Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Simultaneous self-optimisation of yield and purity through successive combination of inline FT-IR spectroscopy and online mass spectrometry in flow reactions

Simultaneous self-optimisation of yield and purity through successive combination of inline FT-IR... Self-optimisation constitutes a very helpful tool for chemical process development, both in lab and in industrial applications. However, research on the application of model-free autonomous optimisation strategies (based on experimental investigation) for complex reactions of high industrial significance, which involve considerable intermediate and by-product formation, is still in an early stage. This article describes the development of an enhanced autonomous microfluidic reactor platform for organolithium and epoxide reactions that incorporates a successive combination of inline FT-IR spectrometer and online mass spectrometer. Experimental data is collected in real-time and used as feedback for the optimisation algorithms (modified Simplex algorithm and Design of Experiments) without time delay. An efficient approach to handle intricate optimisation problems is presented, where the inline FT-IR measurements are used to monitor the reaction’s main components, whereas the mass spectrometer’s high sensitivity permits insights into the formation of by-products. To demonstrate the platform’s flexibility, optimal reaction conditions of two organic syntheses are identified. Both pose several challenges, as complex reaction mechanisms are involved, leading to a large number of variable parameters, and a considerable amount of by-products is generated under non-ideal process conditions. Through multidimensional real-time optimisation, the platform supersedes labor- and cost-intensive work-up procedures, while diminishing waste generation, too. Thus, it renders production processes more efficient and contributes to their overall sustainability. . . . . Keywords Microreactiontechnology InlineFT-IRspectroscopy Onlinemassspectrometry Self-optimisation Organolithium compounds Introduction requirements. Through applying a precise process control that already intervenes at an early stage, manufacturing When industrial production processes are not conducted processes can be rendered more efficient and more sus- under ideal conditions, labor- and cost-intensive work-up tainable, while simultaneously diminishing waste procedures become necessary to fulfill product quality generation. Highlights � Self-optimisation for complex reactions while minimising by-product formation � Successive combination of inline FT-IR and online mass spectrometer, leveraging each method’s advantages � Improved production process efficiency and sustainability through combined DoE and modified Simplex algorithm * Thorsten Röder Institute of Chemical Process Engineering, Mannheim University of t.roeder@hs-mannheim.de Applied Sciences, Paul-Wittsack-Str. 10, 68163 Mannheim, Germany Department of Biochemical and Chemical Engineering, Equipment Merck KGaA, Frankfurter Str. 250, 64293 Darmstadt, Germany Design, TU Dortmund University, Emil-Figge-Str. 68, Institute of Instrumental Analytics and Bioanalysis, Mannheim 44227 Dortmund, Germany University of Applied Sciences, Paul-Wittsack-Str. 10, 68163 Mannheim, Germany 286 J Flow Chem (2021) 11:285–302 Several optimisation strategies exist for this purpose [1, Experimental section 2]. Their high significance for chemical process develop- ment, both in lab and in industrial applications, has been Reactions demonstrated by the extant literature [3–11]. In industrial contexts, however, process optimisation often proceeds Organometallic synthesis through one-by-one optimisation [12, 13] instead of a more efficient multidimensional approach [14, 15]. In ac- In a first step, the exothermic deprotonation reaction of a CH- ademic research, such systematic, multidimensional opti- acidic compound 1 in tetrahydrofuran THF (anhydrous max. misation strategies have been studied in detail [16–28]. 0.005% H O, Merck, Germany) with n-butyllithium 2 leads to They have been proven to constitute a valuable tool for a non-isolable, unstable, lithiated intermediate compound 3 process optimisation, especially when being integrated in- [55–57]. This deprotonation step is followed by a nucleophilic to fully-automated microreactor platforms [29–35]. In addition including the lithiated intermediate’sreaction with an combination with online analysis, real-time reaction mon- electrophilic compound 4. The resulting intermediate 5 is itoring becomes possible, where intermediates and by- quenched with methanol (for synthesis, >99%, Carl Roth, products can be observed as well [9]. In particular, inline Germany) leading to a stable product 6 (Scheme 1). FT-IR spectroscopy and online mass spectrometry consti- The starting material n-butyllithium 2 was chosen from −1 tute promising analysis techniques that enable rapid quan- Sigma Aldrich, Germany, with a concentration of 1.6 mol L tification of reactants (analysis duration <1 min) [24, in n-hexane. Initial concentrations of the CH-acidic compound −1 36–49]. 1 and the electrophilic compound 4 amounted to 0.8 mol L . While the extant literature has applied inline FT-IR spectroscopy [20, 21, 50, 51]aswellas onlinemassspec- Synthesis of terminal epoxide trometry [24] in self-optimisation settings only for rather simple reactions and merely in isolation, this work ex- The epoxide 10 is synthesized from acetophenone 8 tends the prior ones by presenting an enhanced self- (ReagentPlus®, 99%, Sigma Aldrich, Germany) via in-situ optimising platform that integrates a successive combina- generated (bromomethyl)lithium (Scheme 2). Here, lithium- tion of inline FT-IR spectrometer and online mass spec- halogen exchange of dibromomethane 7 (99%, Sigma trometer. This self-optimising platform enables model- Aldrich, Germany) with methyllithium as its lithium bromide free autonomous optimisation without the need for human complex leads to (bromomethyl)lithium, which immediately intervention and is utilized to experimentally identify op- reacts with the carbonyl group of acetophenone, generating its timal reaction conditions of organic syntheses that involve bromomethyl alkoxide. In a following step, the alkoxide cy- complex reaction mechanisms. For this purpose, inline clizes to epoxide 10 [58]. FT-IR measurements are used to monitor the reaction’s The reaction was carried out in tetrahydrofuran THF (an- main components, whereas the high sensitivity of a mass hydrous max. 0.005% H O, Merck, Germany). Initial concen- spectrometer provides insights into the formation of by- trations of dibromomethane 7 and acetophenone 8 amounted products. Unlike prior works (e.g., [23, 28, 52–54]), no −1 to 0.8 mol L . Methyllithium lithium bromide complex solu- chromatographic separation is conducted before MS anal- tion was chosen from Sigma Aldrich, Germany, with a con- ysis, thus accelerating the analysis process drastically. −1 a centration of 1.5 mol L in diethyl ether . A novel approach is developed for solving intricate multidimensional optimisation problems, aiming at Experimental setup maximising product yield and purity. A modified Simplex algorithm as well as Design of Experiments are Microreactor experiments applied to identify ideal reaction conditions, while at the same time delivering an in-depth process understanding. The high flexibility of the chosen set-up is demonstrat- In case of the organometallic synthesis with n-butyllithium, a plate microreactor was directly connected to a capillary ed by means of optimisation of two different reaction types that are of great industrial significance, namely an microreactor, allowing to maintain two independent tem- perature levels (Fig. 1a). The deprotonation step (reaction organometallic reaction with n-butyllithium [55–57]and an epoxide synthesis [58]. Both studied reactions serve as of the CH-acidic compound 1 with n-butyllithium 2)was starting points for a multitude of further synthesis steps. Within certain limits, the concentration of organometallic reagents might be Thus, a broad spectrum of chemical reactions can be cov- subject to minor deviations. However, pretests described in supporting infor- ered, acting as basic building blocks for organic-chemical mation A.1 indicate that this limitation had only a negligible effect on the compounds of industrial relevance [59–65]. study’sfindings. J Flow Chem (2021) 11:285–302 287 Scheme 1 Organometallic synthesis. 1 CH-acidic compound deprotonation ∙ 2 n-butyllithium ∙ 3 lithiated in- Li termediate ∙ 4 electrophilic com- pound ∙ 5 intermediate ∙ 6 product 1 2 3 hydrolysis nucleophilic + MeOH addition - MeOLi 45 6 carried out in the plate microreactor, which had been de- Nuevo, Huber, Germany) allowed for adjusting the tempera- signed and manufactured by mechanical precision milling ture of the heating/cooling fluid. of stainless steel (Fig. 1b). It consisted of three stainless The nucleophilic addition was carried out in a coiled steel plates layered one on top of each other, where channels 1/16 in. stainless steel capillary microreactor. Precooling of for mixing and residence time were milled into the middle the electrophilic compound 4 was conducted in a capillary reactor plate (each channel had a quadratic cross-section). with an inner diameter of 0.5 mm. The precooling capillary To avoid bypass flow, the plates were evenly pressed onto was directly connected to the outlet of the plate reactor via a each other (ensuring equal pressure between the plates) stainless-steel T-mixer (inner diameter 0.5 mm), followed by a using numerous screws. Bore holes that were located on reaction channel that had an inner diameter of 0.5 mm and a the lateral surface of the middle reactor plate were utilized total volume of 0.59 mL. The chosen microreactor setup per- as inlet and outlet connections for the reactants. Channels mitted residence times between 0.2 and 1 min. Temperature control of precooling, mixing, and reaction was achieved for precooling (0.5 mm × 0.5 mm), mixing (0.5 mm × 0.5mm), and reactionwerearrangedin anarrow-shaped using a bath thermostat (Ministat, Huber, Germany). geometry. The reaction channel was divided into two sec- Regarding epoxide synthesis, the microreactor setup tions: at the entrance, a 0.5 mm × 0.5 mm channel enabled consisted of coiled 1/16 in. PFA tubing (Fig. 2). Reactants enhanced heat transfer; with increasing conversion, an in- were precooled and mixed in T-mixers (0.5 mm inner bore creased channel size (1 mm × 1 mm) was employed, en- hole). The reaction mixture then passed two modular reactor abling high reactant conversion at longer residence times. pieces that were connected to each other. In the first capillary, The entrance region of the reaction channel had a length of which had a total volume of 0.5 mL (inner diameter 0.5 mm), 1.1 m, the region with increased channel size had a length of residence time remained constant at 0.5 min. Temperature was 9.11 m, resulting in a total reactor volume of 7.37 mL. varied between −35 °C and − 10 °C (bath thermostat Huber As the kinetics of the deprotonation step had already been Tango Nuevo). In the second capillary, which had a total vol- studied in detail [55–57], deprotonation of the CH-acidic com- ume of 2 mL (inner diameter 0.75 mm), the reaction mixture pound 1 was performed at a constant residence time of 8 min waswarmedto20°C(1minresidencetime, baththermostat and a temperature of −35 °C, ensuring full conversion of the n- Huber Ministat). Thus, cyclization of the bromomethyl alkox- butyllithium 2. In order to avoid clogging, the CH-acidic com- ide intermediate took place. A back-pressure regulator (3 bar) pound 1 was provided in marginal excess; the stoichiometric ensured light overpressure within the whole microreactor. ratio of n-butyllithium: CH-acidic compound amounted to Mixing of reactants occurred through one out of three dif- 0.8. Temperature control was achieved through heat carrier ferent cases. In the first case, dibromomethane 7 and channels (3 mm × 10 mm) that were milled on the bottom of acetophenone 8 had been premixed manually (feed stream 1) the middle reactor plate, where two Pt100 resistance thermom- eters could directly be inserted into the inflowing and When investigating epoxide synthesis, the microreactor setup used PFA outflowing thermal fluid via bore holes located at the lateral tubing (instead of stainless-steel tubing) to avoid pitting corrosion caused by surface of the middle plate. A thermostat (Unistat Tango bromide ions. 288 J Flow Chem (2021) 11:285–302 Scheme 2 Preparation of epoxide 10 from acetophenone 8 via in situ generated (bromomethyl)lithium 7a and then combined with methyllithium (feed stream 2; Fig. (SyrDos2, HiTec Zang GmbH, Germany). Temperature and 2a). In the second case, dibromomethane 7 and acetophenone flow rates were controlled by a laboratory automation system 8 (Fig. 2b) entered through two separate feed streams and, (LabManager, HiTec Zang GmbH, Germany). once they had been mixed, were subsequently combined with methyllithium (feed stream 3), thus allowing to variably adjust Inline FT-IR and online MS measurements the stoichiometric ratio of 7 and 8. In the third case, dibromomethane 7 and methyllithium (Fig. 2c) entered Both described syntheses (organometallic reaction with n- through two separate feed streams and, once they had been butyllithium; epoxide synthesis) were continuously monitored mixed, were subsequently combined with acetophenone 8. at the reactor outlet. The analysis of the respective product Dosage of all starting materials within 1 mL glass syringes stream was conducted through a successive combination of was ensured by continuously working syringe pumps an inline FT-IR spectrometer and an online mass spectrometer FT-IR MS/GC a) PI R deprotonation nucleophilic addition CH-acidic compound P1 M2 M3 PI R M1 R1 R2 product n-BuLi τ = 8 min 6 τ = 0.2 to 1 min P2 T = −35 °C PI R TIR 3 1 electrophilic compound P3 T = −35 °C to −25 °C TIR PI R MeOH quench P4 b) c) Fig. 1 Microreactor setup for organometallic synthesis with n-butyllithium, process flow chart (a), plate microreactor (b), coiled capillary microreactor (c) J Flow Chem (2021) 11:285–302 289 Fig. 2 Microreactor setup for epoxide synthesis, process flow charts: a setup reduced to two feeds with a premixed solution of acetophenone and CH Br , b setup with three feeds and premixing of acetophenone and CH Br , c setup with premixing of CH Br and MeLi � LiBr 2 2 2 2 2 2 (Fig. 3). Inline FT-IR measurements were used to monitor the by-product formation (as MS possesses significantly higher reaction’s main components (starting materials and product), sensitivity). Unlike prior works (e.g., [23, 28, 52–54]), chro- whereas online MS measurements provided information about matographic separation had not been conducted before MS 290 J Flow Chem (2021) 11:285–302 Fig. 3 Experimental setup with successive combination of an inline FT- continuous syringe pumps; 2 – plate reactor for deprotonation; 3 – IR spectrometer and an online mass spectrometer (microreactor setup cooling bath for nucleophilic addition; 4 – inline FT-IR spectrometer; exemplarily provided for organometallic synthesis). Legend: 1 – 5–6/2 port valve; 6 – online MS. −1 analysis, as real-time monitoring without considerable time 696 cm . Analytical IR spectra and details on the integration delayneededto be achieved(m/zvaluesand their intensities method are provided in supporting information A.2. were accumulated over a measurement time of 60 s, resulting As the reactant concentrations (amounting to roughly −1 in a new mass spectrum every 60 s). 