Next Generation NMR Flow Tube Optimized for Small Volume NMR Reaction Monitoring in Preclinical Drug Discovery

Abstract

We developed a 3 mm NMR flow tube that can be inserted into a 5 mm NMR probe converting a high field NMR spectrometer into a reaction monitoring system with a minimum reaction volume of 2 mL. The system uses a standard LC pump to flow the reaction mixture from a reaction vial in an integrated heating block to the center of the magnet for detection before returning the mixture back to the vial. The NMR flow tube has a temperature range of −20 to 120 °C and the NMR instrument can be switched from NMR tubes to the flow tube mode in a few minutes. We report the use of this flow NMR small volume system to monitor the progress of two chemical reactions that resisted study by more common analytical techniques.

Of the common approaches to conducting reaction monitoring by NMR, online NMR monitoring is the preferred method over static NMR tubes, stopped flow or rapid injection NMR. − Online NMR reaction monitoring closely replicates the regular non-monitored experimental setup, since it achieves control over temperature, mixing rate, and reagent addition(s) which are the parameters to control when running a chemical reaction. While NMR reaction monitoring can be done with a benchtop system, the reduced resolution and limited temperature capability has limited the widespread use of this system for many reactions. The two other NMR reaction monitoring techniques: rapid-injection and stopped flow, focus on identifying reaction intermediates. Since our focus was monitoring the formation of target compounds, we chose online reaction monitoring for this study, focusing on using a high field NMR instrument to achieve the desired resolution and expanded temperature capability for pharmaceutical NMR reaction monitoring.

Online NMR reaction monitoring technology has been an extremely successful addition to the Process Analytical Technology (PAT) tools available to synthetic organic chemistry groups, particularly within pharmaceutical development groups. , The development of a 5 mm NMR flow tube for a high field system to enable this technology was documented previously. Our goal was to modify the 5 mm NMR flow tube used previously to identify a flow setup that is better suited for early pharmaceutical research. At this stage of pharmaceutical research, a lead compound has not yet been selected, so conventional PAT does not apply due to the rapidly changing direction of the program to identify a lead compound. PAT processes are introduced at a later stage, when lead compounds have been identified and scale-up begins. In preclinical research, samples quantities are often limited and speed is often prioritized over more detailed studies. Reaction monitoring in early research tends to focus on three questions: can the target compound be synthesized, was the correct compound synthesized and what was the yield? This targeted focus often limits the collection of additional data highlighted in many previous PAT references and may contribute to the limited literature available documenting NMR reaction monitoring in an early research environment.

The engineered flow tube enables reaction monitoring by transporting reaction mixtures from a reaction vessel to detection in the coil area of the NMR probe, without the need to use a specialized NMR flow probe. The key benefit of a flow tube is that it can be used across the range of available spectrometers at varying magnetic field strengths. The flow tube is prepared for use by inserting the flow tubing into a standard 5 mm probe. This reaction monitoring setup is less expensive than systems which require the purchase of a separate NMR flow probe. The flow tube can also be removed from the 5 mm conventional NMR quickly (∼ 5 min), allowing the instrument to be converted back to the analysis of 5 mm NMR tubes as needed. Unfortunately, typical reactor volumes in the 5 mm flow tube setup range from 30 mL to 100 mL, with a minimum reactor volume of 15 mL required. While this may not be an issue during later development, sample quantities are often quite limited in early preclinical pharmaceutical discovery, effectively limiting the use of the 5 mm flow tube in the discovery setting.

Another type of flow system that should work with the limited sample quantities we describe is the microfluidic system, but this system has not been described as having the quick conversion from tubes to flow and back that the flow tube enables. We ultimately wanted to use a system that would eventually be commercially available to minimize downtime and optimize the availability of replacement parts. For these reasons, we collaborated with Bruker to reduce the required volume in this next generation NMR flow tube, making it better suited for preclinical discovery research.

A next generation flow tube has been developed through a collaboration with Bruker which reduces the volume of solvent needed to fill the NMR flow tube (Figure ). A more than 60% reduction in the total fill volume required was achieved by reducing the round-trip tube length from 8 meters to 3 meters. This required us to move the reaction assembly from a conventional fume hood located away from the NMR instrument to an elevated platform close to the top of the magnet (Figure , Supporting Information). To achieve this, we worked with PlasLabs to construct a novel fume hood cabinet that could be placed at the top of the elevated non-magnetic platform. This allowed for the reaction to take place closer to, and at the same height, as the top of the magnet. This fume hood and platform was also portable, facilitating moving the reaction monitoring setup to a different instrument as needed.

