Measuring solubility automatically with vision

Solubility measurements are critical to the development of any chemical process. Recently in iScience, Shiri et al. describe a modular, “human-out-of-the-loop” platform that uses visual data rather than typical analytical techniques such as high-performance liquid chromatography determine when compound has dissolved. “If only you could see what I’ve seen with your eyes.”—Roy Batty Blade Runner Measuring solubility various solvents is fundamental first step process across multiple industries, whether it be commodities, fine chemicals, pharmaceuticals, or materials. Both determining how dissolve order perform some reaction purification and understanding solvent conditions allow for downstream processing isolation start this measurement.1Chen C.-K. Singh A.K. A “bottom-up” approach development: application physiochemical properties products toward direct-drop processes.Org. Process Res. Dev. 2001; 5: 508-513Crossref Scopus (34) Google Scholar In addition, continuous processes increase importance pharmaceutical manufacturing, finding common system suitable range steps operations great importance.2Guidi M. Seeberger P.H. Gilmore K. How flow chemistry.Chem. Soc. Rev. 2020; 49: 8910-8932Crossref PubMed Various methods have been developed measuring solubility; them can conducted semi-autonomous, high-throughput manner, but all rely upon fairly similar steps.3Black S. Dang L. Liu C. Wei H. On measurement solubility.Org. 2013; 17: 486-492Crossref (26) Although variety standard then deployed make actual (the amount solute given solution), most (HPLC). predictions made progress provide trends rough estimates,4Qiu J. Albrecht Janey J.M. Simple method (CHEM-SP) predict from 2-D structures.Org. 2021; 25: 75-81Crossref (2) experimentally determined value still desired. Altogether, complex workflow entails high cost terms either labor (if manual) equipment on an automated HPLC). fact, industrial consortium even issued challenge build take these issues.5Enabling Technologies ConsortiumNext Generation Automated Platform.https://www.etconsortium.org/automatedd-solubility-platformDate: 2018Google al.6Shiri P. Lai V. Zepel T. Griffin D. Reifman Clark Grunert Yunker L.P.E. Steiner Situ al.Automated screening using computer vision.iScience. 24: 102176Abstract Full Text PDF (4) pragmatic, modular solution making by “off-the-shelf” components open-source software enable fully workflow. number innovative features described platform, particular note clever use gauge turbidity thereby at which point enough added solute. simple webcam custom algorithm advance possible. The entire itself brings together common, unit operations, namely (1) robotic arm vial manipulations, handler, (3) solid dosing balance, stir plate, (5) webcam. To bring subunits together, mounted deck orchestrated their Python script. Importantly, they showed individual components, case arm, easily exchangeable agnostic manufacturer. Moreover, able measure absence human interventions. net effect truly open-sourced alternative more sophisticated commercial systems. practice, mimics much would “excess solvent” type relies end addition through observation complete dissolution weighed Figure 1 provides step-by-step summary operations: out Mettler-Toledo Quantos weigh station, handler (in Tecan Cavro syringe pump N9 controller) delivers initial charge while sample sits (3–6) webcam, associated algorithm, captures images small region interest continues add stepwise manner until appears solids dissolved, (7) gets transferred back balance final weight (with both mass volume added), report measured values calculated generated, finally (8) “next experiment” determined. transport between stations, authors (SCARA) 6-axis robot Kinova Gen3 7-axis used interchangeably script controllers. reporting was accomplished real-time experimental updates Slack. Also, depicted 1, there logic gate 6 dissolved needs return 2 solvent. benchmarked performance caffeine’s five different versus established solid” manual slurry filtration method. observed variations accuracy precision, obtained were certainly sufficient guide development. They went show equivalent caffeine tetrabenazine (TBZ) solvents. Certainly, additional refinement its environment improve results. do current limitations system: color contrast relative background, lack controlled temperature lighting, ability run experiments parallel. Improvements recognition along lighting background stage address aforementioned limitations. Environmental controls, automated, interchangeable dispenser heads, multiplexing vials later other challenges throughput. continued advances semi-empirical extrapolated predictions, solvents, mixtures, temperatures will become less pressing thus few, select valuable. This impressive built simple, readily available code. confluence ever-more machine-learning techniques, automation handling, stitch pieces replicate laboratory procedures humans harbinger future where design, execution, analysis enabled intelligent, laboratories.7Mullin R. lab now. Chemical & Engineering News, March 28, 2021.https://cen.acs.org/business/informatics/lab-future-ai-automated-synthesis/99/i11Date: 2021Google Further, barrier building operating systems clearly limited creativity researcher not complexity visionShiri al.iScienceFebruary 11, 2021In BriefOrganic chemistry; organic synthesis; green engineering; Full-Text Open Access