AI-Driven Automated Column Screening and Gradient Optimization for LC Method Development

Significance of the Topic

Liquid chromatography (LC) method development is a fundamental process in analytical laboratories, requiring repeated experiment setup and deep chromatographic expertise. Automating gradient optimization can dramatically reduce development time, enhance reproducibility and lower the barrier for less‐experienced users.

Study Objectives and Overview

This article introduces LabSolutions MD software, which employs an AI‐driven algorithm to automate column screening and gradient condition optimization. The goal is to identify column–gradient combinations that meet predefined criteria for peak resolution and analysis time.

Methodology and Instrumentation

LabSolutions MD was applied on a Shimadzu Nexera X3 UHPLC system using three C18 columns: Shim-pack Velox C18, Shim-pack Scepter C18-120 and Shim-pack GIST C18-AQ (all 100 mm × 3.0 mm, 1.9 µm). A six‐compound mixture (hydrocortisone, furosemide, ketoprofen, probenecid, diclofenac, indomethacin) served as the model sample. The mobile phase consisted of 0.1% formic acid in water (pump A) and acetonitrile (pump B). Analytical parameters were set to 40 °C, 0.7 mL/min flow rate, 5 µL injection volume and 254 nm detection. Optimization criteria were a minimum resolution of 1.5 and a maximum elution time of 10 minutes for the last peak.

Main Results and Discussion

The AI algorithm iteratively improved gradient profiles by alternating condition scouting and correction analysis. Initial conditions on columns A and B yielded suboptimal resolution, which was enhanced to meet the criteria. Column C failed to fully resolve two peaks. Among the tested options, Shim-pack Velox C18 achieved the highest resolution (Rs=2.6) within the time constraint. This demonstrates the software’s capability to automate and streamline method development across multiple columns without manual fine-tuning.

Benefits and Practical Applications

• Significant reduction in labor and time for LC method development.
• Minimized reliance on operator expertise for gradient optimization.
• Applicability to both new method creation and improvement of existing methods.
• Consistent achievement of target resolution and analysis time criteria.

Future Trends and Opportunities

• Expansion of AI‐driven method optimization to additional separation modes (e.g., HILIC, GC).
• Integration with real‐time monitoring and adaptive feedback loops.
• Cloud‐based platforms for collaborative method development and data sharing.
• Advancements in machine learning models for more complex sample matrices.

Conclusion

LabSolutions MD’s AI‐powered workflow effectively automates gradient optimization for LC, achieving target separation performance across multiple columns and offering substantial efficiency gains in chromatographic method development.

References

• Shimadzu Technical Report C190-E309: Software for supporting method development with LabSolutions MD.
• Shimadzu Technical Report C190-E284: Efficient Method Development Based on Analytical Quality by Design with LabSolutions MD Software.