cells
Solid state sensitized heterojunction photovoltaic cells are presently under intense investigation because they present a promising avenue towards cost-effective high efficiency solar power conversion. These devices use molecular dyes or semiconductors in form of quantum dots (QD) or extremely thin absorber (ETA) layers as light harvesting agents. The sensitizer is deposited as a molecular or QD layer at the interface between a hole and electron conducting material, often a large band gap oxide semiconductor of mesoscopic structure. Following light excitation, the light harvester injects negative and positive charge carrier in the respective electronic transport materials, which subsequently are collected as photocurrent at the front and back contacts of the cell. The photo-voltage is given by the difference in the quasi-Fermi level under illumination of the electron- and hole-conducting solids.
Recently research in this field has accelerated; new efficiency records being attained at short intervals. Thus, after years of struggling to get over the 5% PCE barrier, an ETA cell based on Sb2S3 sensitized mesosocopic TiO2 films reached a PCE of 6.3% while dye sensitized solid state heterojunctions have reached 7.2 percent. It is noteworthy that an open-circuit photovoltage of 1.02 V was recently demonstrated from the organic dye loaded TiO2 film combined with spiro-MeOTAD. A further substantial gain in efficiency pushing the PCE to 8.5% was achieved very recently by combining the N719 dye with the p-type semiconductor CsSnI3.
(CH3NH3)PbI3 perovskite nanocrystals have attracted attention as a new class of light harvesters for mesososcopic solar cells. Impressive PEC values of up to 6.54% have been obtained with liquid junction cells based on iodide/triiodide redox couple. However a rapid degradation of performance was witnessed due to dissolution of the perovskite in the electrolyte. Because the (CH3NH3)PbI3 nanocrystals exhibit a one order of magnitude higher absorption coefficient than the conventional N719 dye, they offer advantages for use in solid state sensitized solar cells where much thinner TiO2 layer are employed than in liquid junction devices.
Here we report a new solid-state mesoscopic solar cell employing (CH3NH3)PbI3 perovskite nanocrystals as a light absorber and spiro-MeOTAD as a hole-transporting layer, . A strikingly high PCE of 9.7% was achieved with submicon thick films of mesoporous anatase under AM 1.5G illumination along with excellent long term stability.
Figure 1. Solid-state device and its cross-sectional meso-structure
Perovskite (CH3NH3)PbI3 characterization
Optical band gap and valence band maximum were determined based on reflectance and ultraviolet photoelectron spectroscopy (UPS) measurements. . show the diffuse reflectance spectrum and the transformed Kubelka-Munk spectrum for the (CH3NH3)PbI3-sensitized TiO2 film. The optical absorption coefficient (α) is calculated using reflectance data according to the Kubelka-Munk equation, F(R) = α = (1−R)2/2R, where R is the percentage of reflected light. The incident photon energy (hν) and the optical band gap energy (Eg) are related to the transformed Kubelka-Munk function, [F(R)hν]p = A(hν – Eg), where Eg is the band gap energy, A is the constant depending on transition probability and p is the power index that is related to the optical absorption process. Theoretically p equals to ½ or 2 for an indirect or a direct allowed transition, respectively. Eg of the bare TiO2 film is determined to be 3.1 eV based on the indirect transition, which is consistent with data reported elsewhere. Eg for the (CH3NH3)PbI3 deposited on TiO2 film is determined to be 1.5 eV from the extrapolation of the liner part of [F(R)hν]2 plot ( ), which also indicates that the optical absorption in the perovskite sensitizer occurs via a direct transition. shows UPS spectrum for the (CH3NH3)PbI3 sensitizer deposited on TiO2 film, where the energy is calibrated with respect to He I photon energy (21.21 eV). Valence band energy (EVB) is estimated to −5.43 eV below vacuum level, which is consistent with the previous report. From the observed optical band gap, its conduction band energy (ECB) is determined to −3.93 eV that is slightly higher than the ECB for TiO2. The schematic band alignment is sketched in , where the band positions are well aligned for charge separation
Figure 2. Diffuse reflectance and UPS spectra for (CH3NH3)PbI3 perovskite sensitizer
The photo-induced absorption (PIA) spectra of the (CH3NH3)PbI3 films coated with spiro-MeOTAD show the signature of the oxidized spiro-MeOTAD, featuring a broad absorption peak at 1340 nm, characteristic of the hole being localized on the triaryl amine functionality. This reductive quenching of the (CH3NH3)PbI3 occurs efficiently on this time scale and is observed for both the TiO2 and Al2O3 films. The negative peak is an emission neat the band gap most likely arising from electron-hole recombination, which is consistent with the UPS result and the luminescence results shown in .
