Chapter 1: Engineering Factors for Efficient Flow Processes in Chemical Industries

Continuous chemical processes integrated via energy and material flows are forming the basis of a highly successful petrochemical industry. Effectively all petrochemical processes, starting from crude oil heating, hydrotreating, cracking, refining and further synthesis of bulk products are performed in continuous flow reactors and separators. The same applies to other large-scale processes, for example, the synthesis of ammonia and sulfuric acid. The scale of production and the close integration of materials and energy are the key attributes of traditional continuous flow processes that contribute to their remarkable efficiency.

The introduction of continuous flow processes in smaller-scale manufacturing such as speciality chemicals, chemical intermediates, pharmaceutical intermediates, active ingredients in agrochemicals and pharmaceuticals, nutraceuticals, fragrances, surfactants, etc. faces significant challenges due to the reliance of these industries on sunken capital—the existing infrastructure of batch multipurpose plants and the slow introduction of the suitable scale technologies. Only recently have the compact and microreactor systems been developed that could begin to replace the traditional batch multipurpose plants.^1–16  However, the advantages of continuous processing are clear enough. The processes are generally more efficient than batch ones and offer much higher throughput per unit volume and per unit time. Reactants are introduced continuously, react on contact within a smaller reaction space with better defined temperature and flow fields, and are removed continuously from the reaction space. There is better control of process variables and the risk of side reactions is reduced. The reactor volume is determined by the flow rate and residence time of the materials rather than vice versa; therefore, vessels can be smaller and heat transfer and mixing are easier to control. Waste levels are generally also lower.

The areas in which flow processes have been developing at a rapid pace are biotechnology and biomedicine.^17,18  In these areas, the closer relation to living systems (which can be said to be ‘flow systems’ and highly material/energy integrated systems) gives stronger impetus to the exploitation of the functionality of the flow reactors.¹⁹  Such features of small continuous flow systems as the extensive use of in situ analytics,²⁰  sequential operations, use of weaker fields such as electric and magnetic separations,²¹  microwave heating²²  and sonication as well as parallelisation and automation for increase in productivity have already found numerous applications in biotechnology and are rapidly penetrating into chemical processes.

This chapter considers the engineering basis for the design of continuous flow chemical and biochemical reactors at different scales. The emphasis is on new and emerging areas of process intensification (PI), flow chemistry, and compact and microreactors; process engineering of petrochemical reactors is well covered in earlier literature and some aspects are discussed in Chapter 3.^23,24  One of the main differences between large-scale and micro-scale flow processes, to which we pay particular attention, is the more significant role of surface–fluid interactions and hence the need to account for solid–fluid physico-chemical interactions in the reactor design. The issues of scale up of small-scale flow reactors are also considered.

The process intensification concept that emerged in industry initially aimed to reduce the physical footprint of plants, and hence reduce capital investment and improve safety.¹⁵  This concept is now widely accepted in the broader meaning of the reduction in the overall impact of chemical processes over their entire life cycle. The different tools of PI are shown in Figure 1.1.

In flow chemistry, a chemical reaction is run in a continuously flowing stream; liquids (normally reagent/substrate solutions) are driven through a reactor which is often a capillary or tubing. In recent years, flow chemistry has emerged as a viable means for performing many types of chemical transformations. Within industry, flow chemistry is already having a major impact: large pharmaceutical companies have teams of chemists and chemical engineers active in the field. On the macro scale, flow processes are being developed for the manufacture of active pharmaceutical ingredients where a series of synthesis reactions, work-up steps and crystallisation of the final active pharmaceutical intermediate (API) are performed in a sequence of flow modules as shown schematically in Figure 1.2.

Reactions in systems where at least one reactant is solid play a major role in the materials processing industries, encircling a broad range of operations such as extractive metallurgy (e.g. ore leaching), coal gasification (or more generally combustion of solid fuels: coal, lignite, etc.), pyrolysis of lignocellulosic products, incineration of municipal waste and catalyst regeneration.^25,26  Most of these reactions can be represented by a general stoichiometric equation:

1. external (gas phase) mass transfer;

2. diffusion inside the pores (if solid is porous);

3. chemical reaction(s) between gaseous and solid reactants (may involve adsorption of reactant(s) and desorption of reaction products);

4. diffusion or reaction(s) product(s) from the reaction site towards the external surface of the solid;

5. external mass transfer of formed reaction product(s) away from the solid interface.

The diffusion of reaction products through the pore system of a solid material and external mass transfer—forming an integral part of the process—are important if the reaction is reversible. Although the process steps listed above occur in series, any one or more of these could be rate limiting.

In slow reacting systems, the overall dynamics will be limited by the surface kinetics (intrinsic rate); the increase of reaction rate may change the limitation to the pore diffusion. For faster exothermic reactions, the temperature gradient across the particle or fluid film might become the controlling factor. In the case of very fast chemical reactions, the mass transfer in the external fluid film becomes the rate limiting step.

An important difference that distinguishes fluid–solid reactions from their catalytic counterparts is that, in non-catalytic systems, a solid is also involved as a reactant. Continuous consumption of the solid phase during the reaction leads to structural changes of the solid bed morphology, and the reactor system is always in the transient state.

The rate of the overall process for external mass transfer limitation can easily be obtained from the knowledge of mass transfer around solid particles and several correlations for fixed or moving solid particles are reported in the literature.^25,26  The challenges in the mathematical description of these types of reactors concern the molecular diffusion in the pore systems of a solid phase. Continuous changes of a solid's morphology (pore shrinkage or closure, swelling, sintering, softening or cracking of the particles) affect the effective diffusivity with the progress of reaction.

The primary consideration in the design and analysis of such systems is the mode of contact of the phases. Fixed, fluidised and moving bed techniques appear to be the most common mode of phase contacting. Horizontal moving bed, pneumatic conveyers, rotating cylinders and flat hearth furnaces are less common.

In catalytic fluidised bed reactors, the problems of inhomogeneity of the fluidised bed when the gas phase is used as a fluidising agent could be overcome by using an external magnetic force and magnetisable catalyst particles.^27,28  Fluidisation of magnetisable particles by a gas stream in the presence of a uniform applied magnetic field oriented parallel to the flow prevents the hydrodynamic instability that otherwise leads to bubbles and turbulent motion within the medium. The fluidised emulsion phase expands uniformly in response to gas flow velocity.

