Matter42 researcher guide

Matter42 is a research workspace for AI-assisted analysis of 2D materials data. It is built for scientists working with Raman maps, photoluminescence (PL) maps, microscopy images, tabular measurements, papers, and growth simulations.

The app stores your work in projects, keeps the original files next to the chat history, and lets the research agent call analysis tools that return structured results and interactive figures. Use the book icon in the app header to return to this guide.

What You Can Do

• Parse Raman or PL hyperspectral maps, then inspect spectra, spatial maps, and summary statistics.
• Cluster pixels into spectroscopically distinct defect populations using physics-informed Raman and PL features.
• Estimate Raman-derived defect density from E2g linewidth broadening against MLIP calibration curves.
• Rank likely defect families from E2g and A1g peak positions, linewidths, and asymmetry.
• Estimate PL defect activity from intensity quenching, trion enhancement, and sub-gap emission.
• Segment maps into interior, transition, and damaged regions before running quantitative analysis.
• Simulate TMD CVD growth with a kinetic Monte Carlo model to see how temperature, flux ratio, and nucleation density affect defects.

Start With Data

Most research sessions begin by uploading a file to a project, then asking the agent to parse and explore it. The main parser accepts LabSpec spectral maps, single spectra, numeric tables, documents, and images. Once parsed, a dataset ID links later analysis calls to that exact file.

Good first prompts are specific about the measurement and scientific scope:

Parse this WS2 Raman map. It is a PFIB sample, so use a 2.5 um boundary buffer, then show the exploratory maps.

I uploaded matched PL and Raman maps from the same MoS2 flake. Explore both, then cluster the defect populations using the paired datasets.

Estimate defect density only in the interior region. Use the Raman dataset and correct the E2g linewidth with instrument_fwhm=1.8 cm^-1.

Common Workflows

Raman Defect Quantification

Upload a Raman map, run explore_data, optionally preview clean regions with segment_regions, then run estimate_defect_density and classify_defect_type. This workflow is strongest for MoS2 and WS2 maps with measurable E2g and A1g modes.

Paired PL And Raman

Upload both maps from the same sample. Ask the agent to parse each file, explore each dataset, and use the PL dataset as primary with the Raman dataset as auxiliary for cluster_defects. The tools align the grids by nearest-neighbor matching; you do not need to manually reshape or merge coordinates.

PFIB Or Boundary-Damaged Samples

Set boundary_buffer_um during parsing or analysis, then run segment_regions before heavier tools. Use region_mode="interior" for intact material, region_mode="transition" for the boundary halo, and selection_policy="largest_component" when the usable film is fragmented.

Growth Hypothesis Testing

Use request_kmc_params or run_kmc when you want to explore CVD conditions rather than analyze an uploaded file. The KMC model supports MoS2, WS2, and WSe2 and returns an animated growth trajectory plus defect statistics.

Read Next

• Data guide explains supported uploads, parser hints, and what the app extracts from each file type.
• Research workflow walks through projects, files, chat, figures, and good scientific prompting patterns.
• Tool reference summarizes each tool, required inputs, outputs, and caveats.
• Methods describes the spectroscopy features, calibration assumptions, region logic, and interpretation limits.