0.25 mol L ) were suitable for inline FT-IR analysis, no dilu- The use of an inline FT-IR spectrometer (Bruker ALPHA, tion step was required beforehand. Following the inline FT-IR United States) allowed for real-time reaction tracking with analysis, however, continuous quenching of the reaction mix- time delay <1 s, circumventing the need to quench the reac- ture took place (compact dual piston pump AZURA P4.1S, tion. The reactor outlet was directly connected to the spec- Knauer, Germany). A subsequent split and dilution step (refer trometer’s flow cell through a very short capillary of 3 cm to supporting information A.3 for more details) allowed to with an inner diameter of 0.5 mm [55]. The measuring cell adjust reactant concentrations to render them suitable for anal- of the FT-IR spectrometer with a volume of 40 μLenabled ysis via online mass spectrometry, as the online mass spectrom- extremely fast measurement times, shorter than 2 ms (flow eter’s high sensitivity requires reactant concentrations of −1 −1 rates ≥0.31 mL min ). The employed FT-IR spectrometer around 0.02 mol L to not overload the spectrometer. −1 has an optical wavelength resolution of 4 cm .Infraredspec- Afterwards, aliquots of the reaction mixture were periodically −1 tra (500–1700 cm ) were collected through single reflection introduced to the mass spectrometer using a 6/2 port sample ATR (diamond crystal). valve (Analytical HPLC Continuous-Flow Injector, VICI, Characteristic IR bands allowed for calculating product United States). With the 6/2 port valve in load position, the yield based on previously determined calibration curves. The diluted reaction mixture filled a 5 μL loop fitted across two reaction product of the organometallic synthesis with n- of the valve ports and was thereafter automatically injected into butyllithium 6 was identified by means of a characteristic IR the mass spectrometer, switching the valve to its inject position −1 −1 band at 1241 cm to 1230 cm , whereas epoxide 10 was (a 1/16 in. stainless-steel tubing, 0.5 mm i.d., connected the −1 identified by means of a characteristic IR band at 711 cm to reactor outlet with the valve and the mass spectrometer). J Flow Chem (2021) 11:285–302 291 With the 6/2 port valve in inject position, another high-pressure optimisation is steered by a fully automated experimental se- pump (Agilent Technologies, United States) flushed the loop quence coded in MATLAB, which assumes control over op- −1 with a continuous methanol stream (300 μLmin , hypergrade timisation strategies and the calculation of the objective func- for LC-MS, Sigma Aldrich, Germany) and thus passed the tion. Simultaneously, MATLAB transfers the set points for sample into the mass spectrometer for analysis. The spectrom- pumps and thermostats to the automation system. The plat- eter used was an Advion Expression CMS operating in positive form is compatible with industrial production conditions (ori- atmospheric pressure chemical ionisation mode (APCI). A de- ented on NAMUR standards), ensuring a high level of process tailed description of the online MS setup is provided in safety. Integrated safety features, such as pressure and temper- supporting information A.3. ature monitoring, comply with industrial standards. More de- Overall, by combining these two techniques (FT-IR spec- tails on the experimental self-optimising workflow are provid- troscopy and online MS), the experimental set-up aimed at ed in supporting information A.4. leveraging methodological synergies. Specifically, while FT- Analytical results of inline FT-IR measurements are IR spectroscopy constitutes a fast, broadly applicable method transferred to MATLAB through an OPC interface (real- that reduces barriers to implementation, it is less suitable for time communication with time delay <1 s). Thus, product monitoring by-products unless they exceed the parts per mil- yield can directly be calculated incorporating the respec- lion (ppm) level [10, 37, 66]. tive calibration curve. Conversely, online MS is particularly well-suited for the Regarding online MS measurements, however, the transfer characterization and quantification of analytes up to trace of analytical results proceeds through extraction of CDF-files levels, even in complex mixtures [40, 42, 43, 47]. However, that are gained as output from the mass spectrometer, contain- it may only provide relative (instead of absolute) numbers due ing an accumulated scan time, m/z-values (mass-to-charge- to ionization suppression or enhancement effects [67–71]as ratio), and their respective intensities. As a result, each in- the components may affect each other’s ionization efficien- volved reactant that can be analyzed via APCI, can be identi- cies. For this reason, an additional internal standard (product fied by its characteristic m/z value, with the peak’s intensity intensity) was used to compensate for these biases (further indicating the reactant’squantity (Fig. 4). details regarding the exact procedure are provided in the fol- Especially in case of the organometallic synthesis with lowing chapter). n-butyllithium, pure substances of all reactants expected Compared to traditional HPLC, or even UPLC measure- to be involved in the complex reaction mechanism includ- ments [72], the chosen analytical techniques do not incur a ing by-product formation were available, hence, MS cali- time-delay of several minutes between the initial measurement bration curves could be determined for each individual and the computation of the corresponding analytical result. As component, see supporting information A.5.However, a result, unstable or reactive intermediates and products can be the actual reaction mixture leaving the reactor constitutes identified almost immediately, due to the involved real-time a complex matrix including solvent, main product, by- reaction monitoring leading to a time-delay of 1 min at most products, and not yet consumed starting materials. [10, 37, 40]. Hence, there is a high methodological fit between Ionization suppression or enhancement effects [67–71] FT-IR spectroscopy, online MS, and the objective of efficient may no longer be ruled out, as the components may affect process development, particularly in terms of continuous pro- each other’s ionization efficiencies. cesses. Yet, the absence of chromatographic separation adds In order to circumvent such elusive effects, following the an additional level of complexity to the interpretation of the extraction of CDF-files containing all m/z values, intensities resulting data [40]. of all involved by-products are calculated when product inten- To conclude, the experimental set-up applies FT-IR spec- sity reaches its peak. The sum of all those by-product intensi- troscopy and online MS in a highly complementary manner. ties is then determined, and the ratio of product and sum of by- FT-IR spectroscopy (instead of online MS) delivers absolute products is used as input for the objective function (Eq. 1). As values in order to avoid frequently emerging biases, whereas a result, the determined MS ratio contains the product’smass online MS provides the required sensitivity to examine by- intensity as internal standard and moreover compensates for products at the ppm level. concentration fluctuations that result from quenching the re- action with a constant mass flux. Individual calibration curves Self-optimisation are thus not required to calculate the objective function, but are useful to evaluate whether given components are suitable This work relies on a completely automated self-optimising for MS analysis in general. platform [73], which integrates a microreactor with automated To combine the estimated product yield from FT-IR anal- devices (pumps and thermostats) and a successive combina- ysis with information about undesired by-product formation, tion of real-time reaction monitoring through inline FT-IR the FT-IR signal is multiplied with the calculated MS ratio spectroscopy and online mass spectrometry. Real-time (Eq. 2). Aiming at maximising product yield while 292 J Flow Chem (2021) 11:285–302 Fig. 4 Exemplary online MS analysis of organometallic reaction with n-butyllithium. Experimental parameters: a stoichiometric ratio [3, 4]amounting to 1.0; reaction temperature − 10 °C, b stoichiometric ratio [3, 4] amounting to 1.0; reaction temperature − 30 °C simultaneously minimising by-product formation, the Simplex optimisation proceeds through iteratively replac- resulting objective function can easily be implemented in the ing a randomly chosen initial simplex. Specifically, the start described MATLAB code. A detailed description of the pre- simplex was set to random values located on each axis, with tests that had been conducted to scrutinize the applicability of the last remaining corner point always being randomly chosen the chosen evaluation method regarding MS files is provided within the reaction space. Hence, optimal reaction conditions in supporting information A.6. are identified as soon as the simplexes converge to a local optimum, where the value of the objective function does not intensity at of product vary anymore. The maximum number of experiments (per MS ratio ¼ ð1Þ optimisation cycle) was set to 30. intensities at of by−products When applying DoE, a Central Composite Design (CCD) was chosen. During each optimisation, two experimental runs MS ratio↑low proportion of by−products ð2Þ were executed. The first run screened the entire experimental MS ratio↓high proportion of by−products objective function ¼ FT−IR value  MS ratio space. Based on the first run’s experimentally obtained data points, a surface response model was calculated. The mathe- matical optimum of this surface response model was subse- In this work, two optimisation strategies, modified Simplex quently used as central point for the second DoE run. To refine algorithm and Design of Experiments (DoE), are applied, as the search for a global optimum, the size of the second DoE their performances had already been documented and com- run amounted to 20% of the size of the first DoE run. pared in detail [73]. The derived autonomous platform constitutes a modular and flexible system, Results enabling multi-variate and multi-objective optimisations in real-time. Hence, other objective functions (besides multiplying the FT-IR signal with the cal- Organometallic synthesis culated MS ratio, see Eq. 2) can swiftly be integrated. For instance, an objec- tive function consisting of differently weighted parts (e.g., weighting the costs for separating non-converted starting materials against the costs for work-up In order to evaluate the effectiveness of the presented self- procedures required to remove undesired by-products), or a jump function optimising platform incorporating a successive combination characterizing by-product formation, could also be investigated, depending on the issue being addressed. of real-time reaction monitoring through inline FT-IR J Flow Chem (2021) 11:285–302 293 spectroscopy and online mass spectrometry, two different op- As a result, an optimal stoichiometric ratio close to 1.0 was timisation studies were compared: In the first study, self- identified, whereas the optimal temperature could not be de- optimisation was only based on inline FT-IR measurements, termined exactly, but only be located within a certain range thus aiming at maximisation of main product yield. In the between −23 °C and − 15 °C. After three repetitions, the rela- second study, the estimated product yield from FT-IR analysis tive standard deviation amounted to 4.1% (stoichiometric ra- was combined with information from MS analysis aiming at tio) respectively 22% (temperature). maximising product yield and purity. The organometallic syn- In terms of DoE, the surface response models obtained by thesis with n-butyllithium was used as proof of concept, as the first and second run while optimising stoichiometric ratio kinetics and mechanism had already been studied in detail and temperature, are provided in Fig. 6. Experimental data [55–57, 74]. Pure substances of all involved compounds were points, from which the respective surface response model available meaning that calibration curves, and thus reference was built as best fit, are displayed as red dots. values, had already been known in advance. While the visualisation in Fig. 6 clearly indicates an opti- mal stoichiometric ratio near 1.0, the ideal reaction tempera- ture cannot be determined exactly. Instead, the resulting sur- Product yield maximisation face response models merely point to a broad temperature range. The second DoE run does not improve upon the result Regarding the optimisation procedure based exclusively on of the first one. FT-IR measurements, a modified Simplex algorithm and To assess optimisation accuracy, the average deviation of Design of Experiments were implemented as optimisation the experimental data points from the associated surface re- strategies. The second step of the organometallic synthesis sponse model was determined. It amounted to 15% for the with n-butyllithium, namely the nucleophilic addition, was first DoE run and to 7% for the second DoE run. The relative optimised, aiming at maximum product yield. Two optimisa- standard deviations of calculated optimal product yield after tion parameters were examined: the stoichiometric ratio of six repetitions amounted to 4.2% respectively 3.5%. lithiated intermediate 3: electrophilic compound 4 as well as Table 1 summarizes the obtained results for both optimisa- reaction temperature. The stoichiometric ratio was varied in tion strategies (Simplex and DoE). For each strategy, the op- the range between 0.5 and 2.5, while temperature was varied timal reaction conditions and the required number of experi- between −35 °C and − 10 °C. Note that, as residence time and ments are presented. Note that, in case of DoE optimisation, stoichiometric ratio are intertwined as far as the reaction at the results of the second DoE run are provided. hand is concerned, these two variables may not be varied The results of DoE optimisation are comparable to independently from each other. those of Simplex optimisation. The optimal product yield Figure 5 depicts the experimental results of Simplex optimi- is reached at a stoichiometric ratio of 1.0, which is asso- sation as coloured dots with the colours indicating the corre- ciated with a residence time of 0.3 min. Note that, for sponding values of product yield, and further displays the prog- residence times smaller than 0.3 min, full conversion of ress of the objective function over the course of the iterations. starting materials cannot be accomplished (leading to re- Three repetitions of the same optimisation procedure were duced product yield). However, the optimal reaction conducted, differing only with regard to their start simplexes. Fig. 5 Simplex optimisation of organometallic synthesis. Product yield product yield as objective function, exclusively based on inline FT-IR calculated based on compound 4. Figure illustrates first Simplex measurements. b Progress of objective function across experiments. optimisation (out of three overall). a Results of optimisation with Legend: ○ values of start simplex 294 J Flow Chem (2021) 11:285–302 Fig. 6 Optimisation of organometallic synthesis using DoE. Results are based exclusively on inline FT-IR measurements. Product yield calculated based on compound 4. a First DoE run for screening of whole experimental space. b Second DoE run to refine optimisation temperature is difficult to quantify, when only product (supporting information A.5). The ratio of MS product inten- yield is examined, as temperature optimisation is expected sity and the sum of MS intensities of all four by-products was to predominantly affect by-product formation, with the used as input for the objective function together with the main amount of created by-products in the experiment remain- product’s FT-IR signal, Eq. 2. Again, a modified Simplex ing below FT-IR’s sensitivity threshold. algorithm and Design of Experiments were applied as optimi- sation strategies. Nucleophilic addition was optimised aiming Product yield maximisation with simultaneous minimisation at maximising product yield while simultaneously minimising of by-product formation by-product formation. Figure 7 displays the experimental results of Simplex opti- To expand upon the results of self-optimisation based exclu- misation with stoichiometric ratio and temperature as optimi- sively on inline FT-IR measurements, a successive combina- sation parameters, and further presents the progress of the tion of real-time reaction monitoring through inline FT-IR objective function over the course of iterations. The results spectroscopy and online mass spectrometry was implemented of Simplex optimisation are illustrated as coloured dots, with within the self-optimising platform. As mass spectrometry the colours indicating the corresponding values of the objec- constitutes an analytical technology that possesses significant- tive function, i.e. the combination of main product’sFT-IR signal and MS ratio, including intensities of main product and ly higher sensitivity than FT-IR spectroscopy, and as reaction temperature may exert considerable influence on the forma- sum of four by-products. The stoichiometric ratio of lithiated tion of by-products that are present in the reaction mixture to a intermediate 3: electrophilic compound 4 was varied in the much lesser extent compared to the main product, such a range between 0.5 and 2. The temperature was varied between combination of both measurement techniques was expected −35 °C and − 10 °C. to provide additional insights. The Simplex optimisation with successive combination of The kinetics [55–57] and mechanism of by-product forma- real-time reaction monitoring through inline FT-IR spectros- tion [74] of the presented organometallic synthesis had previ- copy with online mass spectrometry identified an optimal stoi- ously been studied in detail. Thus, it has been known in ad- chiometric ratio close to 1.0, which is in line with the result vance that four characteristic by-products can be formed. All obtained by the prior optimisation solely based on FT-IR mea- four by-products were available as pure substances, from surements, and an optimal reaction temperature of −30 °, which MS calibration curves could be determined whereas the prior optimisation could not determine optimal Table 1 Comparison of experimental results of Simplex and DoE optimisation. Results are based exclusively on inline FT-IR measurements Parameters Simplex optimisation DoE optimisation (2nd run) (experimentally-obtained) Optimal product yield [%] 98.7 99.0 Temperature [°C] insensitive (−23 °C to −15 °C) insensitive (−35 °C to −20 °C) Stoichiometric ratio (lithiated intermediate: electrophilic compound) 1.0 1.0 Residence time [min] 0.3 0.3 Number of experiments 17 9 per run(18 overall) J Flow Chem (2021) 11:285–302 295 Fig. 7 Simplex optimisation of organometallic synthesis. a Results of optimisation based on inline FT-IR and online MS measurements. b Progress of objective function over course of experiments. Legend: ○ values of start simplex reaction temperature accurately. After three repetitions, the strategy instead of the Simplex algorithm to obtain greater relative standard deviation of the stoichiometric ratio understanding of the underlying processes. amounted to 3.9%, and the one of temperature amounted to The epoxide synthesis proceeded through in situ generated 3.0%. (bromomethyl)lithium, which had been generated by lithium- Figure 8 provides the corresponding results of DoE opti- halogen exchange of dibromomethane, and which constitutes misation, based on the combination of FT-IR and MS signals. a highly unstable lithiated intermediate. Even though The surface response models obtained by the first and second microreactors have been proven to be suited