Chemical reactions between a gas and a solute dissolved in a liquid are very common in industry.²⁹  Examples of important processes performed in gas–liquid reactors include:

• absorption of acid gases;

• oxidation of organic compounds by oxygen or air;

• chlorination;

• hydrogenation of organic compounds.

In such reactions, a gaseous component(s) is dissolved in the liquid phase where it reacts with other reagent(s). In the catalytic reactions (homogeneous catalysis), the liquid phase contains a catalyst together with liquid reactant(s). Slightly different scenarios may occur for a biphasic (liquid–liquid) mode of operation. For example, a liquid reagent will be dissolved in the other liquid phase containing the catalyst.

The fundamental analysis of two-phase reactors is complex due to the coupling of simultaneously occurring diffusion and reactive processes. In addition, the hydrodynamic conditions of the reactive two-phase system are difficult to define.²⁹ 

For a chemical reaction taking place in the laminar film and bulk liquid, starting from elementary mass balance of a reactant A, the expression for calculating the overall reaction rate can be developed as shown in eqn (1.1):

where: c_Ab=bulk concentration of A in the liquid phase; H=Henry constant; Ha=Hatta number; k_Ag=mass transfer coefficient in the gas phase; k_Al=mass transfer coefficient in the liquid phase; p_A=partial pressure of A in the gas phase; x=distance from the interface; z=dimensionless length ; δ_l=thickness of the laminar layer.

Hatta number, or more precisely Ha² , is a dimensionless number being a ratio of the maximum rate of the reaction in the liquid laminar film and the maximum rate of transport through the liquid film. For a first order chemical reaction is defined in eqn (1.3) as follows:

where: D_Al=diffusion coefficient of A in the liquid phase; k_A=reaction rate constant; k_Al=mass transfer coefficient in the liquid phase.

For analysis of such coupled fluid–fluid systems (which may include two liquid phases), it is useful to distinguish between three regimes of reaction rate which are characterised by different Ha values and the enhancement factor E (Table 1.1). The mass transfer rate between two phases is compared with that for pure physical adsorption via enhancement factor (E) as shown in eqn (1.3):

For slow reactions (Ha<0.3), the overall rate of mass transfer is not enhanced by the chemical reaction (which takes place mainly in the bulk of reaction phase), and the enhancement factor becomes approximately 1.³⁰  For the intermediate range of Hatta number (0.3<Ha<3), the overall rate of mass transfer is improved by the chemical reaction . In the case of high Hatta number (Ha>3), the reaction is very fast and proceeds only within the laminar boundary layer (E = Ha)

In gas–liquid reactions, yield and selectivity could also be affected by mass transfer, the nature of gas–liquid contact and the residence time distribution. Table 1.2 gives orders of magnitude of mass transfer parameters for various two-phase reactors.

Some typical examples of two-phase (gas–liquid or liquid–liquid) reactors (see Figure 1.4) include:

• stirred tanks reactors;

• bubble or spray columns;

• packed columns;

• Venturi reactors

The presence of the homogeneous catalyst mainly influences the rate of chemical reaction; however some other effects may appear if the catalyst has interfacial properties (e.g. in micellar and phase transfer catalysis).

Three-phase continuous catalytic processes involving gas, liquid and a solid catalyst are widely used in industrial practice including the manufacturing of commodity chemicals.^25,31,32  The most common example includes liquid phase catalytic hydrogenations, which have been carried out industrially for a very long time.²⁵  Other process examples³³  include:

• desulfurisation;

• hydrocracking;

• refining of crude oil products in petrochemistry;

• synthesis of butynediol from acetylene and formaldehyde;

• reduction of adiponitrile to hexamethylenediamine.

The liquid phase, often acting as a solvent in such types of reactors, not only dissolves the reactants, but also provides a liquid layer around the catalyst particles, which may help to:

1. avoid deactivating deposits (i.e. guarantee higher catalyst effectiveness);

2. achieve better temperature control due to higher heat capacity of liquids; and

3. modify active catalytic sites to promote or inhibit certain reaction pathways.³⁴ 

In most applications, the reaction occurs between a dissolved gas and a liquid phase in the presence of a solid catalyst. However, in some cases, when a large heat sink is required for highly exothermic reactions (e.g. in the Air Products methanol synthesis process), the liquid is an inert medium and the reaction takes place between the dissolved gases at a solid interface.

In practice, in implementing three-phase reaction systems, several alternatives are available for bringing the three phases into contact. Generally, one can classify these systems based on whether the catalyst is suspended in the reactor (more precisely in the liquid phase—slurry reactors) or is present in the form of a packed bed of catalyst particles (fixed bed reactors).

Reactors with the catalyst dispersed in a liquid phase may exist in three forms: (a) bubble columns, (b) mechanically stirred tanks, and (c) three-phase fluidised beds (see Figure 1.5).

Reactors with the catalyst placed in a fixed bed mode can operate as (a) trickle bed reactors, and (b) packed bubble column reactors. In the first mode, the gas phase comprises the continuous phase of the reactor; in the second, two-phase (gas–liquid) flow occurs through the fixed bed of catalyst particles (see Figure 1.6).

When comparing the various possible reactors offered to users for a given application, it is necessary to consider both the main characteristic features of each type of reactor (Table 1.3) and a number of appreciation criteria of varying importance in order to perform a given reaction in the reactor of choice effectively (Table 1.4).²⁵ 

The modelling and design of three-phase reactors, including the various mass transfer and reaction steps of the process³⁷  is shown in Figure 1.7.

The first stage of the modelling process considers the following five steps:

1. 

   Equation 1.4

   where: a_i=gas–liquid interfacial area/volume of bed; c_A(g) = bulk gas phase concentration of A; c_Ai(g) = interfacial concentration of A; =unit of reaction rate r′_A in moles per unit of the mass of catalyst and per time unit; ε_b=bed porosity (gas+liquid); ρ_c=density of catalyst pellet.
2. 

   Equation 1.5
3. 

   Equation 1.6

   where: c_Ai = concentration of A in liquid at interface.
4. 

