Balancing efficiency and transparency in organic transparent photovoltaics

The study addresses how to balance transparency and power conversion efficiency (PCE) in transparent photovoltaics (TPVs), a key challenge that limits practical deployment on windows, wearables, and building-integrated systems. While TPVs with AVT >50% have reached ~10% PCE using visible-absorbing materials or tandem designs, achieving AVT ≥80% typically causes steep efficiency losses. Using near-infrared (NIR) harvesting organic semiconductors (PTB7-Th:IEICO-4F) and highly transparent electrodes, the authors propose and test the hypothesis that active-layer thickness predominantly dictates AVT, whereas the donor–acceptor (D–A) ratio primarily tunes efficiency via morphology and charge transport, enabling precise control over the transparency–efficiency trade-off in solid-state TPVs.

Prior work shows UV-selective TPVs readily achieve AVT >70% but are limited to ~1% PCE due to the low UV photon flux (theoretical limit <7%). NIR-harvesting organic semiconductors promise higher efficiencies because over half of solar energy lies in the NIR, but previous semitransparent devices often used materials with significant visible absorption (e.g., PM6:Y6) or required liquid electrolytes, constraining transparency or practicality. Semitransparent organic solar cells have demonstrated good performance at AVT <50%, and optical screening/simulation frameworks exist, but reports with AVT >70% typically struggle to balance aesthetics and output. This work leverages a low-visible-absorption blend (PTB7-Th:IEICO-4F) and highly transparent electrodes (Ag nanowires, PH1000) to extend and systematize the balance toward AVT up to ~85% in solid-state TPVs.

Materials: ITO-coated glass substrates, PEDOT:PSS (Clevios P VP.AI 4083), PTB7-Th (donor), IEICO-4F (acceptor), ZnO nanoparticles (prepared per reference), Ag nanowires (Ag NWs) ink, PH1000 (conductive polymer), MoO3, SDS sacrificial layer, SEBS and SU-8 for ultraflexible substrates, PDMS for PH1000 lamination. Device architectures: Normal TPV: Glass/ITO/PEDOT/PTB7-Th:IEICO-4F/ZnO/Ag NWs. Inverted TPV: Glass/ITO/ZnO/PTB7-Th:IEICO-4F/MoO3/PH1000. Ultra-flexible modules: SEBS/SU-8/ITO/PEDOT/PTB7-Th:IEICO-4F/ZnO/Ag NWs, fabricated on glass with SDS sacrificial layer and then peeled to free-standing. Active layer preparation: PTB7-Th and IEICO-4F dissolved in chlorobenzene with 4% CN additive. Two key knobs were varied: (1) D–A ratio (1:1 to 1:10), and (2) total solution concentration (5–25 mg/mL), to tune film thickness and visible absorption. Spin coating parameters: typically 2000 rpm for 60 s; thermal anneals at 100 °C for 10 min. PEDOT:PSS was spin coated at 4000 rpm, 25 s; annealed at 120 °C for 10 min. Transparent electrodes: Ag NWs spin coated at 2000 rpm for 50 s as top electrode (normal architecture). PH1000 was transferred by lamination onto evaporated 10 nm MoO3 (inverted architecture). Both top electrodes were patterned by pulsed laser to define device areas and module interconnects. Laser patterning: For small-area devices (active area 5 mm²), Ag NWs were etched by pulsed laser (4.5 W, 1000 mm/s, 80 kHz). For modules, three laser lines P1 (ITO: 10.5 W, 400 mm/s, 20 kHz), P2 (after PEDOT/active/ZnO: 4.5 W, 1000 mm/s, 20 kHz), and P3 (Ag NWs: 6 W, 1000 mm/s, 80 kHz) enabled series interconnection. For ultraflexible modules, reduced laser powers/frequencies were used (P2: 0.3 W, 1000 mm/s, 55 kHz; P3: 0.3 W, 1000 mm/s, 80 kHz). Characterization: Film and device transmission/reflection (Cary 6000i UV–Vis–NIR); thickness (profilometer). Morphology and structure via SEM, TEM (Talos F200X G2), and GIWAXS (Cu Kα, sample-detector distance ~70 mm). J–V under AM1.5G (450 W Xe lamp, calibrated to Si reference; 50 mV/s scan, 30 ms dwell). EQE (Newport Quant X-300). AVT and CRI computed from spectral data; model fitting of AVT/PCE/LUE contours performed in Origin 2021. Theoretical AVT modeled using Beer–Lambert law with empirical ~8% reflection loss and human eye response weighting over 430–670 nm. Ultra-flexible devices: Built on glass/SDS with SEBS (~5 µm)/SU-8 (~1 µm) substrate; ITO sputtered (180 nm). Devices peeled off to free-standing structures. Three-cell modules fabricated with same stack and patterning on flexible substrates.

