The increasing academic and industrial interest in biocatalytic processes (chemical reactions catalyzed by an isolated enzyme, immobilized enzyme, or whole cell containing one or more enzymes) is to a large extent driven by the need for selective chemistry . Even more remarkable is that such selectivity is achieved with enzymes under mild reaction conditions. While high selectivity may be easily achievable using biocatalysis, for implementation in industry, it is also necessary to develop a process that is sufficiently efficient that it can be economically feasible. For example, for a pharmaceutical intermediate, a product concentration over 50 g/l must leave the reactor and a high yield of product on biocatalyst (termed biocatalyst yield) must also be achieved [2, 3]. The exact threshold values depend on the type of catalyst and the industry sector (or more accurately the selling price of the product relative to the cost of the substrate). However, almost without exception, a new biocatalytic process studied in the laboratory will not fulfill these requirements, since enzymes are usually evolved to convert natural substrates at low concentrations. This presents an interesting challenge for process chemists and engineers, since the wish to implement a new (non-natural) substrate at high concentrations can only be addressed by a concerted development effort with a combination of biocatalyst modification and process modification. The driver for such process development is economic and while targets can be evaluated in a given case, there remains a further problem, because there are many options to choose from and different routes to solve a given problem . While some solutions are more effective than others, and some are easier to implement than others, there remain many choices to be made. One consequence of such complexity is that to date such an analysis has inevitably been carried out on a case-by-case basis, meaning that often the final scale-up and implementation does not even take place, because it takes too long and is too difficult. Furthermore, in many cases, at an early stage it is not clear which way to develop the process and where to put the research effort. In order to overcome this, one potential vision for the future could be a systematic procedure for automated data collection, followed by testing of a more limited number of alternatives at a miniature scale, such that operations can be carried out with a reduced reagent inventory and potentially even in parallel. Indeed, such schemes already exist for chemical synthetic systems and while the level of complexity with biocatalysis is frequently greater, it is also the case that their value might be the greater. At the very least, it would enable more process options to be evaluated in a shorter time (see Figure 1 for a schematic representation of the philosophy).
Combined with process modeling techniques (Krühne et al., 2013, submitted for publication), this could provide a way to map the solution space and enable design decisions to be made more rapidly and with greater confidence. This is one of the main objectives of the EC-funded BIOINTENSE project. In this brief article, the rationale behind high throughput biocatalytic process development will be discussed, together with the challenges and opportunities such an approach can bring. One of the most important application areas of biocatalytic processes is in the synthesis of pharmaceutical intermediates, where speed of development (and integration with the neighboring catalytic steps) is of the utmost importance.
2 Biocatalytic process development in the pharmaceutical sector
In the pharmaceutical industry, process development time is critical (both for chemical as well as biocatalytic synthetic steps) and therefore it is essential to evaluate and screen process options rapidly. For biocatalytic processes, in order that resources spent on development are used in the most efficient manner possible, a systematic method is necessary to help identify the process constraints (reaction related constraints as well as biocatalyst related constraints). The constraints form the basis of a methodology to identify suitable improvement strategies.
For biocatalytic processes, several strategies are available to improve the process from the initial laboratory reaction, so that it is suitable for industrial application. Strategies focused on reducing the cost contribution of the biocatalyst include fermentation technology (e.g., optimization of the production host platform, carbon feeding strategy, oxygen supply and media composition) to reduce the cost of producing the biocatalyst, as well as protein engineering and biocatalyst immobilization to ensure that the biocatalyst (irrespective of its cost) is subsequently used in the most effective way possible (maximum biocatalyst yield: kg product/kg biocatalyst). Strategies focused on reducing the other cost contributions include reaction engineering (e.g., addition of an organic solvent or use of substrate excess), reactor engineering (e.g., substrate feeding), or process engineering (e.g., in situ product removal), to enable the process to run as effectively as possible (maximum reaction yield, biocatalyst yield and product concentration). Additionally, it is important to recognize the interaction between the strategies.
