A fundamental gap exists in our understanding of Earth’s climate history: the Mid-Pleistocene Transition (MPT), which occurred approximately 1.2 to 0.8 million years ago. During this period, the periodicity of ice age cycles shifted from 41,000 to 100,000 years despite no changes in astronomical forcing, suggesting a major shift in the climate system’s internal response. While this transition is recorded in some archives, the longest continuous ice core record currently available (EPICA Dome C) only spans the last 800,000 years, missing the MPT. To address this, the European flagship project "Beyond EPICA: Oldest Ice Core" (BE-OI) has successfully drilled an ice core in Antarctica to retrieve a unique record extending back at least 1.2 million years. However, a significant problem arises in the deepest layers: the ice is extremely thinned, where a single meter can represent over 20,000 years of climate history. In such thinned ice, chemical impurity signals, which archive information on past dust, sea ice, and atmospheric patterns, are highly vulnerable to post-depositional alterations. Processes such as grain growth, impurity diffusion, and chemical reactions can disturb or even destroy the original stratigraphic layering, leading to potential misinterpretation of the climate record.
The AiCE project aims to overcome these challenges by establishing a "novel 2D vision" for high resolution ice core impurity analysis. The core objective is to determine whether the oldest ice still preserves its original paleoclimatic message and to retrieve that impurity signal with unprecedented confidence. To achieve this, the project is built on three innovative pillars:
Next-Generation 2D Imaging: Moving beyond conventional 1D meltwater analysis, AiCE utilizes Laser-Ablation Inductively-Coupled Plasma Mass Spectrometry (LA-ICP-MS) to image chemical distributions in solid ice at a micro-metric scale. By developing a new large cryocell and integrating Time-Of-Flight (TOF) technology, the project can simultaneously map a wide range of marine and terrestrial proxies (e.g. Na, Mg, Ca, Fe, S, Cl) across large ice sections.
Artificial Intelligence for Automated Analysis: AiCE introduces deep learning and machine learning algorithms for LA-ICP-MS 2D mapping. These tools are designed to help us generate the data we need and to recognize patterns indicative of stratigraphic integrity or disturbance.
Signal Retrieval and Validation: By identifying the minimum scale at which signals are preserved, the project will acquire high-resolution climate records while ruling out corrupted sections, thereby avoiding scientific misinterpretation.
The results of AiCE are expected to have a profound impact on both ice core science and analytical technology. By providing reliable, high-resolution impurity records from the BE-OI core, AiCE will contribute insights into the causes of the MPT. The development of the large-volume cryocell and AI-driven high-throughput establishes a new framework for chemical imaging on ice cores. The project aims to revolutionize the interpretation of the oldest ice paleoclimate signals with potential transferability of the technology also to the analysis of other natural archives like stalagmites or corals.
