Tell AI your goal. Watch scientific tools in action.
The complete conversation and short Agent video demonstrate task interpretation, execution and output.
Complete AI Chat analysis See how natural language organizes data analysis.
FIRST MESSAGE
Provide known, relevant context in your first message.
Give relevant background together to reduce follow-ups and repeated context. Explicitly mark unknown conditions rather than guessing to save a conversation round.
Sample background Materials, preparation, known elements or phases
Experimental data Imported data, source / wavelength and scan range
Analysis goal Phases, quantification, refinement, trends or diagnostics
Known structures & Constraints CIFs, completed steps, fixed parameters or key structural information
Requested output Tables, evidence, anomalies and recommendations
Complete, relevant context can reduce total conversation usage. Avoid unrelated detail.
PROMPT LIBRARY
Prompt templates: replace the bracketed fields.
Choose the template for your task and fill relevant information before sending. Ask AI to list missing critical information together.
Phase identification
Describe sample background and known elements, then ask AI to organize search and result checks.
Analyze the currently imported XRD data.
Sample: [materials, preparation and known elements]
Conditions: [source / wavelength and scan range; mark unknown items]
Known phases or composition: [state if none are known]
Goal: check data quality, find peaks, identify candidate phases and their evidence, and highlight unexplained peaks.
Output: summarize results, uncertainties and next steps.
If critical information is missing, list your questions together. Do not guess unknown conditions.
Rietveld refinement
Include structural models, current results and constraints.
Continue Rietveld refinement of the current data.
Sample and phases: [materials and identification results]
Models: [imported CIF / structure files; identify missing models]
Conditions: [source / wavelength and scan range]
Completed steps: [preprocessing, search or refinement status]
Constraints and goals: [fixed parameters, known composition or structural priorities]
Other characterization, if available: [XAS/EXAFS, UV–vis–NIR, TPR, SEM/TEM, ICP/EDS/XRF, XPS, Raman/FTIR, NMR, Mössbauer, magnetization, BET, neutron/PDF, etc.; provide size, morphology, stress/strain, composition, coordination, valence, oxygen vacancies, reduction, occupancy or magnetic information as constraints or testable priors]
Organize parameter release, track Rwp, Rp, χ² and the difference pattern, and assess physical plausibility.
Report results, phase fractions or cell parameters and remaining concerns. Ask for missing information together.
Abnormal peaks & Fit diagnostics
Specify the anomalous region and current model rather than asking only what is wrong.
Inspect the current fit.
Anomalies: [difference peaks, poorly fitted regions or implausible parameters]
Model: [included phases, main parameters and release status]
Other characterization, if available: [XAS/EXAFS, UV–vis–NIR, TPR, SEM/TEM, ICP/EDS/XRF, XPS, Raman/FTIR, NMR, Mössbauer, magnetization, BET, neutron/PDF, etc.; provide size, morphology, strain, composition, coordination, valence, oxygen vacancies, reduction, occupancy or magnetic information as constraints or testable priors]
Check the data, profiles, settings and possible missing phases. Distinguish evidence from hypotheses and recommend the order of further checks.
Batch sample comparison
Specify sample groups, variables and output format.
Compare the currently imported sample batch.
Groups: [controls, variables or processing conditions]
Metrics: [phases, fractions, cell parameters, peak positions or stress]
Output: [table columns, trends and anomaly labels]
Use consistent methods, summarize results, compare trends, flag anomalies and items requiring review, and provide a concise interpretation.
TOKEN & COST
Reduce follow-up rounds and control total usage.
AI APIs charge for input and output tokens. Complete, relevant context often reduces follow-ups and repeated information, but longer inputs also consume tokens. Total cost depends on task complexity, model pricing and actual usage.
¥ 0.09–3.5Illustrative AI API cost for Rietveld refinement with limited conversation
An estimate, not a fixed fee. More conversation rounds or complex tasks may cost more.
Work step by step. Inspect each scientific result directly.
Start with import, background treatment, peak finding and phase search, then continue to indexing or refinement as needed.
Import experimental data Open a data file and inspect the pattern and measurement information.
Suggested learning path
Prepare & Import
Confirm sample information and conditions, then import and inspect the data.
Preprocess & Find peaks
Choose background treatment and other preprocessing, then inspect peaks and the peak list.
Phases & Structures
Check candidates; continue with indexing and refinement according to your goal.
Review & Export
Check fits, differences and parameter plausibility, then save reviewable results.
Graphical operation preserves direct parameter and tool control. Both modes use the same scientific capabilities. Videos show operations; the learning path explains their purpose.
Tutorials and Exercises
Choose case materials. Complete an analysis yourself.
Use the bundled cases or NJUST demonstrations and data for hands-on XRD analysis.
01 / Hands-on cases
Bundled cases
Use the bundled data for hands-on import, background treatment, peak finding and batch comparisons.
Where are the files?
After installing iPowder, open the example folder in the installation directory.
02 / Hands-on cases
NJUST case resources
Explore phase, quantitative, microstructure and macroscopic stress analysis through the center’s demonstrations.
How do I use them?
Visit the case page to watch demonstrations and download practice data.
