Quantum barriers engineering toward radiative and stable perovskite photovoltaic devices

Metal–halide perovskite photovoltaic devices (PPVs) have reached high power conversion efficiencies, but further progress toward the detailed balance limit requires that devices operate as near-ideal light emitters. Recent PPVs report electroluminescence quantum efficiencies (ELQEs) exceeding 10%, yet interfacial non-radiative recombination—especially at charge-extraction contacts—limits radiative performance and open-circuit voltage. Contrary to common assumptions favoring thin emissive layers for outcoupling, the authors hypothesize that PPVs with thick perovskite absorbers can become brighter than typical perovskite LEDs due to enhanced photon recycling, provided interfacial quenching is suppressed. They propose engineering multiple quantum well (MQW) structures at the perovskite top interface using long (~3 nm) oleylammonium (OLA, C18) spacers introduced via an L-site exchange process to form thick, stable, and charge-selective barriers that suppress interfacial recombination, thereby boosting radiative efficiency, ELQE, Voc, and stability without prohibitive transport penalties.

• PPVs have primarily been optimized for light incoupling and charge collection, whereas PeLEDs optimize outcoupling; as a result, device architectures differ.
• ELQE >10% has recently been demonstrated in high-efficiency PPVs, but PeLEDs typically surpass PPVs in ELQE due to outcoupling design advantages.
• Photon recycling and microcavity effects can improve outcoupling, yet their quantitative contributions in reabsorbing perovskite thin films have been difficult to model; recent optical models resolve divergences in reabsorbing emitters and enable angular-resolved analyses.
• Organic surface treatments forming Ruddlesden–Popper 2D perovskite layers (L2An−1BnX3n+1) atop 3D perovskites are widely used for trap passivation and forming 3D/2D junctions to facilitate charge transfer. Most high-performance PPVs use short (~1 nm) spacers such as octylammonium (OA, C8) or phenethylammonium (PEA) that yield thin barriers with good conductivity.
• OLA-based thick barriers offer excellent crystalline stability but have been less used in high-efficiency PPVs due to poor conductance, especially when n=1 phases form via direct OLA iodide treatment, which severely hinders charge transport.
• Non-radiative quenching at charge-extraction interfaces (e.g., doped spiro-OMeTAD) introduces sub-gap states and dominates over intrinsic perovskite traps, necessitating interfacial strategies to suppress recombination.

• Device architecture: glass/FTO (600 nm)/SnO2 (~80 nm)/3D perovskite (~600 nm)/2D perovskite MQW (C8 or C18)/spiro-OMeTAD (~260 nm)/Au, with an antireflection coating on top.
• Perovskite absorber: (FAPbI3)0.97(MAPbBr3)0.03, spun from DMF/DMSO with MACl additive; antisolvent quench with ethyl ether; annealed at 150 °C for 20 min.
• MQW formation (L-site exchange): sequential spin-coating of OAI in chloroform (forms OA-based 2D perovskite) followed by oleylamine in octane to exchange OA (C8) with OLA (C18) at the L-site, yielding OLA-based MQWs while maintaining n=2 octahedral thickness (dQW ≈ 1.2 nm) and increasing barrier thickness from ≈1.3 nm (C8) to ≈2.6 nm (C18).
• Characterization of structure and energy levels: XRD to verify MQW periodicity (peaks at multiples of 3.56° for C8 and 2.33° for C18; lattice parameters 2.5 and 3.8 nm, respectively); comparison with separately prepared OLA-based n=2 films; UPS/IPES to determine band alignment and barrier heights.
• Optical measurements: PL and PLQE of neat films and perovskite/HTL stacks (continuous-wave 510 nm excitation ~0.1 W cm−2); transient PL decays via TCSPC (470 nm pulsed, 500 kHz, ~6×10^14 cm−3 initial carrier density) from the HTL side to assess charge transfer; spatial PL/reflection mapping via wide-field microscopy.
• Electroluminescence: ELQE measured using calibrated photodiode setup assuming Lambertian emission; EL spectra via spectrometer; current–voltage characterization in the dark.
• Optical modeling: transfer-matrix formalism with a recent reabsorption-stable method to compute outcoupling, parasitic absorption, scattering (empirical scattering coefficient added to k), and recursive photon recycling with internal radiative efficiency ηrad; identical internal spectra used to compare thick PPV vs thin PeLED structures.
• MQW charge transport modeling: one-dimensional transfer-matrix quantum calculation of carrier probability densities across stacked QWs for C8 vs C18 barrier thicknesses; effective masses mh=me=0.15 m0; perovskite wells Ec≈−4.00 eV, Ev≈−5.40 eV; organic barriers Ec≈−0.28 eV, Ev≈−6.57 eV.
• Photovoltaic testing: AM1.5G (Class AAA simulator), Si reference calibrated; J–V scans forward/reverse; aperture area 0.0957 cm2; EQE measurements; certification at Newport for selected devices. Stability: air storage (RT and 60 °C) without encapsulation; encapsulated MPP tracking under 1 sun in air.

