Surface dipole assisted charge carrier extraction in inverted architecture perovskite solar cells

Hybrid metal halide perovskites have emerged as a highly attractive material for high efficiency solar cells. To date, an impressive power conversion efficiency (PCE) of 25.5% and 29.8% was demonstrated for single junction perovskites and perovskite/silicon tandem solar cells, respectively.¹ To further improve the PCE and stability of perovskite solar cells, numerous approaches have been adopted² among which the most noteworthy are perovskite composition engineering,^3–5 metal oxide transport layer modification,^6–8 incorporation of additives into the perovskite precursor solution,^9–13 and reduction of surface defects by the use of a passivation layer.^14–18 Another approach is related to the management of the energetic misalignment at the device interfaces, which can impact charge extraction as well as lead to charge accumulation and recombination that further limit the performance of perovksite solar cells.^19–23 Incorporating materials that form interfacial dipoles is a facile method to reconfigure the energetic alignment in perovskite solar cells.²⁴ Most commonly, these dipoles are introduced either at the interface with the electrodes^25–27 of the device or as modifiers at the perovskite/oxide interface in standard architecture devices.^8,28–32 Few examples exist in which the work function of the perovskite layer was directly tuned by forming a dipole on top of the perovskite active layer. For example, Di Carlo and co-workers have utilized titanium-carbide MXenes to modify the work function of both the TiO₂ extraction layer and that of the perovskite absorber layer.³³ Canil et al. have recently reported that by depositing self-assembled monolayers on top of the perovskite active layer, its energy levels can be effectively tuned by forming either a negative or positive dipole.³⁴ The authors have shown that by applying a negative dipole (i.e., by reducing the perovskite work function), the performance of standard architecture devices can be enhanced. Dong et al. have utilized conjugated aniline modifiers to tune the energy alignment at the interface between the perovskite layer and the phenyl-C₆₁-butyric acid methyl ester (PCBM) electron transport layer (ETL), which resulted in an increase in the work function of the perovskite layer accompanied by the formation of a 2D perovskite capping layer, which led to the improvement in the device performance of inverted architecture solar cells.³⁵ The authors demonstrated that the increase in efficiency is mostly related to the formation of the 2D perovskite, as has also been observed in standard architecture 3D/2D heterostructure devices.^36–38 

Herein, we demonstrate that the efficiency of inverted architecture perovskite photovoltaic devices can be enhanced by moderately enhancing the work function of the perovskite layer without forming a 2D capping layer. We explore four derivatives of dicationic phosphonium-bridged ladder stilbenes (PYMC)³⁹ and deposit them on top of the perovskite absorber to modify its work function. We find that those derivatives whose energy levels match well the energetics at the perovskite/PCBM interface can form a dipole that leads to an improved charge extraction and an enhanced device performance, while others cause a mismatch in the energetic alignment, thus increasing recombination and reducing the device performance.

The device architecture of the perovskite solar cells investigated in this work is shown in Fig. 1(a). In short, a triple cation perovskite Cs_0.05(FAI_0.83MAI_0.17)_0.95Pb(I_0.85Br_0.15)₃ was used as the active layer and deposited following previously developed procedures.⁴⁰ A self-assembled monolayer of [2-(3,6-dimethoxy-9H-carbazol-9-yl)ethyl]phosphonic acid (Meo-2pacz) was utilized as a hole transport layer (HTL), while PCBM and bathocuproine (BCP) served as the ETL and the hole blocking layer, respectively. A detailed fabrication procedure can be found in the supplementary material. The chemical structure of the four different derivatives of PYMC used in this work is summarized in Fig. 1(b). Their synthesis and detailed characterization can be found in the supplementary material (Figs. S1–S39 and Tables S1–S6).

Figure 2(a) displays the box plot of the photovoltaic device performance of reference solar cells as well as those modified by the different PYMC derivatives when characterized under simulated AM 1.5 sunlight at 100 mW/cm² irradiation under ambient conditions. The reference devices exhibit an average PCE of 17.8% and a relatively large spread in photovoltaic performance (ranging from approximately 15% to 20%). The incorporation of PYMC derivatives at the perovskite/PCBM interface has a clear impact on the device performance. PYMC-Me and PYMC-H derivatives show a higher average PCE than that of the reference devices. The main contribution to the higher PCE in PYMC-H modified devices is due to the increased FF and J_sc, while V_OC is only slightly increased by approximately 30 mV. Together these changes lead to an average PCE of 19.8%. The champion solar cell modified with PYMC-H has reached an open-circuit voltage (Voc) of 1.075 V, a short-circuit current (J_SC) of −23.60 mA/cm², and a fill factor (FF) of 79.67%, resulting in a PCE of 20.5%. In the case of PYMC-Me modified devices, high FFs (∼80%) contribute to the increase in PCE (an average of 18.8%). The halogenated derivatives (PYMC-F and PYMC-CF₃) lead to a decreased photovoltaic performance, predominantly due to a reduced FF. It is noteworthy that in all cases, the modified devices exhibit a smaller spread in their photovoltaic performance parameters. Figure 2(b) shows the J–V curves of photovoltaic devices whose performance is closest to average. It clearly shows that PYMC-Me and PYMC-H modified devices lead to an increase in FF. Exemplary external quantum efficiency spectra and J–V characteristics in the dark for reference and one halogenated derivative (PYMC-CF₃) and non-halogenated derivative (PYMC-Me) are shown in the supplementary material (Fig. S40).

