Device engineering of organic solar cells based on a boron subphthalocyanine electron donor molecule

A boron subphthalocyanine molecule has been employed as a novel electron donor in organic solar cells (OPVs), and optimized in terms of composition and device structure in small molecule solar cells. It is demonstrated that the power conversion efficiency (PCE) of the devices obtained by solution-processing in bulk heterojunction solar cells could be improved by one order of magnitude by changing the fabrication method to vacuum deposition, which promotes a better morphology in the OPV active layers. Importantly, upon insertion of an additional pristine C70 thin interlayer between the active layer and the hole transport layer the PCE was further improved, highlighting the importance of interfacial layer engineering in such subphthalocyanine small molecule OPVs.


Introduction
With the increasing living standards and a growing population, the energy consumption of the world today is on the rise. In combination with climate change arising from carbon emissions and the use of fossil fuels, this places forward a great need for novel schemes of green electricity generation. Here, organic solar cells (OPVs) are important elements in this green energy transition, as they provide highly sustainable energy technology with unique features [1]. They are compatible with industrial mass production techniques (e.g. through Roll-to-Roll technology), making their cost lower than that of their inorganic counterparts [1]. They are thin, lightweight, highly transparent, and mechanically flexible, all of which enables new methods for photovoltaic panel integration. Most importantly, their power conversion efficiencies (PCEs) have increased appreciably in the last year, reaching a current record of 19.2% [2] for single junction and above 20% [3] for multi-junction OPV. The development of novel donor and acceptor molecules, as well as the development of optimal device architectures and processing conditions, are the key enabling factors for this progress. Due to the rigid structure of the subphthalocyanine molecules, their well-matched optical properties, non-planar bowl-shape that gives rise to a large extinction coefficient in the visible spectral region, and deep highest occupied molecular orbital (HOMO) energy level enabling large open-circuit voltages (V OC ), subphthalocyanines are considered promising donor materials for OPVs [4][5][6][7]. Different synthetic modifications have been investigated, such as the effects of the central metal atom substitutions of boron-chloride group (Fe, Co, Ni, Cu, Zn and more) on the optical and physical properties of SubPc/fullerene solar cells [8][9][10]. However, the highest efficiencies achieved with this class of molecules are still rather low (5.1% for subPc/C 60 , and 5.5% for subPc/C 70 single junction OPV [11,12], and 7.7% in tandem architecture [13]). In this study, we report on the improvement in PCE of a boron subphthalocyanine donor molecule (SubPc-Ar) in vacuum-deposited OPVs, as compared to their solution-processed counterparts, due to an improved morphology of the active layer in the OPVs. In

