Detecting phase transitions in high-pressure CO2 and supercritical fluids was first attempted in the nineteenth century. By contrast, Green Chemistry, the design and implementation of cleaner methods of manufacturing and processing chemicals, is barely 20 years old. Now, the use of CO2 as an environmentally more acceptable replacement for traditional solvents for greener chemical reactions is creating the need for new, more rapid methods for elucidating high-pressure phase behaviour. This paper describes the advantages and limitations of a number of approaches, developed in Nottingham, to meet this need, including acoustic measurements, shear-mode quartz sensors, the fibre-optic reflectometer, the use of holey fibres, attenuated total reflectance infrared spectroscopy and pressure drop measurements.
It is nearly two centuries since Baron Cagniard de la Tour (1822) first began to answer the question of how a liquid behaves when heated in a sealed vessel. He tried to do this by sealing a ball into a gun barrel and listening to the sound of the ball rolling inside as the barrel was heated; the sound changed in the region of what is now known to be the critical temperature, Tc. Forty years later, Andrews (1869) used an inverted glass capillary tube with a mercury piston to understand the behaviour of high-pressure CO2 and to define the critical points. Since that time a number of techniques have been developed to study the phase behaviour of binary and more complicated mixtures of CO2. The large majority of these methods are quite time-consuming and thus pose a serious difficulty in the context of Green Chemistry where rapid evaluation of phase behaviour of dynamic systems is required.
Green Chemistry was invented in the 1990s as a strategy for preventing the generation of waste, toxic or otherwise, in chemical manufacture and processing (Anastas & Warner 1998). It has caught the imagination of chemists worldwide and has led to increased collaboration between chemists and chemical engineers. In particular, Green Chemistry has highlighted the environmental problems and waste associated with the petroleum-based solvents conventionally used in the majority of chemical reactions and separation processes. Supercritical CO2 (scCO2) and to a lesser extent supercritical H2O (scH2O) have been identified among the possible environmentally more beneficial replacement solvents. Although scCO2 has a rather poor solvent power in comparison with many traditional solvents, it has been successfully employed in a wide range of reactions (Darr & Poliakoff 1999; Beckman 2004; Akien & Poliakoff 2009), which have often been carried out in continuous fixed bed catalytic reactions largely because this simplifies possible scale-up for industrial use (Licence et al. 2003). However, the effective design of such reactors even on a laboratory scale requires knowledge of the phase state of the reaction mixture. Such mixtures are usually more complex than those normally studied by specialists in phase behaviour, and the reaction temperatures and pressures are often above the limit of most conventional phase behaviour analysers. The situation is further exacerbated by the fact that the composition of their reaction mixtures changes during a chemical reaction and indeed chemists often change compositions at frequent intervals as they try to optimize the yields of the products. Thus, green chemists need methods for monitoring high-pressure phase behaviour which are relatively quick, can operate at high temperature and pressures and can also handle multi-component mixtures. Such requirements are quite challenging. This paper describes several different approaches to phase analysis that we have developed in Nottingham to address this challenge.
2. General strategy
The most convenient phase diagram to describe a mixture with specific composition is the pressure–temperature (p–T) projection (figure 1). The phase boundary between the single- and two-phase regions consists of two parts: the bubble-point line and the dew-point line, which are continuous curves meeting at the critical point. All our methods use sensors to construct these phase diagrams by detecting an abrupt change in sensor output when the phase boundary is crossed by varying either temperature or pressure, or both. Based on this concept, there are three searching strategies for the phase boundary (isothermal, isobaric and isochoric; figure 1). For example, isothermal searching normally begins with a single-phase, isoplethic mixture at a constant temperature; the system pressure is then gradually changed by using a syringe pump or a back-pressure regulator (BPR) until phase separation occurs. This separation is usually observed as a discontinuity in the output of the sensor, such as a maximum, a minimum, a sudden increase, a sudden decrease, etc. Different searching methods are required to locate the phase transitions in the different parts of the p–T boundary, e.g. the isobaric method is the most effective method to locate the maxcondentherm point (i.e. the maximum temperature on the p–T envelope).
