As automated synthesis reactors are adopted in research labs, chemists are enabled to be more productive and efficient in their daily work. Modern automated lab reactors and software create platforms for safe, unattended control of reaction parameters and collection of reaction data. This opens the possibility for long experiments to begin at any time of day, and to run continuously through the day and overnight. Yet, until now, it has not been possible for chemists to reliably sample their reactions throughout the duration of the experiment.
The integrated use of in situ PAT tools such as ReactIR™ (in situ FTIR) and Raman provide comprehensive analysis of the reactions in real time. However, critical information concerning low-level impurities is limited with these techniques. To study and monitor low-level impurities in chemical reactions it is standard practice to take samples of reactions for offline analysis by HPLC, UPLC, or GC. Incorporating traditional manual reaction sampling into the daily lab routine is cumbersome, and long reactions cannot be sampled throughout the duration of the experiment. Moreover, manual sampling of chemical reactions can be difficult, and is not always a precise or reproducible process. Researchers may be confronted with sampling challenges with reactions at elevated or sub-ambient temperatures, elevated pressures, heterogeneous reactions, and air- or water-sensitive reaction mixtures. Due to these challenges, researchers may take fewer samples than desired, gain unrepresentative samples of the reaction, or affect the reaction progression while sampling, leading to the need for additional repeat reactions to gain a complete data set. Lack of quality data leads to poor process understanding, poor decisions, and longer development times.
To overcome these sampling challenges, a group of pharmaceutical scientists partnered with METTLER TOLEDO to develop an advanced automated sampling technology called EasySampler™. EasySampler provides continued reaction sampling 24 hours a day, without affecting reaction progression. The patented capture and quench technique provides representative and reproducible samples, even from thick reaction slurries. The inline probe samples in an unattended manner from reactions at elevated pressures, air-sensitive, and toxic reactions and quenches the sample at reaction conditions, without the need to open the reactor, eliminating the risk of altering the reaction or sample during the sampling process. This ensures representative samples for accurate analysis, and reactions that remain unaffected by the sampling process for accurate monitoring of reaction progression.
This article presents three case studies where Pfizer and Servier investigators demonstrated the reliability, robustness, and precision of EasySampler for reactions posing specific sampling challenges, making them difficult or impossible to sample using traditional methods. The case studies provide examples of an oxygen-sensitive C-H activation reaction, an Ullman reaction, and an ester reduction.
Case Study 1. Impurity Profiling for Accurate Reaction End-point
Brian Vanderplas, Senior Scientist, Pfizer
Sampling this C-H Activation reaction is particularly difficult as it is air sensitive. Introduction of 5000 ppm of oxygen into the reactor headspace stalls the reaction, and results in a 50% increase in reaction time. Samples introduced to air during the sampling process before reactive species are quenched are altered, and their associated data reflects this change.
The aim of this experiment was to study the kinetics profile, mechanism of impurity formation, and assign a reasonable endpoint for the C-H activation reaction. Automated in situ sampling with EasySampler was applied to gain representative samples from the reaction, without introducing air into the reactor headspace or the sample before quenching reactive species. In addition, the unattended sampling by EasySampler offered the advantage of sampling the long reaction over a 24 hour period.
Scheme 1 shows the C-H activation reaction with a palladium catalyst, run in the presence of potassium acetate, in an RC1 reaction vessel (Figure 1). The reaction was run at a concentration of 33 mg/mL in t-amyl alcohol at 102 °C. Twelve samples were acquired over a 24 hour period, and analyzed by UPLC.
From a single experiment, the complete reaction profile was gained by sampling over 24 hours (Figure 2). The primary conversion and low-level impurity profiles give insight into the interconnectivities of the impurities observed in the mixture, and especially highlight the need for a tight reaction cycle time. The reaction profile shows that when the product reaches a maximum conversion at approximately 7 to 8 hours, the undesired Des-CN impurity suddenly forms. This impurity cannot be removed easily in subsequent work-up and reaction steps and should be avoided.
Based on the information in the reaction profiles achieved using the automated representative sampling capabilities of EasySampler, the researchers were able to quickly implement in-process controls to relate reaction time with primary conversion and impurity levels to quickly cool the reaction to 20 °C to avoid the Des-CN impurity formation.
