High performance polymer tandem solar cell

A power conversion efficiency of 9.02% is obtained for a fully solution-processed polymer tandem solar cell, based on the diketopyrrolopyrrole unit polymer as a low bandgap photoactive material in the rear subcell, in conjunction with a new robust interconnecting layer. This interconnecting layer is optically transparent, electrically conductive, and physically strong, thus, the charges can be collected and recombined in the interconnecting layer under illumination, while the charge is generated and extracted under dark conditions. This indicates that careful interface engineering of the charge-carrier transport layer is a useful approach to further improve the performance of polymer tandem solar cells.

Scientific RepoRts | 5:18090 | DOI: 10.1038/srep18090 (DMF) into PEDOT:PSS in order to improve its mechanical properties and conductivity, which lead to good transparency and conductivity 14,17 . In their study, a PCE of 5.84% has been reported. Two years later, Dou et al. demonstrated and significant improvement, where the PCE improved to 8.62% utilizing a modified PEDOT:PSS/ ZnO ICL 14 . In short, a perfect ICL is needed to avoid any potential drop, otherwise the V OC of the tandem device will be affected. Here, we introduced a new concept ICL, combining our commonly used p-type hole transport material PEDOT:PSS mixed graphene oxide (GO) paired with an n-type material lithium zinc oxide (LZO). The fabricated device employing this ICL illustrates a high V OC of about 1.60 V.

Results and Discussion
Here we demonstrate solution processable polymer tandem solar cells consisting of two different active materials. Our polymer tandem solar cells, which consist of two subcells along with complimentary absorption spectra 18,19 , allow us to harvest as many photons as possible (Fig. 1a). These two subcells are physically separated by PEDOT:PSS:GO/LZO as the ICL. The ICL functions based on various different reasons including (i) a charge recombination zone, and (ii) a shielding layer for the front subcell during the deposition of the rear subcell. In general, because these subcells have completely different absorption spectra, high-energy photons are absorbed in the front subcell, while low-energy photons are absorbed in the rear subcell; thus the whole visible region of the solar spectrum can be covered. The subcells of our tandem device are electronically connected in series, in which the V OC of the polymer tandem solar cells is the summation of the V OC 's of the front and rear subcells. Attenuation of incident light on the rear subcell, caused by absorption and the optical cavity of the front subcell, leads to a lower photocurrent density in the rear subcell, which determines the photocurrent density of the polymer tandem solar cells. Therefore, we introduce in this study a commercially available DPP unit-based low bandgap (E g ≈ 1.44 eV) polymer, poly{2,6′ -4,8-di(5-ethylhexylthienyl)benzo [1,2-b;3,4-b]dithiophene-alt-5-dibutyloctyl -3,6-bis(5-bromothiophen-2-yl)pyrrolo [3,4-c]pyrrole-1,4-dione} (PBDTT-DPP) [10][11][12] (Fig. 1b) incorporating a wide-bandgap (E g ≈ 1.9 eV) polymer, poly(N-9′ -heptadecanyl-2,7-carbazole-alt-5,5-(4′ ,7′ -di-2-thienyl-2′ ,1′ ,3′ -benzothiadiazole) (PCDTBT) (Fig. 1b) 20 , as the front cell, and PEDOT:PSS:GO/LZO layer as the interconnecting layer (ICL) into tandem solar cells; the devices achieve PCEs of 9.02%, with a J SC of 8.53 mA/cm 2 , VOC of 1.60 V, and fill factor (FF) of 66.14%. The results show that PBDTT-DPP is an ideal low bandgap polymer to fabricate high efficiency solution-processed polymer tandem solar cells. Moreover, the results elucidate that the interfacial engineering of the charge-carrier recombination layer is a useful approach to further improving the PCEs of polymer tandem organic solar cells. Figure 1b shows the absorption spectra of PCDTBT:PC 70 BM and PBDTT-DPP:PC 71 BM layers as active layers for the front and rear subcells. The absorption of the PCDTBT:PC 70 BM layer covers from 300 to 700 nm with strong absorption located at the range of 350-600 nm. The absorption of the PCDTBT:PC 70 BM layer covers the visible spectrum range from 400 to 650 nm, just falling in the valley of the PBDTT-DPP:PC 71 BM absorption spectrum. In order to ensure that more photons pass through the middle ICL and efficient light absorption in the rear subcell, the PBDTT-DPP:PC 71 BM film is employed as the active layer of the rear subcell. The complementary absorption of the PCDTBT:PC 70 BM and PBDTT-DPP:PC 71 BM layers in the polymer tandem solar cells may efficiently improve