Copper thiocyanate: An attractive hole transport/extraction layer for use in organic photovoltaic cells

Thin film solution-processed solar cells are recognized as a promising technology that might provide society with a clean and renewable source of energy.¹ Recent advances in synthetic design and device fabrication have yielded significant increases in power conversion efficiency (PCE). For example, researchers have recently achieved PCE > 10% for polymer:fullerene blends² and 17%–20% for perovskite based hybrid systems.³ Typically, these devices consist of multiple layers that are deposited sequentially from orthogonal solvents, with the most common device architecture comprising an active (donor-acceptor blend) layer sandwiched between hole (HTL) and electron (ETL) collection/transport layers, in turn sandwiched between the anode and cathode contacts. Efficient power conversion on a commercial scale will not only depend on the charge separation efficiency of the active layer but also depend on utilization of robust and cost-effective HTLs and ETLs. The characteristics of an archetypal HTL are: (1) facile deposition and subsequent post-processing conditions, (2) negligible parasitic absorption across the solar spectrum, (3) Ohmic contact with anode and active layer donor materials, and (4) appropriate energy levels for electron carrier blocking from the active layer acceptor material.^4,5 Importantly, a ubiquitous HTL should retain these characteristics reasonably independently of the choice of active layer materials.

Since the initial development of efficient solution processed organic solar cells, poly(3,4-ethylenedioxythiophene):polystyrenesulfonate (PEDOT:PSS) has been used as the standard HTL. The benefits of PEDOT:PSS include facile coating with mild processing conditions that produces smooth films (i.e., RMS roughness less than 2 nm)⁶ over large areas and allows Ohmic contact with a wide variety of materials. However, in spite of its many desirable properties, there is still significant room for improvement. One specific challenge related to PEDOT:PSS is that the work function and surface composition of the film are sensitive to the processing conditions.⁷ The challenge is highlighted by the work of Koch et al., who found that the work function of PEDOT:PSS (Baytron P CH8000) can vary by as much as 0.6 eV (i.e., from 5.05 eV to 5.65 eV) when drying the films in vacuum or with a combination of vacuum and temperature.⁸ This study also found that the surface composition of PEDOT:PSS varies from 1:8 to 1:24, possibly leading to non-optimal electrical contact between the active layer and PEDOT:PSS. The effect on device performance of both PEDOT:PSS film thickness and thermal annealing has also been reported.⁹ A further concern with PEDOT:PSS is that it is not a good electron blocking layer and has a tendency to trap electrons leading to changes in device characteristics over time.¹⁰ Removal of PSS is possible using vapour phase polymerization routes and offers the potential for significantly enhanced conductivity, potentially allowing use as an anode replacement electrode.¹¹ Another approach has been to use dipolar self-assembled monolayers.¹² Finding a HTL material that can simultaneously address these issues and thereby enable reliable production of high PCE solar cells insensitive to fabrication environment remains a compelling goal.

We have recently reported copper(I) thiocyanate (CuSCN) as an alternative HTL for both solar cells^6,13 and light emitting diodes.¹⁴ This inorganic molecular semiconductor is highly soluble in solvents such as diethylsulfide that are orthogonal to those commonly used to process typical organic active layers and can be readily processed to yield thin films with ultra-high transparency across the solar spectrum. Furthermore, CuSCN was found to have relatively high hole mobilities (0.01–0.1 cm² V⁻¹ s⁻¹), achieved after mild heat treatment (∼100 °C) compatible with inexpensive plastic substrates. Furthermore, CuSCN acts as an efficient HTL for a variety of active layer material systems, ranging from polymer or small molecule:fullerene bulk heterojunctions⁶ to hybrid perovskite solar cells.¹⁵ Owing to CuSCN's superior optical transparency, we were previously able to achieve up to a 25% enhancement in the short circuit current density (J_SC) in low band gap polymer:fullerene organic photovoltaic (OPV) cells, improving the PCE from 7% to 8%.⁶ 