Interestingly, several recent reviews about the application of protein engineering strategies to solve biocatalytic process challenges have argued that the advances in protein engineering now make it possible to ‘fit’ the biocatalyst to the process [5, 6], as originally proposed by Burton and co-workers . Therefore, once initial activity for the desired reaction has been detected, the enzyme performance can indeed be enhanced by protein engineering, to improve the desired properties, such as substrate repertoire and selectivity, as well as activity and stability . Today, there are many examples where new biocatalytic routes have been established through significant improvement of an existing enzyme, via iterative rounds of mutagenesis and screening [5, 6, 9, 10]. However, despite the remarkable advances in protein engineering, we are yet to be convinced that it is possible to fit the biocatalyst to all process conditions. For example, while optimal operating conditions for a biocatalyst can be expanded significantly from pH 7 and ambient temperature, enzymes still have limitations when compared to chemical catalysts (which in general operate at high concentrations of substrates and products, as well as elevated temperatures ), meaning that operation under extreme conditions may not be possible. However, of even greater importance is the fact that the thermodynamic constraints of the process cannot be addressed by biocatalyst modifications directly. While in nature, enzymes usually catalyze thermodynamically favorable reactions, for non-natural substrates as well as reactions run in synthetic mode, this is frequently not the case. Thus, the design of any process needs to also consider the likely operating space for the biocatalyst and the implication of changing key parameters on the process feasibility and cost .
3 Process development using microfluidic miniaturized systems
Microfluidic technologies concern the use of fluids in small compartments (e.g., with a size in the order of µl volumes and with dynamic flow driven by pressure gradients or other methods). Such technologies are sometimes referred to as micro unit operations (MUOs), where the basic concept is to have conventional large scale equipment mimicked at a micro scale (e.g., reactors and separators ). However, microfluidic devices also enable novel process development methods [13–15]. At small scale, different physical effects dominate the flow compared to larger scale technologies. Microfluidic technologies exploit these effects in a way that simply cannot be achieved at a larger scale. Often these dominant effects are described by dimensionless numbers :
In microfluidic devices, the Reynolds number (Re), the ratio of convective to viscous forces, is low (Re<100 and usually around 1) indicating that viscous forces are dominating and thereby laminar flows are obtained:
where ρ is the fluid density, ν is the fluid velocity, dh is the hydraulic diameter (4A/P, where A is the cross sectional area and P the wetted perimeter) and μ is the fluid viscosity.
The Péclet (Pe) number, the ratio of mass transfer rate due to convection compared to that of diffusion, becomes small in microfluidic devices, indicating that the rate of mass transport is dominated by diffusion:
where D is the diffusion coefficient and a is the radial length scale.
The bond number (Bo), the ratio of gravitational forces to those caused by surface tension, is small in microfluidic devices, as a consequence of dominant surface tension forces, i.e., Bo<<1:
where g is the gravitational acceleration (9.81) and γ is the surface tension.
The Damköhler number (Da) is another important dimensionless number for the characterization of microfluidic systems. This number is used to relate the chemical or biochemical reaction timescale to other phenomena that occur in miniaturized systems. This can, for instance, be the material transport due to diffusion, interphase transport and fluid dynamic convective driving forces. The mathematical description is omitted here due to the dependency on the specific case considered.
At a larger scale these effects do not have such a significant impact, which may result in problems when transferring processes from micro to large scale and vice-versa. However, it is quite common with conventional technologies to experience problems when transferring knowledge obtained at the lab scale to the industrial scale. Alternatively, rather than scaling-up by increasing dimensions, microfluidic systems can be numbered-up/parallelized in order to obtain the desired process throughput (although clearly there is a cost penalty since ‘economies of scale’ are lost). Indeed, this scaling strategy is, in many cases, not straightforward due to operating and handling issues of many systems in parallel .
Nevertheless, for screening of reactions, biocatalysts and processes, many possibilities exist and therefore, even with the potential limitations for scale-up of processes developed in microfluidic systems, there are many motivators for using microfluidic systems for process development. Indeed, in our opinion it seems most likely that process development will benefit most from the application of miniaturized systems. There is a growing group of bioprocess practitioners that share this view, working not only on development problems related to applied biocatalysis [18–20], but also fermentation  and protein recovery for biopharmaceutical applications . Some of the key motivators are reduced development costs and accelerated process development, compared to conventional technologies in the ml scale. In many cases, microfluidic technologies have been applied for chemical synthesis, for example, where otherwise difficult syntheses have been operated and controlled under new and in some cases extreme conditions [23, 24]. However, there is an increasing interest in applying microfluidic technologies for the development of biocatalytic processes, due to the many general benefits and advantages highlighted in the scientific literature . Examples of potential, advantages and benefits for process development based on microfluidic devices are discussed below, where special attention is given to how this will influence the development of new biocatalytic processes.