the solar light harvesting. Combining these active materials, the absorption spectrum of the tandem photovoltaic cell covers almost the entire visible spectrum as well as the near infrared region. Thus, we anticipate that the improved absorption spectrum results in a larger number of photogenerated excitons and causes a higher photocurrent. Transmittance of the ICL of the PEDOT:PSS:GO/LZO, which electrically connects the front and rear subcells, is also shown in Fig. 1c. The transmittance is higher than 95% and below 650 nm. When the wavelength is above 650 nm, the transmittance is slightly less than 95%. The device structure and the corresponding energy diagram are shown in Fig. 1d. Lithium doped zinc oxide (LZO) was used as the electron-transport material because their work function matches well with the acceptors and high electron mobility 21 . PEDOT:PSS:GO was used as the hole-transport material for PCDTBT, and MoO 3 was used for PBDTT-DPP because of its good work function alignment with polymer and its high hole mobility. Ultraviolet photoelectron spectroscopy (UPS) was used to examine the work functions of the ICLs, including PEDOT:PSS:GO and LZO (Fig. 1d). The energy difference between the different layers was minimized by material selections to ensure good charge transport.
We carried out thickness optimization experiments of single junction OPVs based on PCDTBT:PC 70 BM (see Fig. 2a). The single junction OPV was fabricated on the ITO substrate of 0.2 cm × 0.2 cm dimension using a traditional sandwich structure of ITO/LZO (30 nm)/PCDTBT:PC 70 BM (95 nm)/PEDOT:PSS:GO (10 nm)/Ag (100 nm). The active layer thickness varied between 95 to 115 nm. All constructed single junction OPVs were characterized under standard spectral condition AM1.5 G at 100 mW/cm 2 . The PCDTBT:PC 70 BM thickness experiment shows major influence of the thickness on the J SC and FF values (up to 11.13 mA/cm 2 and up to 70.78%). The highest J SC of 11.13 mA/cm 2 was obtained from the device with a thickness of 95 nm. As shown in Table 1, the J SC decreases as the active layer thickness increases. The J SC,exp demonstrates a 8.53% decrease from 11.13 mA/cm 2 (95 nm) to 10.18 mA/cm 2 (120 nm), with increases in the active layer thickness. As we can see from Table 1, the J SC,exp values are in good agreement with the J SC,sim values obtained from numerical simulations, where the ratio of the J SC,exp over J SC,sim is virtually unity. Despite higher absorption in the thicker active layer, the J SC significantly decreased to 10.18 mA/cm 2 from 11.13 mA/cm 2 with increased active layer thicknesses. This phenomenon can be attributed to at least two possibilities. First, the electric field inside the active materials slightly decreased as we increased the thickness from 95 to 120 nm. Hence, the dissociation rate of the excitons will decline at lower electric fields because this process relies upon the electric field (Table 1). Secondly, a thicker active layer also provides a long pathway for charge collection at their respective top and bottom electrodes. Consequently, one can also expect that there is a high probability for the separated charges to be recombined again before they arrive at their respective electrodes. The maximum J SC clearly indicates that maximum optical interference is responsible for a field redistribution to increase absorption within the active layer 22 . The PCE decreases for the device with 120 nm due to a clearly reduced J SC and reduced FF (see Table 1). The FF also decreased from 70.78% (95 nm) to 69.97% (120 nm) with an increased active layer thickness. The decrease in FF indicates that the probability of photogenerated charge carriers become  trivial with thicker active material. It is also worth noting that at a short-current condition, where the maximum power point is smaller, there will be a lower dissociation rate as well as a higher recombination rate 23 . As mentioned above, in thicker active materials there will be longer pathways for the separated charges to be collected at the top and bottom electrodes. Thus, if the higher probabilities of these charges are recombined, it will result in a lower FF. The best single junction OPV based on PCDTBT:PC 70 BM depicts a PCE of 6.93% (see Table 1), which outperformed the previously published work of 6.2% 24 . Considering the rather narrow absorption band of the PCDTBT:PC 70 BM at short wavelengths (λ max ≈ 450 nm, almost no absorption > 800 nm, see Fig. 1b). It is reported that the PCE of 6.93% is quite high and among the highest reported for the shorter wavelength range of solar irradiation 25,26 .