Herein, we report that, in addition to its desirable optical transparency, CuSCN possesses a deep valence band energy level, E_VB ∼ −5.35 eV, that can help to maximize open circuit voltage (V_OC). The large optical gap ensures that good electron-blocking characteristics are nevertheless maintained. This combination results in significant enhancements in PCE values and we demonstrate these benefits for cells with active layers comprising the deep-lying highest occupied molecular orbital (HOMO) energy (−5.15 (±0.03) eV) donor polymer poly(di(2-ethylhexyloxy)benzo[1,2–b:4,5–b′]dithiophene-co-octylthieno[3,4–c]pyrrole-4,6-dione) (PBDTTPD) blended with the archetypal fullerene electron acceptor phenyl-C₆₁-butyric acid methyl ester (PC₆₁BM) [Figures 1(a) and 1(b)].¹⁶ We show that OPV cells based on a CuSCN HTL produce ∼0.15 V higher V_OC relative to PEDOT:PSS-based cells. Furthermore, because CuSCN is an intrinsic semiconductor,¹³ we show that the energy levels lead to superior electron blocking characteristics. Ultimately, because of the enhanced light transmission and advantageous energy levels, we were able to fabricate PBDTTPD:PC₆₁BM-based OPV cells with maximum PCE = 8.07%, a significantly higher value than the 7.3% value reported previously for PEDOT:PSS-based cells.¹⁷ The CuSCN's chemical stability combined with its unique electronic properties and inexpensive nature makes it an ideal candidate for application in large-volume plastic optoelectronics.

The workfunction energies for CuSCN and PEDOT:PSS were determined using an ambient pressure APS02 air photoemission system (KP Technology) (APS) employing Fowler's analysis.¹⁸ HTL films of 45 nm thickness were deposited on clean 20 Ω/sq ITO coated glass substrates by spin coating filtered solutions of 20 mg/ml CuSCN in diethylsulfide and the as-sold PEDOT:PSS aqueous suspension (Clevios PVP AI 4083). They were measured in dry air after annealing for 20 min at 100 °C for CuSCN and 170 °C for PEDOT:PSS.¹⁹ Figure 1(c) shows the normalized cube root of the photoelectron yield, (J_PE)^1/3, as a function of photon energy for CuSCN and PEDOT:PSS layers deposited onto conductive ITO electrodes. Detailed analysis of the measured data is provided in Figures S1 and S2¹⁹ From the intercept of the tangential (dotted line) fit with the (J_PE)^1/3 background axis, we derive a work function (ϕ) value for CuSCN of −3.35 (±0.11) eV, which is ∼0.35 eV deeper than the work function for our PEDOT:PSS films. We note that the value for ϕ extracted for the 45 nm-thick CuSCN layers by the APS method is highly reproducible and remains largely independent of the underlying electrode/interlayer material that the layer is deposited onto. As we will show, these differences in energy level have a profound effect on solar cell PCE.