   Equation 1.7

   where: a_p = external specific area of pellet; c_As = concentration of A at solid–liquid interface; k_c = liquid–solid mass transfer coefficient.
5. 

   Equation 1.8

   where: c_Bs = concentration of B at solid–liquid interface; k= η= effectiveness factor defined as the ratio of the reaction rate to the rate in the absence of diffusion.

Combining equations and rearranging gives eqn (1.9), which allows the calculation of the so-called combined mass transfer resistance [expression in the denominator of eqn (1.9)]:

In modelling the three-phase catalytic reactor, the next stage is to develop material and energy balances. Two approaches are possible: homogeneous or heterogeneous. In the first approach, only the gas phase is considered in the material balance [eqn (1.9) is used]; in the second, one, two or more phases are included in the material balance. For a continuous reactor, these balances are calculated around a differential ‘slice’ of the reactor (see Figure 1.8) of cross-section S and thickness dz. The material and energy balances are coupled via a temperature-dependent rate constant (Arrhenius law) [eqn (1.10)]:

The correlations necessary to calculate mass transfer coefficients, hold-ups, pressure drops and other model parameters can be found in a number of reviews.^{35–36,38,39}  In addition, more than 30 Microsoft® Excel simulators developed by the group of Professors Faïçal Larachi and Bernard Grandjean allow the calculations of various model parameters necessary to model trickle bed as well as packed bed reactors. These simulators can be found on the web pages of the Department of Chemical Engineering, Université Laval, Quebec, Canada.⁴⁰ 

One of the major arguments for the transition from conventional large reactor technologies to intensive small reactors is the perceived simplicity of scale-up of the latter. Issues relating to the scale-up of micro- and compact reactors are considered in Section 1.6. Here, we briefly consider the scale-up of conventional reactors, in particular to highlight the differences with the microchemical technologies in the later section.

Within a traditional sequence of process development, the scale-up process is normally performed in several discrete scale steps—from scientific idea to laboratory tests, followed by validation of lab data in a mini-plant, followed by characterisation of fluid dynamics, recycling and stability at a pilot scale and in cold flow models and demonstration units, and only then transferring processes to production-scale units.

The number of steps or the scale-up ratio (defined as a ratio of plant throughputs or as a ratio of linear dimensions of a reactor at two scales) is determined by the confidence in the measured performance characteristics, complexity of the overall process (e.g. presence of recycles or integration of heat and mass flows with other units) and whether the reactor is operated in a mass transfer or a kinetic control regime. The latter can be illustrated using the overall reaction rate equations.

For a reaction in a kinetic controlled regime, the change in the concentration of a reactant A for a simple A to B first order scheme can be given by:

where: k = kinetic rate constant.

No terms in eqn (1.11) depend on the length-scale of a reactor. Therefore, such a process, which is entirely under kinetic control, can be scaled up with very large scaling factors. The only limitations to the scale-up ratios are due to:

1. the level of accuracy of the measured parameters (if the accuracy of, say measured concentrations, is within 10%, a scale-up ratio of 10 may potentially lead to a 100% error); and

2. any commercial/business reasoning such as risk vs. cost.

In the opposite case [eqn (1.12)], the rate constant k_D is a mass transfer rate constant and therefore necessarily involves distance in its expression. The processes governed by such rate equations require a much more careful scale-up due to the significant dependence of mass transfer steps on the geometry and overall size of the reactors, especially in the case of liquid–solid or three-phase reactions.

Thus, for example, if the rate constant in eqn (1.12) represents aeration of a gas–liquid agitated vessel, the actual expression could take the form of the following functional dependence on the reactor geometry:

where: a = specific mass transfer area [m⁻¹]; P/V = ratio of agitator power to vessel volume; υ_g = superficial gas velocity in the reactor.

In this case in order to achieve the same production yield at different scales (i.e. to keep k_Da the same at different reactor scales), it is necessary to maintain the ratios of P/V, υ_g and the reactor length/diameter ratio. This example identifies the basic scale-up principle for large reactors—the principle of similarity.

The geometrical similarity (e.g. maintaining constant ratios of reactor length to diameter at different scales) is applicable in simple cases when the key mass transfer step is a simple function of a single parameter; for instance, superficial fluid velocity as in the case of gas–liquid bubble columns . This is also the case when one considers the relationship between heat removal and surface area of the reactor or the heat transfer elements.

A more complex similarity principle is kinetic similarity, when the preserved quantity is, for example, residence time. In more general terms, the rules of similarity can be represented by the ratios of the dimensionless numbers, which express the key mass or heat transfer mechanisms controlling the reaction production at different scales.^25,41–43 

A particular difficulty of the scaling process is the situation when flow hydrodynamics are the controlling factor. In such cases, simple similarity rules may not produce a satisfactory result. Even in the relatively simple gas–solid reaction system, the complexity of fluid dynamics may not be obvious. Thus, computational fluid dynamics (CFD) simulation of the flow of a gaseous mixture in a chemical vapour deposition reactor reveals a complex flow pattern within the laminar flow range dominated by the buoyancy-induced vortexes, which controls the process performance.⁴⁴  Complex fluid flow patterns can also be revealed via experimental techniques, such as positron particle tracking, as in the case of gaseous or particulate pneumatic flow through packed beds.⁴⁵ 

The list of scale-up issues typically addressed in the case of conventional large-scale processes (batch and continuous) include:

• presence of impurities that were not considered/studied in the lab or during the pilot scale tests;

• the fact that explosion limits of mixtures measured in small-scale lab equipment can be narrower than those in the large-scale equipment due to lower rates of heat transfer in the latter;

• shape (agitation, fluid short-circuiting, stagnation zones);

• surface-to-volume ratios, flow patterns;

• construction materials;

• flow stability;

• heat removal;

• wall, edge and end effects.

Microchemical reactor technologies address most of these issues, but especially heat and mass transfer scaling, flow stability, homogeneity of the flow and wall/edge effects. One most obvious differences between large-scale and microchemical reactor technologies is the heat regimes: large-scale reactors are typically operating in an adiabatic regime (apart from multi-tubular reactors such as, for example, steam reforming), whereas microchemical systems can attain a truly isothermal regime, or even be operated with steep temperature gradients or temperature swings.