The increase in FF and J_sc may arise from an improved charge carrier extraction at the PYMC-modified perovskite/PCBM interface. To investigate this, we characterized the steady-state photoluminescence (PL) of glass/perovskite/PYMC samples. Figure 3(a) shows that the deposition of either of the four PYMC derivatives on the perovskite absorber lead to the quenching of its PL quantum efficiency (PLQE) from 9.95% for the unmodified perovskite layer to below 4% for PYMC/perovskite samples. These results could indicate either an increase in non-radiative recombination or an efficient charge extraction into the PYMC molecules. Considering that the V_OC of the photovoltaic devices modified with PYMC is even slightly increased, it is unlikely that they increase the non-radiative recombination in the perovskite layers. Thus, we associate the decrease in PLQE to arise from charge extraction, which quenches excitations in the perovskite layer. This quenching is also observed after the deposition of the PCBM ETL [Fig. 3(b)]. While these PL measurements serve as an indication for improved charge extraction, they alone cannot explain why certain derivatives lead to an increase in FF, while others cause a decrease.

To further understand the mechanism of enhancement or worsening of the performance of PYMC modified perovskite solar cells, the devices were characterized by impedance spectroscopy (Fig. 4). The Nyquist plots measured on the perovskite solar cells with PYMC materials are composed of a depressed circle in the high frequency regime, followed by an arc in the low frequency regime, which can be attributed to the accumulation of ions and electronic charges at the interfaces with the perovskite layer.⁴¹ The curves can be well-represented by a simple circuit with resistors R1 and R2 and a non-ideal capacitor constant phase element (CPE), R1, R2, and CPE represent the series resistance, recombination resistance, and correlated capacitance, respectively,^42,43 and are summarized in Table I. The series resistance R1 is very slightly reduced upon modification with the various PYMC derivatives. However, the biggest changes are observed in the recombination resistance, R2. In comparison to the reference devices, in the case of modification with PYMC-Me, R2 is significantly enhanced up to 4934 ± 33.9 Ω. Similarly, the R2 value of PYMC-H modified devices is also increased to 4540 ± 30.1 Ω in comparison to the reference (4280 ± 38.34 Ω). On the other hand, the recombination resistance of the halogenated derivatives is greatly reduced to 3052 ± 17.52 and 1689 ± 9.63 Ω for the PYMC-F and PYMC-CF₃, respectively. This trend is entirely consistent with that of the FF as observed in Fig. 2(a). Despite the similar improvement in charge carrier extraction by the PYMC materials, devices with the PYMC-H and PYMC-Me derivatives result in a reduction in charge carrier recombination and an increase in FF, while the other two derivatives show the opposite effect.

We note that the physical origin of negative values of impedance remains uncertain.⁴¹ It has been suggested that the interaction of ionic and electronic effects by charge accumulation at the interface could contribute to negative impedance;⁴⁴ however, the study of the origin of this effect is beyond the scope of this paper.