1.

A schematic of the 3 mm flow NMR reaction monitoring setup.

The new flow insert uses a 3 mm flow cell instead of the 5 mm flow cell used previously. The new setup uses 0.5 mm ID × 1/32” OD for the inlet tubing between the pump and the flow cell. This tubing is readily available in the U.S. which is an advantage when replacement parts are needed. In addition, this size tubing displaces only ∼12 μL of solvent in the flow cell. This diameter of tubing is very small and any reactions monitored with this setup need to be monitored for formation of precipitates to minimize clogging of the lines. We routinely use 10 micron porous filters to avoid solids entering the flow line. Alternatively, the system can be set up using 1/16” OD inlet tubing, but this size tubing displaces ∼70 μL in the flow cell, significantly decreasing the signal to noise of the flow cell. The NMR flow tube consists of two layers: an inner transfer line that moves the sample in/out of the magnet, and an outer layer that circulates a temperature regulated fluid to maintain the reaction mixture at the desired temperature. The flow transfer lines are rated for use from −20 to 120 °C depending on the temperature rating of both the NMR probe and the regulated fluid in the chiller. Lastly, the use of small volume reaction vials resulted in a greatly reduced reaction volume (from 15 mL to 2 mL) which better matched the smaller sample sizes available in an early discovery pharmaceutical setting.

To show that the new flow tube yielded acceptable and reproducible spectra, several standard sample concentrations were evaluated, and the line widths were measured at various peak heights (Table ). In addition, the reproducibility of a spectrum was examined by collecting repeated spectra on the same sample as it flowed through the flow tube (Supporting Information). Review of this data showed that the 3 mm flow tube yielded well resolved spectra over a wide range of concentrations.

1. NMR Line Shape Results Using the New 3 mm Flow Tube at 500 MHz .

Sample	Method	50%	0.55%	0.11%
1% CHCl3 in acetone-d 6	5 mm NMR tube	0.45	7.7	12.8
3% CHCl3 in acetone-d 6	3 mm flow tube static	1.92	10.4	20.3
3% CHCl3 in acetone-d 6	3 mm flow tube at 1 mL per minute flow	1.73	10.2	20.9

^aFor comparison, the width of the sample peak is shown at three different peak heights. The values are given in Hertz. These results can be compared to data obtained previously using a 5 mm flow tube on a 400 MHz system.

We selected the base catalyzed reaction of sulfolene with ammonia for our first study (Scheme ). The transformation was well known in the literature but had not been subjected to systematic study. The reaction lent itself well to study by our NMR reaction monitoring system since it was carried out in aqueous solution and the ammonia ¹H resonances were equilibrated and minimized through exchange with the solvent. Hence, there was a single solvent signal to suppress and the only ¹H resonances to be observed would be derived from sulfolene (1), the isomerized intermediate (2), the product (3) of ammonia addition to 2, and the by-product (4) resulting from solvent addition to 2. We had authentic samples of 3 and 4 from previous work, and knew that the ¹H resonances that we expected to monitor were well separated from the solvent signal that we planned to suppress. Further, we anticipated that subsequent studies could be carried out with aqueous solutions of methylamine or dimethylamine in place of ammonia to inform subsequent, more general synthetic work with primary and secondary amines. These experiments would result in two signals to suppress, the solvent and the amine CH₃ groups.

1. Reaction of Amines with Sulfolene 1 .

^a With R₁, R₂ = H, H; H, CH₃; CH₃, CH₃; H, C₂H₅; H, n-butyl; H, CH₂C₆H₅; -(CH₂)₅-.

The base catalyzed reaction of sulfolene with amines was first described in 1938, with subsequent work being focused on preparative synthesis. − Reactions were typically conducted under heat and pressure in autoclave equipment with large excesses of the amine. This makes the execution of the reaction inconvenient and inaccessible to those without pressure equipment. There was no evidence that these conditions were necessary, and the literature was at times self-contradictory. For example, one document stated that the reaction occurs at atmospheric pressure but asserted that it is advantageous to conduct the reaction at 50 to 100 psi. It is not surprising that the reaction has not been subjected to detailed study since there is no chromophore for UV detection, limiting compatibility to observe by LC. In addition, aqueous reactions are frequently poorly suited to direct observation by gas chromatography (GC), and we found that the product 3 transited our GC with difficulty. Previously, we found that we could approximately monitor reaction progress by removing aliquots from the reaction and monitoring conversion by ¹³C NMR by observing the loss of the signals due to the olefin carbon atoms in 1.

