200,000+ products from a single source!
sales@angenechem.com
Home > Each intermediate was followed by online FT- IR analysis of characteristic signals
As expected, with the serial dilution that occurs on the stepwise addition of each reagent, the adiabatic temperature rise of each consecutive step decreases. Of particular note, however, is the very high adiabatic temperature rise of 110 °C following the n- butyl lithium addition. During this experiment, the concentration profile of each intermediate was followed by online FT- IR analysis of characteristic signals, Figure 3. A complete overview of the trending concentrations could be obtained, which was particularly pleasing since few, if any, of these intermediates were expected to be stable enough to be obtained as isolated analytical references.
In order to render the data acquired by FT-IR quantitative, off-line 1H NMR was also used to track key intermediates and byproducts. Once the mol % at a particular time point could be determined by NMR with an internal standard, the peak height by FT-IR could be correlated to this concentration at the same time point. In this way, the continuous data stream coming from the noninvasive FT-IR could be used to build a kinetic model of the reaction progress at a variety of temperatures and concentrations. Particularly insightful was the plot shown in Figure 4 which depicts the dose-controlled transition from compound 6 to compound 7. We determined that, at the slow dosing rate of n-butyl lithium in this experiment, both reactions were addition-controlled, but the rate of conversion of 6 into 6a had to be much faster than that for 6a to 7 in order to account for the sharp transitions observed in the concentration profiles of 6a and 7. As such the first equivalent of n-butyl lithium was almost entirely being employed in deprotonation of 6 to 6a, with very little being made available for ring closure of 6a to 7 (via ortho-lithiation and aryne formation).
In order to explore deviation scenarios, we devised an NMR experiment where dosing of n-butyl lithium was significantly faster and a local hot spot was certain to exist. At a faster dosing rate, we reasoned that we might shift away from a purely dose- controlled addition and begin to accumulate n-butyl lithium which could then be employed for not only selective deprotonation of 6 to 6a but also a parallel ring closure from 6a to 7. This would put lithiated benzoxazole 7 in the presence of starting material 6, a situation that we expected to compromise the selectivity of the reaction due to proton transfer between the acidic 6 and basic 7.15 The half amount of
n-butyl lithium (1.05 instead of 2.10 equiv) was added quickly to the starting material at −35 °C and a sample quenched with deuterated methanol at various time points. Figure 5 shows the 1H NMR spectra at 0.5 and 2.5 h where lithiated starting material 6a, lithiated benzoxazole 7, and other compounds can be identified as deuterium-quenched species (compound 6 could not be tracked by NMR due to background deuterium exchange of the acid NH proton). At 0.5 h (Figure 5a) compound 6a was the predominant product as expected, but already present was 10% of compound 7 which had evidently formed in parallel due to non-dose-controlled conditions brought about by accumulation of n-butyl lithium. Since a 9:1 ratio of 6a to 7 was integrated, by subtraction this required that residual starting material 6 was still present in an approximately 5% amount to account for the 1.05 equiv of n-butyl lithium initially added (otherwise a ratio of 10.0:0.5 of 6a to 7 would have been observed if all 6 had been consumed). By 2.5 h, Figure 5b, the newly identified proton-quench byproduct 10 had appeared in a 1:1 ratio with cyclized compound 7, as would be expected by the complete proton transfer from residual starting material 6 (initially 5%) to compound 7 (reduced now from 10 to 5%), with Figure 5c showing reference spectra of the proton-quenched byproduct 10 prepared separately. Based on this NMR study we concluded that if the chemistry was conducted under conditions where there was the coexistence of compounds 6 and 7 we would always generate byproduct 10 by proton transfer if sufficient time was given.16 In this way, we determined that a slow dosing rate would be a critical parameter if we wanted to avoid a local-zone effect, i.e. an accumulation of n-butyl lithium in the dosage area in batch. This was certainly a compatible criterion for the batch equipment which would otherwise reliably generate a hot spot in the dosage area based on the calorimetry data outlined above.