In recent years, a significant effort has been devoted to the development of novel techniques and equipment leading to compact, safe, energy-efficient and environmentally friendly, sustainable processes.^15,46,47  These developments focus on ‘process intensification’; an approach that has been around for quite some time but has emerged in the past few years as an independent and interesting field of chemical engineering.¹⁵ 

Initially, process intensification was defined as a strategy for making a dramatic reduction in the size of a chemical plant whilst still reaching a given production objective.⁴⁸  Such reduction can come from shrinking the size of individual equipment and also from cutting the number of unit operations or apparatuses involved in the entire process. Ramshaw⁴⁸  considered volume reduction in the order of 100 or more to be classed as intensification, which is quite a challenging number.

However, this definition describes process intensification exclusively in terms of the reduction in plant or equipment size. More recently, Stankiewicz and Moulijn¹⁵  defined process intensification as:

According to Stankiewicz and Moulijn,¹⁵  the process intensification concept can be divided into two areas:

• process-intensifying equipment (e.g. novel reactors, intensive mixing, heat transfer and mass transfer devices);

• process intensifying methods (e.g. new or hybrid separations, integration of reaction and separation, heat exchange, phase separation, techniques using alternative energy sources, and new process control methods).

In general, any chemical engineering development that leads to a substantially smaller, cleaner and more energy-efficient technology is process intensification. Examples include:

• novel reactors that provide high surface areas per unit of volume;

• intense mixing devices;

• equipment that performs several unit operations;

• alternative ways of delivering energy to process equipment (e.g. via ultrasound, microwaves, light, etc.).

These technologies can greatly increase the rate of physical and chemical processes, allowing for high productivity from a smaller volume of a reactor.

Additionally, the important issue is that the strategies for process intensification overlap with those for inherently safer process design.^46,49,50  Reducing the size of a chemical plant generally improves safety by reducing both the quantity of hazardous material that can be released in the case of loss of containment, and the potential energy contained in the equipment (i.e. high temperature, high pressure or heat of reaction).

Relevant examples of process-intensifying equipment are briefly described below.

Static mixers are fine examples of process-intensifying equipment (see, for example, www.sulzerchemtech.com). Figure 1.9 shows an example of a static mixer.

Static mixers offer a more size- and energy-efficient method for mixing and can be applied in reaction engineering as static mixer reactors (SMR).⁵¹  Such reactors, which have mixing elements made of heat transfer tubes, can successfully be applied in processes in which simultaneous mixing and intensive heat removal or supply are necessary (e.g. nitration, neutralisation reaction, etc.). One of the disadvantages of SMRs is their sensitivity to clogging by solids, making their utility for reactions involving a slurry catalyst limited.

Microreactors, as a novel concept in chemical technology, enable the introduction of new reaction procedures in chemistry, molecular biology and pharmaceutical chemistry.^{1–3,5,10,52}  Generally, microreactors are realised as miniaturised continuous systems or reaction vessels with typical channel or chamber widths in the range of 10–150 μm. The reduction of characteristic dimensions, resulting in a reaction zone with a small volume, allows application of high temperatures or steep reactant concentration gradients, as well as significantly improved process control and enhanced heat management.

Thereby, the unique advantages of microreaction systems are to carry out chemical reactions in unusual process regimes or under isothermal conditions. In addition, microreactors enable the production of toxic or explosive chemicals on site or on demand with an inherent safety.

Recently, microreactors have been successfully applied for the synthesis of vitamin precursors by a homogeneously catalysed two-phase reaction⁵²  as well as for partial oxidation of propane to metastable acrolein,⁵³  ammonia oxidation,⁵⁴  water gas shift reaction⁵⁵  and many others. A more detailed analysis of the use of microreactors can be found in Chapter 4.

The monolith honeycomb structure (defined uniform cross-sectional shape) is widely used as a catalyst or catalyst support for gas treatment applications as well as for performing three-phase catalytic reactions.^56–58  For the latter applications, particular interest has been focused on catalytic reactions such as hydrogenation, oxidation and bioreactions.

The most important features of the monoliths are:

• very low pressure drop in single or two-phase flow;

• high specific surface area;

• high catalytic efficiency due to very short diffusion paths;

• good performance in processes in which selectivity is influenced by mass transfer resistances.

As shown in Figure 1.1, most process-intensifying methods fall into three areas:

• integration of reaction and one or more unit operations into so-called multifunctional reactors;

• development of new hybrid separations;

• use of alternative forms and source of energy for processing.

The provision of the right amounts of reactants at the reaction site, the establishment and the maintenance of the adequate reaction conditions, and the in-time removal of the reaction products are tasks that are not necessarily solved optimally in standard reactor configurations. Novel designs to improve the interaction of transport and reaction have therefore attracted considerable interest in recent years and have separated into the new field of reaction engineering–multifunctional reactors.^59–63 

The term ‘multifunctional reactor’ can be defined as reaction equipment in which performance is synergistically enhanced by means of integrating one or more additional process functions.⁶²  A widely known example of integrating reaction and heat transfer in a multifunctional unit is the reverse flow reactor.^64,65  In such a reactor, the periodic flow reversal allows perfect utilisation of the heat of exothermic reactions by keeping it within the catalyst bed and, after reversion of the flow direction, exploiting stored energy to preheat the cold reactant inlet gases.

Another widely adopted multifunctional reactor is the combined reactor/heat exchanger for fast exothermic chemical reactions, aimed at improving product selectivity and reducing the risk of explosions.^66–68  An example of such a reactor is the diffusion-bonded three-phase compact reactor integrating the functionalities of static mixing, reaction and heat transfer in a single monolithic block; Figure 1.10 shows a schematic representation.^69–71 

An additional functionality that can easily be realised in such reactors is the staged injection of reactants. This is particularly attractive for controlling the selectivity of reactions such as selective oxidations or reductions. Staged injection can be achieved by, for example, incorporating membranes such as the oxygen sensitive ‘chemical valve’ membranes,⁷²  where there is a continuum flux of oxygen through a reactor wall, as shown in Figure 1.11. However, it was subsequently shown that it was sufficient to have a small number of injection points to increase the product yield.⁷³  The latter is easier to implement in microchemical reactor systems.^71,74 

The concept of combined reactors/heat exchangers can be readily extended to the combination of exothermic and endothermic reactions in a single reactor system.^62,75,76  Such coupled reactions can also exert a favourable influence on the equilibrium, kinetics and selectivity of the synthesis reactions.