Meanwhile, Fig. 2b shows the external quantum efficiency (EQE) spectra for the corresponding single junction OPVs, which demonstrate a wider response from 350 up to 850 nm, with an average EQE of 47%. The integrated J SC 's from the EQE spectra are within the 3% error; 11.02, 10.82, and 9.94 mA/cm 2 , respectively. Figure 3a and Table 2 demonstrate the J-V characteristics of the rear subcell OPVs under 100 mW/cm 2 at AM1.5 G illumination, with a device configuration ITO/LZO (30 nm)/PBDTT-DPP:PC 71 BM/MoO 3 (15 nm)/Ag (100 nm) with PBDTT-DPP:PC 71 BM variation thicknesses of 135, 145, and 155 nm. In the PBDTT-DPP:PC 71 BM thickness study, V OC remains constant at around 0.74 V for all thicknesses studied (Fig. 3b). Similar to that of PCDTBT:PC 70 BM devices, the J SC,sim decreases with increased PBDTT-DPP:PC 71 BM thickness (see Table 2), as expected from the increased absorption. With 100 nm of PBDTT-DPP:PC 71 BM, the device performance enhance to FF = 65.69% and PCE = 6.36%. For an identical device, with a different ETL (ZnO:TiO x = 30 nm), the device demonstrated a FF = 55.35% and a PCE of 2.52%. It is worth noting that for LZO ETL, both J SC,exp and J SC,sim are in agreement, leading to a ratio of ≈ 1. In addition, for thicker active layers (145 and 155 nm), the ratios decreased below 0.9. Thus, we conclude that as one increases the active layer thickness, the absorption increases but the J SC continues to decrease, demonstrating that the increase in light absorption does not lead to a higher photocurrent under short-circuit conditions. This is accompanied with a plunge of FF slightly from 65.69 to 63.50%. This phenomenon occurred because of poor charge transport properties even though it is in relatively thick active layers. Above all, the thinnest device, of 100 nm, demonstrated the best performance with a high PCE of 6.36%, resulting from strong absorption of the PBDTT-DPP:PC 71 BM, even in thin films (see Fig. 1b). Figure 3b depicts the EQE spectra of a PBDTT-DPP:PC 71 BM single junction OPVs with different active layer thicknesses. As we can see from the spectra, the J SC values obtained from the integrated EQE are in agreement with the J SC values obtained from the J-V characteristics (Fig. 3a). The values are 12.97, 12.37, and 11.88 mA/cm 2 for 135, 145, and 155 nm, respectively.
These observations of PBDTT-DPP:PC 71 BM are a significant improvement compared to the previous study 14 , where they achieved a PCE of 6.12%, using PEDOT:PSS as the hole transport layer.

Table 2. Device performance of PBDTT-DPP:PC 71 BM-based inverted single junction solar cells with different BHJ layer thicknesses and different ETLs.
In the tandem geometry, the front and rear single junction organic solar cells are stacked in series, which implies that, for a well-performing tandem cell, the V OC of the polymer tandem solar cell is equal to the sum of the V OC 's of both the front and rear subcells. The J SC of the polymer tandem solar cells is limited by the lowest J SC of the two individual subcells. For maximum performance of the polymer tandem solar cells, the J SC of each subcell has to be matched. It is worth mentioning that, the front and rear subcells of the polymer tandem solar cells can also be measured individually by contacting the front (anode) and middle (cathode) electrodes for the front subcell, and the middle (anode) and top (cathode) electrodes for the rear subcell. As explained above, the combination of 95 nm PCDTBT:PC 70 BM for the front subcell (large bandgap) and 135 nm PBDTT-DPP:PC 71 BM for the rear subcell (small bandgap) results in an optimized optical and electronic coupling for the tandem cells in series. The 135 nm thickness of the active layer (PBDTT-DPP:PC 71 BM) of the rear subcell is optimized for performance. The structure of the polymer tandem solar cells is shown in Fig. 1d.