for handling such DoE run, while optimising stoichiometric ratio and tempera- temperature-sensitive syntheses [58, 75–79] due to enabling ture, are displayed with experimental data points being repre- efficient mixing and fast heat transfer, several undesired side sented as red dots. Moreover, individual surface response reactions cannot be ruled out entirely. This includes (1) nucle- models resulting from solely evaluating the FT-IR respective- ophilic addition of the alkyllithium reagent to the carbonyl ly MS signals were calculated and are provided in supporting group of acetophenone, (2) formation of polymers from a information B.2. coupling of bromomethyl lithium with dibromomethane or The surface response models illustrated in Fig. 8 clearly (3) a premature quench of the reaction mixture before the indicate an optimal stoichiometric ratio near 1.0 and an opti- cyclization step could occur [58, 80, 81]. Thus, the applied mal reaction temperature of −30 °C, comparable to the results self-optimisation procedure once again aimed at maximising of Simplex optimisation, when the latter is based on a combi- of main product yield, while simultaneously minimising by- nation of FT-IR and MS measurements. The average devia- product formation. More details on potential side-reactions are tion of experimental data points from the associated surface provided in supporting information C.1. response model amounted to 10% in case of the first DoE run, In contrast to the organometallic reaction with n- and to 5% in case of the second DoE run. The relative standard butyllithium, pure substances of by-products were not avail- deviations of the calculated values of the objective function able in terms of epoxide synthesis. Thus, pre-tests under var- after six repetitions amounted to 10.1% and 9.8% for the first iation of reaction temperature were conducted to identify char- and second run, respectively. acteristic by-products of epoxide synthesis. A detailed de- Table 2 summarizes the performance of both optimisa- scription of pre-tests as well as MS spectra are provided in tion strategies (Simplex and DoE) in terms of maximising supporting information C.2. As a result, the ratio of MS prod- product yield while simultaneously minimising by-product uct intensity and the sum of MS intensities of six characteristic formation. Synthesis of terminal epoxide DoE allowed to screen the entire experimental space and to build surface response models, thus providing further insights into the reaction mechanism In order to evaluate the platform’s versatility in terms of reac- (and where to find the global optimum). Conversely, applying the modified tion types, an epoxide synthesis was also investigated. In con- Simplex algorithm for such a comparatively unknown reaction would have exposed the study to the risk of the algorithm getting stuck at a local (rather trast to the reaction examined before, this time less informa- than global) optimum. Due to the lack of a priori information on the reaction, tion regarding reaction mechanism and kinetics had been no adequate assessment could have taken place to determine whether a given available a priori. Thus, DoE was applied as optimisation optimum is a global or local one. 296 J Flow Chem (2021) 11:285–302 Fig. 8 Optimisation of organometallic synthesis using DoE. Results are based on inline FT-IR and online MS measurements. a First DoE run for screening of whole experimental space. b Second DoE run to refine optimisation m/z values (by-products) was used as input for the objective evaluation of FT-IR respectively MS signals are provided in function together with the main product’s FT-IR signal, Eq. 2. supporting information C.3. Three different dosing options were investigated, differing in their respective type of reactants mixing based on two or Epoxide synthesis proceeding through three feed streams three reactant feeds. In all three cases, residence time in the first part of the reactor was kept constant at 0.5 min. In the Second, the experimental setup was enhanced to handle three following cyclization step, the reaction mixture was then held feed streams: dibromomethane 7 and acetophenone 8 entered at 20 °C for 1 min. through two separate feed streams and, once they had been mixed, were subsequently combined with methyllithium (feed stream 3, see Fig. 2b). Thus, three parameters were optimised: Epoxide synthesis proceeding through two feed streams stoichiometric ratio of acetophenone 8: dibromomethane 7 in the range between 0.1 and 1.5, stoichiometric ratio of Initially, the epoxide synthesis proceeded through combina- methyllithium: premixed starting materials in the range be- tion of a manually-premixed solution of dibromomethane 7 tween 0.6 and 1.1, and reaction temperature in the range be- and acetophenone 8 (1.1 eq. dibromomethane) with the tween −35 °C and − 10 °C. methyllithium, resulting in two feed streams (Fig. 2a). Thus, In the third case, dibromomethane 7 and methyllithium two parameters were optimised: stoichiometric ratio of entered through two separate feed streams and, once they methyllithium: premixed starting materials in the range be- had been mixed, were subsequently combined with tween 0.1 and 1.1, and reaction temperature in the range be- acetophenone 8 (Fig. 2c). Again, three parameters were tween −35 °C and − 10 °C. optimised, albeit now, stoichiometric ratio of methyllithium: In Fig. 9, the resulting surface response models obtained by dibromomethane 7 in the range between 0.1 and 1.1, stoichio- the first and second DoE run are illustrated with manual metric ratio of lithiated intermediate: acetophenone 8 in the premixing of dibromomethane 7 and acetophenone 8. range between 0.1 and 1.1, and reaction temperature in the Results are based on the combination of FT-IR and MS sig- range between −35 °C and − 10 °C. nals. Experimental data points are represented as red dots. For both cases including three feed streams, Fig. 10 repre- The optimal stoichiometric ratio of methyllithium: sents the results of DoE optimisation as experimental data premixed starting materials and optimal reaction temperature points, whose colour scheme represents the objective function were shown to be close to 0.9 respectively −35 °C (Fig. 9). combined from FT-IR and MS signals (see Eq. 2) dependent on Individual surface response models resulting from an the respective variable parameters. The corresponding surface response models derived for the optimisation of three variable Note that, there is a minor deviation between, on the one hand, the results of parameters can be found in supporting information C.3. the DoE optimisation and, on the other hand, the findings described in the Investigating the second synthesis route variable premixing extant literature, where a one-dimensional optimisation of reagent amount and concentration led to 1.5 equiv. of methyllithium under ideal conditions [58]. of dibromomethane 7 and acetophenone 8, DoE optimisation However, the applied optimisation strategies cannot be compared directly, as resulted in an optimal stoichiometric ratio of acetophenone: the setup described in this work allows for multidimensional self-optimisation dibromomethane amounting to 0.9. Thus, under optimised based on calculations of optimisation algorithms, whereas the extant work applies a one-by-one optimisation based on an evaluation of HPLC yields. conditions, dibromomethane 7 is provided in slight excess. Moreover, in the current work, the formation of by-products was considered This finding is in accordance with studies described in the within the objective function, thus resulting in an intricate optimisation prob- extant literature [58]. Again, the best compromise between lem where product yield is maximised with simultaneously reducing the the highest possible main product yield and the lowest level amount of undesired by-products. J Flow Chem (2021) 11:285–302 297 Table 2 Comparison of Parameters Simplex optimisation DoE optimisation experimental results of Simplex and DoE optimisation. Results are (2nd run) based on inline FT-IR and online MS measurements (experimentally-obtained) Optimal result 28 27 Temperature [°C] −30 −30 Stoichiometric ratio 1.0 1.0 (lithiated intermediate: electrophilic compound) Residence time [min] 0.3 0.3 Number of experiments 15 9 per run (18 overall) of by-product formation, was found at a stoichiometric ratio of options are discussed). For every case, optimal reaction con- 0.9 (methyllithium: starting materials) and a reaction temper- ditions are displayed, which were calculated based on the ature of −35 °C. respective surface response model of the second DoE run. Whereas the second synthesis route simply constituted an Moreover, the average deviation of the experimental data extension to three feed streams compared to the first route, a points from the associated surface response model was deter- completely different reaction control was applied in the third mined, allowing to assess each optimisation’saccuracy. case. Herein, external quenching [77, 82] was conducted, in- Comparing the results of DoE optimisation for all three stead of “Barbier”-like internal quenching [83–85]. Given the investigated dosing options, it becomes apparent that the instability of lithium carbenoids, which are likely to suffer objective function assumes a significantly higher value in from immediate thermal decomposition, the generation of its optimum, when applying external quenching. This indi- carbenoid species followed by external trapping with electro- cates a noticeably smaller proportion of undesired by- philes has been less researched compared to internal products compared to reaction control through internal quenching [81]. The enhanced heat transfer in microreactors quenching as maximum main product yield is similar across permits the handling of such thermolabile carbenoids [31, 58, all three investigated cases, see supporting information C.3. 76–79, 86]. The optimisation studies conducted in this work However, while the examination of both internal quenching identified optimal conditions at a slight shortage of routes resulted in surface response models from which op- methyllithium (stoichiometric ratio of methyllithium: timal reaction conditions could be predicted in a reliable and replicable manner as indicated by comparatively low aver- dibromomethane 7 amounting to 0.9) and slight excess of acetophenone (stoichiometric ratio of lithiated intermediate: age deviations of the experimental data points from the as- acetophenone 8 amounting to 0.9). A reaction temperature sociated surface response models, a sufficiently precise rep- of −35 °C once again proved to be ideal for minimisation of resentation of experimental data resulting from the external by-product formation. quenching route through surface response models is much Table 3 summarizes the obtained results of DoE optimisa- harder to achieve. This is due to the fact that, during DoE tion for epoxide synthesis (all three investigated dosing optimisation, experimental plans need to cover a Fig. 9 Optimisation of epoxide synthesis using DoE including manual temperature. Results are based on inline FT-IR and online MS measure- premixing of acetophenone and dibromomethane with two variable ments, a First DoE run for screening of whole experimental space. b parameters, stoichiometric ratio (MeLi: premixed starting materials) and Second DoE run to refine optimisation 298 J Flow Chem (2021) 11:285–302 Fig. 10 Optimisation of epoxide synthesis using DoE with experimental acetophenone: dibromomethane, stoichiometric ratio of MeLi: premixed setup enhanced to handle three feed streams. Results are based on inline starting materials, and temperature, b Premixing of dibromomethane 7 FT-IR and online MS measurements. Experimental data of second DoE and MeLi (external quenching) with three variable parameters, stoichio- run is demonstrated, respectively. a Premixing of dibromomethane 7 and metric ratio of MeLi: dibromomethane, stoichiometric ratio of lithiated acetophenone 8 with three variable parameters, stoichiometric ratio of intermediate: acetophenone, and temperature comparatively broad temperature range (first run: −35 °C up optimum), thus enabling investigation of smaller tempera- to −10 °C; second run: −35 °C up to −25 °C) from which ture steps compared to DoE optimisation. surface response models can be built. Yet, applying external Figure 11 displays the results of this Simplex optimisation quenching requires a longer lifetime of the lithium and further presents the progress of the objective function over carbenoid species compared to its in-situ generation during the course of its iterations. The experimental results are illus- internal quenching. However, due to the high thermal insta- trated as coloured dots, with the colours indicating the corre- bility of lithium carbenoids [82, 87–89], at temperatures sponding values of the objective function (combination of above −35 °C, they are likely to decompose before a reac- main product’s FT-IR signal and MS ratio). Three parameters tion with acetophenone can occur. To overcome this chal- were optimised: stoichiometric ratio of methyllithium: lenge, an additional Simplex optimisation was conducted dibromomethane 7 in the range between 0.8 and 1.0, stoichio- for the external quenching route based on the findings of metric ratio of lithiated intermediate: acetophenone 8 in the the first DoE run (start simplex was located around its range between 0.9 and 1.1, and reaction temperature in the Table 3 Experimental results of epoxide optimisation using DoE. Results are based on inline FT-IR and online MS measurements Parameters Premixing of acetophenone + Premixing of acetophenone + Premixing of MeLi + dibromomethane: dibromomethane: dibromomethane two variable parameters three variable parameters (external quenching) (experimentally-obtained) 2.0 2.1 16.7 Optimal result Stoichiometric ratio 1 0.9 0.9 0.9 (MeLi: premixed (acetophenone: dibromomethane) (MeLi: starting materials) dibromomethane) Stoichiometric ratio 2 – 0.9 0.9 (MeLi: starting materials) (lithiated intermediate: acetophenone) Temperature [°C] −35 −35 −35 Number of experiments 18 (each run 9) 22 (each run 11) 22 (each run 11) relative standard deviations 8.6 10.7 9.5 after six repetitions [%] 8.5 11.8 9.8 1st point 2nd point average deviation experimental 12 17 30 data points – surface response mod- 9 9.6 21 el [%] first DoE run second DoE run J Flow Chem (2021) 11:285–302 299 Fig. 11 Simplex optimisation of epoxide synthesis with premixing of dibromomethane and MeLi (external quenching). a Results of optimisation based on inline FT-IR and online MS measurements. b Progress of objective function over course of experiments. Legend: ○ values of start simplex range between −35 °C and − 25 °C. The results of a second in this work, which was capable of successfully solving the Simplex run, used to verify the obtained results, are provided presented problems within a single working day and without in supporting information C.3. any human intervention. Due to its modular and flexible na- The ideal conditions identified through Simplex optimisa- ture, the set-up moreover accelerates process development tion are comparable to those identified by DoE optimisation. while also diminishing the obstacles that would otherwise Again, the optimal stoichiometric ratio of methyllithium: hamper the transfer from lab to pilot scale. dibromomethane 7 and the optimal stoichiometric ratio of The developed approach incorporates a successive combina- lithiated intermediate: acetophenone 8 were both shown to tion of an inline FT-IR spectrometer and an online mass spec- be close to 0.9 with a slight shortage of methyllithium respec- trometer. There, inline FT-IR spectroscopy is utilized to moni- tively slight excess of acetophenone. Additionally, a mini- tor main product yield, whereas the mass spectrometer’shigh mum amount of by-products is generated at a reaction tem- sensitivity provides insights into the formation of by-products. perature of −35 °C. Examining Fig. 11, it becomes apparent Both techniques entail individual benefits and drawbacks: that the objective function reaches particularly high values Inline FT-IR measurements permit the calculation of product when reaction temperature is kept at a low level. This is in yield based on calibration curves. However, given the lower accordance with the assumption that external quenching re- sensitivity of FT-IR spectroscopy compared to mass spectromet- quires reaction temperatures near −35 °C (but at least lower ric analysis, by-products cannot directly be monitored when they than −33 °C), as otherwise thermal decomposition of the lith- remain on or below a parts per million (ppm) level. Conversely, ium carbenoid species occurs, preventing its reaction with online mass spectrometry illuminates by-product formation, but acetophenone. may only provide relative (instead of absolute) numbers due to ionization suppression or enhancement effects [67–71], as the components may affect each other’s ionization efficiencies. Discussion Overall, by applying both techniques in a manner where they nullify each other’s drawbacks, the presented platform When optimisation aims at maximising product yield while leverages methodological synergies. It enables the multidi- simultaneously minimising by-product formation, the exami- mensional optimisation of all involved process parameters. nation of multi-stage organic syntheses that involve complex Additionally, this work provides further evidence that, while reaction mechanisms leads to intricate multidimensional opti- FT-IR spectroscopy by itself already constitutes a powerful misation problems. Examples include the organic syntheses tool for self-optimisation, performance can be improved fur- investigated