Case Study 2. Removing Barriers for Impurity and Kinetics Profiling in an Ullman Reaction
Kristin Wiglesworth, Pfizer
To test the sampling capabilities of EasySampler, scientists at Pfizer selected the Ullman Reaction since it is typically a difficult reaction to sample manually. The reaction solution changes consistency and thickens over the course of the experiment, it is a dark mixture with known insoluble reagents, making it difficult to see what was being sampled, and manual sampling tools are easily clogged with the insoluble solids in toluene. The reaction continues to thicken as it progresses, making it difficult to sample reliably using manual methods. Sampling needed to occur over a 30 hour period, so the unattended capabilities of EasySampler were required.
The reaction is shown in Scheme 2; an aryl bromide reacts with an aryl phenol, with copper catalyst and cesium carbonate present, at 85 °C. The dark colored reaction (Figure 3) was at a high concentration (125 mg/ mL), and difficult to sample due to insoluble reagents in toluene.
As the cesium carbonate dispersed over the course of the experiment, the consistency of the reaction changed. EasySampler was inserted into a 10 mL glass tube reactor in an EasyMax synthesis workstation, and captured samples overnight at pre-programed time intervals. Scientists optimized the quench solvent to 1 v/v % water in DMSO to ensure complete dissolution of all reaction components, and to be ready for accurate offline analysis.
All samples were analyzed by UPLC. EasySampler successfully sampled this Ullman reaction, and provided representative samples throughout the 30 hour experiment. The conversion data shows that product formation stalls after 18 hours (Figure 4a), yet impurity formation continues through the course of the reaction (Figure 4b).
This study clearly shows that by using EasySampler, accurate reaction kinetics and impurity profiles, and end-point information can be collected for reactions that are typically difficult to sample using traditional sampling methods.
Case Study 3. End-point detection of an Ester Reduction
Mathieu Grandjean, Servier
At ORIL Industrie (Servier), this specific ester reduction is an important reaction, which needs accurate monitoring in order to determine key information on reaction kinetics that influence product yield, including product degradation. However, this reaction is viscous and air-sensitive, therefore particularly difficult to sample reliably by traditional manual methods. Poor quality samples lead to inaccurate and variable data when analyzed by UPLC. To eliminate these challenges, scientists at Servier applied EasySampler to automatically sample this challenging reaction, and provide high quality representative samples.
The reaction is shown in Scheme 3. Using EasySampler, eleven samples were collected from the viscous reaction (Figure 5) over a 12 hour time period. Sample collection was scheduled for every 80 minutes after the end of hydride addition; this ensured that the maximum conversion point was not missed. Samples were analyzed by UPLC.
The data presented in Figure 6 illustrates that product formation is maximum at 8.5 hours. In addition, the data shows product degradation with increasing reaction time. As a result, Servier were able to reduce the reaction time significantly, thus increasing productivity by 40%, as well as enjoying an increase in product yield of 20%. This directly translates to savings in both time and costs.
The manual sampling method used at Servier was inaccurate, and resulted in imprecise analytical monitoring via UPLC. Manual sampling was difficult due to the viscous heterogeneous and air-sensitive reaction, making a representative sample difficult or impossible to capture. With the application of EasySampler, Servier now has a robust method to reliably sample and accurately monitor these types of ester reductions. Improved reaction understanding directed further process optimization studies, resulting in increased product yield and purity, shorter reaction time, and cost savings.
Semi-automated lab reactors, coupled with unattended sampling, present an opportunity to control and monitor reaction parameters, and sample these reactions without affecting reaction progression. The ability to take representative samples of reactions throughout the reaction, and from reactions that are typically difficult or impossible to sample, enabled the researchers to gain complete and accurate reaction information from a single experiment – they had never been able to do this before. Ultimately, this saves time, and increases productivity and cost savings.
The case studies presented here clearly show that accurate kinetics and impurity profiles, and end-point information can be collected using EasySampler, even when the investigators are away from the lab. The representative samples gained with EasySampler provides high quality samples for accurate reaction information, leading to good process decisions.
Like this article? Click here to subscribe to free newsletters from Lab Manager