The solid state device based on the (CH3NH3)PbI3 perovskite NPs deposited on a 0.6 μm thick mesoporous TiO2 film shows a high short-circuit photocurrent density of 17.6 mA/cm2, an open-circuit voltage of 888 mV and a fill factor (FF) of 0.62 under AM 1.5G solar illumination, corresponding to a PCE of 9.7% (). This strikingly high efficiency can be achieved with submicron thick TiO2 films due to the large optical absorption cross section (absorption coefficient of 1.5×104 cm−1 at 550 nm) of the pervoskite nanoparticles and the well-developed interfacial features including complete pore filling by the hole conductor as can be seen in . The incident photon-to-electron conversion efficiency (IPCE) reaches a broad maximum at 450 nm remaining at a level over 50% up to 750 nm ( ). The appearance of a IPCE plateau indicates that the (CH3NH3)PbI3 NP's embedded in the 0.6 μm-thick mesoporous TiO2 film harvest efficiently the incident photons, converting them with a high quantum yield to electric current. The photocurrent density of perovskite sensitized solid state cell is linearly proportional to light intensity ( ), which indicates that the (CH3NH3)PbI3-sensitized TiO2/spiro-MeOTAD junction is a non space-charge limited structure, associated with little difference in electron and hole mobility.
Figure 3. Photovoltaic characteristics of (CH3NH3)PbI3 perovskite sensitized solar cell
Dependence of photovoltaic performance on TiO2 film thickness
shows that, the photocurrent density (JSC) is not strongly dependent on film thickness, where JSC's of 16–17 mA/cm2 can be obtained within the range of film thicknesses of 0.6–1.4 μm. Open-circuit voltage (VOC) is however more significantly influenced by changing the film thickness. The VOC decreases from ~0.9 V to ~0.85 V as the film thickness increases to 0.8 μm and further decreases to around 0.8 V when the film is greater than 1.2 μm. VOC starts to decline significantly from 1.5 μm. This decrease of VOC is expected as the dark current augments linearly dependant with film thickness lowering the electron concentration under ilumination and hence their quasi Fermie level. The FF is gradually decreased with increasing the film thickness, which is a consequence of the lower Voc and an increase of the electron transport resistance. Due to the diminishing VOC and FF, PCE (η) is clearly decreased with increasing the TiO2 film thickness. The thinnest film of 0.6 μm can deliver a PCE of over 9% and more than 8% can be achieved from thicknesses less than 1 μm.
Figure 4. Effect of TiO2 film thickness on the key photovoltaic performance parameters
To elucidate the relation between thickness of the TiO2 layer and the photovoltaic performance, impedance spectra (IS) were measured. The frequency domain in the Nyquist plot which belongs to the recombination process dominating the dark and the photocurrent could be easily processed and is presented in . Three different thicknesses of mesoporous TiO2 layers were investigated (0.6, 1.15 and 1.4 µm). The JSC and FF were similar for all 3 cells but the VOC dropped from about 920 to 880 to 850 mV with increasing TiO2 thickness. The dark current scaled nearly linearly with thickness of the mesoporous TiO2 layer ( ) and is well mirrored by the behavior of the recombination – or charge transfer-resistance RCT ( ). The RCT near short circuit is dominated by the interface between the hole conductor and the under-layer as apparent from the small potential dependence of the resistance. The behavior of RCT changes as soon as the conductance in the photoanode increases due to the rise of the Fermi level in the photo-anode under forward bias (Vapplied > 500mV). RCT drops steeply with increasing forward bias because the dark current is now dominated by the flow of electrons across the photo-anode interface to the hole conductor and no longer by the underlayer/hole conductor interface.
Figure 5. IS measurements as TiO2 thickness: red 0.6 µm, blue 1.15 µm and green 1.4 µm
A similar picture is observed for the IS response under illumination (). The increase of the recombination current with higher surface area of the thicker photoanodes leads to a faster reduction of the RCT and ultimately to a lower VOC. The capacitance (CA) near JSC, which is associated to the capacitance of the under layer/hole conductor interface, shows nearly no change. It increases as the mesoscopic TiO2 film is filled with the electrons induced by the applied potential showing CA values comparable to other mesoporous solid state devices and therefore about 100 times lower than in liquid DSC devices. Another feature in common with solid state devices like BHJ or ETA solar cells is the drop of the capacitance at even higher forward bias. The origin of this behavior is not fully understood so far. Bisquert et al. mentioned that the balance with the Helmholtz capacitance might be a reason. Alternatively also the direct faradaic current flow could also lead to the overall reduction of the capacitance. Finally, we can see that the calculated electron lifetime (τn = CA× RCT) shows a faster decline in τn at higher forward bias with increasing TiO2 thickness leading to the observed overall reduction in the VOC ( ).