in this work: Both the organometallic synthesis ther by a complementary application of online mass spectrom- with n-butyllithium (deprotonation followed by nucleophilic etry. Together, they provide comprehensive insights into com- addition) and epoxide synthesis (reaction and subsequent cy- plex reaction mechanisms. clization) constitute consecutive reactions. A considerable Yet, when conducting multi-stage syntheses, it has to be amount of by-products is generated if either of these syntheses considered that process parameters can be intertwined and, in is not conducted under ideal process conditions, diminishing such cases, may not be varied independently from each other. process efficiency, i.e. leading to decreased product yield and The examined optimisation algorithms in this work can only necessitating work-up procedures. deal with such circumstances as far as certain restrictions are Such intricate optimisation problems may be solved applied, otherwise the number of experiments required to solve the respective optimisation problem would skyrocket. through the versatile self-optimisation approach developed 300 J Flow Chem (2021) 11:285–302 For instance, in this work, constant process parameters were Two different reaction types, organolithium and epoxide applied for the deprotonation step of the organometallic syn- synthesis, were examined in this work, outlining the plat- thesis with n-butyllithium (first step of the synthesis, followed form’s broad range of applicability. All studied optimisa- by nucleophilic addition) to reduce the number of variable tion problems were successfully solved within a single parameters. In terms of epoxide synthesis, the residence times working day. Thus, researchers and industry alike may of both synthesis steps (reaction and cyclization) were held consider its implementation as efficient, reliable and ver- constant and the three examined dosing schemes were inves- satile self-optimisation tool. tigated independently from each other. Supplementary Information The online version contains supplementary To conclude, despite their limitations, both optimisation material available at https://doi.org/10.1007/s41981-021-00140-x. strategies applied in this work, the modified Simplex algo- rithm as well as Design of Experiments, successfully identi- Acknowledgements The authors would like to thank Marius Graute and fied optimal reaction conditions, while at the same time Fabian Wolff (Mannheim University of Applied Sciences) for great tech- granting an in-depth process understanding. A detailed discus- nical support, and the analytics department at Merck KGaA as well as Jörg Sedelmeier for valuable discussions. sion of by-product formation is provided in supporting infor- mation B.3 and C.4. Funding Open Access funding enabled and organized by Projekt DEAL. This work was funded by the German Federal Ministry of Education and Research (BMBF), programme FH Impuls - Partnership for Innovation 2 2 M Aind, project SM all (grant No. 13FH8I01IA). Conclusion Declaration Through its use of multidimensional optimisation of reaction Conflicts of interest On behalf of all authors, the corresponding author parameters, the self-optimising system derived in this work au- states that there is no conflict of interest. tonomously guides chemical processes towards ideal reaction conditions, whilst reducing the need for human intervention. Abbreviations 1, [−] CH-acidic compound; 2, [−] n-butyllithium; 3, [−] Several benefits are provided: The successive combination of lithiated intermediate; 4, [−] electrophilic compound; 5, [−] intermediate; FT-IR spectrometer and mass spectrometer permits investigating 6, [−] product of organometallic synthesis with n-butyllithium; 7, [−] dibromomethane; 8, [−] acetophenone; 9, [−] methyllithium as its lithium intricate optimisation problems, aiming at maximising product bromide complex; 10, [−]epoxide yield while simultaneously minimising by-product formation. Experimental data, which is collected in real-time, can be used Open Access This article is licensed under a Creative Commons as immediate feedback to decide on the next experimental con- Attribution 4.0 International License, which permits use, sharing, adap- ditions, as the use of inline FT-IR spectroscopy and successive tation, distribution and reproduction in any medium or format, as long as online mass spectrometry without former chromatographic sep- you give appropriate credit to the original author(s) and the source, pro- aration means that no waiting times between measurement and vide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included evaluation thereof occur. Thus, both measurement techniques’ in the article's Creative Commons licence, unless indicated otherwise in a unique advantages are leveraged. credit line to the material. If material is not included in the article's Furthermore, going beyond real-time reaction monitor- Creative Commons licence and your intended use is not permitted by ing and the observation of intermediates and by-products, statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this the platform derived in this work fully exploits the poten- licence, visit http://creativecommons.org/licenses/by/4.0/. tial of online analytics by autonomously solving optimi- sation problems through a model-free experimental inves- tigation. Applying a modified Simplex algorithm and References Design of Experiments, optimal reaction conditions of or- ganic syntheses that involve complex reaction mecha- 1. Wehrens R, Buydens LM (2006) In: Meyers RA (ed) Encyclopedia nisms and a large number of variable parameters can be of Analytical Chemistry. John Wiley & Sons, Chichester identified, thus obtaining in-depth process understanding. 2. Schlick T (1992) In: Lipkowitz KB, Boyd DB (eds) Reviews in Computational Chemistry. VCH Publishers, New York The screening of the entire experimental space through 3. Reizman BJ, Jensen KF (2012). Org Process Res Dev 16(11): DoE and subsequent Simplex optimisation constitutes a 1770–1782. https://doi.org/10.1021/op3001838 reliable approach that is particularly useful when a priori 4. McMullen JP, Jensen KF (2010). Annu Rev Anal Chem (Palo Alto, information on the reaction mechanism is not available. Calif) 3:19–42. https://doi.org/10.1146/annurev.anchem.111808. 5. Rasheed M, Wirth T (2011). Angew Chem 123(2):374–376. The mass spectrometer used in this work supported only standard resolution https://doi.org/10.1002/ange.201006107 and could thus not be applied for determination of molecular structures. Therefore, by-product formation was implemented as sum parameter into the 6. Fabry DC, Sugiono E, Rueping M (2014). Isr J Chem 54(4):341– 350. https://doi.org/10.1002/ijch.201300080 objective function. J Flow Chem (2021) 11:285–302 301 7. Ley SV, Fitzpatrick DE, Myers RM, Battilocchio C, Ingham RJ 32. Hessel V, Kralisch D, Kockmann N (2015) Novel process win- dows: innovative gates to intensified and sustainable chemical pro- (2015). Angew Chem Int Ed Eng 54(35):10122–10136. https:// doi.org/10.1002/anie.201501618 cesses. Wiley, Weinheim 33. Jensen KF (2001). Chem Eng Sci 56(2):293–303. https://doi.org/ 8. Houben C, Lapkin AA (2015). Curr Opin Chem Eng 9:1–7. https:// 10.1016/S0009-2509(00)00230-X doi.org/10.1016/j.coche.2015.07.001 34. Kockmann N, Gottsponer M, Zimmermann B, Roberge DM 9. Fabry DC, Sugiono E, Rueping M (2016). React Chem Eng 1(2): (2008). Chem Eur J 14(25):7470–7477. https://doi.org/10.1002/ 129–133. https://doi.org/10.1039/c5re00038f chem.200800707 10. Sans V, Cronin L (2016). Chem Soc Rev 45(8):2032–2043. https:// 35. Kockmann N (2008) Transport Phenomena in Micro Process doi.org/10.1039/c5cs00793c Engineering, heat and mass transfer. Springer-Verlag, Berlin 11. Bédard A-C, Adamo A, Aroh KC, Russell MG, Bedermann AA, 36. Qian Z, Baxendale IR, Ley SV (2010). Chem Eur J 16(41):12342– Torosian J, Yue B, Jensen KF, Jamison TF (2018). Science 12348. https://doi.org/10.1002/chem.201002147 361(6408):1220–1225. https://doi.org/10.1126/science.aat0650 37. Carter CF, Lange H, Ley SV, Baxendale IR, Wittkamp B, Goode 12. Gooding OW (2004). Curr Opin Chem Biol 8(3):297–304. https:// JG, Gaunt NL (2010). Org Process Res Dev 14(2):393–404. https:// doi.org/10.1016/j.cbpa.2004.04.009 doi.org/10.1021/op900305v 13. Murray PM, Bellany F, Benhamou L, Bučar D-K, Tabor AB, 38. Moore JS, Jensen KF (2012). Org Process Res Dev 16(8):1409– Sheppard TD (2016). Org Biomol Chem 14(8):2373–2384. 1415. https://doi.org/10.1021/op300099x https://doi.org/10.1039/c5ob01892g 39. Chung R, Hein JE (2017). Top Catal 60(8):594–608. https://doi. 14. Leardi R (2009). Anal Chim Acta 652(1–2):161–172. https://doi. org/10.1007/s11244-017-0737-9 org/10.1016/j.aca.2009.06.015 40. Browne DL, Wright S, Deadman BJ, Dunnage S, Baxendale IR, 15. Carlson R, Carlson JE (2005) Design and optimization in organic Turner RM, Ley SV (2012). Rapid Commun Mass Spectrom synthesis, data handling in science and technology. Elsevier, 26(17):1999–2010. https://doi.org/10.1002/rcm.6312 Amsterdam 41. Adamczyk M, Fishpaugh J, Gebler J, Mattingly P, Shreder K 16. Fernanda Giné M, Tuon RL, Cesta AA, Paula Packer A, Reis BF (1998). Eur J Mass Spectrom 4(1):121. https://doi.org/10.1255/ (1998). Anal Chim Acta 366(1–3):313–318. https://doi.org/10. ejms.198 1016/S0003-2670(98)00158-5 42. Fabris D (2005). Mass Spectrom Rev 24(1):30–54. https://doi.org/ 17. McMullen JP, Jensen KF (2008). Proc Int Conf Miniat Syst Chem 10.1002/mas.20007 Life Sci 12:1907–1909 43. Bristow TWT, Ray AD, O'Kearney-McMullan A, Lim L, 18. McMullen JP, Jensen KF (2010). Org Process Res Dev 14(5): McCullough B, Zammataro A (2014). J Am Soc Mass Spectrom 1169–1176. https://doi.org/10.1021/op100123e 25(10):1794–1802. https://doi.org/10.1007/s13361-014-0957-1 19. McMullen JP, Stone MT, Buchwald SL, Jensen KF (2010). Angew 44. Roscioli KM, Zhang X, Li SX, Goetz GH, Cheng G, Zhang Z, Siems WF, Hill HH (2013). Int J Mass Spectrom 336:27–36. Chem Int Ed Eng 49(39):7076–7080. https://doi.org/10.1002/anie. 201002590 https://doi.org/10.1016/j.ijms.2012.12.004 45. Mathieson JS, Rosnes MH, Sans V, Kitson PJ, Cronin L (2013). 20. Skilton RA, Parrott AJ, George MW, Poliakoff M, Bourne RA Beilstein J Nanotechnol 4:285–291. https://doi.org/10.3762/ (2013). Appl Spectrosc 67(10):1127–1131. https://doi.org/10. bjnano.4.31 1366/13-06999 46. Ray A, Bristow T, Whitmore C, Mosely J (2018). Mass Spectrom 21. Fitzpatrick DE, Battilocchio C, Ley SV (2016). Org Process Res Re v 37(4):565–579. https://doi.org/10.1002/mas.21539 Dev 20(2):386–394. https://doi.org/10.1021/acs.oprd.5b00313 47. Fleischer H, Do VQ, Thurow K (2019). SLAS technology 24(3): 22. Cortés-Borda D, Wimmer E, Gouilleux B, Barré E, Oger N, 330–341. https://doi.org/10.1177/2472630318813838 Goulamaly L, Peault L, Charrier B, Truchet C, Giraudeau P, 48. Pulliam CJ, Bain RM, Osswald HL, Snyder DT, Fedick PW, Rodriguez-Zubiri M, Le Grognec E, Felpin F-X (2018). J Ayrton ST, Flick TG, Cooks RG (2017). Anal Chem 89(13): Organomet Chem 83(23):14286–14299. https://doi.org/10.1021/ 6969–6975. https://doi.org/10.1021/acs.analchem.7b00119 acs.joc.8b01821 49. Haven JJ, Vandenbergh J, Junkers T (2015). Chem Commun 23. Reizman BJ, Jensen KF (2015). Chem Commun 51(68):13290– 51(22):4611–4614. https://doi.org/10.1039/c4cc10426a 13293. https://doi.org/10.1039/c5cc03651h 50. Ingham RJ, Battilocchio C, Hawkins JM, Ley SV (2014). Beilstein 24. Holmes N, Akien GR, Savage RJD, Stanetty C, Baxendale IR, JOrg Chem 10:641–652. https://doi.org/10.3762/bjoc.10.56 Blacker AJ, Taylor BA, Woodward RL, Meadows RE, Bourne 51. Rueping M, Bootwicha T, Sugiono E (2012). Beilstein J Org Chem RA (2016). React Chem Eng 1(1):96–100. https://doi.org/10. 8:300–307. https://doi.org/10.3762/bjoc.8.32 1039/c5re00083a 52. Garcia-Egido E, Spikmans V, Wong SYF, Warrington BH (2003). 25. Echtermeyer A, Amar Y, Zakrzewski J, Lapkin A (2017). Beilstein Lab Chip 3(2):73–76. https://doi.org/10.1039/B302381H JOrg Chem 13:150–163. https://doi.org/10.3762/bjoc.13.18 53. Colombo E, Ratel P, Mounier L, Guillier F (2011). J Flow Chem 26. Nieuwland PJ, Koch K, van Harskamp N, Wehrens R, van Hest 1(2):68–73. https://doi.org/10.1556/jfchem.2011.00009 JCM, Rutjes FPJT (2010). Chem Asian J 5(4):799–805. https://doi. 54. Delville MM, van Gool JJ, van Wijk IM, van Hest JC, Rutjes FP org/10.1002/asia.200900705 (2012). J Flow Chem 2(4):124–128. https://doi.org/10.1556/JFC- 27. Reizman BJ, Wang Y-M, Buchwald SL, Jensen KF (2016). React D-12-00008 Chem Eng 1(6):658–666. https://doi.org/10.1039/c6re00153j 55. Fath V, Kockmann N, Röder T (2019). Chem Eng Technol 42(10): 28. Reizman BJ, Jensen KF (2016). Acc Chem Res 49(9):1786–1796. 2095–2104. https://doi.org/10.1002/ceat.201900074 https://doi.org/10.1021/acs.accounts.6b00261 56. Fath V, Szmais S, Lau P, Kockmann N, Röder T (2019). Org 29. Wegner J, Ceylan S, Kirschning A (2011). Chem Commun 47(16): Process Res Dev 23(9):2020–2030. https://doi.org/10.1021/acs. 4583–4592. https://doi.org/10.1039/C0CC05060A oprd.9b00265 30. Yoshida J-I (2008) Flash chemistry: fast organic synthesis in 57. Fath V, Lau P, Greve C, Kockmann N, Röder T (2020). Org microsystems. Wiley, Weinheim Process Res Dev 24(10):1955–1968. https://doi.org/10.1021/acs. 31. Hessel V, Kralisch D, Kockmann N, Noël T, Wang Q (2013). oprd.0c00037 58. von Keutz T, Cantillo D, Kappe CO (2019). Org Lett 21(24): ChemSusChem 6(5):746–789. https://doi.org/10.1002/cssc. 10094–10098. https://doi.org/10.1021/acs.orglett.9b04072 201200766 302 J Flow Chem (2021) 11:285–302 59. Ulrich W, Steinbild M (2000) Ullmann's Encyclopedia of Industrial 76. Hartwig J, Metternich JB, Nikbin N, Kirschning A, Ley SV (2014). Org Biomol Chem 12(22):3611–3615. https://doi.org/10.1039/ Chemistry: Lithium and Lithium Compounds. Wiley-VCH, Weinheim C4OB00662C 60. (2004) In: I Marek, Z Rappoport (eds.), The chemistry of 77. Degennaro L, Fanelli F, Giovine A, Luisi R (2015). Adv Synth organolithium compounds, Patai series, Wiley, Chichester Catal 357(1):21–27. https://doi.org/10.1002/adsc.201400747 61. Wu G, Huang M (2006). Chem Rev 106(7):2596–2616. https://doi. 78. Hafner A, Filipponi P, Piccioni L, Meisenbach M, Schenkel B, org/10.1021/cr040694k Venturoni F, Sedelmeier J (2016). Org Process Res Dev 20(10): 62. Kamienski CW, McDonald DP, Stark MW, Papcun JR (2000) 1833–1837. https://doi.org/10.1021/acs.oprd.6b00281 Kirk-Othmer Encyclopedia of Chemical Technology, vol 6. John 79. Hafner A, Mancino V, Meisenbach M, Schenkel B, Sedelmeier J Wiley & Sons, Inc, Hoboken (2017). Org Lett 19(4):786–789. https://doi.org/10.1021/acs. 63. Jacobsen EN (2000). Acc Chem Res 33(6):421–431. https://doi. orglett.6b03753 org/10.1021/ar960061v 80. Pace V, Castoldi L, Holzer W (2014). Adv Synth Catal 356(8): 64. Padwa A, Murphree SS (2006). Arkivoc 37:6–33 1761–1766. https://doi.org/10.1002/adsc.201301042 65. Schneider C (2006). Synthesis 2006(23):3919–3944. https://doi. 81. Pace V, Holzer W, de Kimpe N (2016). Chem Rec 16(4):2061– org/10.1055/s-2006-950348 2076. https://doi.org/10.1002/tcr.201600011 66. Kessler RW, Kessler W, Zikulnig-Rusch E (2016). Chem Ing Tech 82. Köbrich G, Akhtar A, Ansari F, Breckoff WE, Büttner H, Drischel 88(6):710–721. https://doi.org/10.1002/cite.201500147 W, Fischer RH, Flory K, Fröhlich H, Goyert W, Heinemann H, 67. Kruve A, Rebane R, Kipper K, Oldekop M-L, Evard H, Herodes K, Hornke I, Merkle HR, Trapp H, Zündorf W (1967). Angew Chem Ravio P, Leito I (2015). Anal Chim Acta 870:29–44. https://doi. Int Ed Eng 6(1):41–52. https://doi.org/10.1002/anie.196700411 org/10.1016/j.aca.2015.02.017 83. Cainelli G, Tangari N, Ronchi A (1972). Tetrahedron 28(11):3009– 68. Kruve A, Rebane R, Kipper K, Oldekop M-L, Evard H, Herodes K, 3013. https://doi.org/10.1016/0040-4020(72)80015-2 Ravio P, Leito I (2015). Anal Chim Acta 870:8–28. https://doi.org/ 84. Pace V, Castoldi L, Holzer W (2013). J Organomet Chem 78(15): 10.1016/j.aca.2015.02.016 7764–7770. https://doi.org/10.1021/jo401236t 69. Taylor PJ (2005). Clin Biochem 38(4):328–334. https://doi.org/10. 85. Pace V, Holzer W, Verniest G, Alcántara AR, De Kimpe N (2013). 1016/j.clinbiochem.2004.11.007 Adv Synth Catal 355(5):919–926. https://doi.org/10.1002/adsc. 70. Bonfiglio R, King RC, Olah TV, Merkle K (1999). Rapid Commun Mass Spectrom 13(12):1175–1185. https://doi.org/10.1002/(SICI) 1097-0231(19990630)13:12<1175:AID-RCM639>3.0.CO;2-0 86. Schwolow S, Ko JY, Kockmann N, Röder T (2016). Chem Eng Sci 71. King R, Bonfiglio R, Fernandez-Metzler C, Miller-Stein C, Olah T, 141:356–362. https://doi.org/10.1016/j.ces.2015.11.022 Am J (2000). Soc Mass Spectrom 11(11):942–950. https://doi.org/ 87. Köbrich G, Fischer RH (1968). Tetrahedron 24(11):4343–4346. 10.1016/S1044-0305(00)00163-X https://doi.org/10.1016/0040-4020(68)88194-3 72. Swartz ME, Liq J (2005). Chromatogr Relat Technol 28(7–8): 88. Köbrich G (1972). Angew Chem Int Ed Eng 11(6):473–485. 1253–1263. https://doi.org/10.1081/JLC-200053046 https://doi.org/10.1002/anie.197204731 73. Fath V, Kockmann N, Otto J, Röder T (2020). React Chem Eng 89. Kirmse W (1965). Angew Chem Int Ed Eng 4(1):1–10. https://doi. 5(7):1281–1299. https://doi.org/10.1039/D0RE00081G org/10.1002/anie.196500011 74. Winkler S (2007) Untersuchungen zur Herstellung und Umsetzung fluorsubstituierter Phenyllithiumverbindungen, Doctoral Thesis, Publisher’snote Springer Nature remains neutral with regard to jurisdic- TU Darmstadt tional claims in published maps and institutional affiliations. 75. von Keutz T, Cantillo D, Kappe CO (2020). Org Lett 22(19):7537– 7541. https://doi.org/10.1021/acs.orglett.0c02725 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Flow Chemistry Springer Journals

Simultaneous self-optimisation of yield and purity through successive combination of inline FT-IR spectroscopy and online mass spectrometry in flow reactions

Download PDF
18 pages

Loading next page...