• Optical design insight: Thick perovskite PPVs exhibit much higher photon reabsorption (Freabs ~88.5%) and lower parasitic absorption (Fpara ~6.6%) than thin PeLEDs (Freabs ~36.7%, Fpara ~45.9%). At the radiative limit (ηrad=100%), PPV ELQE is predicted to reach ~42%, surpassing thin PeLEDs.
• Interfacial engineering: L-site exchange converts OA (C8) MQWs to OLA (C18) MQWs while maintaining n=2 phase (dQW≈1.2 nm) and thickening barriers to dQB≈2.6 nm; XRD confirms periodicity (2.33° for C18, 3.56° for C8). UPS/IPES and quantum calculations show increased electron-blocking selectivity with C18 (barrier asymmetry ~0.33 eV vs 0.30 eV for C8).
• Suppression of interfacial quenching: PLQE of 3D perovskite drops to ~0.7% upon contact with doped spiro-OMeTAD; insertion of MQWs increases PLQE to 11.2% (C8) and 16.5% (C18), indicating dominant suppression of interfacial non-radiative recombination. Transient PL decays slow with C18 MQWs, consistent with hindered hole extraction due to larger valence band offset (Ev ~5.95 eV for C18 MQWs).
• Electroluminescence: ELQE increases from 16.8% (C8) to 19.7% (C18) at peak; at Jph=26.0 mA cm−2, ELQE increases from 13.5% (C8) to 17.8% (C18). Direct outcoupled emission (excluding recycling) is only 3.9% (ηrad80%) for C8 and 4.2% (ηrad86%) for C18, yet photon recycling nonlinearly amplifies external ELQE (32% relative gain from a 7.2% internal emission increase).
• Photovoltaic performance: Compared to 3D-only devices, C8 MQWs raise Voc from 1.164 V to 1.184 V and PCE to 25.79% (champion). C18 MQWs further increase Voc to 1.193 V with PCE=26.04% (Jsc=25.98 mA cm−2; FF=83.97%); certified PCE=25.16%. Series resistance increases (RA from 1.08 to 1.55 Ω cm2) with a slight FF decrease (84.15%→83.88%), but average PCE still improves (25.57%→25.81%). Voc improvement is mainly from enhanced photon recycling rather than direct emission.
• Stability: Encapsulated C18 devices retain 92% (500 h) and 83% (1,150 h) of initial PCE under continuous 1 sun in air, versus 77% and 66% for C8. At 60 °C in air, C18 retains 84% after 300 h, while C8 drops to 68% within 18 h. Long-term air storage: ηEL(ph) ~12% after 2 months and PCE ~22% after 2 years for C18 devices. XRD indicates C18 MQWs maintain crystallinity on 3D perovskites, whereas C8 MQWs deform within a week.
• Fundamental benchmark: Devices with C18 reach ~96.4% of the detailed balance Voc limit, approaching GaAs performance.

The study shows that when interfacial quenching is minimized, thick perovskite PPVs benefit strongly from photon recycling, overturning the conventional preference for thin emissive layers in LED design. By introducing thick, stable, and electron-selective OLA (C18) MQW barriers via L-site exchange, interfacial non-radiative recombination at the perovskite/HTL interface is substantially reduced. Although thick barriers modestly increase series resistance and slightly reduce FF, the concurrent increase in internal radiative efficiency yields a disproportionate rise in external ELQE and Voc due to nonlinear photon recycling. Optical modeling quantitatively separates direct emission from recycled contributions, confirming that most Voc gains originate from recursive photon recycling rather than incremental direct outcoupling. The L-site exchange maintains an n=2 phase with moderate conductivity, overcoming the transport limitations typically associated with thick (C18) barriers formed via direct OLA iodide deposition (which often yields n=1 phases). The approach reconciles radiative performance and operational stability: OLA-based MQWs stabilize interfacial crystallinity, leading to significantly improved photo-, thermal-, and air-stability. These results redefine design rules for radiative PPVs: prioritize suppression of interfacial quenching and maximize photon recycling, while carefully managing transport trade-offs. With further optical optimization (e.g., reducing electrode absorption), near-unity ELQE and vanishing non-radiative voltage losses are plausible.

By engineering multiple quantum-well interfacial barriers using an L-site exchange to form thick OLA (C18) spacers, the authors realize radiative and stable PPVs that leverage strong photon recycling. This yields peak ELQE of 19.7% (17.8% at 1-sun equivalent), a champion PCE of 26.04% (certified 25.16%), and robust operational and thermal stability. The optical gains from reduced interfacial quenching and enhanced photon recycling outweigh modest electrical penalties (slightly increased series resistance and reduced FF). The devices approach 96.4% of the detailed balance Voc limit, illustrating that PPVs can surpass thin PeLED architectures in brightness at high internal radiative efficiencies. Future work should focus on optimizing optical architectures (e.g., minimizing electrode absorption, tailored cavities/scatterers) and integrating with state-of-the-art transport layers to further elevate ELQE toward unity, eliminate non-radiative voltage loss, and achieve theoretical efficiency limits.

• Electrical trade-offs: Thicker C18 barriers increase series resistance and slightly reduce FF; charge transfer to the HTL is hindered (no direct electrical advantage), requiring careful phase control to maintain conductivity (n=2) and avoid n=1 formation.
• Approach specificity: Direct OLA iodide treatments can form n=1 phases that severely impair transport; the benefits rely on the L-site exchange process which may require precise processing conditions.
• Optical/outcoupling constraints: Direct outcoupling remains low (~4%), with performance heavily dependent on photon recycling; further improvements require reduction of parasitic absorption (e.g., electrodes) and scattering optimization.
• Measurement/model assumptions: EL analyses assume unity charge balance and Lambertian emission; optical modeling treats non-radiative near-field coupling as fully recycled to avoid divergence; results may vary with different device stacks or material properties.
• Generalizability: Demonstrations are in n–i–p architecture with spiro-OMeTAD; extension to other transport layers or inverted architectures may exhibit different transport and stability behaviors.
• Passivation on neat films: L-site exchange shows no additional trap passivation benefit over C8 on neat perovskite films; gains arise primarily at interfaces with extraction layers.