Selective removal of products from the middle of an adsorptive reactor is analogous to the withdrawal of a hot gas side stream,⁷⁷  and offers an interesting technique for achieving high conversions (e.g. for the Claus process⁶² ). The in situ adsorption of water produced by the reaction on zeolite pellets mixed into the catalyst bed can be used to displace the equilibrium in favour of product formation. A similar principle has been proposed for increasing the yield of hydrogen in the steam reforming reaction by means of adsorptive removal of carbon dioxide.^78,79 

Reactive distillation is the other well-known example of integrating reaction and separation in a continuous apparatus.⁸⁰  A reactive distillation column is usually split into three sections: (i) the reactive section, in which the reactants are converted into products, and where, by means of distillation, the products are separated out of the reactive zone; the tasks of the rectifying (ii) and stripping (iii) sections depend on the boiling points of the reactants and products.

The advantages of catalytic distillation units, besides the continuous removal of reaction products and higher yields due to the equilibrium shift, consist mainly of reduced energy requirements and lower capital investment.⁸¹  The incorporation of internally finned monoliths into multifunctional reactors for catalytic distillation⁸²  combines both areas of process intensification.

The other examples of combined reactions and separation processes are:

• reactive extraction;^83,84 

• reactive crystallisation;⁸⁵ 

• reactive sorption, i.e. chromatographic reactors^86–88  and pneumatic transport reactors.^79,89 

In all these processes, the incorporation of a separation unit shifts equilibrium towards product formation.

Membrane reactors have found utility in a broad range of applications including biochemical, chemical, environmental and petrochemical systems.^90,91  A variety of membrane separation processes, the novel characteristics of membrane structures and the geometrical advantages offered by the membrane modules have been employed to enhance and assist reaction schemes to attain higher performance levels compared to conventional approaches.

Membranes perform a wide variety of functions—often more than one function in a given context. They can be employed to:

• introduce/separate/purify reactant(s) and products;

• provide the surface for a reaction;

• provide a structure for the reaction medium;

• retain a specific catalyst.

One of the interesting applications of membranes is the direct bubble free oxygen supply.^92–95 

Control of the reaction pathway is a major concern in reaction engineering. Partial oxidation reactions of hydrocarbons are especially relevant here. Using a membrane to introduce oxygen in a controlled fashion into the reactor can facilitate achievement of the desired reaction conditions (i.e. to avoid over-oxidation).

A spinning disk reactor (SDR)⁹⁶  is a horizontally circular plate rotating with the speed of 100 to about 6000 rpm. The rotating plate can be cooled or heated with a heat exchange fluid which flows inside the plate. The applied centrifugal force produces thin liquid films on the surface of the rotating disks. Reactants, which are introduced through the centre of the disk, move across the surface forming thin film (the chemical reaction occurs during this step) and are collected on the edge of the disk.

The small dimension of the film (typically 100 μm) is responsible for very high heat transfer rates between the film and the disk (14 kW m⁻²  K⁻¹) as well as the high mass transfer between the liquid streams and/or between the film and the gas in the surrounding atmosphere. Additionally, SDRs provide very intense mixing in the thin liquid film and can therefore maintain uniform concentration profiles within a rapidly reacting fluid. The residence time is short and, as a result, may allow the use of higher processing temperatures.

The reactor is characterised by a plug flow. It operates in a safe mode due to the small reactor hold-up and the excellent control of fluid temperature. The apparatus is low fouling and is easy to clean. SDRs have been successfully used to perform fast organic reactions and precipitations,⁹⁶  polymerisations⁹⁷  or production of nano- and micron-size particles.^96,98 

One of the key differences between conventional large-scale reactors and most micro- and compact reactor systems is the difference in the flow regimes, or more precisely, the difference in the dominant forces within the flow system.

Most large-scale reactors are dominated by the forces of inertia or large Reynolds numbers (Re) and the force of gravity or large Bond numbers (Bo), with a smaller importance of viscoelastic interactions and very small effects of surface interactions—unless phenomena at a fluid–solid interface, especially for porous solids, are of significance. Contrary to that, flow in the channels of micro- and compact reactors is largely independent of Re numbers, and is dominated by viscoelastic and fluid–surface interactions. For a comprehensive analysis of the physics of fluid flow in microchannels, reader are directed to two very good reviews.^12,99  Table 1.5 lists the various dimensionless numbers useful for describing fluid behaviour in microchannels.

The relative importance of the different forces namely inertia, gravity, surface tension, viscoelastic interactions are illustrated in Figure 1.12. The yellow plane depicts the range of operating conditions and reactor channel sizes where surface interactions dominate over inertia and gravity; the low fluid velocities and small channel diameters are characteristic of microchemical systems. At the micro-level, different multiphase flow structures could be attained in narrow flow channels due to the interplay between effects of pressure, interfacial and intermolecular forces. Thus, pressure-driven behaviour leads to the formation of segmented flow (see Figure 1.13); dominance of capillary forces leads to the formation of dispersed small droplets in the so-called ‘flow focusing’ set up; alternatively, wavy liquid–liquid interfaces could be generated at relatively low velocity differences in the annular flow regime with two immiscible liquids via Kelvin–Helmholtz instability.¹² 

One of the consequences of the low inertia forces in empty microchannels is the long mixing time, dominated by molecular diffusion. This can be exploited in specific measurement and separation applications, as illustrated in ref. 99. In the case of chemical reactor applications, the long mixing time could be problematic. However, the very high specific surface areas of microreactors, which can be patterned to induce directional changes in flow, allow the design of different types of static mixers such as, for example, the ‘chaotic mixer’ concept.¹⁰⁰  Even simple static mixer elements, such as periodic baffles, have been shown to be effective in promoting mixing in microchannels.⁷¹  Alternatively, molecular diffusion mixing can be enhanced by further thinning the fluid streams either by using concentric or parallel steams, or in the so-called interdigital laminar micromixers commercialised by the Institut für Mikrotechnik Mainz GmbH (IMM).¹ 

The flexibility in design, access to a wide range of construction materials and the ability to utilise capillary forces and surface–fluid interactions result in a large number of possible arrangements for phase contacting in compact and microreactors.¹⁰¹ 

Dispersive contacting is typically achieved in slug (Taylor) flow, foam flow or bubble flow, as well as in the form of impinging jet reactors (see Figure 1.13). In particular, Taylor flow has been intensively studied for the cases of multiphase monolith, membrane and microreactors; see, for example, refs. 56,95,102–109.