• Transparency–efficiency control: AVT of TPVs precisely tuned from ~40% to ~85% by controlling active-layer thickness (via solution concentration) and D–A ratio. PCEs up to 4.06% (AVT ~70.6%), 3.46% (AVT ~75.6%), and 2.38% (AVT ~80.4%). All devices with AVT >70% exhibited CRI >85.
• Dominant role of thickness in transparency: Film/device AVTs are strongly correlated with active-layer thickness; thresholds: AVT >50% around ~115 nm; AVT >60% (<100 nm), >70% (<70 nm), >80% (<30 nm). D–A ratio has minimal impact on AVT at fixed thickness but significantly affects PCE via morphology/transport.
• Electrode transparency: Device AVT closely tracks film AVT (within ~2%); highly transparent electrodes (Ag NWs, PH1000) and charge transport layers contribute minimal additional loss, enabling high AVT devices.
• Morphology–performance link: GIWAXS/TEM show optimal π–π stacking and continuous D–A networks at D–A ≈ 1:2, correlating with higher JSC/EQE and PCE. At ≥1:4, ordering weakens and phase separation increases, degrading charge transport and efficiency.
• Performance limits at extreme thinness: Active-layer thickness <30 nm causes sharp efficiency drops; devices below ~15 nm perform poorly, setting a practical AVT ceiling ~82% for PTB7-Th:IEICO-4F.
• Double-sided operation: Devices operate from both sides; for an ~80.5% AVT device, illumination from ITO side yielded higher efficiency than from Ag NW side due to higher photocurrent (EQE-confirmed).
• Modules on rigid substrates: Series-connected 3-cell (~1.05 cm²) modules: PCE ~2.93% (AVT ~70%) and 1.37% (AVT ~80%). Eight-cell (~10.3 cm²) modules: PCE ~2.31% (AVT ~70%) and ~1.12% (AVT ~80%). Diffuse reflector (paper) behind module increased JSC by ~50% and boosted PCE from 1.37% to 1.90% for an ~80% AVT module. Output voltages >5 V demonstrated smartphone charging.
• Ultra-flexible modules: Three-cell ultraflexible module on glass showed Voc 2.04 V, JSC 2.87 mA cm⁻², FF 52.0%, PCE 3.04% (AVT ~70–80% sets). After peel-off to free-standing, PCE ~2.44% due to JSC reduction; AVT changed slightly (e.g., 70.8% to 70.0%). Devices are ultra-thin, lightweight, and conformable.
• Stability: Unencapsulated devices (rigid and ultraflexible, AVT ~80%) stored in glovebox for two weeks retained >80% of initial efficiency after an initial slight drop.

The study resolves the transparency–efficiency dilemma in solid-state organic TPVs by decoupling the roles of active-layer parameters: film thickness sets visible-light transmission (AVT), while D–A ratio tunes morphology and charge transport to maintain efficiency at a given transparency. Modeling and extensive device statistics (>200 devices) provide contour maps for AVT, PCE, and LUE, enabling target-driven design (e.g., AVT ≥80% with optimized D–A around 1:1.5–1:2 and suitable thickness). GIWAXS and TEM confirm that near 1:2 D–A yields strong π–π stacking and continuous pathways, maximizing EQE and PCE; donor-lean blends (≥1:4) lose ordering and continuity, reducing IQE and JSC. Highly transparent electrodes (Ag NWs, PH1000) and charge transport layers contribute little additional optical loss, so device transparency closely follows film transparency, validating the thickness-first strategy. Practical demonstrations with large-area and ultraflexible modules indicate that the achieved balance supports real-world functionalities (e.g., >5 V output for charging, high CRI for aesthetics), reinforcing the relevance of the approach for building-integrated and wearable applications. The findings delineate a practical AVT ceiling (~82%) for this material system due to performance collapse at ultrathin active layers, framing clear design boundaries.

This work establishes a systematic framework to balance transparency and efficiency in NIR-harvesting organic TPVs. By identifying active-layer thickness as the primary determinant of AVT and D–A ratio as the key lever for photovoltaic performance via morphology and transport, the authors realize TPVs with AVT from 40% to >80% and PCE up to 4.06% (>70% AVT) and 2.38% (>80% AVT). The approach translates to large-area (~10 cm²) and ultraflexible modules with high CRI and useful power output, demonstrating applicability to smart windows and wearable electronics. Future research should focus on discovering and engineering semiconductors with stronger invisible-light (UV/NIR) harvesting but minimal visible absorption, optimizing morphology at high transparency (e.g., around 1:2 D–A), and developing more flexible, stable transparent electrodes to further raise the AVT ceiling and durability.

• Material-system ceiling: For PTB7-Th:IEICO-4F, practical AVT is limited to ~82% because active layers thinner than ~30 nm cause sharp efficiency losses and below ~15 nm devices perform poorly.
• Morphology sensitivity: Efficiency degrades at donor-lean ratios (≥1:4) due to weakened ordering and phase separation; achieving optimal morphology at very thin films is challenging.
• Electrode constraints: Ultrathin flexible devices rely on sputtered ITO, which can crack or degrade upon peeling, contributing to JSC loss; more robust flexible transparent electrodes are needed.
• Stability assessment: Preliminary stability was conducted only in a glovebox, unencapsulated, over two weeks; long-term outdoor durability and encapsulation effects remain untested.
• Generalizability: Findings are demonstrated for one donor–acceptor pair; while principles should transfer, exact thresholds and performance may vary with other materials and device stacks.