 
/lp/springer-journals/simultaneous-self-optimisation-of-yield-and-purity-through-successive-JCkAFtRplC

References (89)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
ISSN
2062-249X
eISSN
2063-0212
DOI
10.1007/s41981-021-00140-x
Publisher site
See Article on Publisher Site

Abstract

Self-optimisation constitutes a very helpful tool for chemical process development, both in lab and in industrial applications. However, research on the application of model-free autonomous optimisation strategies (based on experimental investigation) for complex reactions of high industrial significance, which involve considerable intermediate and by-product formation, is still in an early stage. This article describes the development of an enhanced autonomous microfluidic reactor platform for organolithium and epoxide reactions that incorporates a successive combination of inline FT-IR spectrometer and online mass spectrometer. Experimental data is collected in real-time and used as feedback for the optimisation algorithms (modified Simplex algorithm and Design of Experiments) without time delay. An efficient approach to handle intricate optimisation problems is presented, where the inline FT-IR measurements are used to monitor the reaction’s main components, whereas the mass spectrometer’s high sensitivity permits insights into the formation of by-products. To demonstrate the platform’s flexibility, optimal reaction conditions of two organic syntheses are identified. Both pose several challenges, as complex reaction mechanisms are involved, leading to a large number of variable parameters, and a considerable amount of by-products is generated under non-ideal process conditions. Through multidimensional real-time optimisation, the platform supersedes labor- and cost-intensive work-up procedures, while diminishing waste generation, too. Thus, it renders production processes more efficient and contributes to their overall sustainability. . . . . Keywords Microreactiontechnology InlineFT-IRspectroscopy Onlinemassspectrometry Self-optimisation Organolithium compounds Introduction requirements. Through applying a precise process control that already intervenes at an early stage, manufacturing When industrial production processes are not conducted processes can be rendered more efficient and more sus- under ideal conditions, labor- and cost-intensive work-up tainable, while simultaneously diminishing waste procedures become necessary to fulfill product quality generation. Highlights � Self-optimisation for complex reactions while minimising by-product formation � Successive combination of inline FT-IR and online mass spectrometer, leveraging each method’s advantages � Improved production process efficiency and sustainability through combined DoE and modified Simplex algorithm * Thorsten Röder Institute of Chemical Process Engineering, Mannheim University of t.roeder@hs-mannheim.de Applied Sciences, Paul-Wittsack-Str. 10, 68163 Mannheim, Germany Department of Biochemical and Chemical Engineering, Equipment Merck KGaA, Frankfurter Str. 250, 64293 Darmstadt, Germany Design, TU Dortmund University, Emil-Figge-Str. 68, Institute of Instrumental Analytics and Bioanalysis, Mannheim 44227 Dortmund, Germany University of Applied Sciences, Paul-Wittsack-Str. 10, 68163 Mannheim, Germany 286 J Flow Chem (2021) 11:285–302 Several optimisation strategies exist for this purpose [1, Experimental section 2]. Their high significance for chemical process develop- ment, both in lab and in industrial applications, has been Reactions demonstrated by the extant literature [3–11]. In industrial contexts, however, process optimisation often proceeds Organometallic synthesis through one-by-one optimisation [12, 13] instead of a more efficient multidimensional approach [14, 15]. In ac- In a first step, the exothermic deprotonation reaction of a CH- ademic research, such systematic, multidimensional opti- acidic compound 1 in tetrahydrofuran THF (anhydrous max. misation strategies have been studied in detail [16–28]. 0.005% H O, Merck, Germany) with n-butyllithium 2 leads to They have been proven to constitute a valuable tool for a non-isolable, unstable, lithiated intermediate compound 3 process optimisation, especially when being integrated in- [55–57]. This deprotonation step is followed by a nucleophilic to fully-automated microreactor platforms [29–35]. In addition including the lithiated intermediate’sreaction with an combination with online analysis, real-time reaction mon- electrophilic compound 4. The resulting intermediate 5 is itoring becomes possible, where intermediates and by- quenched with methanol (for synthesis, >99%, Carl Roth, products can be observed as well [9]. In particular, inline Germany) leading to a stable product 6 (Scheme 1). FT-IR spectroscopy and online mass spectrometry consti- The starting material n-butyllithium 2 was chosen from −1 tute promising analysis techniques that enable rapid quan- Sigma Aldrich, Germany, with a concentration of 1.6 mol L tification of reactants (analysis duration <1 min) [24, in n-hexane. Initial concentrations of the CH-acidic compound −1 36–49]. 1 and the electrophilic compound 4 amounted to 0.8 mol L . While the extant literature has applied inline FT-IR spectroscopy [20, 21, 50, 51]aswellas onlinemassspec- Synthesis of terminal epoxide trometry [24] in self-optimisation settings only for rather simple reactions and merely in isolation, this work ex- The epoxide 10 is synthesized from acetophenone 8 tends the prior ones by presenting an enhanced self- (ReagentPlus®, 99%, Sigma Aldrich, Germany) via in-situ optimising platform that integrates a successive combina- generated (bromomethyl)lithium (Scheme 2). Here, lithium- tion of inline FT-IR spectrometer and online mass spec- halogen exchange of dibromomethane 7 (99%, Sigma trometer. This self-optimising platform enables model- Aldrich, Germany) with methyllithium as its lithium bromide free autonomous optimisation without the need for human complex leads to (bromomethyl)lithium, which immediately intervention and is utilized to experimentally identify op- reacts with the carbonyl group of acetophenone, generating its timal reaction conditions of organic syntheses that involve bromomethyl alkoxide. In a following step, the alkoxide cy- complex reaction mechanisms. For this purpose, inline clizes to epoxide 10 [58]. FT-IR measurements are used to monitor the reaction’s The reaction was carried out in tetrahydrofuran THF (an- main components, whereas the high sensitivity of a mass hydrous max. 0.005% H O, Merck, Germany). Initial concen- spectrometer provides insights into the formation of by- trations of dibromomethane 7 and acetophenone 8 amounted products. Unlike prior works (e.g., [23, 28, 52–54]), no −1 to 0.8 mol L . Methyllithium lithium bromide complex solu- chromatographic separation is conducted before MS anal- tion was chosen from Sigma Aldrich, Germany, with a con- ysis, thus accelerating the analysis process drastically. −1 a centration of 1.5 mol L in diethyl ether . A novel approach is developed for solving intricate multidimensional optimisation problems, aiming at Experimental setup maximising product yield and purity. A modified Simplex algorithm as well as Design of Experiments are Microreactor experiments applied to identify ideal reaction conditions, while at the same time delivering an in-depth process understanding. The high flexibility of the chosen set-up is demonstrat- In case of the organometallic synthesis with n-butyllithium, a plate microreactor was directly connected to a capillary ed by means of optimisation of two different reaction types that are of great industrial significance, namely an microreactor, allowing to maintain two independent tem- perature levels (Fig. 1a). The deprotonation step (reaction organometallic reaction with n-butyllithium [55–57]and an epoxide synthesis [58]. Both studied reactions serve as of the CH-acidic compound 1 with n-butyllithium 2)was starting points for a multitude of further synthesis steps. Within certain limits, the concentration of organometallic reagents might be Thus, a broad spectrum of chemical reactions can be cov- subject to minor deviations. However, pretests described in supporting infor- ered, acting as basic building blocks for organic-chemical mation A.1 indicate that this limitation had only a negligible effect on the compounds of industrial relevance [59–65]. study’sfindings. J Flow Chem (2021) 11:285–302 287 Scheme 1 Organometallic synthesis. 1 CH-acidic compound deprotonation ∙ 2 n-butyllithium ∙ 3 lithiated in- Li termediate ∙ 4 electrophilic com- pound ∙ 5 intermediate ∙ 6 product 1 2 3 hydrolysis nucleophilic + MeOH addition - MeOLi 45 6 carried out in the plate microreactor, which had been de- Nuevo, Huber, Germany) allowed for adjusting the tempera- signed and manufactured by mechanical precision milling ture of the heating/cooling fluid. of stainless steel (Fig. 1b). It consisted of three stainless The nucleophilic addition was carried out in a coiled steel plates layered one on top of each other, where channels 1/16 in. stainless steel capillary microreactor. Precooling of for mixing and residence time were milled into the middle the electrophilic compound 4 was conducted in a capillary reactor plate (each channel had a quadratic cross-section). with an inner diameter of 0.5 mm. The precooling capillary To avoid bypass flow, the plates were evenly pressed onto was directly connected to the outlet of the plate reactor via a each other (ensuring equal pressure between the plates) stainless-steel T-mixer (inner diameter 0.5 mm), followed by a using numerous screws. Bore holes that were located on reaction channel that had an inner diameter of 0.5 mm and a the lateral surface of the middle reactor plate were utilized total volume of 0.59 mL. The chosen microreactor setup per- as inlet and outlet connections for the reactants. Channels mitted residence times between 0.2 and 1 min. Temperature control of precooling, mixing, and reaction was achieved for precooling (0.5 mm × 0.5 mm), mixing (0.5 mm × 0.5mm), and reactionwerearrangedin anarrow-shaped using a bath thermostat (Ministat, Huber, Germany). geometry. The reaction channel was divided into two sec- Regarding epoxide synthesis, the microreactor setup tions: at the entrance, a 0.5 mm × 0.5 mm channel enabled consisted of coiled 1/16 in. PFA tubing (Fig. 2). Reactants enhanced heat transfer; with increasing conversion, an in- were precooled and mixed in T-mixers (0.5 mm inner bore creased channel size (1 mm × 1 mm) was employed, en- hole). The reaction mixture then passed two modular reactor abling high reactant conversion at longer residence times. pieces that were connected to each other. In the first capillary, The entrance region of the reaction channel had a length of which had a total volume of 0.5 mL (inner diameter 0.5 mm), 1.1 m, the region with increased channel size had a length of residence time remained constant at 0.5 min. Temperature was 9.11 m, resulting in a total reactor volume of 7.37 mL. varied between −35 °C and − 10 °C (bath thermostat Huber As the kinetics of the deprotonation step had already been Tango Nuevo). In the second capillary, which had a total vol- studied in detail [55–57], deprotonation of the CH-acidic com- ume of 2 mL (inner diameter 0.75 mm), the reaction mixture pound 1 was performed at a constant residence time of 8 min waswarmedto20°C(1minresidencetime, baththermostat and a temperature of −35 °C, ensuring full conversion of the n- Huber Ministat). Thus, cyclization of the bromomethyl alkox- butyllithium 2. In order to avoid clogging, the CH-acidic com- ide intermediate took place. A back-pressure regulator (3 bar) pound 1 was provided in marginal excess; the stoichiometric ensured light overpressure within the whole microreactor. ratio of n-butyllithium: CH-acidic compound amounted to Mixing of reactants occurred through one out of three dif- 0.8. Temperature control was achieved through heat carrier ferent cases. In the first case, dibromomethane 7 and channels (3 mm × 10 mm) that were milled on the bottom of acetophenone 8 had been premixed manually (feed stream 1) the middle reactor plate, where two Pt100 resistance thermom- eters could directly be inserted into the inflowing and When investigating epoxide synthesis, the microreactor setup used PFA outflowing thermal fluid via bore holes located at the lateral tubing (instead of stainless-steel tubing) to avoid pitting corrosion caused by surface of the middle plate. A thermostat (Unistat Tango bromide ions. 288 J Flow Chem (2021) 11:285–302 Scheme 2 Preparation of epoxide 10 from acetophenone 8 via in situ generated (bromomethyl)lithium 7a and then combined with methyllithium (feed stream 2; Fig. (SyrDos2, HiTec Zang GmbH, Germany). Temperature and 2a). In the second case, dibromomethane 7 and acetophenone flow rates were controlled by a laboratory automation system 8 (Fig. 2b) entered through two separate feed streams and, (LabManager, HiTec Zang GmbH, Germany). once they had been mixed, were subsequently combined with methyllithium (feed stream 3), thus allowing to variably adjust Inline FT-IR and online MS measurements the stoichiometric ratio of 7 and 8. In the third case, dibromomethane 7 and methyllithium (Fig. 2c) entered Both described syntheses (organometallic reaction with n- through two separate feed streams and, once they had been butyllithium; epoxide synthesis) were continuously monitored mixed, were subsequently combined with acetophenone 8. at the reactor outlet. The analysis of the respective product Dosage of all starting materials within 1 mL glass syringes stream was conducted through a successive combination of was ensured by continuously working syringe pumps an inline FT-IR spectrometer and an online mass spectrometer FT-IR MS/GC a) PI R deprotonation nucleophilic addition CH-acidic compound P1 M2 M3 PI R M1 R1 R2 product n-BuLi τ = 8 min 6 τ = 0.2 to 1 min P2 T = −35 °C PI R TIR 3 1 electrophilic compound P3 T = −35 °C to −25 °C TIR PI R MeOH quench P4 b) c) Fig. 1 Microreactor setup for organometallic synthesis with n-butyllithium, process flow chart (a), plate microreactor (b), coiled capillary microreactor (c) J Flow Chem (2021) 11:285–302 289 Fig. 2 Microreactor setup for epoxide synthesis, process flow charts: a setup reduced to two feeds with a premixed solution of acetophenone and CH Br , b setup with three feeds and premixing of acetophenone and CH Br , c setup with premixing of CH Br and MeLi � LiBr 2 2 2 2 2 2 (Fig. 3). Inline FT-IR measurements were used to monitor the by-product formation (as MS possesses significantly higher reaction’s main components (starting materials and product), sensitivity). Unlike prior works (e.g., [23, 28, 52–54]), chro- whereas online MS measurements provided information about matographic separation had not been conducted before MS 290 J Flow Chem (2021) 11:285–302 Fig. 3 Experimental setup with successive combination of an inline FT- continuous syringe pumps; 2 – plate reactor for deprotonation; 3 – IR spectrometer and an online mass spectrometer (microreactor setup cooling bath for nucleophilic addition; 4 – inline FT-IR spectrometer; exemplarily provided for organometallic synthesis). Legend: 1 – 5–6/2 port valve; 6 – online MS. −1 analysis, as real-time monitoring without considerable time 696 cm . Analytical IR spectra and details on the integration delayneededto be achieved(m/zvaluesand their intensities method are provided in supporting information A.2. were accumulated over a measurement time of 60 s, resulting As the reactant concentrations (amounting to roughly −1 in a new mass spectrum every 60 s). 