Thick LiF Cathode Buffer Layer Enhanced by a C60 Interlayer Boosts Polymer Solar Cell Efficiency

Abstract

Lithium fluoride (LiF) is a proven cathode buffer layer (CBL) in bulk‑heterojunction polymer solar cells (PSCs). Conventional LiF layers are restricted to ~1 nm because of their insulating nature, which complicates deposition control and offers insufficient protection to the underlying active layer. Here we demonstrate that a much thicker LiF layer can be employed without sacrificing efficiency by inserting a thin C60 interlayer between the active material and LiF. PSCs with a C60 (3 nm)/LiF (5 nm) bilayer achieve a peak power‑conversion efficiency (PCE) of 3.65 %, twice the 1.79 % obtained with a 5‑nm LiF layer alone. The enhanced performance originates mainly from improved electrical conductivity at the C60/LiF interface due to intermixing, the formation of a P3HT/C60 sub‑cell, and the optical spacer effect of C60. When the LiF thickness is increased to 8 nm, the C60/LiF bilayer still delivers a PCE of 1.10 %, whereas a single LiF layer shows negligible activity. These results confirm that a C60/LiF bilayer is a robust alternative to thin LiF, offering high tolerance to thickness variations.

Background

Solution‑processed bulk‑heterojunction polymer solar cells have attracted considerable attention for their low cost, flexibility, and scalability. However, their power‑conversion efficiencies (PCEs) remain below those of silicon photovoltaics. Recent breakthroughs, reaching 11–13 %, stem from novel donor polymers and non‑fullerene acceptors. In addition, inserting anode or cathode buffer layers (ABLs/CBLs) can further enhance device performance.

In conventional PSCs, the ITO anode is complemented by a low‑work‑function cathode buffer such as Ca. Unfortunately, Ca oxidises readily, compromising stability. Lithium fluoride (LiF) is a widely used CBL because it creates an interfacial dipole that reduces the cathode work function. Nevertheless, LiF’s insulating character limits its practical thickness to less than 2 nm, typically ~1 nm, making precise deposition challenging and offering inadequate protection during metal evaporation.

To overcome these constraints, we previously reported a multilayer C60/LiF stack that preserved high conductivity even with thick LiF layers, but the deposition sequence was complex and costly. In this work, we simplify the architecture by employing a single C60/LiF bilayer. By depositing a C60 interlayer before LiF, we can use a thicker LiF without sacrificing efficiency. Devices with C60 (25 nm)/LiF (1–6 nm) maintain ~3 % PCE, while a 5‑nm LiF alone drops sharply. The peak efficiency of 3.77 % for the bilayer is ~23 % higher than the 3.06 % of LiF‑only devices, demonstrating the superiority of the bilayer.

Methods

Fabrication of PSCs

ITO‑coated glass substrates (Delta Technologies, LTD) were cleaned in acetone and isopropyl alcohol (IPA) by sonication (5 min each) and then treated with O2 plasma (60 s) to render the surface hydrophilic. A filtered PEDOT:PSS (Clevios PH 500, H.C. Starck) solution was spin‑coated at 2000 rpm for 50 s and baked at 110 °C for 20 min under nitrogen. The samples were then transferred to a N2‑purged glovebox (<0.1 ppm O2/H2O) for spin‑coating the photoactive layer.

P3HT (Rieke Metals Inc., 4002‑EE, 91–94 % regioregularity) and PCBM (American Dye Source, >99.5 % purity) were dissolved in chlorobenzene at a 1:1 weight ratio. The solution was filtered (0.45 µm) and spin‑coated on top of the PEDOT:PSS at 1000 rpm for 50 s, followed by thermal annealing at 130 °C for 20 min. This process yields a ~160 nm active layer (measured by Dektek profilometer). Sequential thermal evaporation (base pressure 1 × 10^−6 mbar) deposited C60, LiF, and Al (75 nm). A 1 mm circular shadow mask defined the active area before Al deposition.

Characterization

Current density–voltage (J–V) characteristics were recorded with a Keithley 2400 under simulated AM 1.5 G (100 mW cm^−2) illumination, calibrated against a reference silicon cell. Measurements were performed inside the glovebox. Atomic force microscopy (AFM) images were captured in tapping mode using a Veeco Dimension‑Icon. UV/Vis absorption spectra were obtained with a Varian Cary 50. Photo‑induced charge extraction by linearly increasing voltage (Photo‑CELIV) was performed on PSCs under ambient conditions: a pulsed N2 laser (337.1 nm, 1.4 ns) generated carriers, followed by a reverse‑bias ramp after a 100 µs delay. Current transients were recorded on a digital oscilloscope (50 Ω input). Charge carrier mobility (μ) was calculated from the standard Photo‑CELIV equation: μ = (2d^2)/(3At_max^2[1+0.36Δj/j(0)]).