The first obvious benefit of performing process development in microfluidic systems is the reduced consumption of valuable and scarce resources. The reduced consumption of resources makes it possible to obtain greater process knowledge with the available resources and at the same time reduce the development costs. For biocatalytic processes, this is especially important, since the availability of a generally expensive biocatalyst is initially limited and will continue to be so until the process has been validated. For example, when improving the performance of biocatalysts through protein engineering, only small quantities of different putative mutants need to be tested for their performance before larger scale production is initiated. The reduced consumption is in general, especially for the fine chemicals and pharmaceutical industries, a major driver for using microfluidic systems. Development costs can therefore be reduced since resources are so valuable.
Process development requires the testing and optimization of different biocatalyst and process options [e.g., reactors and downstream unit operations (separations)], which can in principle be performed relatively easily in microfluidic systems. For example, scientific literature can be found on membrane based microfluidic separation units . Furthermore the liquid-liquid extraction in microsystems has also been proven to be successful , especially operated in a continuous way. The extraction in microsystems in two phase systems is also being investigated more . Furthermore, the most promising microfluidic unit operations can easily be tested in combination, to get an indication of how they influence one another. It should though be mentioned that individual reaction systems or processes benefit differently from miniaturization and in some cases it will not be advantageous to use microsystems. In the scientific literature, it has been argued, with good justification, that micro-reactors benefit faster reactions . However, there are also examples where slower reaction systems, e.g., biocatalytic reaction systems, have proven to greatly benefit from being operated at a micro scale . The easy testing and optimization of process options in microfluidic systems opens the possibility of greatly accelerating the development of new processes, which is especially important in intellectual property (IP)-dominated industry sectors, such as pharmaceuticals. Assuming that miniaturized microfluidic systems contribute to easy testing and optimization of processes, such systems open the possibility of greatly accelerated process development, realized through parallelization and automation of the microfluidic systems. Operating the systems in parallel potentially increases screening and testing throughput. This potentially makes it possible to test different process conditions and options relatively quickly, thus generating knowledge that can be used to select and focus on feasible process options, eliminating infeasible processes. The information collected could also serve well the regulatory needs for Quality-by-Design (QbD) of the US Food and Drug Administration . However, a certain degree of automation will be required in order to run the systems in parallel and ensure high throughput, and certainly there is still a major effort in software development required in order to reach automated and parallelized experimental microfluidic platforms . Nevertheless, in principle at least, microfluidic systems already require a certain degree of automation in order to be operated. For example, it is not possible to achieve controlled flows through the devices without automated pumps. Automated systems will also aid in increasing the throughput of the parallel systems, since they in principle are able to operate continuously, with minimum downtime. Automated systems also have the advantage of having consistent systematic errors, making results comparable, unlike manual sample handling which may vary from operator to operator and from day to day.
Furthermore, microfluidic systems can be manufactured in a modular way, thus allowing the user to combine the different fluidic modules to test the influence of different process steps on the process efficiency [33–35]. It will therefore be possible to test the entire miniaturized process before making any efforts to scale-up the best process option.
Microfluidic systems have the advantage of enhanced process control (e.g., controlled flow scenarios and with rapid heat and mass transfer). The characteristic high surface-to-volume ratio in microfluidic systems enables fast and highly controlled heat and mass transfer. This opens up possibilities for dynamic process scenarios (e.g., fast transition between hot and cold regions for reactions operated in cascades). Likewise, laminar flows in microfluidic systems make it possible to operate with different flow scenarios (e.g., parallel, plug flow, slug flow). This can be very useful in order to precisely control mass transfer in these systems and enables the possibility of obtaining valuable mass transfer knowledge for the processes of interest. Also, it makes it much easier to simulate and model the processes in a microfluidic device.
Having laminar flows also enables easy liquid separation in the systems, based on capillary forces or controlled phase (or flow) splitting. This is very useful for extractive purposes and provides an option to operate biocatalytic processes in new ways. For example, this could enable the possibility of having substrate(s) continuously fed to the reaction stream. Other possibilities are in situ product removal or in situ co-product removal operating scenarios, where an auxiliary phase is used to continuously remove products or co-products from the reaction stream. For biocatalytic processes, these scenarios could potentially be useful in order to improve process feasibility by shifting unfavorable reaction equilibria and overcoming the inhibitory effects of substrates and products on the biocatalyst.