First, we studied the reaction of 1 with ammonia in aqueous solution. We used a concentration of 1 at 0.67 M in water, containing varying equivalents of ammonia. Experiments confirmed that we could run the reaction in water using WET (Water Elimination by Tailoring) solvent suppression (Supporting Information). The ammonia was provided to the reaction mixtures as concentrated ammonium hydroxide, with 0.14 equiv of sodium hydroxide added last as catalyst to initiate the reaction. We examined the use of 18, 10, and 3 equiv of ammonia and monitored the reactions for 5 days at 20 °C (Table ). As expected, we could observe and distinguish all four components of the reaction: 1, 2 (at very low levels), 3, and 4. A plot of the relative integrals for 1, 3, and 4 versus time was generated for each experiment (Supporting Information).

2. Results of Continuous NMR Monitoring of the Reaction of 1 with Ammonia.

Equivalents NH3	Time to apparent completion	Mole fraction 3	Mole fraction 4
18	70 h	69	31
10	110 h	30	12
3	> 140 h	8	3

^aThe remaining fraction was 1.

The reaction proceeded to completion at 20 °C and atmospheric pressure in about 3 days with 18 equiv of ammonia, and therefore the application of heat and pressure as documented previously only served to shorten the reaction time and were not necessary to obtain 3 (Figure ). Addition of water to 1 was a competing side reaction, but the use of aqueous ammonia is so much more convenient than the use of anhydrous ammonia and pressure equipment that the production of 4 as a side product was acceptable in our view.

2.

¹H reaction monitoring of 1 in water at 20°C flowing at 1 mL/min. A reaction sequence of 1 is shown over time in the bottom figure and was monitored using MestreNova software with the reaction monitoring plugin.

Extension of this study to methylamine and dimethylamine (18 equiv of aqueous solution at 0.67 M of 1), with suppression of the water and amine CH₃ resonances, showed that these amines underwent reaction at significantly faster initial rates and with less formation of 4 (Supporting Information). This is consistent with these amines being more basic (pK _a 10.6) than ammonia (pK _a 9.21) in aqueous solution. Further work showed that these reactions went to completion in 48 h at 20 °C with as little as 3 equiv of the amine, with minimal formation of byproduct 4.

Based on these results, we conducted preparative experiments on a 50 mmol scale (Supporting Information), using 3 equiv of amine and 0.14 equiv of NaOH in aqueous solution at a final reaction concentration of 0.8 M 1. These reactions were allowed to proceed for 48 h at 20 °C and atmospheric pressure without monitoring, based on the results obtained with methylamine and dimethylamine. In all cases except the reaction with ammonia, the reactions proceeded to completion, the isolated yields of analytically pure products exceeded 80%, and the results were consistent with the percentage conversions observed in the NMR monitoring experiments.

We next chose a more ambitious reaction that was applicable to an ongoing medicinal chemistry project that had significant need for in situ monitoring. The reaction encompassed three steps to be carried out without isolation of intermediates, as shown in Scheme . Each step had to be completed before the start of the next step. This reaction sequence was selected for NMR reaction monitoring from among several possibilities that were considered since the full sequence could not be monitored satisfactorily by any other techniques, including in situ IR monitoring. This was due to the combination of several factors: a lack of chromophores, low molecular weights of the reactants and products yielding poor ionization in the mass spectrometer, the use of reagents in excess, weak absorbance of the alkyne group by IR, and the inability to distinguish between azide groups by IR.

2. Three Step Azide–Alkyne Cycloaddition Sequence .

^a Substances in brackets are not isolated.

The first step encompassed the preparation of the sulfonyl azide diazo transfer reagent 7 from trifluoromethanesulfonyl chloride 6 and sodium azide. Upon completion, cyclopropylamine 8 was added to react with the sulfonyl azide to afford cyclopropyl azide 9. In the last step, benzylacetylene 10 was added along with cuprous iodide as catalyst to undergo a [3 + 2] cycloaddition to afford the 1,2,3-triazole product 11. The alkyne 10 was selected as a model for a proprietary alkyne which was the product of a prior multistep synthesis and was therefore the most valuable component, hence it was used as the limiting reagent in the sequence.