In order to guide the fabrication of the tandem solar cells, optical modeling using the transfer matrix formalism 27 was performed on relevant device architectures. Figure 4a Figure 4a shows that the simulations data as a function of different thicknesses for the front and rear subcells. One could achieve a maximum efficiency of 10% with suitable front and rear subcell thicknesses. Figure 4b presents the measured J-V characteristics of the front and rear subcells, along with the polymer tandem solar cells under 100 mW/cm 2 AM1.5 illumination. The values of the V OC , J SC , and FF extracted from the measurements are summarized in Table 3. The front subcell single   Table 3. Device performance of front, rear, and inverted tandem cells.
junction OPV demonstrates a V OC of 0.88 V, J SC of 11.13 mA/cm 2 , FF of 70.78%, and AM1.5 PCE of 6.93%. The rear subcell single junction OPV shows a V OC of 0.74 V, J SC of 13.08 mA/cm 2 , FF of 65.69%, and AM1.5 PCE of 6.36%. The ideal polymer tandem solar cells should demonstrate a V OC equal to the summation of the V OC 's of the front and rear subcells. The polymer tandem solar cells with PEDOT:PSS:GO/LZO ICL exhibit a V OC of 1.60 V, which is 0.02 V less than the ideal summation of the V OC 's of the front and rear single junction OPVs. The polymer tandem solar cells provide an AM1.5 PCE of 9.02%, compared to the 6.93 and 6.36% values achieved by the front and rear subcells, respectively. For reference, we fabricated the polymer tandem solar cells that do not benefit from the ICL. This tandem device demonstrated a significantly lower V OC of 0.73 V (data not shown), as well as a notably low J SC compared to the PEDOT:PSS:GO/LZO device. We speculate that this unacceptable performance is because of the formation of an undesired barrier to electron flow in the intended cascade from PC 71 BM to PCDTBT. Figure 4a and Table 3 demonstrate that the tandem structure improves the performance of the individual subcells (bottom and top cell), since the efficiency of the tandem cells is 30% higher than that of the front subcell and 42% higher than that of the rear subcell. The front subcell generates a photocurrent 28,29 lower than the rear subcell and limits the performance of the tandem cells in series configuration.
The EQE spectra (Fig. 4c) of the constituent single-junction solar cells further confirms current balancing. EQE measurements of the polymer tandem solar cell structures require special precaution due to the coupled light absorption and photocurrent-generation processes in each cell 30 . EQE measurements were taken with two excitation light sources. A 700 nm light optical bias light beam was used to excite only one of the subcells, while a 550 nm light was used to measure the EQE of the other subcell. The EQE spectra demonstrated an excellent balance in photocurrents generated by the front and rear subcells. The EQE spectra closely follow the absorptance spectra of the front and rear subcells, confirming that the photocurrents render from photoactive layers.
In this study, we have fabricated 43 inverted tandem polymer solar cells and measured them using optimized front and bottom subcell thicknesses. Figure S3 demonstrates the histograms of the photovoltaic parameters and it shows that our tandem devices are highly reproducible.
One of the most critical issues for series connected polymer tandem solar cells is current balancing in each subcell. It is accepted that, the rear subcell absorbs the light that is not absorbed by the front subcell and is illuminated under lower light intensities. However, as shown above, the front subcell produces lower photocurrents. This implies that in our tandem, the structure of the extracted photocurrent is almost the same as the photocurrent of the subcell that generates the lowest photocurrent. If the front subcell generates much more photocurrent, the excess electrons cannot recombine with the electrons from the rear subcell and will charge the ICL. This charge will partially compensate for the built-in voltage across the front subcell, until the photocurrent of the front subcell matches the photocurrent of the rear subcell. This results in deteriorated polymer tandem solar cell performance. Thus, to take full advantage of the tandem architecture, the photocurrent generated in the front subcell has to balance the photocurrent of the rear subcell. The photocurrent of the front subcell has to be adjusted until both photocurrents are almost identical.