Time resolve single photon and femto-second laser spectroscopy studies
A powder of (CH3NH3)PbI3 shows a band edge emission which is centered at 780 nm, Figure SI2. The emission decay was examined by single photon counting technique and showed a multi-exponential decay with lifetimes of 78 ns, and 350 ns which is assigned to radiative decay of excitons in the (CH3NH3)PbI3.
Femtosecond transient absorption (TAS) studies were performed to elucidate the mechanism of charge separation in the device. Samples were subjected to pulsed excitation (λexc = 580nm) and were probed by white light continuum (WLC) pulses, whose spectrum covered the 440–740 nm domain. Comparison was made between samples containing the perovskite material deposited on Al2O3 () and TiO2 ( ), respectively. Due to the mismatch of the energies of the conduction bands of (CH3NH3)PbI3 and Al2O3 (ECB = −0.1 eV vs vacuum), no electron transfer is expected between these two materials upon excitation of the light absorber.
Figure 6. Femtosecond transient absorbance spectra with λexc = 580 nm and WLC probe of
a) (CH3NH3)PbI3/Al2O3, b) (CH3NH3)PbI3/TiO2,
c) Spiro/(CH3NH3)PbI3/Al2O3, and d) Spiro/(CH3NH3)PbI3/TiO2,
recorded at various time delays after excitation
The negative signal peaking at 483 nm was attributed to the bleaching of the absorption of the perovskite. A positive peak that has not been assigned yet was also observed with a maximum at 638 nm. The same bleaching as well as the same positive peak could be observed on both Al2O3 and TiO2 samples. Furthermore, a large negative feature above 700 nm was identified in both samples and assigned to the stimulated emission of excited (CH3NH3)PbI3. This signal matches the spectrum shown in Figure SI2 obtained by steady-state emission and corresponds to the excitonic absorption edge of the material.
Relaxation of the positive and negative bands observed in the 450–540 nm and 630–700 nm regions is completed on a 1 ns timescale, suggesting that both spectral features are associated with the same excited state of the material, which half-lifetime can be estimated to be ~ 50ps. No major difference could be observed between samples based on Al2O3 and TiO2: the recovery of the initial state in 1 ns for both samples shows that no significant charge injection from the excited state of the perovskite into TiO2 can be evidenced. This is also confirmed by the observation of the decay of the stimulated emission, which does not appear to be quenched by TiO2.
Samples covered by the hole-transporting material spiro-OMeTAD were measured to further investigate the working mechanisms of the perovskite-based cell. On Al2O3, (), the amplitude of the bleaching signal at 483 nm was found to be smaller than on a sample deprived of the HTM. The positive absorption signal in the 630–700 nm region completely disappeared, concomitantly with a strong quenching of the stimulated emission above 700 nm. These results suggest a rapid reductive quenching of the excited state of the perovskite by the hole-transporting material. The absorption spectrum of oxidized form of the spiro-OMeTAD generated during this process extends in the visible region and could thus interfere with other spectral features observed between 400 and 700 nm.
The transient spectrum of the sample HTM/(CH3NH3)PbI3/TiO2 () exhibits the same features observed on the one deprived of spiro-OMeTAD ( ), with the bleaching peak in the 480 nm-region being less pronounced than in the case of the perovskite alone on TiO2, as discussed previously for the Al2O3 sample. Moreover, the stimulated emission peak is clearly attenuated in the presence of the HTM, also pointing towards the reductive quenching of the perovskite.
The apparent weaker quenching of the (CH3NH3)PbI3 excited state by the HTM in the TiO2 sample compared to the alumina-based layer might be rationalized by a different morphology of both oxides mesoporous films. In Al2O3 layers, a better pore filling by the spiro-OMeTAD could offer a better contact between the perovskite and the HTM and consequently yield a stronger reductive quenching of the photoexcited state. Further studies are needed to quantify the extent of this effect on different metal oxides and measurements in the near-IR (>1000 nm) would be required to unambiguously time-resolve the appearance of the oxdized form of the spiro-OMeTAD which however was unambiguously identified in the PIA spectra shown in Figure SI1 for both Al2O3 and TiO2 supported NPs in contact with the spiro-OMeTAD.
As previously reported, the (CH3NH3)PbI3 peroskite nanoparticles are unstable in iodide-contained liquid electrolyte due to rapid dissolution. Contrary to the liquid cell, stability is remarkably improved in the solid-state device as can be seen in which shows the results from ex-situ long-term stability tests for over 500 h, where the devices are stored in air at room temperature without encapsulation. The Jsc shows only a slight decrease during the first 200 h attaining a plateau thereafter. The Voc remained stable while the FF improved and stabilized with time. Thus the initial PCE improved by about 14% after 200 h, which is mostly due to the increase in FF. The PCE remained stable during the rest of the test.
Figure 7. Performance