Results and discussion
Subphthalocyanines is a well-investigated family of molecules used in organic photovoltaics [15][16][17][18][19][20]. The electron donor molecule SubPc-Ar, and the matched electron acceptor molecules studied in this work, are both presented in figure 1. The molecule was previously characterized [14] in solution via UV-vis and photoluminescence spectroscopy to give insight into the molecule's absorption and emission properties, as shown in figure 2. The absorption maximum of the molecule in toluene is found at 563 nm, and its emission maximum at 573 nm. The fluorescence quantum yield has been determined [21] to 0.30 in CHCl 3 , and the fluorescence lifetime to 1.84 ns. SubPc-Ar can be electrochemically reduced and oxidized, and using cyclic voltammetry [21], lowest unoccupied molecular orbital (LUMO) and HOMO energy levels were identified as −3.3 and −5.1 eV, respectively. This is in good agreement with the quantum chemical modeling presented in figure 3. Although the absorption spectrum of this molecule appears relatively narrow, which will limit the current output of the solar cells, we note that such molecules could in the future find their interest particularly in semi-transparent solar cells, e.g. for windows or agrivoltaic applications.
Obtaining a favorable morphology in solution-processed all-small-molecule solar cells, which is a prerequisite for achieving highly efficient devices, can present a great challenge [22,23] (see supplementary info SI1). In order to find a way around the pronounced affinity of small molecules to self-aggregate already in the solution phase, as shown in supplementary info SI1, in this study, we modified the processing of the active layers to physical vapor deposition, which allows forming a bi-continuous interpenetrating network straight from the deposition step. Furthermore, it allows device engineering approaches involving the insertion of additional thin material layers, which can affect both the charge generation and transfer processes, thus enhancing the performance of the solar cell devices. In this study, we have followed the approach by Feng et al, in which a thin layer of the C 70 acceptor molecule is inserted in between the active layer and the anode of the inverted devices, which significantly boosted the device performance as compared to inverted solar cell devices without the additional C 70 layer [24]. This concept at first might seem counterintuitive, as the presence of C 70 at the anode contact, in terms of the energy levels from the individual molecules, presents a barrier for hole transfer from the donor molecule to the anode (figure 4). However, the presence of a C 70 layer appears to indeed enhance the hole extraction efficiency under operating conditions, and together with the decent hole transport abilities of the fullerenes [25][26][27], this can lead to further improvement in the performance of the devices.   As a model system, and based on our past work [28][29][30][31], the devices in ITO glass/MoO x (10 nm)/C 60 (X nm)/SubPc-Ar:C 70 (Y nm)/C 70 (20 nm)/BCP (10 nm)/Ag (100 nm) layer stack configuration were fabricated, where MoO x (molybdenum oxide) is employed as the hole transport layer and BCP  [33,34], SubPc-Ar [21], C60 [35] and MoOx [36].
(2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline as the electron transport layer. Due to the strong interaction of Ag with the BCP molecules, the LUMO level of the Ag-BCP complex aligns well with the LUMO level of C 70 and the electron transport occurs through the LUMO of the complex [32]. The energy levels of the device stack, as given for the individual molecules, are given in figure 4.
In order to optimize the device performance, we focus on the following steps: composition of the active layer, presence of the underlying C 60 anode interface layer, active layer thicknesses, and the presence of the overlying C 70 layer to stepwise improve the performance of the vacuum evaporated solar cells.
In the first step, the composition of the active layer, donor:acceptor ratio was varied with the compositions of 2:3, 1:4, and 1:9. For each of the tested ratios, the effect of the C 60 layer was tested. Hereby the total thickness of the active layer and C 60 was kept at 40 nm, so that in the presence of the 4 nm of C 60 anode interlayer, the active layer thickness is kept at 36 nm. Likewise, when no C 60 anode interlayer is employed the active layer thickness is kept at 40 nm. Solar cell parameters of the devices from this optimization are shown in table 1, and the respective J(V) curves are shown together in SI2. The J(V) curves of the optimized device that corresponds to 1:4 ratio with underlying 4 nm of C 60 are shown in figure 5. The optimal performance was obtained with 1:4 ratio, where the average short-circuit current density and fill factor is at maximum, indicating slightly better charge transport properties and reduced recombination effects compared to the other rations. We note that the difference in device performance for the dilute donor ratio of 1:9 is small, and larger donor ratios indeed is more optimal for these cells. This has also been demonstrated for many other small molecule solar cells [37]. Notably, for each of the ratios tested, the presence of the underlying C 60 layer contributed to further improvement in all device parameters. For the dilute donor case (1:9), this even led to further improved V oc and J sc compared to the 1:4 ratio, however, with a slightly lower fill factor. For the 1:4 ratio, the V oc was also here significantly improved with the C 60 anode interlayer, followed by an improvement in fill factor too, which indicates less interface recombination compared to devices without the anode interlayer, as thus improved hole extraction properties. This could be explained in terms of the ability of molybdenum oxide to pin to the contacting organic layer [38]. When C 60 is in contact with MoO x , the HOMO level of C 60 (6.2 eV) will be pinned to the Fermi level of MoO x . In the case of no anode interlayer, it is the effective HOMO level of the active SubPc-Ar (5.1 eV):C 70 (6.1 eV) layer in contact with MoO x that will pin to the MoO x Fermi level. Irfan et al have shown that as a result of this pinning effect, when fullerene is deposited on top of MoO x , the energy levels of fullerene are lifted up due to the high work function of MoO x [39]. This results in p-type doping of the fullerene and band bending at the MoO x /fullerene interface, which extends several nanometers into the fullerene/active layer. This band bending could in the case of the C 60 anode interlayer align the HOMO levels for efficient hole extraction, while providing an electron/exciton blocking layer (C 60 ) at the anode contact, reducing interface recombination. This could explain the improved performance observed in the SubPc-Ar solar cells with the fullerene anode interlayer, although further studies would be needed to make a final conclusion on this mechanism. A similar effect was recently observed, although for a different molecular system [40]. Table 1. Optimization of the SubPc-Ar:C70 active layer ratio, with and without the underlying C60 layer. The error is expressed in terms of the standard deviation of the measured devices: without C60 layer: 2:3 ratio 11 devices, 1:4 ratio 12 devices, 1:9 ratio 21 devices; with C60 layer: 2:3 ratio 15 devices, 1:4 ratio 24 devices, 1:9 ratio 19 devices.   In the second step, the thickness of the active layer was fine-tuned in the range from 20 to 80 nm, as shown in table 2 and figure 6. All the J(V) curves are shown together in SI3. The open circuit voltage (V oc ) demonstrated independence from active layer thickness, while the short circuit current (J sc ) increased with the increase of the active layer thickness, and decreased after reaching the optimum thickness. With the increasing thickness of the active layer, the generation of charges will increase due to an increased amount of available chromophores. However, for very thick active layers, the built-in electric field within the device will decrease, and related, the thick active will lead to more recombination effects upon charge transport, thus decreasing the charge collection efficiency and lowering the J sc and fill factor, and from that the device efficiency.
To further investigate the effect of the active layer thickness on the performance of the devices, in the last step, the thickness of C 70 layer on top of the bulk heterojunction layer was varied. The extra C 70 layer acts as an electron-transport and hole-blocking layer, efficiently extracting the electrons from the active layer to the BCP/Ag cathode. As reflected in the performance parameters, figure 7 and table 3, the decrease in the C 70 layer thickness causes a strong reduction in J sc and fill factor, pointing to the importance of avoiding the recombination effects at the active layer/cathode interface using a thick C 70 interfacial layer. This could indeed be affected by the roughness of the active layer too, and may be due to an interplay between the C 70 and BCP thicknesses at the cathode contact. The J(V) curves for different C 70 layer thicknesses are shown together in SI2.

Conclusions
In this study, we have demonstrated the implementation of SubPc-Ar as a novel donor molecule in OPVs, and studied the importance of interfacial layer engineering in such devices. We have optimized the composition and thickness of the active layers in small molecule evaporated OPVs based on the new boron subphthalocyanine molecule, using a standard device configuration architecture. Importantly, we have shown that the performance of the solar cells based on the SubPc-Ar system can be enhanced by the insertion of an additional fullerene layer at the anode contact, which can significantly enhance the charge (hole) collection efficiency, and thus the device performance. Although further tuning of the device and/or molecular system would be needed to reach high-performance levels with this novel molecule in OPVs, these approaches are of high importance for the development of optimized layer stacks yielding a boost in optimized solar cell performance.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.

Funding
This research was funded by The Independent Research Fund Denmark, Technology and Production Sciences Projects PhotostablePV (Grant No. 0136-00081B) and ReactPV (Grant No. 8022-00389B), as well as by Villum Foundation Project CompliantPV (Grant No. 13365).