The laminar flows correspond to having a membrane free separation or supply system. It is also possible to inject an auxiliary phase between two reacting phases (i.e., liquid membrane operation using hydrodynamic focusing, and thereby control the reaction rate). It is, however, also possible to implement ordinary membranes into these systems, as for example demonstrated by Cervera-Padrell and co-workers . The driving force for the laminar flow and membrane operations is the concentration gradient between the different fluids.
Biocatalytic processes are operated in different ways dependent on the formulation of the biocatalyst, i.e., free, surface immobilized, immobilized in/on support particles, or in whole cell form. In relation to microfluidic systems, the different immobilization scenarios can be exploited in order to perform controlled sequential cascade reactions, or actually replicate metabolic pathways. For example, in Figure 2A, a micro packed bed reactor performing a cascade reaction is illustrated and in Figure 2B, an illustration of a packed bed reactor can be seen, where laminar side-by-side flow is used to perform continuous adsorption and desorption of products.
One of the most important functionalities in pharmaceutical molecules is the amine group and in recent years, therefore, routes to optically pure chiral amines have attracted considerable academic and industrial interest. Of the possible routes for synthesis of such molecules, which include selective crystallization and chemical catalytic methods, biocatalysis is particularly attractive. Biocatalytic methods offer high selectivity, under mild conditions with a renewable and tunable catalyst. In principle, several biocatalytic options exist, but the use of ω-transaminases (EC 2.6.1.X) in synthetic mode has driven significant research to find not only S-selective, but also R-selective enzymes for specific applications, and process routes to effectively implement the technology. Despite the excellent selectivity of this reaction and its unique ability to create a chiral center, in principle with 100% yield, in reality the ω-transaminase is one of the more challenging of the biocatalytic reactions; the substrates and products are often poorly water-soluble, the equilibrium is frequently unfavorable  and the substrate(s) and product(s) are more often than not inhibitory to the reaction (see Table 1) [38, 39]. This means that at first glance such a process is not only economically infeasible, but indeed far away from the targets which would be required for economic industrial exploitation . Interestingly, in common with many other biocatalytic reactions, via a combination of protein engineering and clever use of reaction, reactor and process engineering, a cost effective process can be established (see Figure 3), and excellent precedent has already been set with the synthesis of sitagliptin by Merck and Co (USA) [42, 43], and other examples by Cellgene/Cambrex (USA and Sweden)  and Astra Zeneca (UK and Sweden).
However, there are many other potential molecules to be synthesized using ω-transaminases, where the challenges have not yet been overcome and in general no standardized procedure exists to design an appropriate reaction, reactor and process for a given transaminase conversion. For this reason, we decided to use this reaction as a test system for the microfluidic development platform in the BIOINTENSE project.
Transaminases catalyze the transfer of an amine (-NH2) group from a donor molecule, usually an amino acid or a simple non-chiral amine such as 2-propylamine, to a pro-chiral ketone acceptor, yielding a chiral amine as well as a co-product ketone (or alpha-keto acid) (Figure 4). The enzyme requires the cofactor pyridoxal phosphate (PLP) to act as a shuttle to transfer the amine group. The cofactor is tightly bound to the enzyme and therefore does not pose the cofactor regeneration problems so often encountered in biocatalytic oxidation and reduction reactions [46, 47].
The asymmetric synthesis of chiral amines by ω-transaminase consists of three major steps (Figure 5); fermentation, biocatalytic reaction and product recovery. In order to avoid unnecessary costs, the biocatalyst is used in the crudest possible form (either as whole cells or cell free extract). Immobilization of the enzymes can be used to facilitate recovery and recycle, thereby improving the biocatalyst yield (g products/g biocatalyst).
After the reaction is complete, the biocatalyst is removed (biocatalyst separation) and the product is isolated from the substrate (which may also be recycled dependent upon the cost contribution to the process) prior to purification.
There are many challenges inherent to transaminase processes that need to be dealt with and numerous reports have been published that address one or more of these challenges. Frequently, the suggested strategies solve more than one problem, for instance the use of an auxiliary phase may solve issues related to substrate and product inhibition as well as low water solubility; by contrast, the solution might pose other problems, such as lower biocatalyst stability. An overview of transaminase process challenges has been compiled in Table 1, along with the suggested technologies and strategies used to overcome these, as well as the further implications of using a specific technology.
Although there is a great potential for the application of microfluidic miniaturized systems in process development, there are also several challenges related to their operation.