Regarding the stability of compound 7, similar NMR experiments with a slower dosing of 2.1 equiv of n-butyl lithium (i.e., seeking to eliminate the possible presence of residual starting material 6 in the presence of lithiated 7, even during the dosing period) demonstrated that at −15 °C the lithiated intermediate 7 could be held for up to 1.5 h without significant degradation. However, if the temperature was raised to 0 °C then up to 70 mol % of a new byproduct identified as dimeric 11 was formed within 1.5 h, Table 1. The data strongly suggested that dimeric 11 was being formed from the desired intermediate 7 at this temperature.17 In complement, FT-IR tracking at three temperatures (10, 0, and −10 °C) confirmed the temperature sensitivity, and the kinetic data were used to set a maximal internal temperature of −10 °C for the ideal process.
In conclusion to these mechanistic studies, we took the NMR and FT-IR data together and realized that in order to retain the high purity of lithiated intermediate 7 the ideal process would require a long n-butyl lithium dosing time and good mixing to avoid a local-zone effect and the resulting issues of proton transfer from residual 6, as well as hot spots. However, the holding time of 7 should be as short as possible to avoid decomposition in general and dimerization to 11 in particular.
Our conclusion was that this paradoxical situation excluded the possibility of using batch equipment on manufacturing scale since a long dosing time would equate to a long holding time of 7 during dosing, and the mixing efficiency of large batch reactors was expected to decrease on increasing volume. In flow equipment, the dosing time is quasi-equivalent to the mixing time and can be rapid with judicious choice of flow rates and mixer type. Heat removal can also be excellent due to the higher surface-area-to-volume ratio compared to batch. Finally, the holding time of 7 could also be short in continuous flow since the next reagent would intercept the reaction mixture after a short time defined by the flow rates and volume of the selected equipment. The combination of calorimetry, FT-IR, and NMR studies therefore consistently pointed us toward carefully designed continuous flow equipment to handle the first step of the sequence toward compound 1.
NMR-based stability experiments on the lithium sulfinate 8 resulting from sulfur dioxide quench showed that it was entirely stable at the low temperature required for the ortho-lithiation. The first signs of decomposition according to SITARAM calorimetry began at 97 °C. Equally, sulfonyl chloride 9 was found to be stable in the reaction medium, with NMR holding experiments at room temperature over 3 days showing no sign of degradation. Background hydrolysis of 9 by the aqueous solution of the Weinreb amine was to be expected, however, and screening of the base required to deprotonate the Weinreb amine salt (vide supra) had already led us to select dipotassium hydrogen phosphate. Further inorganic carbonate and phosphate bases were included in the extended screening, but dipotassium hydrogen phosphate led to the least decomposition of compound 9 over an extended period at 30 °C according to quantitative HPLC, Table 2.
Sulfonyl chloride 9 was found to hydrolyze to the sulfonic acid only at elevated pH. Conversely, low pH led to ring-opening hydrolysis of the oxazole moiety in compound 1 to provide a phenolic compound. As such we avoided extreme acidity or basicity by only dosing 40% of the full 10 equiv of aqueous dipotassium hydrogen phosphate solution to the sulfonyl chloride, the remaining 60% coming premixed with the Weinreb amine in aqueous solution.
Having determined the kinetics for each of the byproduct pathways, summarized in Figure 6, these pathways were incorporated into the Dynochem-based kinetic model to enable an overall simulation of the reaction within the flow reactor equipment (see the Supporting Information).
In order to effectively select the appropriate flow reactor, we needed to tackle the issue of solid-generation which had been a critical issue on lab scale. Lithium chloride is a prime suspect as a precipitant in all chemistries between organolithium reagents and chloro-aromatics. Commercial solutions of lithium chloride are available at 0.5 M in tetrahydrofuran, and the maximum possible concentration in our chemistry was calculated at 0.3 M (1.0 equiv). However, since our solution contained dissolved substrates, and we operated at sub-ambient temperatures, we initiated a seeding experiment to determine if we were operating under conditions of supersaturation. Small, but successive amounts of lithium chloride (0.03 eq. then 0.04 equiv) were spiked into the mixture of starting material 6 and n-butyl lithium at −30 °C, but gratifyingly no crystallization event was observed over a 2 h aging period.19
Lithium sulfinate intermediate 8 is a known solid, and so as a next step, we prepared it in a separate experiment and recirculated a concentrated suspension by peristaltic pump between two glass vessels with a gravity overflow between them.20 Over 24 h no accumulation of solids at the bottom of the vessels or at the overflow junction was observed.
© 2019 Angene International Limited. All rights Reserved.