Pickering emulsion droplet-based biomimetic microreactors for continuous flow cascade reactions

A continuous flow cascade of multi-step catalytic reactions is a cutting-edge concept to revolutionize stepwise catalytic synthesis yet is still challenging in practical applications. Herein, a method for practical one-pot cascade catalysis is developed by combining Pickering emulsions with continuous flow. Our method involves co-localization of different catalytically active sub-compartments within droplets of a Pickering emulsion yielding cell-like microreactors, which can be packed in a column reactor for continuous flow cascade catalysis. As exemplified by two chemo-enzymatic cascade reactions for the synthesis of chiral cyanohydrins and chiral ester, 5 − 420 fold enhancement in the catalysis efficiency and as high as 99% enantioselectivity were obtained even over a period of 80 − 240 h. The compartmentalization effect and enriching-reactant properties arising from the biomimetic microreactor are theoretically and experimentally identified as the key factors for boosting the catalysis efficiency and for regulating the kinetics of cascade catalysis.

(Volume ratio is 0.7:1:0.3) was stirred for 12 h at room temperature to reach equilibrium. After phase separation, the concentration of benzaldehyde in the oil phase was determined by GC. The concentration of benzaldehyde in the IL phase is calculated from the difference between the initial concentration in the oil and the determined concentration after equilibrium. This experiment was repeated three times to obtain an average value. The partition coefficient α of benzaldehyde between the IL phase and the oil phase is the ratio of the benzaldehyde concentration in the ionic liquid to that in the oil at equilibrium. By using this method, the partition coefficients of benzaldehyde, acetyl cyanide and O-acylated cyanohydrin were determined to be 4.2, 28 and 26, respectively (average values from different initial concentrations in oil).

Notes:
The fine-tuning of microreactor droplet size was achieved by varying the amount of emulsifier.
As the amount of emulsifier was increased from 1 to 3 and 6 wt%, the average microreactor droplet radius decreased from ca. 27 to 14.5 to 7.5 μm ( Supplementary Fig. 14a). was purchased from Novozymes. Water used in this study was de-ionized water.

Characterization
Nitrogen-sorption analysis was performed at −196 °C on a Micromeritics ASAP 2020 analyzer. Before measurement, samples were out gassed at 120 o C under vacuum for 6 h. The specific surface area was

Material synthesis
Preparation of mesoporous silica nanospheres (MSNs). MSNs were prepared according to a previously reported method 1 . 0.36 g TEA was added into 120 mL CTAC aqueous solution (10 wt%).
The resultant mixture was gently stirred at 60 °C for 1 h. Then, a solution of TEOS in cyclohexane (20 v/v%, 40 mL) was slowly added to the above suspension, which was maintained at 60 °C in a water bath for 12 h under magnetic stirring. The solid product was collected by centrifugation and washed for several times with ethanol. After calcination in air at 550 °C for 5 h (ramping rate, 2 °C min −1 ), MSNs were obtained.

Preparation of Ti(Salen)-containing sub-compartments (Ti/SCs).
A dichloromethane solution of 0.08 g of Ti(Salen) (synthesized according to a reported method 2 ) and 0.03 g of 4dimethylaminopyridine (as cocatalyst) was added dropwise to a dispersion of MSNs (0.2 g). The resultant mixture was further sonicated for 20 min. After evaporation of dichloromethane, Ti/SCs was produced.
Preparation of CALB-containing sub-compartments (CALB/SCs). Before enzyme immobilization, the MSNs were modified with hydrophobic organosilane. In a typical procedure, 1.0 g of as-