Possible approaches (Dohrn et al. 2010) to measuring phase equilibria can be divided into synthetic methods that use mixtures prepared with a specific composition before the start of the experiment and analytic methods, where the composition of the different phases is established by sampling. In the following sections, we will describe six methods for studying fluid phase equilibria of mixtures using the synthetic approach.
Since the p–T boundary is smooth near the critical point, it is not possible to locate the critical point merely from the (p, T) data without differentiating between the bubble-point and dew-point lines. However, most of our sensor methods can distinguish between bubble points and dew points by examining the characteristic of the change in sensor outputs when crossing a phase boundary. In addition, the acoustic method (see §3) measures the speed of sound, which is sensitive to the critical fluctuations, and therefore gives a direct indication of critical points. The attenuated total reflection (ATR) method is the only technique that provides information on the compositions of each component in the liquid phase, which allows us to detect bubble-point transitions directly.
3. The acoustic method
Our development of the acoustic method has been focused on the determination of the critical point of fluids and fluid mixtures by measuring the speed of sound in the near-critical region. The speed of sound is related to two key thermodynamic properties (Rowlinson & Swinton 1982): isothermal compressibility (KT=−(1/V)(∂V/∂p)T) and isochoric heat capacity (CV; see also equation (3.1)). For pure substances, the speed of sound reaches a minimum very slightly offset from the critical point (Oag et al. 2004) because (∂p/∂V)T must be zero to satisfy the criticality conditions. When critical fluctuations are further taken into account, CV is weakly divergent at the critical point (Griffiths & Wheeler 1970; Levelt Sengers 1991). Therefore, neglecting absorption and dispersion, the speed of sound at zero frequency (C0) should vanish at the critical point.
The zero frequency speed of sound for mixtures (C0,x) is given by the following equation: 3.1 where Vm is the molar volume, M is the average molar mass of the mixture, and CV,x,m is the isochoric molar heat capacity at a specific composition. One can see that, unlike for pure substances, ] does not necessarily reach a minimum (i.e. zero) at the critical point because, apart from azeotropic mixtures, (∂p/∂Vm)T,x≠0 at the critical point (Young 1972). However, when critical fluctuations in mixtures are considered, CV,x,m comes to a finite maximum at a sharp cusp approaching Tc (Scott 1972; Reis et al. 2006). This behaviour is also supported by recent experimental work (Polikhronidi et al. 2006) on CO2 + n-decane mixtures, which demonstrates that the one-phase CV,x,m shows a local maximum at Tc along the coexistence curve. Since (∂p/∂Vm)T,x approaches smoothly to a finite negative quantity (∂p/∂Vm)T,x<0) at Tc along the coexistence curve, the sharp cusp in CV,x,m leads to a minimum in the speed of sound very close to the critical point of the mixture.
Our measurements have been carried out in static cells (Popov et al. 1994; Kordikowski et al. 1996a, 1997), the key components of which are two identical ultrasonic transducers placed in parallel (figure 2a): the first feeding a sound wave into the fluid and the second acting as a ‘microphone’ to convert the sound wave into electric signals. We measure the transit time (Δt) of the sound wave, which is inversely proportional to the speed of sound. The basic experimental steps involve the isothermal method for recording Δt–p traces at a series of temperatures in the near-critical region (see figure 2b obtained from pure CO2), followed by plotting the maximum transit time on each trace as a function of T. Tc is indicated clearly by the global maximum shown in figure 2c. Once Tc has been obtained, the critical pressure (pc) can be easily determined from the data shown in figure 2b. The overall error of our acoustic method was less than ±0.2 K for Tc and less than ±0.3 bar for pc, respectively, depending on the system being studied; for the systems studied so far, these errors apply to multi-component mixtures as well as pure substances.