The attractive features of segmented flow include:

• relatively high mass transfer coefficients between capillary wall and liquid phase, and between gas and liquid phase;^95,104 

• narrow residence time distribution.¹⁰⁶ 

In particular the latter, in combination with enhanced mixing in meandering rather than straight channels, has been exploited for the synthesis of particles with a narrow particle size distribution via a sol–gel route.¹⁰⁶  The enhanced mass transfer is sufficient to produce a marked increase in the observed reaction rates of common chemical reactions.¹¹⁰ 

Bubble flow can be efficiently realised in micro-packed beds in the case of three-phase reactions. Such a system has been proposed for the selective oxidation of alcohols in a compact multifunctional reactor.^69,71  An idealised two-dimensional representation of bubble flow is shown in Figure 1.13: microspherical supports with a uniform particle size provide a convenient way to pack millimetre-scale reactor channels and also result in high Péclet numbers (Pe) (i.e. behaviour close to plug flow), and good hydrodynamic stability (low and stable pressure drop), see Figure 1.14.^70,71  Conversely, randomly shaped catalyst support particles result in larger pressure drops and unstable hydrodynamics. In both cases (segmented and dispersed bubble flows), the continuous phase is characterised by a laminar flow regime, leading to a relatively slow mixing.

A very different approach to dispersive multiphase reactors, which can be realised also in microchemical technology, is impinging jet reactors.^111–115  Collision of high-velocity fluids streams in a confined space creates high turbulence regimes, promoting mixing and heat transfer. Because of the enhanced heat transfer, there is a significant interest and a large body of literature on the applications of impinging jets in heat transfer and burning, but these are outside the scope of this chapter.

Non-dispersive contacting can be arranged in an annular or pipe flow,^116,117  with the thin liquid film flowing along the walls of narrow channels and gas flowing in the core of the channel, or with the much lower fluid linear velocities in the falling film reactors¹¹⁸  and in membrane contactors.^119,120  A comparison of the same reaction performed in dispersed Taylor flow and in annular flow regimes showed an appreciable increase in the reactor throughput.¹¹⁷  Microfabricated membranes, such as the glass plate shown in Figure 1.13, allow very tight control over mass transfer across the interface due to uniformity of pores. However, membrane contactors may also be developed using less expensive techniques, for example, via the use of thin porous plate membranes¹²⁰  or hollow fibre membrane modules.⁹⁰ 

The majority of well-studied reactions in micro- and compact reactors are either uncatalysed stoichiometric or homogeneously catalysed single and two-phase reactions. These do not represent particular challenges for micro- or compact reactors, apart from those cases when more viscous fluids are being used (e.g. ionic liquids) or when precipitation of an insoluble product/by-product is possible.

Heterogenisation of catalysts in microchemical systems on another hand is a challenge. There are four key approaches developed in the literature:

• coated wall reactors with a catalytic ‘washcoat’;^121–124 

• grafted or non-covalently bonded molecular catalysts on the walls of empty channels;

• micro-packed beds;^69,125 

• channels filled with a structured packing, e.g. polymeric porous materials.^126,127 

Coated wall microcapillaries and structured reactors are widely investigated for petrochemical-type reactions (e.g. reforming, shift, etc.) as well as fine chemistry applications.

With such reactors, there is an inevitable compromise between the considerable advantages of microreactors, namely:

• high rates of mass transfer;

• good control over flow distribution hence selectivity;

• high rates of heat transfer;

• possibility for staged injection of reactants;

• possibility for using magnetic and electric fields;

• flow manipulation by controlling interfacial tensions;

• use of in situ analytics etc,

And a rather long list of drawbacks, including most obvious:

• difficulty in creating perfectly adhered catalytic layers on microstructures;

• high cost of manufacturing;

• single use (difficult to remake catalytic layers after deactivation);

• cost of scaling-up (or scaling-out).

However, there is significant potential for developing such reactor systems if the material science problems (adherence) and cost of manufacturing (see below) could be resolved. Several recent studies have demonstrated effective synthesis of coated wall porous supports and catalysts with apparent high stability and catalyst effectiveness. Thus, Pd nanoparticle catalysts or bi-metallic catalysts deposited into a porous titania layer in a fused capillary microreactor is described in ref. 122. Another class of oxide supports that is frequently used in catalysis is mesoporous silica. It has been demonstrated in a number of studies (see, for example, ref. 128) that the open mesoporous structure of a support is beneficial for mass transfer and efficiency of the supported catalysts or chemisorbents. Preparation of a templated mesoporous silica inside microcapillaries was described recently in ref. 129.

Grafting different inorganic or organic structures to attach active catalyst sites, ion exchange or other bonding sites is a well-investigated technique and we will not spend much time on discussing its applications for microreactors. However, two topics are worth mentioning. Recently, particular interest has emerged in the use of porous silicon as a matrix for anchoring active groups, or as a reactive template for the synthesis of metal nanoparticles.¹³⁰  Porous silicon can also be formed inside microreactor channels and used as a template for grafting active groups or depositing catalysts. The interested reader is directed to recent papers on this topic.^131–133  The second topic is flow manipulation by using responsive polymers deposited onto reactor walls.¹³⁴  This development offers tremendous opportunities for active fluid manipulation in compact and microreactors using temperature, electric potential, light or specific chemical traces as controlling signals.

Of the four identified types of heterogeneous compact and microreactor systems, the use of micro-packed beds is arguably the most versatile approach since it utilises the wealth of knowledge in the catalysis community on the design of porous heterogeneous catalysts.

