Disentangling the origin of degradation in perovskite solar cells via optical imaging and Bayesian inference

This study employs a novel approach combining photoluminescence imaging, drift-diffusion simulations, and Bayesian inference to map the spatially non-uniform degradation of perovskite solar cells, successfully distinguishing between bulk and interface defects and demonstrating that amino-silane passivation effectively suppresses interfacial degradation.

The Gist

Imagine a solar cell as a bustling city where sunlight is the energy supply, and electricity is the traffic flowing through the streets. For this city to work perfectly, the "roads" (the materials inside the cell) must be smooth, and the "traffic lights" (the interfaces where different layers meet) must function flawlessly.

This paper is about figuring out exactly why some of these solar cities start to crumble over time, and how to fix them. The researchers used a clever combination of high-tech cameras, computer simulations, and a statistical method called "Bayesian inference" (think of it as a super-smart detective that weighs all possible clues to find the most likely truth) to solve the mystery.

Here is the breakdown of their discovery:

1. The Problem: The City is Degrading Unevenly

When the researchers let these solar cells age under heat and light (simulating years of sun exposure in a few weeks), they didn't just see the whole city get a little worse. Instead, they saw a patchwork quilt of failure.

• The "Dark Spots": Some areas turned into "ghost towns" where electricity couldn't flow.
• The "Bright Islands": Other areas remained vibrant and efficient.
• The Mystery: Looking at the city from a distance (standard testing) couldn't tell them where the problem was. Was the road itself crumbling (the bulk material)? Or was the traffic light at the intersection broken (the interface between layers)?

2. The Solution: The "Super-Detective" Camera

To solve this, the team didn't just take a photo; they took a movie of the city glowing under different lights. They then fed this data into a computer model that simulates how electricity and ions (tiny charged particles) move inside the cell.

Using their "Bayesian detective" method, they worked backward from the glow to figure out the hidden numbers governing the city. They created a map for every single tiny pixel of the solar cell, revealing:

• How long electrons can survive before dying out (Bulk Lifetime).
• How fast electrons are lost at the top and bottom walls of the city (Surface Recombination Velocity).

3. The Findings: Two Different Ways to Fail

The detective work revealed that the solar cells fail in two very different ways, depending on the location:

• The "Rust in the Roads" (Bulk Degradation): In some areas, the problem was the road itself. The material inside the cell started to degrade unevenly, creating islands of good material surrounded by bad material. It was like the asphalt cracking randomly in some spots but not others.
• The "Broken Traffic Light" (Interface Degradation): In other, more severe areas, the road was fine, but the "traffic lights" at the bottom of the city (where the solar layer meets the electron-transport layer) were broken. This caused electrons to get stuck and lost. Crucially, these failures started as tiny, isolated dots and then spread outward like a stain, eventually swallowing up the whole area.

4. The Fix: The "Molecular Glue"

The researchers tested a special treatment using a molecule called amino-silane. Think of this molecule as a high-tech "molecular glue" or a "patch kit."

• What it did: It specifically glued itself to the "traffic lights" at the bottom of the city, sealing up the cracks and fixing the broken connections.
• The Result: The treated solar cells didn't just last longer; they stayed uniform. They didn't develop those spreading "stains" of failure. The "traffic lights" stayed green, and the roads remained smooth.
• The Proof: By comparing the treated cells to the untreated ones, they proved that the main reason the untreated cells failed was because those bottom "traffic lights" broke down. The glue treatment stopped this specific failure mode, keeping the whole city running smoothly.

The Bottom Line

This paper shows that solar cells don't just "wear out" evenly. They fail in specific, localized ways—sometimes the road crumbles, but often the connections at the edges break first and spread.

By using this new "detective" method, the researchers could pinpoint exactly which part of the solar cell was failing. They then proved that a specific molecular treatment acts like a targeted repair crew, fixing the most critical weak point (the interface) and preventing the entire device from collapsing. This gives scientists a powerful new tool to design solar cells that don't just work well today, but stay strong for years to come.

Technical Summary

Technical Summary: Disentangling the Origin of Degradation in Perovskite Solar Cells via Optical Imaging and Bayesian Inference

Problem Statement
Metal halide perovskite solar cells (PSCs) have achieved high power conversion efficiencies, yet their commercial viability is hindered by stability issues. Degradation in PSCs is complex, involving intrinsic stressors (light, heat, voltage) and extrinsic factors, and rarely proceeds uniformly. It often involves coupled bulk and interfacial processes, particularly at the perovskite/transport layer interfaces. While spatially resolved imaging techniques like photoluminescence (PL) have revealed heterogeneous degradation, previous methods lacked the ability to definitively distinguish between bulk recombination losses and interface-specific recombination (at the electron or hole transport layers) within a single, fully fabricated "mono-facial" device. Standard metrics, such as Quasi-Fermi Level Splitting (QFLS) maps, indicate where degradation occurs but cannot inherently quantify the underlying physical parameters (e.g., bulk lifetime vs. surface recombination velocity) responsible for the observed spatial variations.