To emphasize the role of HTL workfunction, we fabricated solar cells with a bulk heterojunction active layer comprising an 1:1.5 blend of PBDTTPD and PC₆₁BM. We utilized APS to measure the HOMO energy of PBDTTPD, yielding approximately −5.15 (±0.03) eV, i.e., ∼0.15 eV deeper lying than the PEDOT:PSS workfunction (Figure 1(c) and Figure S3).¹⁹ Importantly, the workfunction of CuSCN lies ∼0.2 eV deeper still, which yields solar cells with V_OC values not electronically limited by the HTL. We note that earlier studies on CuSCN-based OPVs did not report similar improvements in V_OC most likely because the workfunction values of CuSCN and PEDOT:PSS were both deeper than the HOMO level of the donor polymer, namely, poly(3-hexylthiophene) (P3HT), used (cf. Figure S4).¹⁹ The HTLs (45 nm-thickness) were first deposited on clean ITO-coated glass substrates, as described above for APS measurements, followed by spin-coating of the PBDTTPD:PC₆₁BM layer from 12 mg/ml chlorobenzene solution under dry nitrogen. Fabrication was completed by evaporation of a bilayer Ca/Al (10/90 nm) cathode. The current-voltage (J-V) characteristics of the resulting solar cells were measured under AM1.5G illumination and are plotted in Figure 1(d). OPV cells fabricated with a PEDOT:PSS HTL exhibited median values (from 6 cells fabricated simultaneously) for short circuit current density (J_SC), V_OC, fill factor (FF), and PCE of 11.17 mA/cm², 0.80 V, 0.66, and 5.94%, respectively, relative to 11.02 mA/cm², 0.95 V, 0.69, and 7.20% for CuSCN-based devices. Even though CuSCN is more transparent than PEDOT:PSS,⁶ both device types produced similar J_SC. This is plausibly due to compensating differences in optical interference within the active layer, resulting from the different HTL refractive indices; behavior that can be further engineered by optimizing the active layer thickness.^6,20

Interestingly, when replacing PEDOT:PSS with CuSCN, the FF is found to increase from 0.66 to 0.69 consistent with improving charge collection. This may in part be due to dipole formation resulting from an energetically favorable electron transfer between PBDTTPD and CuSCN.²¹ However, our data cannot rule out the possible effect from changes in active layer blend composition near the HTL interface.^5,22 Finally, we observe a significant increase in the V_OC from 0.80 V to 0.95 V for OPV cells when PEDOT:PSS is replaced by CuSCN as the HTL [Figure 1(d)]. We note that this significant enhancement is consistent and highly reproducible for a large number of devices fabricated during this study. Importantly, the V_OC enhancement appears to be independent of the atmospheric conditions under which spin casting of the CuSCN layer was performed, hence making the proposed approach robust and potentially highly scalable.

Given the minimal leakage current for both device types (see Figure S5),¹⁹ it appears that the 0.15 V reduction in V_OC for PEDOT:PSS devices is most likely due to the PBDTTPD HOMO being ∼0.15 eV deeper than the PEDOT:PSS workfunction. Although previous studies have reported similarly high V_OC values for the same materials combination, here the high V_OC = 0.95 V was only obtained from CuSCN devices with the PEDOT:PSS cells consistently underperforming. For the CuSCN containing devices, the interfacial energy alignment does not limit V_OC, leading ultimately to a 17% higher PCE. The apparent correlation between CuSCN workfunction and measured V_OC and FF values highlights the importance of HTL material selection and the tremendous potential of CuSCN for application in OPVs.

An ideal HTL should also have the ability to limit the passage of dark injected minority carriers in reverse bias (i.e., for negative bias applied to the anode) from the anode, through the active layer to the cathode. To investigate this, we have electrically characterized standard structure solar cells [Figure 2(a)] in dark conditions. Figure 2(b) displays the measured current-voltage characteristics. The first important observation is the striking difference in the reverse bias characteristics for CuSCN and PEDOT:PSS-based devices. In the latter case, the current density reaches 0.019 mA/cm² at −0.5 V reverse bias, whereas for CuSCN HTL devices it is more than one order of magnitude lower (0.001 mA/cm²). This direct observation confirms that CuSCN does indeed suppresses minority carrier injection from the metal electrode into the PBDTTPD:PC₆₁BM active layer. As a result, CuSCN-based cells exhibit a shunt resistance that is >100 times higher (1.5 × 10⁵ Ω cm) than that measured for PEDOT:PSS-based devices (8.5 × 10³ Ω cm). This is not surprising if one considers the very high LUMO energy of intrinsic semiconducting nature of CuSCN as compared to the semi-metallic nature of PEDOT:PSS.