A micro-packed bed reactor for liquid phase reactions is described in ref. 125. The device contains catalyst filling channels, as well as internal filters, to prevent loss of catalyst particles with the flow of reactants. A larger diameter capillary reactor is described for the direct synthesis of hydrogen peroxide.¹³⁵  In this work, a 1/16 inch tubing with 0.765 mm internal diameter (ID) was used as a reactor. The flow regime was not well characterised and is described as broken Taylor flow (see definitions of multiphase contacting in microreactors in Section 1.5.2).

A millimetre-scale compact multifunctional reactor with a microspherical packed bed is described in detail in a series of papers.^69–71  The microspherical carbon support was used to enable easy catalyst channels loading by free-flowing solid particles, exploiting the flexibility of the design of the phenolic resin based structured micro–meso porous carbons.^136,137  The same reactor packed with randomly shaped alumina or titanate nanotube supports exhibited much higher pressure drops and worse hydrodynamic stability than in the case of the spherical monodispersed particles.

A combination of coated wall and packed bed concepts was proposed on the basis of a microstructured reactor environment.¹²¹  In this case, the high surface/volume ratio required to achieve a high catalytic area is attained by producing structured micro-pillars filling the reactor volume. The structured packing results in both high specific area and low pressure drop—the same effect which is achieved by packing membranes into bundles of narrow capillaries (hollow fibres) or by depositing catalysts onto walls of monoliths with high cell density. The micro-pillar geometry of microreactors is highly efficient, but is expensive to produce.

An alternative packing for microreactors is a functionalised porous polymer matrix either as a porous polymeric monolith or deposited onto porous glass.^{126–127,138,139}  Polymeric support offers broad functionality for attaching catalytic functions, which is particularly attractive for anchoring organometallic catalysts.^126,127  Polymers also offer good flexibility for the manufacture of different flow patterns through the monolith and the resultant framework can be used to deposit the catalyst containing layers, as for example in the case of supported ionic liquids.¹³⁹ 

One of the biggest challenges for rapid implementation of micro- and compact reactor technologies is the use of expensive fabrication methods. Here we describe only several key techniques. For earlier reviews on the topic the reader is directed towards an excellent book.¹ 

The chemical etching of polished thin metal plates to produce groves or through channels and holes, followed by assembly of the pack of plates and their diffusion bonding in a high temperature press/furnace, results in a three-dimensional (3D) structured metal monolith with complex internal structures. This process has been perfected by Chart Heat Exchangers (www.chartindustries.com) and Heatric (www.heatric.com), which use either grooved plates or through channels in plates (shims) to assemble their structures. Heatric produces very large compact heat exchangers based on this technology, but also produces a number of custom-designed research modules, including a pilot plant module for our group.¹⁴⁰ 

This technology can produce complex internal structures in an efficient 3D geometry and is suitable for the development of micro packed bed or coated wall micro- and compact reactors. Because the resultant structures are effectively metal monoliths, the diffusion-bonded reactors can be developed for high-pressure applications. The downsides of the technology are the high cost of manufacturing, reliance on chemical etching which produces a significant amount of chemical waste, and long lead times for developing protocols for extending the range of metals which can be used in the diffusion-bonding process.

A simplified method of assembling individually machined or etched plates is to apply pressure to a stack of plates by bolts or sealing the plates via external welding. Examples of such reactors are the corrugated metal plate reactor described in an early patent,¹⁴¹  a metal plate reactor/heat exchanger for fluorination reactions described in ref. 116 or a metal foil assembly reactor for compact fuel processing.¹⁴²  However, such structures often suffer from internal leaks and buckling of plates, and hence poor flow distribution or even blockages.

There are many descriptions of different microreactor designs in silicon or glass. Examples include photochemical reactors etched in silicon wafers,¹⁴³  or either etched or machined, and then anodically bonded glass plates.¹⁴⁴  This platform is the most widely explored and a number of commercial providers and research institutes offer microreactors etched or machined in silicon, glass and polymers. Most of such devices are open channel structures suitable as residence-time modules or mixing modules for lab-scale flow chemistry studies. However, more complex structures can be developed. For example, a highly regular low pressure drop catalyst support fabricated in a silicon microreactor is described in ref. 121.

Significant advantages over etching, machining and diffusion bonding exist in the polymer and metal fabrication methods based on moulding, stamping and rapid prototyping. Early adoption of microfluidics in medicinal and biochemical applications was due to access to cheap devices on the basis of polydimethylsiloxane (PDMS), which can be cast into moulds or formed using rapid prototyping at very low cost and with sufficient precision in the device's features; see ref. 145 for a good review on PDMS applications in microreactors. Alternative polymers could be used to increase the solvent tolerance of the devices.¹⁴⁶  Moulding and casting techniques can also be used to produce microreactor elements in ceramics.^147,148 

Cheap fabrication methods as moulding and rolling can also be used to form channels of holes in thin metal films, which could later be assembled into microreactors. Rapid prototyping could also be extended to metals via the so-called selective laser melting (SLM) technique, which allows the manufacturing of 3D monolith structures from fine metal powders.

The issues of scale-up in microchemical reactor systems differ radically from the typical scale-up process of conventional large-scale reactors. Thus, heat transfer management to avoid hot spots or mass transfer problems induced by, for example, flow inhomogeneity are of little importance. However, new challenges have appeared. One of the key ideas behind microchemical technology is the ability to maintain the reaction conditions/features of the small-scale reactors at the scale of industrial throughput, which is believed to be attainable by the numbering up of small reactors into large parallel reactor systems, as shown conceptually in Figure 1.15.

New types of scale-up/scale-out issues have emerged for microchemical reactor systems including:

• blockage by particulates of narrow reaction and heat transfer channels;

• inhomogeneous distribution of flow between parallel reaction channels;

• cost of manufacturing reactors;

• precision of flow control, pulse-free flow;

• pressure drop in narrow reaction and heat transfer channels;

• control of large parallel reaction systems.

Below we briefly discuss some of the specific issues of the scale-up/scale-out of compact and microreactor systems.

The problem of solids in microreactors is widely appreciated. Solids originate from:

• industrial quality solutions/reagents;

• debris from other equipment and infrastructure;

• the synthesis of low solubility by-products;

• the synthesis of solids.