First we sought to prepare the diazo transfer reagent trifluoromethanesulfonyl azide 7. Sulfonyl azides are well known as diazo transfer reagents for the synthesis of azides from primary amines, but safety testing has shown that as a class they possess the hallmarks of explosive compounds, including high energy release upon decomposition and shock sensitivity. Even sulfonyl azides that were initially believed to be non-explosive have subsequently been shown to be energetic. Trifluoromethanesulfonyl azide 7 is of particular value, and also concern, due to its reactivity, and revealed its explosive nature as early as 1972. Therefore, it is customary to prepare and consume sulfonyl azides in solution without isolation, and this safety measure has long been applied to 7. The use of flow chemistry provides a further safety benefit, because the hazardous component is generated in solution and is present in very small amounts at any given moment in time, is physically contained within the flow apparatus, and is subsequently consumed in situ as it is formed. Accordingly, the use of flow chemistry has been applied to the in situ generation and subsequent reaction of sulfonyl azides.

Sulfonyl azides are generally prepared from the corresponding sulfonyl chloride and sodium azide During this reaction, the generation of hydrazoic acid, a highly toxic, volatile and explosive substance, cannot be precluded. , Therefore, the use of flow chemistry becomes even more appealing to control this additional hazard.

We performed an analysis of the reaction mixture using a thermal screening unit (TSU) to characterize the thermal and pressure hazards involved in the synthesis and use of 7 to inform our use of it in flow. Sulfonyl azide 7 was prepared as a 0.125 M solution in aqueous acetonitrile, and a sample of the solution was then heated at 2 °C per minute while measuring the changes in internal temperature and pressure (Supporting Information). This showed that evolution of gas began at about 120 °C, while a temperature excursion began at about 135 °C. Upon returning to 25 °C, the internal pressure had increased by 3.1 bar. Based on these results, we concluded that we could perform the reaction safely in flow with a 100 °C safety margin between the operating temperature and the thermal onset temperature.

While 7 is most frequently prepared from trifluoromethanesulfonic anhydride, we planned to use ¹⁹F NMR to monitor the reaction and therefore elected to use trifluoromethanesulfonyl chloride 6 (Scheme ), since the reaction using trifluoromethanesulfonic anhydride generates an equivalent of sodium trifluoromethanesulfonate which would complicate the ¹⁹F NMR work unnecessarily.

3. Reaction of Sulfonyl Chloride 6 with Sodium Azide .

^a Substances in brackets are not isolated.

Accordingly, a mixture of sodium azide (1 equiv) and trifluoromethanesulfonyl chloride 6 (1.5 equiv) was stirred in a mixture of 50% acetonitrile and 50% water (v/v) and monitored by ¹⁹F NMR (Figure ). Control spectra were recorded to establish the resonances of the trifluoromethanesulfonyl chloride 6 (−77.3 ppm) and the potential by-product trifluoromethanesulfonic acid (12, −79.5 ppm) resulting from hydrolysis of either 6 or 7. An excess of 6 was taken to ensure consumption of sodium azide and compensate for hydrolysis to 12. The formation of 7 was monitored by its resonance at −76.5 ppm and judged to be complete in 20 min at 20 °C.

3.

Monitoring the reactions of 6 and 7 using ¹⁹F NMR. at 20 °C.

The second step was the preparation of cyclopropyl azide 9 from cyclopropylamine 8 and 7. Again, it was mandatory that 9 be prepared and used in situ, due to the high vapor pressure, low molecular weight, and explosivity hazard posed by 9. Thus, when the formation of the sulfonyl azide 7 had gone to completion, the second step was initiated by the addition of 4 equiv of 8 (Scheme ).

4. Reaction Products Resulting from Reaction of Cyclopropylamine (8) with 7 .

^a Substances in brackets are not isolated.

The excess of cyclopropylamine was needed to compensate for unavoidable side reactions, including the neutralization of any by-product trifluoromethanesulfonic acid 12 from the first step and the potential formation of N-cyclopropyl trifluoromethanesulfonamide (15) from cyclopropylamine 8 and remaining 6. We observed the rapid formation of trifluoromethanesulfonamide (13), the expected by product of the desired diazo transfer reaction, by ¹⁹F NMR at −80.5 ppm. A by-product 14 was observed at -88.5 ppm that did not increase or decrease in intensity during the subsequent course of the experiment, indicating that it was unreactive to the conditions of the experiment. Compound 14 was also observed in control experiments in the absence of azide (Supporting Information). Not surprisingly, we also observed the formation of 15 at −78.3 ppm, based on control experiments (Supporting Information) and the literature. Compound 15 is not stable to the reaction conditions and decomposes to form additional 13, thus the quantity of 13 produced is proportional to the entire quantity of 6 used in the reaction. The conversion of 8 to 9 could not be monitored by ¹H NMR directly. A large excess of 8 was used which would make any monitoring of 9 by ¹H difficult since only a fraction of 8 would go to 9. The formation of 9 was inferred from the ¹⁹F NMR: when 7 was consumed and 13 stopped increasing, the formation of 9 was therefore complete since only 7 can provide 9. This required about 15 min at 20 °C.