0.25 mol L ) were suitable for inline FT-IR analysis, no dilu- The use of an inline FT-IR spectrometer (Bruker ALPHA, tion step was required beforehand. Following the inline FT-IR United States) allowed for real-time reaction tracking with analysis, however, continuous quenching of the reaction mix- time delay <1 s, circumventing the need to quench the reac- ture took place (compact dual piston pump AZURA P4.1S, tion. The reactor outlet was directly connected to the spec- Knauer, Germany). A subsequent split and dilution step (refer trometer’s flow cell through a very short capillary of 3 cm to supporting information A.3 for more details) allowed to with an inner diameter of 0.5 mm [55]. The measuring cell adjust reactant concentrations to render them suitable for anal- of the FT-IR spectrometer with a volume of 40 μLenabled ysis via online mass spectrometry, as the online mass spectrom- extremely fast measurement times, shorter than 2 ms (flow eter’s high sensitivity requires reactant concentrations of −1 −1 rates ≥0.31 mL min ). The employed FT-IR spectrometer around 0.02 mol L to not overload the spectrometer. −1 has an optical wavelength resolution of 4 cm .Infraredspec- Afterwards, aliquots of the reaction mixture were periodically −1 tra (500–1700 cm ) were collected through single reflection introduced to the mass spectrometer using a 6/2 port sample ATR (diamond crystal). valve (Analytical HPLC Continuous-Flow Injector, VICI, Characteristic IR bands allowed for calculating product United States). With the 6/2 port valve in load position, the yield based on previously determined calibration curves. The diluted reaction mixture filled a 5 μL loop fitted across two reaction product of the organometallic synthesis with n- of the valve ports and was thereafter automatically injected into butyllithium 6 was identified by means of a characteristic IR the mass spectrometer, switching the valve to its inject position −1 −1 band at 1241 cm to 1230 cm , whereas epoxide 10 was (a 1/16 in. stainless-steel tubing, 0.5 mm i.d., connected the −1 identified by means of a characteristic IR band at 711 cm to reactor outlet with the valve and the mass spectrometer). J Flow Chem (2021) 11:285–302 291 With the 6/2 port valve in inject position, another high-pressure optimisation is steered by a fully automated experimental se- pump (Agilent Technologies, United States) flushed the loop quence coded in MATLAB, which assumes control over op- −1 with a continuous methanol stream (300 μLmin , hypergrade timisation strategies and the calculation of the objective func- for LC-MS, Sigma Aldrich, Germany) and thus passed the tion. Simultaneously, MATLAB transfers the set points for sample into the mass spectrometer for analysis. The spectrom- pumps and thermostats to the automation system. The plat- eter used was an Advion Expression CMS operating in positive form is compatible with industrial production conditions (ori- atmospheric pressure chemical ionisation mode (APCI). A de- ented on NAMUR standards), ensuring a high level of process tailed description of the online MS setup is provided in safety. Integrated safety features, such as pressure and temper- supporting information A.3. ature monitoring, comply with industrial standards. More de- Overall, by combining these two techniques (FT-IR spec- tails on the experimental self-optimising workflow are provid- troscopy and online MS), the experimental set-up aimed at ed in supporting information A.4. leveraging methodological synergies. Specifically, while FT- Analytical results of inline FT-IR measurements are IR spectroscopy constitutes a fast, broadly applicable method transferred to MATLAB through an OPC interface (real- that reduces barriers to implementation, it is less suitable for time communication with time delay <1 s). Thus, product monitoring by-products unless they exceed the parts per mil- yield can directly be calculated incorporating the respec- lion (ppm) level [10, 37, 66]. tive calibration curve. Conversely, online MS is particularly well-suited for the Regarding online MS measurements, however, the transfer characterization and quantification of analytes up to trace of analytical results proceeds through extraction of CDF-files levels, even in complex mixtures [40, 42, 43, 47]. However, that are gained as output from the mass spectrometer, contain- it may only provide relative (instead of absolute) numbers due ing an accumulated scan time, m/z-values (mass-to-charge- to ionization suppression or enhancement effects [67–71]as ratio), and their respective intensities. As a result, each in- the components may affect each other’s ionization efficien- volved reactant that can be analyzed via APCI, can be identi- cies. For this reason, an additional internal standard (product fied by its characteristic m/z value, with the peak’s intensity intensity) was used to compensate for these biases (further indicating the reactant’squantity (Fig. 4). details regarding the exact procedure are provided in the fol- Especially in case of the organometallic synthesis with lowing chapter). n-butyllithium, pure substances of all reactants expected Compared to traditional HPLC, or even UPLC measure- to be involved in the complex reaction mechanism includ- ments [72], the chosen analytical techniques do not incur a ing by-product formation were available, hence, MS cali- time-delay of several minutes between the initial measurement bration curves could be determined for each individual and the computation of the corresponding analytical result. As component, see supporting information A.5.However, a result, unstable or reactive intermediates and products can be the actual reaction mixture leaving the reactor constitutes identified almost immediately, due to the involved real-time a complex matrix including solvent, main product, by- reaction monitoring leading to a time-delay of 1 min at most products, and not yet consumed starting materials. [10, 37, 40]. Hence, there is a high methodological fit between Ionization suppression or enhancement effects [67–71] FT-IR spectroscopy, online MS, and the objective of efficient may no longer be ruled out, as the components may affect process development, particularly in terms of continuous pro- each other’s ionization efficiencies. cesses. Yet, the absence of chromatographic separation adds In order to circumvent such elusive effects, following the an additional level of complexity to the interpretation of the extraction of CDF-files containing all m/z values, intensities resulting data [40]. of all involved by-products are calculated when product inten- To conclude, the experimental set-up applies FT-IR spec- sity reaches its peak. The sum of all those by-product intensi- troscopy and online MS in a highly complementary manner. ties is then determined, and the ratio of product and sum of by- FT-IR spectroscopy (instead of online MS) delivers absolute products is used as input for the objective function (Eq. 1). As values in order to avoid frequently emerging biases, whereas a result, the determined MS ratio contains the product’smass online MS provides the required sensitivity to examine by- intensity as internal standard and moreover compensates for products at the ppm level. concentration fluctuations that result from quenching the re- action with a constant mass flux. Individual calibration curves Self-optimisation are thus not required to calculate the objective function, but are useful to evaluate whether given components are suitable This work relies on a completely automated self-optimising for MS analysis in general. platform [73], which integrates a microreactor with automated To combine the estimated product yield from FT-IR anal- devices (pumps and thermostats) and a successive combina- ysis with information about undesired by-product formation, tion of real-time reaction monitoring through inline FT-IR the FT-IR signal is multiplied with the calculated MS ratio spectroscopy and online mass spectrometry. Real-time (Eq. 2). Aiming at maximising product yield while 292 J Flow Chem (2021) 11:285–302 Fig. 4 Exemplary online MS analysis of organometallic reaction with n-butyllithium. Experimental parameters: a stoichiometric ratio [3, 4]amounting to 1.0; reaction temperature − 10 °C, b stoichiometric ratio [3, 4] amounting to 1.0; reaction temperature − 30 °C simultaneously minimising by-product formation, the Simplex optimisation proceeds through iteratively replac- resulting objective function can easily be implemented in the ing a randomly chosen initial simplex. Specifically, the start described MATLAB code. A detailed description of the pre- simplex was set to random values located on each axis, with tests that had been conducted to scrutinize the applicability of the last remaining corner point always being randomly chosen the chosen evaluation method regarding MS files is provided within the reaction space. Hence, optimal reaction conditions in supporting information A.6. are identified as soon as the simplexes converge to a local optimum, where the value of the objective function does not intensity at of product vary anymore. The maximum number of experiments (per MS ratio ¼ ð1Þ optimisation cycle) was set to 30. intensities at of by−products When applying DoE, a Central Composite Design (CCD) was chosen. During each optimisation, two experimental runs MS ratio↑low proportion of by−products ð2Þ were executed. The first run screened the entire experimental MS ratio↓high proportion of by−products objective function ¼ FT−IR value  MS ratio space. Based on the first run’s experimentally obtained data points, a surface response model was calculated. The mathe- matical optimum of this surface response model was subse- In this work, two optimisation strategies, modified Simplex quently used as central point for the second DoE run. To refine algorithm and Design of Experiments (DoE), are applied, as the search for a global optimum, the size of the second DoE their performances had already been documented and com- run amounted to 20% of the size of the first DoE run. pared in detail [73]. The derived autonomous platform constitutes a modular and flexible system, Results enabling multi-variate and multi-objective optimisations in real-time. Hence, other objective functions (besides multiplying the FT-IR signal with the cal- Organometallic synthesis culated MS ratio, see Eq. 2) can swiftly be integrated. For instance, an objec- tive function consisting of differently weighted parts (e.g., weighting the costs for separating non-converted starting materials against the costs for work-up In order to evaluate the effectiveness of the presented self- procedures required to remove undesired by-products), or a jump function optimising platform incorporating a successive combination characterizing by-product formation, could also be investigated, depending on the issue being addressed. of real-time reaction monitoring through inline FT-IR J Flow Chem (2021) 11:285–302 293 spectroscopy and online mass spectrometry, two different op- As a result, an optimal stoichiometric ratio close to 1.0 was timisation studies were compared: In the first study, self- identified, whereas the optimal temperature could not be de- optimisation was only based on inline FT-IR measurements, termined exactly, but only be located within a certain range thus aiming at maximisation of main product yield. In the between −23 °C and − 15 °C. After three repetitions, the rela- second study, the estimated product yield from FT-IR analysis tive standard deviation amounted to 4.1% (stoichiometric ra- was combined with information from MS analysis aiming at tio) respectively 22% (temperature). maximising product yield and purity. The organometallic syn- In terms of DoE, the surface response models obtained by thesis with n-butyllithium was used as proof of concept, as the first and second run while optimising stoichiometric ratio kinetics and mechanism had already been studied in detail and temperature, are provided in Fig. 6. Experimental data [55–57, 74]. Pure substances of all involved compounds were points, from which the respective surface response model available meaning that calibration curves, and thus reference was built as best fit, are displayed as red dots. values, had already been known in advance. While the visualisation in Fig. 6 clearly indicates an opti- mal stoichiometric ratio near 1.0, the ideal reaction tempera- ture cannot be determined exactly. Instead, the resulting sur- Product yield maximisation face response models merely point to a broad temperature range. The second DoE run does not improve upon the result Regarding the optimisation procedure based exclusively on of the first one. FT-IR measurements, a modified Simplex algorithm and To assess optimisation accuracy, the average deviation of Design of Experiments were implemented as optimisation the experimental data points from the associated surface re- strategies. The second step of the organometallic synthesis sponse model was determined. It amounted to 15% for the with n-butyllithium, namely the nucleophilic addition, was first DoE run and to 7% for the second DoE run. The relative optimised, aiming at maximum product yield. Two optimisa- standard deviations of calculated optimal product yield after tion parameters were examined: the stoichiometric ratio of six repetitions amounted to 4.2% respectively 3.5%. lithiated intermediate 3: electrophilic compound 4 as well as Table 1 summarizes the obtained results for both optimisa- reaction temperature. The stoichiometric ratio was varied in tion strategies (Simplex and DoE). For each strategy, the op- the range between 0.5 and 2.5, while temperature was varied timal reaction conditions and the required number of experi- between −35 °C and − 10 °C. Note that, as residence time and ments are presented. Note that, in case of DoE optimisation, stoichiometric ratio are intertwined as far as the reaction at the results of the second DoE run are provided. hand is concerned, these two variables may not be varied The results of DoE optimisation are comparable to independently from each other. those of Simplex optimisation. The optimal product yield Figure 5 depicts the experimental results of Simplex optimi- is reached at a stoichiometric ratio of 1.0, which is asso- sation as coloured dots with the colours indicating the corre- ciated with a residence time of 0.3 min. Note that, for sponding values of product yield, and further displays the prog- residence times smaller than 0.3 min, full conversion of ress of the objective function over the course of the iterations. starting materials cannot be accomplished (leading to re- Three repetitions of the same optimisation procedure were duced product yield). However, the optimal reaction conducted, differing only with regard to their start simplexes. Fig. 5 Simplex optimisation of organometallic synthesis. Product yield product yield as objective function, exclusively based on inline FT-IR calculated based on compound 4. Figure illustrates first Simplex measurements. b Progress of objective function across experiments. optimisation (out of three overall). a Results of optimisation with Legend: ○ values of start simplex 294 J Flow Chem (2021) 11:285–302 Fig. 6 Optimisation of organometallic synthesis using DoE. Results are based exclusively on inline FT-IR measurements. Product yield calculated based on compound 4. a First DoE run for screening of whole experimental space. b Second DoE run to refine optimisation temperature is difficult to quantify, when only product (supporting information A.5). The ratio of MS product inten- yield is examined, as temperature optimisation is expected sity and the sum of MS intensities of all four by-products was to predominantly affect by-product formation, with the used as input for the objective function together with the main amount of created by-products in the experiment remain- product’s FT-IR signal, Eq. 2. Again, a modified Simplex ing below FT-IR’s sensitivity threshold. algorithm and Design of Experiments were applied as optimi- sation strategies. Nucleophilic addition was optimised aiming Product yield maximisation with simultaneous minimisation at maximising product yield while simultaneously minimising of by-product formation by-product formation. Figure 7 displays the experimental results of Simplex opti- To expand upon the results of self-optimisation based exclu- misation with stoichiometric ratio and temperature as optimi- sively on inline FT-IR measurements, a successive combina- sation parameters, and further presents the progress of the tion of real-time reaction monitoring through inline FT-IR objective function over the course of iterations. The results spectroscopy and online mass spectrometry was implemented of Simplex optimisation are illustrated as coloured dots, with within the self-optimising platform. As mass spectrometry the colours indicating the corresponding values of the objec- constitutes an analytical technology that possesses significant- tive function, i.e. the combination of main product’sFT-IR signal and MS ratio, including intensities of main product and ly higher sensitivity than FT-IR spectroscopy, and as reaction temperature may exert considerable influence on the forma- sum of four by-products. The stoichiometric ratio of lithiated tion of by-products that are present in the reaction mixture to a intermediate 3: electrophilic compound 4 was varied in the much lesser extent compared to the main product, such a range between 0.5 and 2. The temperature was varied between combination of both measurement techniques was expected −35 °C and − 10 °C. to provide additional insights. The Simplex optimisation with successive combination of The kinetics [55–57] and mechanism of by-product forma- real-time reaction monitoring through inline FT-IR spectros- tion [74] of the presented organometallic synthesis had previ- copy with online mass spectrometry identified an optimal stoi- ously been studied in detail. Thus, it has been known in ad- chiometric ratio close to 1.0, which is in line with the result vance that four characteristic by-products can be formed. All obtained by the prior optimisation solely based on FT-IR mea- four by-products were available as pure substances, from surements, and an optimal reaction temperature of −30 °, which MS calibration curves could be determined whereas the prior optimisation could not determine optimal Table 1 Comparison of experimental results of Simplex and DoE optimisation. Results are based exclusively on inline FT-IR measurements Parameters Simplex optimisation DoE optimisation (2nd run) (experimentally-obtained) Optimal product yield [%] 98.7 99.0 Temperature [°C] insensitive (−23 °C to −15 °C) insensitive (−35 °C to −20 °C) Stoichiometric ratio (lithiated intermediate: electrophilic compound) 1.0 1.0 Residence time [min] 0.3 0.3 Number of experiments 17 9 per run(18 overall) J Flow Chem (2021) 11:285–302 295 Fig. 7 Simplex optimisation of organometallic synthesis. a Results of optimisation based on inline FT-IR and online MS measurements. b Progress of objective function over course of experiments. Legend: ○ values of start simplex reaction temperature accurately. After three repetitions, the strategy instead of the Simplex algorithm to obtain greater relative standard deviation of the stoichiometric ratio understanding of the underlying processes. amounted to 3.9%, and the one of temperature amounted to The epoxide synthesis proceeded through in situ generated 3.0%. (bromomethyl)lithium, which had been generated by lithium- Figure 8 provides the corresponding results of DoE opti- halogen exchange of dibromomethane, and which constitutes misation, based on the combination of FT-IR and MS signals. a highly unstable lithiated intermediate. Even though The surface response models obtained by the first and second microreactors have been proven to be