One of the main challenges is the large number of samples required for analysis due to the sensitivity of the measurements and manual sample handling for off-line measurements. The implementation of on-line measurements could be a possible solution. However, the standard on-line measurement methods [e.g., near-infrared (NIR) and ultraviolet (UV)] can be quite problematic. The compounds involved in the processes studied by BIOINTENSE, amines and ketones, have peaks appearing in critical regions of the NIR and UV spectra. For instance, the amines are shadowed by water in the NIR spectrum, and in the UV spectrum, the peaks appear in the lower region, where common materials used for fabrication of microfluidic devices will have shadowing effects.
The integration of the hardware such as pumps, valves, analytical equipment and the heating/cooling zone can be quite challenging when working at the micro scale. For this reason, it is necessary to standardize connections to simplify their application. There is a similar constraint related to the available technology that can be applied to process development. Here, there is a need for readily and commercially available platforms, modules and methodologies. For instance, for biocatalytic processes, there is no guidance and there has been a trend towards starting from the very beginning each time. For that matter, methodologies should also cover development and scale-up procedures and/or strategies. This is one of the tasks that will be undertaken in BIOINTENSE.
Likewise, the formation (or use) of solids in microsystems can cause severe channel clogging due to surface adhesion. The large surface to volume ratio supports adhesion and it is difficult to prevent . This is a great bottleneck, since the biocatalyst formulation can vary, e.g., free solubilized enzymes, immobilized enzymes on solid support, or whole cells. Biocatalysts are usually expensive and it is intended to use them in as crude as possible a state, or at least for as many cycles as possible . Another challenge that should be considered is the catalyst immobilization in strategic locations of the micro-reactor surface for topology studies. These studies can involve complex biocatalyst distribution patterns determined by simulations using biocatalyst immobilization and can be difficult to replicate experimentally.
6 Future outlook
In the BIOINTENSE project, we are developing entirely new tools and only time will tell if the results and the performance of the microsystem based platform will reveal a new ‘high throughput’ paradigm. However, based on the preliminary results obtained, it can already now be seen that the developed ‘microtools’ contribute to entirely new results, including deepening the understanding and knowledge of mass transfer parameters (like diffusion velocities of the substrates and products). With the help of this information, it will become possible to understand the complex interactions of the biocatalytic system better and hence it can also be expected that in the long run, this information can contribute to the rapid development of the identified processes. Indeed, we are convinced that it will be necessary to develop a miniaturized toolbox for the investigation and screening of process options. Nevertheless, the exact composition of that toolbox is today unknown. The project will show, in the end, if the full advantages of microsystems can be applied for rapid process development and if this is, from an economic point of view, worthwhile. However, the highest expectations are at the moment to prove if the miniaturized process toolbox will contribute to the acceleration of the process development and thereby to the reduction of development time.
Financial support by the European Union FP7 Project BIOINTENSE – Mastering Bioprocess integration and intensification across scales (KBBE 2012.3.3-03 Grant Agreement Number 312148) is gratefully acknowledged.