The final step of the sequence was the [3 + 2] cycloaddition of cyclopropyl azide 9 with the alkyne 10, catalyzed by copper­(I) iodide (Scheme ). The Huisgen dipolar cycloaddition of azides and alkynes is well known. It has become exceptionally popular since the discovery that catalysis by Cu­(I) ion allows the reaction to be carried out at ambient temperatures in virtually any solvent.

5. Cycloaddition of Azide 9 with Alkyne 10 to Afford 1,2,3-Triazole 11 .

^a Substances in brackets are not isolated.

Addition of a mixture of alkyne 10 (1 equiv) and copper­(I) iodide (0.1 equiv) in acetonitrile triggered the [3 + 2] cycloaddition reaction. This was monitored by ¹H NMR, observing the consumption of 10 as well as the development of the corresponding resonances known to be associated with the 1,2,3-triazole product 11 (Figure ). The reaction attained a 50% conversion of 10 to 11 after about 5.5 h at 20 °C, and a maximum conversion of 84% after about 96 h. Extension of these learnings to a flow chemistry apparatus provided triazole 11 in 91% purified yield (Supporting Information).

4.

Monitoring the reaction of azide 9 with alkyne 10 using ¹H NMR at 20°C. Data points in green are the cyclopropyl CH₂ protons in 11; data points in tan are the triazole ring CH proton in 11, data points in blue are the CH₂–phenyl protons in 11.

A modified, small volume reaction monitoring assembly was constructed to permit in situ, continuous real-time monitoring of reactions on smaller scales in the preclinical pharmaceutical environment than was previously possible. − This setup permits the safer study of reactions that are potentially hazardous or that involve valuable reagents or advanced intermediates that cannot be obtained in sufficient amounts for conventional work. The use of WET solvent suppression eliminated the need for deuterated solvents in these experiments.

We attained proof of concept of the system in the exploration of the base catalyzed reaction of sulfolene with amines under atmospheric pressure. This reaction was successfully monitored by flow-NMR using commercial aqueous solutions of the amines in this study. The data showed that the customary use of elevated temperatures and pressures were unnecessary and established boundary conditions for good yields of the intended products using ammonia, primary amines, and secondary amines.

In addition, a complex transformation encompassing three synthetic steps in one vessel was successfully monitored by flow NMR using both ¹⁹F and ¹H resonances. These studies established the boundary conditions for success and the relative rates of these steps which informed the successful development of a flow chemistry procedure to perform these steps and subsequent inline monitoring of the progress of the flow chemistry reactor by ReactIR. The understanding gained by the flow NMR work allowed the safe synthesis and use of volatile and potentially explosive intermediates by formation and subsequent immediate consumption under well understood conditions.

The 3 mm flow tube and associated tubing described here can be quickly removed from an NMR probe enabling the rapid conversion of the instrument back to conventional NMR tubes. The flow tube can be inserted into any high field magnet, and we have successfully moved our setup to different instruments. We have found this setup to be particularly useful when planning a flow chemistry synthesis or when trying to monitor reactions that did not work well with more traditional monitoring techniques such as HPLC, GC, and/or mass spectroscopy.

Glossary

Abbreviations

GC

Gas Chromatography

LC

Liquid Chromatography

MS

Mass Spectrometry

NMR

Nuclear Magnetic Resonance

PAT

Process Analytical Technology

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.6c00388.

• ¹H NMR spectra for all compounds and additional experimental details, materials, and methods, including photographs of the NMR reaction monitoring setup. (PDF)

¶.

Department of Chemistry, Dartmouth College, Hanover, New Hampshire 03755, United States

The manuscript was written through contributions of all authors.

The authors declare no competing financial interest.

⊗.

Retired

No unexpected safety hazards were encountered. Work with trifluoromethanesulfonyl azide and organic azides must be carried out in solution with precautions due to explosive compounds, including the use of flow chemistry to consume these intermediates as they are formed.

Due to a production error, this paper was published ASAP on September 21, 2026, with an older version of the Supporting Information. The correct version was reposted on September 22, 2026.