(i) Detailed analysis of equilibrium in batch system, and final conversion
The maximum conversion is achieved at equilibrium. To help the analysis of the equilibrium reached in the batch case, we first introduce the following set of parameters for each component. For any where represents the partition coefficient of the component X between IL and the oil phase. In terms of and we have the following relations between the concentrations of the component X S30 in IL phase, in oil and within the whole (total) system: The exchange of the components between different phases in the system occurs much more quickly than the rate at which the reactions proceed. Therefore, it is reasonable to assume that the above relations apply at any time throughout the duration of the reaction. It is easy to check that the average concentration [X] does indeed correspond to Xtot, as seen below Now after a sufficiently long time, when equilibrium is attained, the rates of the backward and forward reactions become the same. Hence Once again, as throughout the paper, it is assumed that the reagents B and E are far in excess. This condition is easily satisfied here for our reactions. Hence, one can take the concentrations of B and E as remaining largely constant at their initial values, unaffected by the progress of the reaction. Also, the rate of the backward reaction, converting A to D in the second step is normally far slower than the forward reaction for the same conversion occurring through the first step. As such then, this can be ignored (hence the absence of k4 in (4)). Expressing all the concentrations in the IL phase in terms of their overall values in the whole system using (2), we have Since every time that an A molecule is consumed it is converted to a C or a D molecule and vice versa, S31 then the total sum of A+C+D in the system is conserved. Furthermore, this should be equal to the initial value of A at the start of the experiment, as introduced within the (1-) liters of oil, per every liter of emulsion. This is to say that Combining equations in (5) with the one above, we arrive at We now recall that the total concentration of B was 0 = (1 − ∅) and similarly 0 = (1 − ∅) . Therefore, equation (7) becomes Now the conversion at equilibrium for the batch system can be calculated as

(ii) Detailed analysis of the conversion in a continuous column reactor
For a sufficiently long reaction column, of length L, the steady-state concentrations inside the droplets at the bottom of the column will correspond to their equilibrium values. We once again have Note once again that any A converted in the column is changed to C or D and vice versa. In a steadystate operation then, the amount of A entering the column through the oil at the inlet should be the same as the total sum of A+C+D in the same amount of oil that is existing the column. Therefore, Combining the above equation with those in (11) allows one to obtain The conversion for the continuous column reactor is then given us

(iii) Detailed analysis of the cascade reaction on the length scale of a single droplet
The diffusion-reaction equations governing the evolution of the concentration of reactants and products, for the model two step cascade reaction introduced above, are as follows within a droplet: In the above equations df is the diffusion coefficient, assumed to be the same for A, C and D, in the IL phase. We also define symbols Under steady state conditions, the concentration of the components stabilizes and ceases to vary with time. Under such circumstances, the above equations reduce to ( ) Now we add the above three equations together to obtain 2 ( ) 0 A C D ∇ + + = which has the general solution ( ) Q A C D q r + + = + within the droplet. Since the total concentration (A+C+D) is finite at the centre of the droplet (r = 0), this implies that Q = 0. Therefore, the total sum of the concentration of the three components A, C and D, remains uniform inside the droplet, equal to (as yet undetermined) constant q. In other words We can use the above result to eliminate A from equations (20) and (21) in favour of C and D. Doing so leads to which more conveniently can be represented in a matrix form ∇ 2 u=(Mu)−v, with the 2 X 2 matrix M ( ) ( ) and vectors u and v defined as The pair of equations in (23) are coupled. In order to solve them one needs to manipulate these so as S ω ω ω = + + Denoting the corresponding eigenvectors as x+ and x− , these are calculated to be Now, by forming the matrix with functions h1 and h2 formed from linear combinations of C and D according to: It should be noted that the two equations for h1 and h2 in (29) To solve equation (29) we express the ∇ 2 operator in their polar spherical coordinates. Taking advantage of the spherical symmetry of the droplet, the decoupled set of equations for h1 and h2 in this coordinate system now read where, as seen from equation (29), constants ε1 and ε2 are (33b) The solution to equations (32a) and (32b) can readily be obtained and read Next we consider the diffusion in the oil phase. It is assumed that no (or extremely slow) reactions occur in oil, since no catalyst or enzyme is present there. Therefore, the mass transport equations in this phase, once the steady state has been achieved, simply become ∇ 2 A=0, ∇ 2 C=0 and ∇ 2 D=0. Solved once again in the spherical polar coordinates, the solution to these equations are  S38   0  2  2  1  1  2  1  2  1  1  2  2 3 sinh( / ) sinh( / ) 5 1 where as before we have denoted the partition coefficient of A, C and D as αA , αC , and αD . For the continuity of the fluxes we have To distinguish the diffusion coefficient in the oil from that in IL, we use the symbol o f d for the former phase. When expressed explicitly, the above equations become These can further be simplified to Next we re-write the first two equations in (37) as ( ) Equations (42a) and (42b) fully express γC and γD, and therefore also q and γA viz. equations (40)