Results and Discussion

Figure 1 illustrates the J–V curves of PSCs with and without a C60 interlayer between the active layer and a 5‑nm LiF buffer. Devices lacking C60 exhibit a pronounced S‑shaped curve, leading to a low fill factor (FF) and consequently reduced PCE, despite normal Jsc and Voc values. The S‑shape arises from LiF’s insulating nature, which increases series resistance (Rs) and decreases shunt resistance (Rsh). Introducing a 3‑nm C60 layer eliminates the S‑shape and raises FF from 32.4 % to 56.3 %. FF peaks at 67 % for an 8‑nm C60 layer before decreasing slightly with further thickness. The corresponding PCE rises to 3.65 %, twice the 1.79 % of the LiF‑only device. Device parameters are highly reproducible (standard deviations <0.2 %), confirming the reliability of these results.

AFM analysis (Figure 2) reveals that the C60 layer forms large spherical aggregates, while LiF deposits as smaller islands. When the 5‑nm LiF is evaporated onto the 35‑nm C60 layer, the underlying C60 aggregates remain partially exposed, leading to intermixing at the C60/LiF interface. This mixed morphology provides a percolation network that enhances electrical conductivity of the bilayer, as evidenced by the improved FF.

Figure 3 compares devices with a single LiF CBL and a C60/LiF bilayer while varying LiF thickness from 0.5 to 8 nm. LiF‑only devices achieve a maximum PCE of 3.06 % at 1 nm, but performance drops sharply to 0.79 % (6 nm) and 0.06 % (8 nm). In contrast, C60 (25 nm)/LiF bilayers retain high efficiencies: 3.77 % at 1 nm, 2.65 % at 6 nm, and 1.10 % at 8 nm. The improvement stems from a higher FF and Jsc, attributable to reduced Rs. Photo‑CELIV measurements (Figure S1) show that the C60/LiF bilayer boosts electron mobility by an order of magnitude compared to LiF alone, confirming enhanced charge transport.

Further analysis shows that the C60 interlayer also enhances Jsc by forming a P3HT/C60 sub‑cell and acting as an optical spacer. Devices with ITO/PEDOT:PSS/P3HT/C60 (25 nm)/LiF/Al and varying P3HT thickness demonstrate that Jsc peaks when the P3HT layer is ~10 nm, matching the exciton diffusion length in P3HT (~10 nm). This effect accounts for an additional ~1 mA cm^−2 increase in Jsc in the bilayer devices.

Simulations of the electric field distribution (Figure S2a) reveal that incorporating C60 weakens the short‑wavelength field but strengthens it at longer wavelengths, leading to a red‑shifted absorption profile. Optical spectra (Figure S2b) confirm that the C60/LiF bilayer does not introduce additional absorption in the visible range, while enhancing absorption in the 400–510 nm and 580–680 nm windows. Incident photon‑to‑current conversion efficiency (IPCE) measurements (Figure S2c) show a modest decrease at short wavelengths (due to C60 parasitic absorption) but a notable increase at longer wavelengths, reflecting the optical spacer effect and sub‑cell contribution.

When the LiF layer reaches 8 nm, the C60/LiF bilayer’s PCE falls to 1.10 % due to increased Rs. AFM images (Figure S3) show that at this thickness the LiF completely covers the C60 aggregates, preventing intermixing and creating a blocking layer that hinders electron extraction. Consequently, the bilayer remains more tolerant to LiF thickness variations than a single LiF layer.

Conclusions

We have shown that a thick LiF buffer can be successfully employed in P3HT:PCBM PSCs by introducing a C60 interlayer. The C60 (25 nm)/LiF (5 nm) bilayer delivers a peak PCE of 3.65 %, double that of the LiF‑only device (1.79 %). The performance gains stem from a higher FF—thanks to the conductive C60/LiF interface—and a modest Jsc increase from the P3HT/C60 sub‑cell and optical spacer effect. Even at 8 nm LiF, the bilayer retains a PCE of 1.10 % versus 0.06 % for LiF alone. The decline at 8 nm is attributed to the lack of intermixing and the resulting electron transport barrier. Overall, the C60/LiF bilayer offers a simple, cost‑effective route to robust, high‑efficiency PSCs with relaxed LiF thickness constraints.