A Neat Idea

Derek Lowe points me at some lateral thinking in the lab

Composition and transformation of substance.

Moderators: Calilasseia, ADParker

A Neat Idea

#1  Postby Calilasseia » Oct 02, 2020 11:52 pm

Special thanks to Derek Lowe for reporting on this neat idea.

Here's your problem. You want to perform some chemical reactions, and have the reaction products separated into different solvent layers, to make extraction of pure product easier. But, doing this in a static container requires thick solvent layers, in order to prevent the equilibrium from being disturbed by various outside influences, and this impacts heavily upon transport between the different solvent layers.

So, a group of chemists came up with an interesting idea. Namely, instead of arranging the solvent layers in a vertical stack in a static container, add them to a cylindrical container that is being rotated at speed. What happens when you do this?

It turns out, courtesy of experiments documented in this paper in Nature, that you can produce stable concentric rings of solvent layers, that are thin enough to enable transport of your reaction products from one ring to another, and which facilitate such joys as being able to perform extraction of acid and base products simulatneously into separate layers after the reaction is complete.

From the abstract:

Cybulski et al, 2020 wrote:Recent years have witnessed increased interest in systems that are capable of supporting multistep chemical processes without the need for manual handling of intermediates. These systems have been based either on collections of batch reactors1 or on flow-chemistry designs2–4, both of which require considerable engineering effort to set up and control. Here we develop an out-of-equilibrium system in which different reaction zones self-organize into a geometry that can dictate the progress of an entire process sequence. Multiple (routinely around 10, and in some cases more than 20) immiscible or pairwise-immiscible liquids of different densities are placed into a rotating container, in which they experience a centrifugal force that dominates over surface tension. As a result, the liquids organize into concentric layers, with thicknesses as low as 150 micrometres and theoretically reaching tens of micrometres. The layers are robust, yet can be internally mixed by accelerating or decelerating the rotation, and each layer can be individually addressed, enabling the addition, sampling or even withdrawal of entire layers during rotation. These features are combined in proof-of-concept experiments that demonstrate, for example, multistep syntheses of small molecules of medicinal interest, simultaneous acid–base extractions, and selective separations from complex mixtures mediated by chemical shuttles. We propose that ‘wall-less’ concentric liquid reactors could become a useful addition to the toolbox of process chemistry at small to medium scales and, in a broader context, illustrate the advantages of transplanting material and/or chemical systems from traditional, static settings into a rotating frame of reference.

The layer organisation is primarily controlled by density, of course, but the authors demonstrate that they can insert and extract entire layers from the concentric stack of rings at will, by altering the rotation speed of the container whilst the procedures are being performed. A wide range of arrangements of stable solvent layers can be produced using this technique, and because the solvent layers are actually in motion at different speeds around the rotation axis, countercurrent extraction becomes possible.

Indeed, the authors provide the requisite explanation:

Cybulski et al, 2020 wrote:The inspiration for our research comes from the so-called density columns5–8 in which immiscible or pairwise-immiscible liquids stack up vertically—if such a multilayered column contained different reagents in different layers, it could potentially drive a multistep reaction of chemicals migrating thought it. In practice, however, static stacks are easily destabilized by capillary forces or by small mechanical disturbances, so the layers must be thick and cannot be internally mixed to speed up transport. Both of these limitations result in very long reagent diffusion times, and thus render the idea impractical.

To avoid such problems, we prepared the stacks not in a static column but in a rotating cylindrical container. In the basic design in Fig.1a, the container is mounted on a vertically aligned shaft of an electric motor. Without rotation the liquids are layered horizontally, with the densest liquid (transparent Fluorinert FC-40) at the bottom. When the container starts to rotate, the liquids adopt slanted profiles with increasing slopes, and ultimately—at the highest rotational speed (ωc ≈ 2,600 rpm)—assume a concentric-layer configuration in which FC-40 forms the outermost ring and the central portion of the container is occupied by air (for theoretical details of layer evolution, see Supplementary Information section 4 and Supplementary Video 8).

Liquids can be added to an already-rotating stack by simply dispens-ing them either near the centre of rotation or via a system of channels embedded in the bottom surface of the container (Fig.1b, c, Supplementary Videos 1, 2). Notably, when the addition of different liquids is sequential, not only all-immiscible but also pairwise-immiscible liquids can be stacked up: with adequate control of liquid addition, stable stacks comprising more than 20 layers can be assembled (Fig. 1d–f). By using channels of different geometries, it is also possible to modify desired, individual layers within the stack: by injecting additional components, drawing small-volume samples, or even removing an entire layer out of the rotating stack (see Supplementary Figs. 1–4, Supplementary Videos 2, 7).

At constant rotational speed and with the liquids subject to rigid-body rotation, the transport processes across each layer are purely diffusive and are therefore slow. A considerably more rapid transport and efficient mixing within (Fig. 2a, b) and between (Fig. 2c) the layers can be achieved—within seconds—by periodically decelerating and accelerating the system, which induces inertial circulation of liquids with respect to the container (along both angular and radial coordinates), deforms the layers and imparts shearing forces (see Supplementary Fig. 5). In particular, the ability to impart localized mixing and/or emulsification near layer boundaries—without disrupting the entire stack or cross-contaminating non-neighbouring layers, even very thin ones—is essential for accelerating reaction sequences and extractions, as described later.