To explore the influence of the modification with PYMC on the energetics at the perovskite/PCBM interface, the samples were characterized by ultra-violet photoemission spectroscopy (UPS). UPS spectra reveal that the deposition of PYMC derivatives leads to the formation of a surface dipole at the perovskite/PYMC interface [Fig. 5(a)],²⁴ represented by a shift (Δ) in the photoemission onset (Table II). Considering the symmetry of the PYMC derivatives, they do not introduce a molecular dipole (Fig. S41) but rather lead to an interfacial dipole formed due to a ground state charge transfer from the perovskite layer to the electron accepting PYMC molecules. As a result, the work function of the perovskite layers is increased, in particular, in the case of the halogenated derivatives. However, to understand the effect on the device photovoltaic performance, one has to consider not only the dipole formation but also the charge transport levels of the PYMC molecules. In particular, the position of the lowest unoccupied molecular orbital (LUMO) of the PYMC molecules will play a role at determining the efficiency of electron extraction at the modified interfaces. To estimate the relative position of the PYMC derivatives' LUMOs, we performed density functional theory (DFT) calculations using B3LYP/def2TZVP. The detailed results of these calculations are shown in Figs. S34–S37 and Table S5 in the supplementary material. Figure 5(b) summarizes the extracted LUMO positions for the PYMC derivatives. The halogenated derivatives result in deeper LUMOs than the PYMC-H and PYMC-Me derivatives—in particular, in the case of the PYMC-CF₃ derivative. This is in agreement with the larger interfacial dipole formed for the halogenated derivatives due to their being stronger electron acceptors. In all cases, the position of the highest occupied molecular orbital (HOMO) of the derivatives is much deeper than both the valence band of the perovskite layer and the HOMO of PCBM and ranges from 6.8 to 7.5 eV. Considering these results together with the PL and impedance spectroscopy measurements suggest that both the formation of the surface dipole and the energy levels of the PYMC derivatives dictate the effect on the device performance. In the case of the reference devices, while the energetic alignment at the perovskite/PCBM interface promotes electron extraction, the relatively low ionization potential of PCBM⁴⁵ does not lead to efficient hole blocking. As a result, some holes can be transported to the PCBM layer, leading to recombination and resulting in an S-shape J–V curve with a low FF.²⁶ Hence, a hole blocking layer BCP is commonly added to the device structure. Modification with either the PYMC-H or PYMC-Me leads to the formation of a small surface dipole at that interface, thus improving the energetic alignment with PCBM. Moreover, the very low lying HOMO of these derivatives effectively blocks the holes from being injected into the PCBM layer, thus further suppressing recombination, while effectively extracting electrons from the perovskite active layer. The halogenated derivatives also lead to a surface dipole formation, but their deeper LUMOs prevent an efficient charge extraction of electrons from the PYMC to the PCBM, thus leading to charge accumulation and enhanced recombination with the holes in the perovskite layer. As a result, the recombination resistance is reduced (Table I) leading to a lower photovoltaic performance [Fig. 2(a)]. These three scenarios are summarized in Fig. 5(c).

Importantly, we can rule out other mechanisms such as surface passivation or tunneling contact formation from contributing to the improvement in device performance. Structural and microstructural characterization (Figs. S42 and S43) reveals that the introduction of PYMC derivatives does not impact neither the crystalline structure nor the microstructure of the perovskite layer. Passivation of defects by PYMC derivatives would have led to higher photoluminescence, but we observe PL quenching due to efficient charge transfer to the PYMC. Similarly, tunneling contact formation can be ruled out since halogenated and non-halogenated derivatives—despite identical processing—lead to very different effects in terms of photovoltaic performance.

We note that our earlier works on introduction of dipoles into the device structure have shown to markedly increase the built-in potential and, correspondingly, the V_OC of the devices by as much as 120 mV.^25–27 This was made possible by the non-ideal structure of these devices that employed a poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) hole extraction layer, which significantly limited the built-in potential of these devices. On the other hand, the Meo-2pacz HTL employed here does not limit the built-in potential, so the increase in V_OC is relatively minor on the order of 30 mV only. This is comparable to the similar increase observed previously when a poly(triaryl-amine) (PTAA) HTL was employed.²⁶ Moreover, theoretical simulations have recently shown that the energetic alignment with the transport layers does not play a significant role in determining the V_OC,⁴⁶ suggesting the manipulation of this alignment by PYMC derivatives is not expected to lead to large changes in V_OC, but rather increase the fill factor due to better charge extraction and reduced recombination.

Beyond the impact on the photovoltaic performance, the modification of the interfacial energetics at the perovskite/PCBM interface by PYMC derivatives might also explain the reduced spread of the photovoltaic parameters observed in Fig. 2(a). While many factors can contribute to the distribution of device performance, large-scale inhomogeneities in the electronic structure of perovskites—especially those fabricated using the solvent engineering method—have been tied to variation in the photovoltaic performance.⁴⁷ It is possible that the modification with PYMC homogenizes the electronic structure at the surface, thus leading to a narrower distribution in the photovoltaic performance, which would be investigated in more detail in future studies.

To conclude, in this work, we demonstrated that the formation of surface dipoles at the perovskite/PCBM interface can result in an improved charge extraction and a reduced recombination, thus leading to enhanced FFs and overall device performance. However, for this to occur, the energy levels of the surface modifiers have to align well with those of the perovskite and the PCBM layers. Our study provides valuable insights required for the development of novel compounds that enable energy level engineering in perovskite photovoltaics and which could potentially be extended to the fields of perovskite light-emitting diodes⁴⁸ and field-effect transistors,⁴⁹ where optimization of charge injection is of key importance.

See the supplementary material for a detailed characterization of the PYMC derivatives and experimental methods.

This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (ERC Grant Agreement No. 714067, ENERGYMAPS) and the Deutsche Forschungsgemeinschaft (DFG) within the framework of SPP 2196, project PERFECT PVs (Project No. 424216076). Q.A. and Y.V. thank the Center for Advancing Electronics Dresden (CFAED) for generous support. J.-M.M. and J.B. thank the Vector Foundation (Grant Number P2019-0089) for funding.