suited for handling such DoE run, while optimising stoichiometric ratio and tempera- temperature-sensitive syntheses [58, 75–79] due to enabling ture, are displayed with experimental data points being repre- efficient mixing and fast heat transfer, several undesired side sented as red dots. Moreover, individual surface response reactions cannot be ruled out entirely. This includes (1) nucle- models resulting from solely evaluating the FT-IR respective- ophilic addition of the alkyllithium reagent to the carbonyl ly MS signals were calculated and are provided in supporting group of acetophenone, (2) formation of polymers from a information B.2. coupling of bromomethyl lithium with dibromomethane or The surface response models illustrated in Fig. 8 clearly (3) a premature quench of the reaction mixture before the indicate an optimal stoichiometric ratio near 1.0 and an opti- cyclization step could occur [58, 80, 81]. Thus, the applied mal reaction temperature of −30 °C, comparable to the results self-optimisation procedure once again aimed at maximising of Simplex optimisation, when the latter is based on a combi- of main product yield, while simultaneously minimising by- nation of FT-IR and MS measurements. The average devia- product formation. More details on potential side-reactions are tion of experimental data points from the associated surface provided in supporting information C.1. response model amounted to 10% in case of the first DoE run, In contrast to the organometallic reaction with n- and to 5% in case of the second DoE run. The relative standard butyllithium, pure substances of by-products were not avail- deviations of the calculated values of the objective function able in terms of epoxide synthesis. Thus, pre-tests under var- after six repetitions amounted to 10.1% and 9.8% for the first iation of reaction temperature were conducted to identify char- and second run, respectively. acteristic by-products of epoxide synthesis. A detailed de- Table 2 summarizes the performance of both optimisa- scription of pre-tests as well as MS spectra are provided in tion strategies (Simplex and DoE) in terms of maximising supporting information C.2. As a result, the ratio of MS prod- product yield while simultaneously minimising by-product uct intensity and the sum of MS intensities of six characteristic formation. Synthesis of terminal epoxide DoE allowed to screen the entire experimental space and to build surface response models, thus providing further insights into the reaction mechanism In order to evaluate the platform’s versatility in terms of reac- (and where to find the global optimum). Conversely, applying the modified tion types, an epoxide synthesis was also investigated. In con- Simplex algorithm for such a comparatively unknown reaction would have exposed the study to the risk of the algorithm getting stuck at a local (rather trast to the reaction examined before, this time less informa- than global) optimum. Due to the lack of a priori information on the reaction, tion regarding reaction mechanism and kinetics had been no adequate assessment could have taken place to determine whether a given available a priori. Thus, DoE was applied as optimisation optimum is a global or local one. 296 J Flow Chem (2021) 11:285–302 Fig. 8 Optimisation of organometallic synthesis using DoE. Results are based on inline FT-IR and online MS measurements. a First DoE run for screening of whole experimental space. b Second DoE run to refine optimisation m/z values (by-products) was used as input for the objective evaluation of FT-IR respectively MS signals are provided in function together with the main product’s FT-IR signal, Eq. 2. supporting information C.3. Three different dosing options were investigated, differing in their respective type of reactants mixing based on two or Epoxide synthesis proceeding through three feed streams three reactant feeds. In all three cases, residence time in the first part of the reactor was kept constant at 0.5 min. In the Second, the experimental setup was enhanced to handle three following cyclization step, the reaction mixture was then held feed streams: dibromomethane 7 and acetophenone 8 entered at 20 °C for 1 min. through two separate feed streams and, once they had been mixed, were subsequently combined with methyllithium (feed stream 3, see Fig. 2b). Thus, three parameters were optimised: Epoxide synthesis proceeding through two feed streams stoichiometric ratio of acetophenone 8: dibromomethane 7 in the range between 0.1 and 1.5, stoichiometric ratio of Initially, the epoxide synthesis proceeded through combina- methyllithium: premixed starting materials in the range be- tion of a manually-premixed solution of dibromomethane 7 tween 0.6 and 1.1, and reaction temperature in the range be- and acetophenone 8 (1.1 eq. dibromomethane) with the tween −35 °C and − 10 °C. methyllithium, resulting in two feed streams (Fig. 2a). Thus, In the third case, dibromomethane 7 and methyllithium two parameters were optimised: stoichiometric ratio of entered through two separate feed streams and, once they methyllithium: premixed starting materials in the range be- had been mixed, were subsequently combined with tween 0.1 and 1.1, and reaction temperature in the range be- acetophenone 8 (Fig. 2c). Again, three parameters were tween −35 °C and − 10 °C. optimised, albeit now, stoichiometric ratio of methyllithium: In Fig. 9, the resulting surface response models obtained by dibromomethane 7 in the range between 0.1 and 1.1, stoichio- the first and second DoE run are illustrated with manual metric ratio of lithiated intermediate: acetophenone 8 in the premixing of dibromomethane 7 and acetophenone 8. range between 0.1 and 1.1, and reaction temperature in the Results are based on the combination of FT-IR and MS sig- range between −35 °C and − 10 °C. nals. Experimental data points are represented as red dots. For both cases including three feed streams, Fig. 10 repre- The optimal stoichiometric ratio of methyllithium: sents the results of DoE optimisation as experimental data premixed starting materials and optimal reaction temperature points, whose colour scheme represents the objective function were shown to be close to 0.9 respectively −35 °C (Fig. 9). combined from FT-IR and MS signals (see Eq. 2) dependent on Individual surface response models resulting from an the respective variable parameters. The corresponding surface response models derived for the optimisation of three variable Note that, there is a minor deviation between, on the one hand, the results of parameters can be found in supporting information C.3. the DoE optimisation and, on the other hand, the findings described in the Investigating the second synthesis route variable premixing extant literature, where a one-dimensional optimisation of reagent amount and concentration led to 1.5 equiv. of methyllithium under ideal conditions [58]. of dibromomethane 7 and acetophenone 8, DoE optimisation However, the applied optimisation strategies cannot be compared directly, as resulted in an optimal stoichiometric ratio of acetophenone: the setup described in this work allows for multidimensional self-optimisation dibromomethane amounting to 0.9. Thus, under optimised based on calculations of optimisation algorithms, whereas the extant work applies a one-by-one optimisation based on an evaluation of HPLC yields. conditions, dibromomethane 7 is provided in slight excess. Moreover, in the current work, the formation of by-products was considered This finding is in accordance with studies described in the within the objective function, thus resulting in an intricate optimisation prob- extant literature [58]. Again, the best compromise between lem where product yield is maximised with simultaneously reducing the the highest possible main product yield and the lowest level amount of undesired by-products. J Flow Chem (2021) 11:285–302 297 Table 2 Comparison of Parameters Simplex optimisation DoE optimisation experimental results of Simplex and DoE optimisation. Results are (2nd run) based on inline FT-IR and online MS measurements (experimentally-obtained) Optimal result 28 27 Temperature [°C] −30 −30 Stoichiometric ratio 1.0 1.0 (lithiated intermediate: electrophilic compound) Residence time [min] 0.3 0.3 Number of experiments 15 9 per run (18 overall) of by-product formation, was found at a stoichiometric ratio of options are discussed). For every case, optimal reaction con- 0.9 (methyllithium: starting materials) and a reaction temper- ditions are displayed, which were calculated based on the ature of −35 °C. respective surface response model of the second DoE run. Whereas the second synthesis route simply constituted an Moreover, the average deviation of the experimental data extension to three feed streams compared to the first route, a points from the associated surface response model was deter- completely different reaction control was applied in the third mined, allowing to assess each optimisation’saccuracy. case. Herein, external quenching [77, 82] was conducted, in- Comparing the results of DoE optimisation for all three stead of “Barbier”-like internal quenching [83–85]. Given the investigated dosing options, it becomes apparent that the instability of lithium carbenoids, which are likely to suffer objective function assumes a significantly higher value in from immediate thermal decomposition, the generation of its optimum, when applying external quenching. This indi- carbenoid species followed by external trapping with electro- cates a noticeably smaller proportion of undesired by- philes has been less researched compared to internal products compared to reaction control through internal quenching [81]. The enhanced heat transfer in microreactors quenching as maximum main product yield is similar across permits the handling of such thermolabile carbenoids [31, 58, all three investigated cases, see supporting information C.3. 76–79, 86]. The optimisation studies conducted in this work However, while the examination of both internal quenching identified optimal conditions at a slight shortage of routes resulted in surface response models from which op- methyllithium (stoichiometric ratio of methyllithium: timal reaction conditions could be predicted in a reliable and replicable manner as indicated by comparatively low aver- dibromomethane 7 amounting to 0.9) and slight excess of acetophenone (stoichiometric ratio of lithiated intermediate: age deviations of the experimental data points from the as- acetophenone 8 amounting to 0.9). A reaction temperature sociated surface response models, a sufficiently precise rep- of −35 °C once again proved to be ideal for minimisation of resentation of experimental data resulting from the external by-product formation. quenching route through surface response models is much Table 3 summarizes the obtained results of DoE optimisa- harder to achieve. This is due to the fact that, during DoE tion for epoxide synthesis (all three investigated dosing optimisation, experimental plans need to cover a Fig. 9 Optimisation of epoxide synthesis using DoE including manual temperature. Results are based on inline FT-IR and online MS measure- premixing of acetophenone and dibromomethane with two variable ments, a First DoE run for screening of whole experimental space. b parameters, stoichiometric ratio (MeLi: premixed starting materials) and Second DoE run to refine optimisation 298 J Flow Chem (2021) 11:285–302 Fig. 10 Optimisation of epoxide synthesis using DoE with experimental acetophenone: dibromomethane, stoichiometric ratio of MeLi: premixed setup enhanced to handle three feed streams. Results are based on inline starting materials, and temperature, b Premixing of dibromomethane 7 FT-IR and online MS measurements. Experimental data of second DoE and MeLi (external quenching) with three variable parameters, stoichio- run is demonstrated, respectively. a Premixing of dibromomethane 7 and metric ratio of MeLi: dibromomethane, stoichiometric ratio of lithiated acetophenone 8 with three variable parameters, stoichiometric ratio of intermediate: acetophenone, and temperature comparatively broad temperature range (first run: −35 °C up optimum), thus enabling investigation of smaller tempera- to −10 °C; second run: −35 °C up to −25 °C) from which ture steps compared to DoE optimisation. surface response models can be built. Yet, applying external Figure 11 displays the results of this Simplex optimisation quenching requires a longer lifetime of the lithium and further presents the progress of the objective function over carbenoid species compared to its in-situ generation during the course of its iterations. The experimental results are illus- internal quenching. However, due to the high thermal insta- trated as coloured dots, with the colours indicating the corre- bility of lithium carbenoids [82, 87–89], at temperatures sponding values of the objective function (combination of above −35 °C, they are likely to decompose before a reac- main product’s FT-IR signal and MS ratio). Three parameters tion with acetophenone can occur. To overcome this chal- were optimised: stoichiometric ratio of methyllithium: lenge, an additional Simplex optimisation was conducted dibromomethane 7 in the range between 0.8 and 1.0, stoichio- for the external quenching route based on the findings of metric ratio of lithiated intermediate: acetophenone 8 in the the first DoE run (start simplex was located around its range between 0.9 and 1.1, and reaction temperature in the Table 3 Experimental results of epoxide optimisation using DoE. Results are based on inline FT-IR and online MS measurements Parameters Premixing of acetophenone + Premixing of acetophenone + Premixing of MeLi + dibromomethane: dibromomethane: dibromomethane two variable parameters three variable parameters (external quenching) (experimentally-obtained) 2.0 2.1 16.7 Optimal result Stoichiometric ratio 1 0.9 0.9 0.9 (MeLi: premixed (acetophenone: dibromomethane) (MeLi: starting materials) dibromomethane) Stoichiometric ratio 2 – 0.9 0.9 (MeLi: starting materials) (lithiated intermediate: acetophenone) Temperature [°C] −35 −35 −35 Number of experiments 18 (each run 9) 22 (each run 11) 22 (each run 11) relative standard deviations 8.6 10.7 9.5 after six repetitions [%] 8.5 11.8 9.8 1st point 2nd point average deviation experimental 12 17 30 data points – surface response mod- 9 9.6 21 el [%] first DoE run second DoE run J Flow Chem (2021) 11:285–302 299 Fig. 11 Simplex optimisation of epoxide synthesis with premixing of dibromomethane and MeLi (external quenching). a Results of optimisation based on inline FT-IR and online MS measurements. b Progress of objective function over course of experiments. Legend: ○ values of start simplex range between −35 °C and − 25 °C. The results of a second in this work, which was capable of successfully solving the Simplex run, used to verify the obtained results, are provided presented problems within a single working day and without in supporting information C.3. any human intervention. Due to its modular and flexible na- The ideal conditions identified through Simplex optimisa- ture, the set-up moreover accelerates process development tion are comparable to those identified by DoE optimisation. while also diminishing the obstacles that would otherwise Again, the optimal stoichiometric ratio of methyllithium: hamper the transfer from lab to pilot scale. dibromomethane 7 and the optimal stoichiometric ratio of The developed approach incorporates a successive combina- lithiated intermediate: acetophenone 8 were both shown to tion of an inline FT-IR spectrometer and an online mass spec- be close to 0.9 with a slight shortage of methyllithium respec- trometer. There, inline FT-IR spectroscopy is utilized to moni- tively slight excess of acetophenone. Additionally, a mini- tor main product yield, whereas the mass spectrometer’shigh mum amount of by-products is generated at a reaction tem- sensitivity provides insights into the formation of by-products. perature of −35 °C. Examining Fig. 11, it becomes apparent Both techniques entail individual benefits and drawbacks: that the objective function reaches particularly high values Inline FT-IR measurements permit the calculation of product when reaction temperature is kept at a low level. This is in yield based on calibration curves. However, given the lower accordance with the assumption that external quenching re- sensitivity of FT-IR spectroscopy compared to mass spectromet- quires reaction temperatures near −35 °C (but at least lower ric analysis, by-products cannot directly be monitored when they than −33 °C), as otherwise thermal decomposition of the lith- remain on or below a parts per million (ppm) level. Conversely, ium carbenoid species occurs, preventing its reaction with online mass spectrometry illuminates by-product formation, but acetophenone. may only provide relative (instead of absolute) numbers due to ionization suppression or enhancement effects [67–71], as the components may affect each other’s ionization efficiencies. Discussion Overall, by applying both techniques in a manner where they nullify each other’s drawbacks, the presented platform When optimisation aims at maximising product yield while leverages methodological synergies. It enables the multidi- simultaneously minimising by-product formation, the exami- mensional optimisation of all involved process parameters. nation of multi-stage organic syntheses that involve complex Additionally, this work provides further evidence that, while reaction mechanisms leads to intricate multidimensional opti- FT-IR spectroscopy by itself already constitutes a powerful misation problems. Examples include the organic syntheses tool for self-optimisation, performance can be