When forward bias is applied, holes are efficiently injected into the active layer once the built in potential (V_bi) is passed. By analyzing the J-V slope in forward bias, we can estimate that the series resistance of the diodes using CuSCN as HTL is 1.8 times higher (2.4 Ω cm) than for PEDOT:PSS (1.3 Ω cm). Furthermore, by fitting the experimental J-V curves with the Mott-Gurney relationship²³ [Figure 2(b)], we estimate that the V_bi value for CuSCN devices is 0.12 V higher than for PEDOT:PSS devices. This is consistent with both APS and OPV data described above, hence further supporting the important role of the materials energetics. Critically, replacing PEDOT:PSS with CuSCN results in an increase in the shunt resistance of the diode. This has a synergistic effect on the ability to collect holes and ultimately improves V_OC, FF, and PCE in the solar cells.

CuSCN-based OPV cells with a PBDTTPD:PC₆₁BM (1:1.5) active layer can be further optimized by varying the donor:acceptor blend film thickness to maximize optical absorption. Because CuSCN has a different refractive index, the dependence of J_SC on the active layer thickness (dictated by interference effects and their dependence on incident wavelength) will differ from that for PEDOT:PSS HTL devices (see Figure S5).¹⁹ Figure 3(a) shows J-V characteristics for optimized cells with an 90 nm-thick PBDTTPD:PC₆₁BM active layer and a 45 nm-thick CuSCN HTL, measured both under illumination (100 mW/cm², AM1.5G) and in the dark. These cells give V_OC = 0.92 V and FF = 0.69 together with an impressive J_SC = 12.74 mA/cm², resulting in a PCE value of 8.07%. The high J_SC value was confirmed by measurements of external quantum efficiency (EQE), which yielded an average 64% across the 400–680 nm range with a peak value of ∼75% at 645 nm [Figure 3(b)]. These results represent the highest PCE reported to date for cells based on PBDTTPD:PC₆₁BM active layer blends and highlight the advantageous characteristics of CuSCN as the HTL. Such a high efficiency is even more notable when taking into account that the active layer was deposited from a simple solvent system (i.e., chlorobenzene) and was utilized as cast, without any post-deposition thermal/solvent vapour annealing step. In addition, no ETL was required to reach these rather impressive values making device manufacturing significantly simpler.

In conclusion, we have shown that CuSCN exhibits highly desirable energy levels for use as a HTL in high efficiency organic solar cells. In addition to its optical transparency, CuSCN possesses a relatively deep workfunction, which facilitates the formation of Ohmic contact with the p-type component of the donor-acceptor blend and does not limit V_OC. This characteristic enabled fabrication of CuSCN-based OPV cells with enhanced performance relative to control cells based on PEDOT:PSS. It was shown that CuSCN limits dark injection of minority carriers (electrons) and increases V_bi relative to PEDOT:PSS-based devices. Careful optimization of the PBDTTPD:PC₆₁BM active layer thickness then allowed simple structure CuSCN HTL solar cells to be fabricated with PCE values exceeding 8%.

N.D.T. acknowledges support from the National Science Foundation International Research Fellowship (OISE-1201915) and European Research Council (ERC) Marie Curie International Incoming Fellowship (Grant Agreement No. 300091) programs. N.Y.-G. and T.D.A. are grateful to the UK Engineering and Physical Sciences Research Council (EPSRC Grant No. EP/J001473/1) and T.D.A. and H.F. to the ERC (AMPRO Project No. 280221) for financial support. N.S. acknowledges support from an ERC Starting Independent Researcher Fellowship (Grant Agreement No. 279587). A.K.P. and D.D.C.B. acknowledge support from the Imperial College EPSRC Impact Acceleration Account and T.D.A., N.S., and D.D.C.B further acknowledge the EPSRC Centre for Innovative Manufacturing in Large Area Electronics (EP/K03099X/1) project. D.D.C.B. is the Lee-Lucas Professor of Experimental Physics.