Methodology
The authors introduce a novel framework that integrates intensity-dependent photoluminescence imaging with drift-diffusion simulations and Bayesian inference to extract spatially resolved physical parameters.

1. Experimental Data: The study utilizes PL imaging of fully fabricated PSCs (ITO/Me-4PACz/Perovskite/C₆₀/BCP/Cr/Au) under accelerated aging (70°C, full-spectrum sunlight, open-circuit). Both control devices and devices treated with an amino-silane (AEAPTMS) passivation layer were analyzed.
2. Simulation and Modeling: The authors employed the IonMonger drift-diffusion simulation package, which explicitly accounts for mobile ionic species. They generated a large database (look-up table) of simulated QFLS responses across varying illumination intensities. This database mapped permutations of four key physical parameters to simulated QFLS values:
   • Bulk carrier lifetime ($\tau$)
   • Surface recombination velocity at the Electron Transport Layer interface ($SRV_{ETL}$)
   • Surface recombination velocity at the Hole Transport Layer interface ($SRV_{HTL}$)
   • Ion vacancy density ($N_0$)
3. Bayesian Inference: Instead of traditional fitting, the team applied Bayesian inference on a per-pixel basis. By comparing experimental QFLS data against the simulation look-up table, they calculated posterior probability distributions for the physical parameters. This approach handles parameter degeneracy (where multiple parameter sets could produce similar outputs) by identifying the most probable values and their uncertainties.
4. Validation: The inferred parameters were cross-validated using Time-Correlated Single Photon Counting (TCSPC) measurements on partially fabricated stacks and compared against previous findings from the same group regarding AEAPTMS passivation.

Key Results
The application of this methodology yielded spatially resolved "inferred maps" of recombination parameters, revealing distinct degradation pathways:

• Heterogeneous Bulk Degradation: In "less degraded" regions of control devices, the analysis showed that spatial variations in QFLS were primarily driven by spatial variations in bulk carrier lifetime ($\tau$). Bright clusters of high QFLS corresponded to regions retaining longer lifetimes, while surrounding areas degraded into non-radiative recombination zones.
• Interface-Driven Catastrophic Failure: In "more severely degraded" regions, the degradation originated from discrete points that expanded over time. The Bayesian inference identified that these regions were characterized by a significant increase in $SRV_{ETL}$ (surface recombination velocity at the electron transport layer), rather than bulk degradation. This was corroborated by a simultaneous drop in charge collection quality ($Q_{col}$). The study notes that while $SRV_{ETL}$ initially decreased slightly in some areas (possibly due to ion redistribution), the catastrophic failure mode was defined by a sharp rise in $SRV_{ETL}$.
• Role of AEAPTMS Passivation: In devices treated with AEAPTMS, the inferred maps showed remarkable spatial uniformity. The treatment suppressed the formation of heterogeneous bulk clusters and, crucially, maintained low $SRV_{ETL}$ values throughout the aging process. The analysis confirmed that the primary mechanism for the enhanced stability of AEAPTMS-treated devices is the passivation of the perovskite/ETL interface, preventing the specific degradation pathway (rising $SRV_{ETL}$) that leads to device failure in control samples.
• Parameter Identifiability: The study highlighted the utility of analyzing full posterior distributions. While bulk lifetime and $SRV_{ETL}$ were well-resolved, the posterior distributions for $SRV_{HTL}$ were broader, and ion density ($N_0$) could not be reliably quantified from intensity-dependent QFLS alone, demonstrating the method's ability to assess parameter identifiability.

Significance and Claims
The paper claims to provide the first robust, spatially resolved, and statistically grounded methodology capable of deconvolving the contributions of bulk, ETL-side, and HTL-side recombination across a single macroscopic photovoltaic device without requiring dual-side measurements.

The authors assert that this approach:

1. Disentangles Degradation Origins: It successfully distinguishes between bulk degradation and specific interfacial degradation (ETL vs. HTL) in fully fabricated devices, a capability previously limited to specialized measurement setups.
2. Quantifies Localized Failure: It moves beyond spatially averaged metrics to identify specific failure modes, such as the expansion of high-$SRV_{ETL}$ regions, which are the primary drivers of catastrophic device failure in control samples.
3. Validates Passivation Mechanisms: It offers a quantitative, a priori confirmation that amino-silane passivation stabilizes devices primarily by suppressing interfacial recombination at the ETL, while also providing a secondary bulk passivation effect.
4. Enhances Data Value: The work serves as an exemplar of how advanced Bayesian inference can significantly increase the value of experimental imaging data, transforming qualitative maps into quantitative, physics-based parameter distributions to accelerate the design of durable energy materials.