improved fur- investigated in this work: Both the organometallic synthesis ther by a complementary application of online mass spectrom- with n-butyllithium (deprotonation followed by nucleophilic etry. Together, they provide comprehensive insights into com- addition) and epoxide synthesis (reaction and subsequent cy- plex reaction mechanisms. clization) constitute consecutive reactions. A considerable Yet, when conducting multi-stage syntheses, it has to be amount of by-products is generated if either of these syntheses considered that process parameters can be intertwined and, in is not conducted under ideal process conditions, diminishing such cases, may not be varied independently from each other. process efficiency, i.e. leading to decreased product yield and The examined optimisation algorithms in this work can only necessitating work-up procedures. deal with such circumstances as far as certain restrictions are Such intricate optimisation problems may be solved applied, otherwise the number of experiments required to solve the respective optimisation problem would skyrocket. through the versatile self-optimisation approach developed 300 J Flow Chem (2021) 11:285–302 For instance, in this work, constant process parameters were Two different reaction types, organolithium and epoxide applied for the deprotonation step of the organometallic syn- synthesis, were examined in this work, outlining the plat- thesis with n-butyllithium (first step of the synthesis, followed form’s broad range of applicability. All studied optimisa- by nucleophilic addition) to reduce the number of variable tion problems were successfully solved within a single parameters. In terms of epoxide synthesis, the residence times working day. Thus, researchers and industry alike may of both synthesis steps (reaction and cyclization) were held consider its implementation as efficient, reliable and ver- constant and the three examined dosing schemes were inves- satile self-optimisation tool. tigated independently from each other. Supplementary Information The online version contains supplementary To conclude, despite their limitations, both optimisation material available at https://doi.org/10.1007/s41981-021-00140-x. strategies applied in this work, the modified Simplex algo- rithm as well as Design of Experiments, successfully identi- Acknowledgements The authors would like to thank Marius Graute and fied optimal reaction conditions, while at the same time Fabian Wolff (Mannheim University of Applied Sciences) for great tech- granting an in-depth process understanding. A detailed discus- nical support, and the analytics department at Merck KGaA as well as Jörg Sedelmeier for valuable discussions. sion of by-product formation is provided in supporting infor- mation B.3 and C.4. Funding Open Access funding enabled and organized by Projekt DEAL. This work was funded by the German Federal Ministry of Education and Research (BMBF), programme FH Impuls - Partnership for Innovation 2 2 M Aind, project SM all (grant No. 13FH8I01IA). Conclusion Declaration Through its use of multidimensional optimisation of reaction Conflicts of interest On behalf of all authors, the corresponding author parameters, the self-optimising system derived in this work au- states that there is no conflict of interest. tonomously guides chemical processes towards ideal reaction conditions, whilst reducing the need for human intervention. Abbreviations 1, [−] CH-acidic compound; 2, [−] n-butyllithium; 3, [−] Several benefits are provided: The successive combination of lithiated intermediate; 4, [−] electrophilic compound; 5, [−] intermediate; FT-IR spectrometer and mass spectrometer permits investigating 6, [−] product of organometallic synthesis with n-butyllithium; 7, [−] dibromomethane; 8, [−] acetophenone; 9, [−] methyllithium as its lithium intricate optimisation problems, aiming at maximising product bromide complex; 10, [−]epoxide yield while simultaneously minimising by-product formation. Experimental data, which is collected in real-time, can be used Open Access This article is licensed under a Creative Commons as immediate feedback to decide on the next experimental con- Attribution 4.0 International License, which permits use, sharing, adap- ditions, as the use of inline FT-IR spectroscopy and successive tation, distribution and reproduction in any medium or format, as long as online mass spectrometry without former chromatographic sep- you give appropriate credit to the original author(s) and the source, pro- aration means that no waiting times between measurement and vide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included evaluation thereof occur. Thus, both measurement techniques’ in the article's Creative Commons licence, unless indicated otherwise in a unique advantages are leveraged. credit line to the material. If material is not included in the article's Furthermore, going beyond real-time reaction monitor- Creative Commons licence and your intended use is not permitted by ing and the observation of intermediates and by-products, statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this the platform derived in this work fully exploits the poten- licence, visit http://creativecommons.org/licenses/by/4.0/. tial of online analytics by autonomously solving optimi- sation problems through a model-free experimental inves- tigation. Applying a modified Simplex algorithm and References Design of Experiments, optimal reaction conditions of or- ganic syntheses that involve complex reaction mecha- 1. Wehrens R, Buydens LM (2006) In: Meyers RA (ed) Encyclopedia nisms and a large number of variable parameters can be of Analytical Chemistry. John Wiley & Sons, Chichester identified, thus obtaining in-depth process understanding. 2. Schlick T (1992) In: Lipkowitz KB, Boyd DB (eds) Reviews in Computational Chemistry. VCH Publishers, New York The screening of the entire experimental space through 3. Reizman BJ, Jensen KF (2012). Org Process Res Dev 16(11): DoE and subsequent Simplex optimisation constitutes a 1770–1782. https://doi.org/10.1021/op3001838 reliable approach that is particularly useful when a priori 4. McMullen JP, Jensen KF (2010). Annu Rev Anal Chem (Palo Alto, information on the reaction mechanism is not available. Calif) 3:19–42. https://doi.org/10.1146/annurev.anchem.111808. 5. Rasheed M, Wirth T (2011). Angew Chem 123(2):374–376. The mass spectrometer used in this work supported only standard resolution https://doi.org/10.1002/ange.201006107 and could thus not be applied for determination of molecular structures. Therefore, by-product formation was implemented as sum parameter into the 6. Fabry DC, Sugiono E, Rueping M (2014). Isr J Chem 54(4):341– 350. https://doi.org/10.1002/ijch.201300080 objective function. J Flow Chem (2021) 11:285–302 301 7. Ley SV, Fitzpatrick DE, Myers RM, Battilocchio C, Ingham RJ 32. Hessel V, Kralisch D, Kockmann N (2015) Novel process win- dows: innovative gates to intensified and sustainable chemical pro- (2015). Angew Chem Int Ed Eng 54(35):10122–10136. https:// doi.org/10.1002/anie.201501618 cesses. Wiley, Weinheim 33. Jensen KF (2001). Chem Eng Sci 56(2):293–303. https://doi.org/ 8. Houben C, Lapkin AA (2015). Curr Opin Chem Eng 9:1–7. https:// 10.1016/S0009-2509(00)00230-X doi.org/10.1016/j.coche.2015.07.001 34. Kockmann N, Gottsponer M, Zimmermann B, Roberge DM 9. Fabry DC, Sugiono E, Rueping M (2016). React Chem Eng 1(2): (2008). Chem Eur J 14(25):7470–7477. https://doi.org/10.1002/ 129–133. https://doi.org/10.1039/c5re00038f chem.200800707 10. Sans V, Cronin L (2016). Chem Soc Rev 45(8):2032–2043. https:// 35. Kockmann N (2008) Transport Phenomena in Micro Process doi.org/10.1039/c5cs00793c Engineering, heat and mass transfer. Springer-Verlag, Berlin 11. Bédard A-C, Adamo A, Aroh KC, Russell MG, Bedermann AA, 36. Qian Z, Baxendale IR, Ley SV (2010). Chem Eur J 16(41):12342– Torosian J, Yue B, Jensen KF, Jamison TF (2018). Science 12348. https://doi.org/10.1002/chem.201002147 361(6408):1220–1225. https://doi.org/10.1126/science.aat0650 37. Carter CF, Lange H, Ley SV, Baxendale IR, Wittkamp B, Goode 12. Gooding OW (2004). Curr Opin Chem Biol 8(3):297–304. https:// JG, Gaunt NL (2010). Org Process Res Dev 14(2):393–404. https:// doi.org/10.1016/j.cbpa.2004.04.009 doi.org/10.1021/op900305v 13. Murray PM, Bellany F, Benhamou L, Bučar D-K, Tabor AB, 38. Moore JS, Jensen KF (2012). Org Process Res Dev 16(8):1409– Sheppard TD (2016). Org Biomol Chem 14(8):2373–2384. 1415. https://doi.org/10.1021/op300099x https://doi.org/10.1039/c5ob01892g 39. Chung R, Hein JE (2017). Top Catal 60(8):594–608. https://doi. 14. Leardi R (2009). Anal Chim Acta 652(1–2):161–172. https://doi. org/10.1007/s11244-017-0737-9 org/10.1016/j.aca.2009.06.015 40. Browne DL, Wright S, Deadman BJ, Dunnage S, Baxendale IR, 15. Carlson R, Carlson JE (2005) Design and optimization in organic Turner RM, Ley SV (2012). Rapid Commun Mass Spectrom synthesis, data handling in science and technology. Elsevier, 26(17):1999–2010. https://doi.org/10.1002/rcm.6312 Amsterdam 41. Adamczyk M, Fishpaugh J, Gebler J, Mattingly P, Shreder K 16. Fernanda Giné M, Tuon RL, Cesta AA, Paula Packer A, Reis BF (1998). Eur J Mass Spectrom 4(1):121. https://doi.org/10.1255/ (1998). Anal Chim Acta 366(1–3):313–318. https://doi.org/10. ejms.198 1016/S0003-2670(98)00158-5 42. Fabris D (2005). Mass Spectrom Rev 24(1):30–54. https://doi.org/ 17. McMullen JP, Jensen KF (2008). Proc Int Conf Miniat Syst Chem 10.1002/mas.20007 Life Sci 12:1907–1909 43. Bristow TWT, Ray AD, O'Kearney-McMullan A, Lim L, 18. McMullen JP, Jensen KF (2010). Org Process Res Dev 14(5): McCullough B, Zammataro A (2014). J Am Soc Mass Spectrom 1169–1176. https://doi.org/10.1021/op100123e 25(10):1794–1802. https://doi.org/10.1007/s13361-014-0957-1 19. McMullen JP, Stone MT, Buchwald SL, Jensen KF (2010). Angew 44. Roscioli KM, Zhang X, Li SX, Goetz GH, Cheng G, Zhang Z, Siems WF, Hill HH (2013). Int J Mass Spectrom 336:27–36. Chem Int Ed Eng 49(39):7076–7080. https://doi.org/10.1002/anie. 201002590 https://doi.org/10.1016/j.ijms.2012.12.004 45. Mathieson JS, Rosnes MH, Sans V, Kitson PJ, Cronin L (2013). 20. Skilton RA, Parrott AJ, George MW, Poliakoff M, Bourne RA Beilstein J Nanotechnol 4:285–291. https://doi.org/10.3762/ (2013). Appl Spectrosc 67(10):1127–1131. https://doi.org/10. bjnano.4.31 1366/13-06999 46. Ray A, Bristow T, Whitmore C, Mosely J (2018). Mass Spectrom 21. Fitzpatrick DE, Battilocchio C, Ley SV (2016). Org Process Res Re v 37(4):565–579. https://doi.org/10.1002/mas.21539 Dev 20(2):386–394. https://doi.org/10.1021/acs.oprd.5b00313 47. Fleischer H, Do VQ, Thurow K (2019). SLAS technology 24(3): 22. Cortés-Borda D, Wimmer E, Gouilleux B, Barré E, Oger N, 330–341. https://doi.org/10.1177/2472630318813838 Goulamaly L, Peault L, Charrier B, Truchet C, Giraudeau P, 48. Pulliam CJ, Bain RM, Osswald HL, Snyder DT, Fedick PW, Rodriguez-Zubiri M, Le Grognec E, Felpin F-X (2018). J Ayrton ST, Flick TG, Cooks RG (2017). Anal Chem 89(13): Organomet Chem 83(23):14286–14299. https://doi.org/10.1021/ 6969–6975. https://doi.org/10.1021/acs.analchem.7b00119 acs.joc.8b01821 49. Haven JJ, Vandenbergh J, Junkers T (2015). Chem Commun 23. Reizman BJ, Jensen KF (2015). Chem Commun 51(68):13290– 51(22):4611–4614. https://doi.org/10.1039/c4cc10426a 13293. https://doi.org/10.1039/c5cc03651h 50. Ingham RJ, Battilocchio C, Hawkins JM, Ley SV (2014). Beilstein 24. Holmes N, Akien GR, Savage RJD, Stanetty C, Baxendale IR, JOrg Chem 10:641–652. https://doi.org/10.3762/bjoc.10.56 Blacker AJ, Taylor BA, Woodward RL, Meadows RE, Bourne 51. Rueping M, Bootwicha T, Sugiono E (2012). Beilstein J Org Chem RA (2016). React Chem Eng 1(1):96–100. https://doi.org/10. 8:300–307. https://doi.org/10.3762/bjoc.8.32 1039/c5re00083a 52. Garcia-Egido E, Spikmans V, Wong SYF, Warrington BH (2003). 25. Echtermeyer A, Amar Y, Zakrzewski J, Lapkin A (2017). Beilstein Lab Chip 3(2):73–76. https://doi.org/10.1039/B302381H JOrg Chem 13:150–163. https://doi.org/10.3762/bjoc.13.18 53. Colombo E, Ratel P, Mounier L, Guillier F (2011). J Flow Chem 26. Nieuwland PJ, Koch K, van Harskamp N, Wehrens R, van Hest 1(2):68–73. https://doi.org/10.1556/jfchem.2011.00009 JCM, Rutjes FPJT (2010). Chem Asian J 5(4):799–805. https://doi. 54. Delville MM, van Gool JJ, van Wijk IM, van Hest JC, Rutjes FP org/10.1002/asia.200900705 (2012). J Flow Chem 2(4):124–128. https://doi.org/10.1556/JFC- 27. Reizman BJ, Wang Y-M, Buchwald SL, Jensen KF (2016). React D-12-00008 Chem Eng 1(6):658–666. https://doi.org/10.1039/c6re00153j 55. Fath V, Kockmann N, Röder T (2019). Chem Eng Technol 42(10): 28. Reizman BJ, Jensen KF (2016). Acc Chem Res 49(9):1786–1796. 2095–2104. https://doi.org/10.1002/ceat.201900074 https://doi.org/10.1021/acs.accounts.6b00261 56. Fath V, Szmais S, Lau P, Kockmann N, Röder T (2019). Org 29. Wegner J, Ceylan S, Kirschning A (2011). Chem Commun 47(16): Process Res Dev 23(9):2020–2030. https://doi.org/10.1021/acs. 4583–4592. https://doi.org/10.1039/C0CC05060A oprd.9b00265 30. Yoshida J-I (2008) Flash chemistry: fast organic synthesis in 57. Fath V, Lau P, Greve C, Kockmann N, Röder T (2020). Org microsystems. Wiley, Weinheim Process Res Dev 24(10):1955–1968. https://doi.org/10.1021/acs. 31. Hessel V, Kralisch D, Kockmann N, Noël T, Wang Q (2013). oprd.0c00037 58. von Keutz T, Cantillo D, Kappe CO (2019). Org Lett 21(24): ChemSusChem 6(5):746–789. https://doi.org/10.1002/cssc. 10094–10098. https://doi.org/10.1021/acs.orglett.9b04072 201200766 302 J Flow Chem (2021) 11:285–302 59. Ulrich W, Steinbild M (2000) Ullmann's Encyclopedia of Industrial 76. Hartwig J, Metternich JB, Nikbin N, Kirschning A, Ley SV (2014). Org Biomol Chem 12(22):3611–3615. https://doi.org/10.1039/ Chemistry: Lithium and Lithium Compounds. Wiley-VCH, Weinheim C4OB00662C 60. (2004) In: I Marek, Z Rappoport (eds.), The chemistry of 77. Degennaro L, Fanelli F, Giovine A, Luisi R (2015). Adv Synth organolithium compounds, Patai series, Wiley, Chichester Catal 357(1):21–27. https://doi.org/10.1002/adsc.201400747 61. Wu G, Huang M (2006). Chem Rev 106(7):2596–2616. https://doi. 78. Hafner A, Filipponi P, Piccioni L, Meisenbach M, Schenkel B, org/10.1021/cr040694k Venturoni F, Sedelmeier J (2016). Org Process Res Dev 20(10): 62. Kamienski CW, McDonald DP, Stark MW, Papcun JR (2000) 1833–1837. https://doi.org/10.1021/acs.oprd.6b00281 Kirk-Othmer Encyclopedia of Chemical Technology, vol 6. John 79. Hafner A, Mancino V, Meisenbach M, Schenkel B, Sedelmeier J Wiley & Sons, Inc, Hoboken (2017). Org Lett 19(4):786–789. https://doi.org/10.1021/acs. 63. Jacobsen EN (2000). Acc Chem Res 33(6):421–431. https://doi. orglett.6b03753 org/10.1021/ar960061v 80. Pace V, Castoldi L, Holzer W (2014). Adv Synth Catal 356(8): 64. Padwa A, Murphree SS (2006). Arkivoc 37:6–33 1761–1766. https://doi.org/10.1002/adsc.201301042 65. Schneider C (2006). Synthesis 2006(23):3919–3944. https://doi. 81. Pace V, Holzer W, de Kimpe N (2016). Chem Rec 16(4):2061– org/10.1055/s-2006-950348 2076. https://doi.org/10.1002/tcr.201600011 66. Kessler RW, Kessler W, Zikulnig-Rusch E (2016). Chem Ing Tech 82. Köbrich G, Akhtar A, Ansari F, Breckoff WE, Büttner H, Drischel 88(6):710–721. https://doi.org/10.1002/cite.201500147 W, Fischer RH, Flory K, Fröhlich H, Goyert W, Heinemann H, 67. Kruve A, Rebane R, Kipper K, Oldekop M-L, Evard H, Herodes K, Hornke I, Merkle HR, Trapp H, Zündorf W (1967). Angew Chem Ravio P, Leito I (2015). Anal Chim Acta 870:29–44. https://doi. Int Ed Eng 6(1):41–52. https://doi.org/10.1002/anie.196700411 org/10.1016/j.aca.2015.02.017 83. Cainelli G, Tangari N, Ronchi A (1972). Tetrahedron 28(11):3009– 68. Kruve A, Rebane R, Kipper K, Oldekop M-L, Evard H, Herodes K, 3013. https://doi.org/10.1016/0040-4020(72)80015-2 Ravio P, Leito I (2015). Anal Chim Acta 870:8–28. https://doi.org/ 84. Pace V, Castoldi L, Holzer W (2013). J Organomet Chem 78(15): 10.1016/j.aca.2015.02.016 7764–7770. https://doi.org/10.1021/jo401236t 69. Taylor PJ (2005). Clin Biochem 38(4):328–334. https://doi.org/10. 85. Pace V, Holzer W, Verniest G, Alcántara AR, De Kimpe N (2013). 1016/j.clinbiochem.2004.11.007 Adv Synth Catal 355(5):919–926. https://doi.org/10.1002/adsc. 70. Bonfiglio R, King RC, Olah TV, Merkle K (1999). Rapid Commun Mass Spectrom 13(12):1175–1185. https://doi.org/10.1002/(SICI) 1097-0231(19990630)13:12<1175:AID-RCM639>3.0.CO;2-0 86. Schwolow S, Ko JY, Kockmann N, Röder T (2016). Chem Eng Sci 71. King R, Bonfiglio R, Fernandez-Metzler C, Miller-Stein C, Olah T, 141:356–362. https://doi.org/10.1016/j.ces.2015.11.022 Am J (2000). Soc Mass Spectrom 11(11):942–950. https://doi.org/ 87. Köbrich G, Fischer RH (1968). Tetrahedron 24(11):4343–4346. 10.1016/S1044-0305(00)00163-X https://doi.org/10.1016/0040-4020(68)88194-3 72. Swartz ME, Liq J (2005). Chromatogr Relat Technol 28(7–8): 88. Köbrich G (1972). Angew Chem Int Ed Eng 11(6):473–485. 1253–1263. https://doi.org/10.1081/JLC-200053046 https://doi.org/10.1002/anie.197204731 73. Fath V, Kockmann N, Otto J, Röder T (2020). React Chem Eng 89. Kirmse W (1965). Angew Chem Int Ed Eng 4(1):1–10. https://doi. 5(7):1281–1299. https://doi.org/10.1039/D0RE00081G org/10.1002/anie.196500011 74. Winkler S (2007) Untersuchungen zur Herstellung und Umsetzung fluorsubstituierter Phenyllithiumverbindungen, Doctoral Thesis, Publisher’snote Springer Nature remains neutral with regard to jurisdic- TU Darmstadt tional claims in published maps and institutional affiliations. 75. von Keutz T, Cantillo D, Kappe CO (2020). Org Lett 22(19):7537– 7541. https://doi.org/10.1021/acs.orglett.0c02725

Journal

Journal of Flow Chemistry – Springer Journals

Published: Sep 1, 2021

Keywords: Microreaction technology; Inline FT-IR spectroscopy; Online mass spectrometry; Self-optimisation; Organolithium compounds

There are no references for this article.