The atmosphere trapped in the cavity of a glass-glass-sealed perovskite solar cell directly shapes its long-term operational behaviour^1-3, yet the precise degradation mechanism remains poorly understood in efficient but unstable Au-based n-i-p PSCs. Here, we systematically compare ambient air, CO₂, and N₂ encapsulation in laser-sealed PSCs based on a triple-cation Cs_0.05(MA_0.17FA_0.83)_0.95Pb(I_0.83Br_0.17)₃ absorber. Champion devices exceeded 19 % PCE prior to encapsulation (average 17.79 ± 1.03 %). The 65 °C laser-sealing step induced a moderate efficiency loss of ca. 7 % (16.59 ± 1.02 %), primarily from photocurrent losses, while simultaneously reducing anomalous hysteresis across all conditions: the hysteresis index dropping from 26.52 % to 7.86 % (CO₂) and 8.51 % (N₂), attributed to thermal passivation of grain boundary defects. Air-sealed devices showed a more modest improvement (23.64 %), hindered by competing oxygen passivation and early-stage moisture-driven degradation.

Under continuous MPP tracking at 1 sun and 45 °C over 160 hours, the encapsulation atmosphere unequivocally dictated device lifespan. Air-sealed devices failed catastrophically within 10 hours: PCE reaching near-zero, with J_SC and FF collapsing below 25 % and V_OC retaining only 25 %, driven by the synergistic photo-oxidative attack of trapped O₂ and moisture. N₂ encapsulation delivered a transformative improvement: after a ca. 40-hour burn-in, PCE stabilised at 50 % while V_OC was preserved at 93 % until reach 160 h of continuous exposure, confirming that excluding oxidative species suppresses deep-level non-radiative recombination centre formation in the bulk absorber. CO₂ encapsulation offered a mechanistically distinct response: during the first 50 hours, CO₂-sealed devices matched the N₂ baseline entirely, attributed to the in-situ formation of a carbonaceous passivation layer at the perovskite surface. V_OC remained stable at ca. 90 % across 160 hours, confirming effective bulk protection. However, after 50 hours, CO₂ devices exhibited progressive FF attenuation absent in N₂ samples, causing the final PCE to stabilize at ca. 33% after 160 h of aging. This selective FF loss indicates a specific interfacial vulnerability under prolonged CO₂ exposure.

Postmortem analysis (SEM, XPS, GDOES, PL, TRPL, and EIS) of exposed perovskite surfaces from dark-stored and light-aged devices revealed atmosphere-specific degradation pathways consistent with the MPP tracking results. SEM showed that air encapsulation shrinks the pervoskite grain size by 34 % during dark storage via moisture-driven boundary degradation, while CO₂ promotes Ostwald ripening with a 22 % grain size increase. Aging induced complete morphological collapse in air-sealed films, while N₂-sealed films retained their granular structure; CO₂-sealed films, by contrast, exhibited only minimal grain size reduction, largely preserving their initial morphology. GDOES depth profiling identified the mechanistic origin of late-stage FF loss in CO₂: the degradation of the PTAA layer compromises its barrier properties, promoting aggressive iodide accumulation at the weakened interface and deep gold migration into the perovskite stack. XPS confirmed the formation of carbonaceous species through the perovskite–CO₂ reaction¹, which passivates the perovskite surface while leaving the overlying PTAA unprotected. PL and TRPL measurements corroborated severe bulk phase degradation in air, intact bulk in N₂, and increased interfacial trap density in CO₂ without changes in the bulk phase. EIS also demonstrated that recombination resistance collapsed from the megaohm range to tens of kilohms in air and CO₂ after aging, while N₂-sealed devices maintained moderate interfacial quality.

These findings establish that complete exclusion of O₂ and moisture is the primary requirement for stable encapsulation and identify CO₂-driven PTAA degradation as a previously unreported pathway for gold migration and interfacial failure in n-i-p PSCs, with critical implications for the design of hermetic encapsulation strategies in perovskite photovoltaics.