Straathof AJJ, Panke S, Schmid A. Curr. Opin. Biotech. 2002, 13, 548–556.Google Scholar
Tufvesson P, Lima-Ramos J, Nordblad M, Woodley JM. Org. Proc. Res. Dev. 2011b, 15, 255–274.Google Scholar
Tufvesson P, Lima-Ramos J, A-Haque N, Gernaey KV, Woodley JM. Org. Proc. Res. Dev. 2013, 17, 1233–1238.Google Scholar
Burton SG, Cowan DA, Woodley JM. Nat. Biotech. 2002, 20, 35–46.Google Scholar
Turner NJ. Nat. Chem. Bio. 2009, 8, 567–573.Google Scholar
Bommarius AS, Blum JK, Abrahamson MJ. Curr. Opin. Chem. Biol. 2011, 2, 194–200.Google Scholar
Reetz MT. Angew. Chem. Int. Ed. 2011, 50, 138–174.Google Scholar
Vennestrøm PNR, Christensen CH, Pedersen S, Grunwaldt J-D, Woodley JM. ChemCatChem 2010, 2, 249–258.Google Scholar
Baraldi PT, Hessel V. Green Proc. Synth. 2012, 1, 149–167.Google Scholar
Wirth T. Microreactors in Organic Chemistry and Catalysis, 2nd ed., Wiley-VCH Verlag GmbH & Co KGaA: Weinheim, Germany, 2013.Google Scholar
Bruus H. Theoretical Microfluidics, Oxford University Press: Oxford, 2008.Google Scholar
Roberge DM, Gottsponer M, Eyholzer M, Kockmann N. Chem. Today 2009, 27, 8–11.Google Scholar
Cull S, Lovick J, Lye G, Angeli P. Bioprocess Biosyst. Eng. 2002, 3, 143–153.Google Scholar
O’Sullivan B, Al-Bahrani H, Lawrence J, Campos M, Cázares A, Baganz F, Wohlgemuth R, Hailes HC, Szita N. J. Mol. Catal. B 2012, 77, 1–8.Google Scholar
Dencic I, de Vaan S, Noël T, Meuldijk J, de Croon M, Hessel V. Ind. Eng. Chem. Res. 2013, 52, 10951–10960.Google Scholar
Betts J, Baganz F. Microb. Cell Fact. 2006, 1, 21.Google Scholar
Jensen KF. Chem. Eng. Sci. 2001, 56, 293–303.Google Scholar
McMullen JP, Jensen KF. Annu. Rev. Anal. Chem. 2010, 3, 19–42.Google Scholar
Žnidaršič-Plazl P, Plazl I. Lab Chip 2007, 7, 883–889.Google Scholar
Roberge DM, Ducry L, Bieler N, Cretton P, Zimmermann B. Chem. Eng. Technol. 2005, 28, 318–323.Google Scholar
Bodla VK, Seerup R, Krühne U, Woodley JM, Gernaey KV. Chem. Eng. Technol. 2013, 36, 1017–1026.Google Scholar
U.S. Food and Drug Administration, Department of Health and Human Services. Pharmaceutical CGMPs for the 21st Century - a Risk-Based Approach: Progress Report. U.S. Food and Drug Administration, Department of Health and Human Services: 2007: http://www.fda.gov/AboutFDA/CentersOffices/OfficeofMedicalProductsandTobacco/CDER/ucm128080.htm#mission.
Perozziello G, Simone G, Candeloro P, Gentile F, Malara N, Larocca R, Coluccio M, Pullano SA, Tirinato L, Geschke O. Micro Nanosyst. 2010, 2, 227–238.Google Scholar
Shaikh KA, Ryu KS, Goluch ED, Nam J, Liu J, Thaxton CS, Chiesl TN, Barron AE, Lu Y, Mirkin CA. Proc. Natl. Acad. Sci. 2005, 102, 9745–9750.Google Scholar
Cervera-Padrell AE, Morthensen ST, Lewandowski DJ, Skovby T, Kiil S, Gernaey KV. Org. Proc. Res. Dev. 2012, 16, 888–900.Google Scholar
Tufvesson P, Jensen JS, Kroutil W, Woodley JM. Biotech. Bioeng. 2012, 109, 2159–2162.Google Scholar
Ward JM, Wohlgemuth R. Curr. Org. Chem. 2010, 14, 1914–1927.Google Scholar
Tufvesson P, Lima-Ramos J, Jensen JS, Al-Haque N, Neto W, Woodley JM. Biotech. Bioeng. 2011a, 108, 1479–1493.Google Scholar
Goldberg K, Edegger K, Kroutil W, Liese A. Biotech. Bioeng. 2006, 95, 192–198.Google Scholar
Truppo MD, Strotman H, Hughes G. ChemCatChem 2012, 4, 1071–1074.Google Scholar
Martin A, DiSanto R, Plotnikov I, Kamat S, Shonnard D, Pannuri S. Biochem. Eng. J. 2007, 37, 246–255.Google Scholar
Frodsham L, Golden MD, Hard S, Kenworthy MN, Klauber DJ, Leslie K, MacLeod C, Meadows RE, Mulholland KR, Reilly J, Squire C, Tomasi S, Watt D, Wells AS. Org. Proc. Res. Dev. 2013, 17, 1123–1130.Google Scholar
Pannuri S, DiSanto R, Kamat S. Biocatalysis. In Kirk-Othmer Encyclopedia of Chemical Technology, Wiley: Hoboken, NJ, USA, 2003.Google Scholar
Hwang BY, Cho BK, Yun H, Koteshwar K, Kim B-G. J. Mol. Catal. B. 2005, 37, 47–55.Google Scholar
Hartman RL. Org. Proc. Res. Dev. 2012, 16, 870–887.Google Scholar
About the article
Published Online: 2013-12-06
Published in Print: 2014-02-01