Keeping the optimized PBDTT-DPP:PC 71 BM thickness of 135 nm constant, Fig. 5 and Table 4 show the variation of J SC with the PCDTBT:PC 70 BM layer thickness in the proposed tandem device. The thicker the PCDTBT:PC 70 BM active layer, the more light that is transmitted and can be absorbed by the PBDTT-DPP:PC 71 BM layer. Thus it is possible to balance the subcell photocurrent by adjusting the front subcell thicknesses. Clearly, the J SC increases with the increase of the PCDTBT:PC 70 BM thickness. Thicker layers will not only increase the series resistance (R S ), but also halt carrier transport. Furthermore, according to the working principles of bilayer OPV's, only the excitons in the PCDTBT:PC 70 BM layer can diffuse the interface for the photocurrent 31 . Therefore, for a very thick PCDTBT:PC 70 BM layer, some excitons will be lost. A very thick PCDTBT:PC 70 BM will also decrease the number of photons reaching the PBDTT-DPP:PC 71 BM layer due to their overlap in the absorption spectra. The ICL of the polymer tandem solar cells must have Ohmic contact in both the front and rear subcells and induce efficient recombination of charge carriers coming from both cells. In devices made by stacking solution processed films, the ICL must also prevent the solution, during the deposition, of the rear subcell from penetrating into the front subcell 17 . Many appealing and interesting ICL concepts have been put forward lately [32][33][34][35][36][37][38][39][40][41][42] . As we know, the ICL not only serves as the charge recombination region for charges coming from the front and rear subcells, but it also ensures the presence of suitable interface energy for efficiently recombining the charges from the subcells. In addition, the ICL must prevent any formation of a reverse built-in potential that will affect the V OC of the polymer tandem solar cells. Hence, the ICL should be optically transparent to avoid any absorption and reflection when projected light passes through the rear subcell back into the front subcell. We expanded our study on polymer tandem solar cells using five different sets of ICLs including i) PEDOT:PSS:NiO x /LZO, ii) PEDOT:PSS:rGO/LZO, iii) PEDOT:PSS:CNT/LZO, iv) PEDOT:PSS/LZO, and (v) PEDOT:PSS:WO x /LZO. Figure 6 exhibits the J-V characteristics of tandem cells under 100 mW/cm 2 AM1.5G illumination with different ICLs. The extracted photovoltaic parameters of the respective tandem cells are summarized in Table 5. As shown in Fig. 6, the tandem cells performance varied significantly with different ICLs. This shows that, apart from current balancing between the front and rear subcells, finding a suitable ICL is another crucial issue in designing high performance polymer tandem solar cells. This data demonstrates that polymer tandem   Table 4. Device performance of inverted tandem polymer solar cells with different front subcell thicknesses.

Interconnecting layer J sc (mA/cm 2 ) V oc (V) FF (%) PCE (%)
PEDOT  various organic solvents, such as chlorobenzene, chloroform, 1,2-dichlorobenzene, and 1,3,5-trichlorobenzene. Figure S1 shows that our new ICL has a good chemical resistance to prevent any penetration during the deposition of the rear subcell. However, the anomalous S-shaped curve was observed in the positive bias regime for PEDOT:PSS:NiO x /LZO ICL. This indicates that the ICL does not form ohmic contact. In the S-shaped curve, the photocurrent levels turn off with the increase of bias and further increase in the V OC region. This anomalous behavior essentially decreases both the FF and PCE. It has been frequently observed and attributed to (i) the local space charge in the multilayer device 43 , (ii) the defects at the cathode interface and other interfacial behavior [44][45][46] , and (iii) the strong imbalance of individual charge carrier mobilities 47 . The latest argument is based on vertical phase separation of the two organic components 48 . Villers and co-workers ascribed that the vertical phase segregation was controlled by subtle factors in the solvent evaporation kinetics during the spin-coating process 48 . Gupta and Glatthaar, along with their respective co-workers attributed that the S-shaped curve depends on the quality of the cathode interface and the presence of p-type impurity doping, respectively 49,50 .