Thus, in our studies of selective oxidation of benzyl alcohol in benzaldehyde,^69–71  slow formation of insoluble benzoic acid resulted in (i) reversible deactivation of catalysts and (ii) increase in the pressure drop across micro-packed beds. However, it was shown that, by careful design of wetting of microchannels, it is possible to avoid clogging even in reactions involving the synthesis of solids.¹⁴⁹  This particular approach is reliant on segmented (Taylor) flow behaviour, in which the dispersed phase does not come into direct contact with the capillary wall. In this case, the solids are confined to the dispersed phase and thus do not build-up on the reactor walls. It is an efficient method of synthesis of the uniform sized micro- or nanoparticles.

One of the key problems in scaling-out of microchemical reactor systems based on the concept of multiple parallel reaction channels is the establishment of a uniform flow/reactants distribution among all the parallel reactor channels. Inability to reach this goal completely undermines the purpose of the whole concept, since good control over reaction selectivity cannot then be achieved. Therefore, considerable attention has been paid to the design of flow headers for connecting different types of compact and microreactors to inlet and outlet flow pipes.

The design of flow distributors for parallel channels in plate reactors or heat exchangers requires full computational fluid dynamics (CFD) simulation to avoid the imbalance of flow, and hence sub-optimal selectivity/performance of microreactors.¹⁵⁰  An example of CFD design of flow distributor geometry for fast flow in a short contact time in a microreactor is given in ref. 151. The geometries used for optimisation of the inlet and outlet fluid headers, reproduced in Figure 1.16, are not necessarily apparent, and the optimum geometry cannot be arrived at without CFD.

A very useful concept for flow distribution in 3D microreactors is the design of a so-called ‘thick-wall’ screen header,¹⁵²  as illustrated in Figure 1.17. Uniformity of flow across all channels of the reactor module (R) is achieved by splitting the flow into thinner slices in two modules (U and D), with perpendicular orientation of thick-wall screens. The effect of the design variables (sizes of screens: h, c, a, d, l_up and l_down) on the screen efficiency have been studied carefully and design equations for the optimal screen geometry developed.¹⁵³ 

Another fluid distribution problem is when a single gas (or more generally a condensable fluid) stream must be separated into multiple reaction or mixing channels. The downstream pressure drop can have a significant influence on the behaviour of the fluid in the entrance region and on the distribution between the channels. In this case, a pressure-drop section and percolating of the channel diameter are frequently employed for the purpose of flow equilibration; Figure 1.18 shows an example of such a design.

Continuous chemical processes characterised by highly integrated material and energy flows are well-established in traditional petrochemical industries. They are the workhorses of current chemical technologies, which were the foundation of the tremendous success of industrialised economies.

The current drive towards developing more sustainable chemical processes has drawn attention towards creating a new range of continuous processes, suitable for smaller scale manufacturing in speciality chemicals, pharmaceuticals, nutraceuticals, food and fragrance industries, etc. Some of these processes are more widely known as ‘flow-chemistry’. It is a new field, but there are already some remarkable success stories about commercial realisations of compact and microreactors, and other intensified reactor technologies.^{1,3,68,154,155}  There are also many new avenues for development of these technologies, which should lead to significant technological advances (e.g. in photochemistry),^{143,144,156–158}  integrating reactors with sensors and smart elements,¹⁵⁹  and integration of reactor technology with design of nanomaterials.¹⁶⁰ 

The engineering of flow reactors for new applications cannot rely on the traditional rule book. Not only is the physics of flow in narrow channels different to that in large channels (see Section 1.5.1), but many manufacturing principles and manufacturing technologies are also rather different. Thus, the conventional pressure regulations for large vessels do not work for microreactors and attempts to use them result in disaster. For example, a combined reactor/heat exchanger with an inbuilt highly efficient micro-heat exchanger, assembled into a metal body with flanges designed according to a conventional pressure vessel handbook would simply not work, since the thermal mass of the device would kill any advantages of the advanced device architecture.

At the same time, advances in manufacturing methods (e.g. the fast evolution of the rapid prototyping techniques) will considerably reduce the development time of novel flow reactors and also reduce their cost. There is a strong argument for close integration of the development of novel chemistry alongside the development of novel processing technologies in a concerted manner, and also involving the full arsenal of reaction engineering: computational fluid dynamics, CAD design, flow visualisation, reactor simulation, in situ analytics, etc. Only in such a manner will a true breakthrough in these technologies be developed.

a_i

gas–liquid interfacial area / volume of bed [m² m⁻³]

a_p

external specific area of pellet [m² gcat⁻¹]

c_A(g)

bulk gas phase concentration of A [kmol m⁻³]

c_Ai(g)

interfacial concentration of A [kmol m⁻³]

c_Ai

concentration of A in liquid at interface [kmol m⁻³]

c_Ab

bulk liquid concentration of A [kmol m⁻³]

c_As

concentration of A at solid–liquid interface [kmol m⁻³]

d

diameter, channel diameter, hydraulic diameter [m]

E

enhancement factor

H

Henry's constant

k

specific reaction constant [m³ of liquid mol⁻¹ gcat⁻¹ s⁻¹]

k_g

gas phase mass transfer coefficient [m s⁻¹]

k_l

liquid phase mass transfer coefficient [m s⁻¹]

k_c

liquid–solid mass transfer coefficient [m s⁻¹]

k_Al

mass transfer coefficient in the liquid phase [m s⁻¹]

k_Ag

mass transfer coefficient in the gas phase [m s⁻¹]

l

length [m]

p_A

partial pressure of A in the gas phase [MPa]

u

linear fluid velocity, superficial velocity [m s⁻¹]

u_d

linear characteristic velocity of the dispersed phase [m s⁻¹]

z

dimensionless length

δ_l

thickness of the laminar layer [m]

ε_b

bed porosity (gas + liquid) [–]

μ

dynamic viscosity [N s m⁻²]

η

effectiveness factor

ρ_c

density of catalyst pellet [kg m⁻³]

ρ

density [kg m⁻³]

(Δρ)

density difference between immiscible fluids [kg m⁻³]

σ

surface tension [N m⁻¹]