Regarding minimal layer thickness, we found that although ‘stretching’ a liquid around another, immiscible liquid always involves a substantial increase in energy due to surface tension, centripetal forces in our macroscopically sized rotors dominate over any surface tension effects even at moderate rotational speeds. In particular, by consider-ing the balance between capillary and centripetal energies9, the minimal thickness of a stable layer of radius r, density ρ, and sandwiched between layers of densities ρ±Δρ (with which it has surface tension γ) can be expressed as d=(2/ω )√[2y/(Δρr)] (where ω is theangular speed in rad s−1; see Fig. 2d, Supplementary Information section 4.2). Below this value, the layer becomes metastable and, in response to large perturbations, may break into an ‘arc’, as illustrated in Fig. 2e, f. The derived dependence also implies that, for liquids with given values of Δρ and γ, thinner layers can be maintained at higher rotational speeds and at larger radial locations. In our experiments, we were able to routinely prepare layers as thin as a few hundred micrometres (for example, 150 μm in Fig. 2f, g, 300–400 μm in Fig. 2h, i; see also Supplementary Video 4), although theory predicts that even 10-μm-thick layers would be stable for realistic experimental parameters (for example, γ=50 mN m−1, Δρ=0.5 g ml−1, a=106g, as in top-speed ultracentrifuges).

Next, the authors inform us what they did with their rapidly spinning rings of solvents ...

Cybulski et al, 2020 wrote:With such precise control of the thickness and stability of the layers, we implemented several multilayered systems in which substrates migrating through the layers undergo sequences of organic reactions. The first system (Fig. 3a) involves no mixing within or between the layers (constant ωc) and is intended to illustrate the ability to control the outcome of reactions solely by the layer thickness and the rate of diffusive transport through the rotating stack. Here, a phosphonium salt generated in situ in toluene layer 3 gradually transits into aqueous layer 2, where it forms a water-insoluble ylide that, in turn, migrates into dichloromethane (DCM) phase 1 to undergo a Wittig reaction. When the aqueous layer is thin (around 1.5 mm), the dominant product is the diester, P2, rather than the monoester, P1 (dark blue versus black bars, Fig. 3b). However, when the aqueous layer is thicker (around 3.5 to 4.5 mm), the monoester product P1 becomes dominant (grey versus violet bars, Fig. 3b). Similar trends in the formation of P1 and P2 are observed at different concentrations, and can be explained by reaction-diffusion arguments and the differences in the flux of the ylide transported into the DCM phase (see caption of Fig. 3c).

The second example, illustrated in Fig. 3d, is a three-layer synthesis of the analgesic N-(4-ethoxyphenyl)acetamide (also known as phenacetin). The product of initial N-acylation transfers to the water phase, in which the phenolic hydroxyl is deprotonated. Phase transfer (mediated by tetra-n-butylammonium bromide (TBAB)) and subsequent reaction with ethyl iodide produce phenacetin in the outermost layer. We note that mixing within the layer—induced by alternating ωc between 750 rpm and 850 rpm every 5 s—is important as it helps to replenish reagents that are consumed near the interfacial areas, and improves the overall isolated yield of the sequence (48%, compared with 37% in the absence of mixing).

In the third example, the anti-amoebic drug diloxanide furoate is made across four layers. This sequence (Fig. 3e) entails deprotonation of N-methyl-p-aminophenol sulfate, phase transfer of the free amine to the dichloromethane/hexane layer, acylation to form an amide, amide transfer to the second water layer (in which phenol is deprotonated), TBAB-mediated transfer of the phenolic anion to the outermost layer, and the final acylation step therein. The overall yield was 25% and, again, was crucially dependent on the variation of velocity between 750 rpm and 850 rpm and the resultant mixing within layers—without mixing, only traces of the product were detected.

I'll let everyone read what hilarity ensued when they introduced silver nanoparticles into their system. The astute chemists among the readership will be enjoying the prospects of finding out how to apply this to some of their own syntheses, no doubt. :)

It remains only for me to provide the full paper citation, in the interests of honest discourse, namely:

Concentric Liquid Reactors For Chemical Synthesis And Separation by Olgierd Cybulski, Miroslav Dygas, Barbara Mikulak-Klucznik, Marta Siek, Tomasz Klucznik, Seong Yeol Choi, Robert J. Mitchell, Yaroslav I. Sobolev & Bartosz A. Grzybowski, Nature, 586: 57-64 (30th September 2020) DOI: 10.1038/s-41586-020-2768-9
Signature temporarily on hold until I can find a reliable image host ...
User avatar
RS Donator
Posts: 22051
Age: 58

Country: England
United Kingdom (uk)
Print view this post

Return to Chemistry

Who is online

Users viewing this topic: No registered users and 1 guest