In order to enable us to evaluate the specific electrical properties of the interfaces, we carried out the impedance spectra measurements in the dark for all fabricated inverted devices with different ICLs. Figure S4  To further understand and access in detail, the surface morphologies of the ICL were studied using atomic force microscopy (AFM). All AFM images were taken after the deposition of the corresponding ICLs on the front subcell. As shown in Fig. 7a- To support our AFM images data, the focused ion beam (FIB) images for all polymer tandem solar cells with different ICLs are shown in Fig. 8. We can see that the PEDOT:PSS:GO/LZO (Fig. 8b) obviously separated the front and rear subcells. Unlike PEDOT:PSS:NiO x /LZO, interconnecting layers do not have a clear separation of subcells, and large voids (bright region) can be observed (Fig. 8a). Based on the AFM and FIB experiments, it is believed that the dense PEDOT:PSS:GO/LZO ICL helps the separation of subcells and forms better interfacial contact between Finally, we concluded our detailed and systematic investigation on polymer tandem solar cells by considering any dependence on aperture. Several authors have previously reported the study of single junction OPVs with and without the presence of aperture or photo-masking 52 . For this purpose, we used an aperture with the same size of our polymer tandem solar cells' active area (0.04 cm 2 ). Figure S2 shows that, in the case without any existence of aperture, the J SC increases about 10% compared to that with aperture (8.53 mA/cm 2 ), which leads to a 10.03% jump in PCE. This is probably due to the light piping, where a remarkably huge number of charge carriers outside of the active area also flow toward the electrode. At the same time it influences the photocurrent and the real active area which contributes to the real photocurrent which would be a lot larger than it should be. All summarized photovoltaic parameters are without the existence of aperture, as shown in Table S1. Thus, one effective way to accurately to measure the OPV performance is by introducing the aperture into the active area.
We have demonstrated systematic studies on DPP containing low bandgap polymer PBDTT-DPP in multiple junction OPVs. Combining various different approaches including numerical simulation, morphological, interface, as well as device engineering, our single junction OPV exhibited an improved PCE of nearly 7% for PCDTBT-based OPVs. Moreover, the tandem cells demonstrated a high PCE of 9.02%, which represents the highest small-scale laboratory efficiency. These encouraging observations demonstrate that PBDTT-DPP is a very promising low bandgap polymer for high performance OPVs.
Tandem Devices Fabrication. The PCDTBT:PC 70 BM was deposited through a similar process as mentioned above. Then the PEDOT:PSS:GO and LZO were spin-coated on the active layer of the front subcell in sequence. The thickness of PEDOT:PSS:GO and LZO are 40 nm and 10 nm, respectively. Later, the PBDTT-DPP:PC 71 BM was spin-coated on the LZO layer, where the thicknesses of the active layer was controlled by the spin coating speed. Finally, the samples were transferred into the evaporation chamber (1 × 10 −7 Torr) to fabricate the MoO 3 (15 nm)/Ag (100 nm) electrode; the device area is 0.04 cm 2 .
Device Characterization. For tandem solar cells, the layers comprising of LZO/PCDTBT:PC 70 BM/ PEDOT:PSS:GO/LZO/PBDTT-DPP:PC 71 BM were electrically isolated using toluene and methanol along the perimeter, as defined by the area of the top electrode. This isolation prevents fringing effects and also eliminates over estimation of the photocurrents generated by the tandem cell. During the measurements and stability tests, a shadow mask (0.04 cm 2 ) with a single aperture was placed onto the tandem solar cells in order to identify its photoactive area. The current density-voltage (J-V) characteristics were recorded with a Keithley 2410 source unit. The EQE measurements were performed using an EQE system (Model 74000) obtained from Newport Oriel Instruments USA and HAMAMATSU calibrated silicon cell photodiodes were used as the reference diode. The wavelength was controlled with a monochromator 200-1600 nm.