We have measured the critical parameters of systems as simple as pure substances (e.g. CO2, C2H6, CF3CH2F) to as complex as six-component mixtures (CO2+H2+CO+propene+iso- and n-butyraldehyde; Popov et al. 1994; Kordikowski & Poliakoff 1998; Ke et al. 2001a; Oag et al. 2004). Among these systems, the binary mixtures provide a fundamental understanding of the behaviour of mixed supercritical solvents (e.g. CO2+hydrofluorocarbons) (Kordikowski et al. 1996a). The critical parameters measured acoustically for CO2+He gave a method to determine the He content in He-pressurized carbon dioxide cylinders, showing that the effect of He in CO2 cannot be ignored when this type of cylinder is used in supercritical fluid (SCF) studies (Kordikowski et al. 1996b). Furthermore, the studies on the multi-component systems give an insight into the phase behaviour of complicated mixtures for chemical reactions (such as hydrogenations, Ke et al. 2002; hydroformylations, Ke et al. 2001b; and polymerizations of aliphatic polyketones, Kordikowski et al. 1997) in SCFs. For example, the Tc and pc data for the reaction mixtures of the hydroformylation of propene in CO2 demonstrated how the critical points change as the reaction proceeds from zero conversion to completion.
There are two major limitations. First, for mixtures with permanent gases (e.g. He, H2 and CO), the maximum in the transit time becomes less pronounced and broadens with increasing mole fractions of such gases in the mixtures (Kordikowski et al. 1997). Although the critical parameters of a few key mixtures obtained by the acoustic method have been validated by using more conventional visual methods, this limitation restricts our measurements to low concentrations of permanent gases (e.g. xCO<0.4). Secondly, the acoustic method is unable to locate the whole p–T phase boundary because the maximum transit time at a given temperature does not necessarily correspond to a phase transition. An example is given in figure 2c, in which the dashed line could not possibly be a phase separation line because the system temperature is above the Tc of pure CO2. We shall show that these limitations can be overcome by using the shear-mode and fibre-optic sensors.
4. Shear-mode quartz sensor
A shear-mode quartz sensor is essentially an AT-cut, thin quartz plate with metallic coatings over the central areas of the opposing faces. The electrical equivalent circuit of the sensor is affected through acoustic coupling by the properties of the fluid in immediate contact with its surface. The relationship between the impedance minimum () of the equivalent circuit of the sensor and the fluid properties is given by (Oag et al. 2003): 4.1 where ρ and η are the density and viscosity of the fluid, respectively. The density and viscosity of fluids are generally higher for liquids than for vapours, thus the quartz sensor gives a high value of in a liquid-like environment and a low value in a gas-like fluid.
Figure 3a,b shows two typical traces recorded isothermally as a function of pressure, corresponding to the behaviour at the bubble and dew point, respectively. When the system reaches the bubble point, small bubbles are formed and rise to the top of the cell. Although these bubbles cannot be detected by the sensor, the sensor registers a modest increase in in the bulk of the fluid owing to the liquid being enriched in heavy components. Therefore, the curve versus pressure shows the ‘tick shape’. In contrast, a thin film of liquid is formed on the surface of the sensor at the dew point, resulting in a sudden increase in . As the system pressure decreases further, more and more liquid is formed and collected in the bottom of the cell. This liquid is continually thrown to the sensor by the stirrer so that a liquid film is remained on the surface of the sensor. Since the quartz sensor is a surface-sensitive device and responds to a very thin layer of fluid (less than 200 nm), a small amount of the newly formed liquid is sufficient to produce a sudden jump (), which represents the difference in (ρη)1/2 between the original bulk gas phase and the newly formed liquid phase (Ke et al. 2005).
Figure 3d shows a series of –p traces obtained using the isothermal method for the mixture of CO2+MeOH with xMeOH=0.103. A minimum on each isotherm indicates the onset of the phase transition. The p–T phase boundary is then simply obtained by plotting the corresponding pressure at the minimum against the temperature, as in figure 3e. More importantly, one can easily distinguish between the bubble and dew points by inspecting the shape of –p traces (‘tick’ shape or ‘jump’ shape) and hence to locate the critical point. We have also developed a numerical method to measure Tc more precisely than by visual inspection of the traces. The closer the system is to the critical point, the smaller the difference in (ρη)1/2, and the smaller the jump (). A jump with a zero height must indicate the critical point because both the density and viscosity of two coexisting phases are identical at the critical point. As shown in figure 3c, Tc is located at the temperature where extrapolates to zero. This approach to locating the critical point can be generalized to any sensor that only responds to a property of the fluid in immediate contact with its surface (e.g. the optic fibre—see below).
The shear-mode sensor has been used for measuring the p–T phase boundary and the critical point of a wide range of binary mixtures (Ke et al. 2004, 2005, 2007), including a type I mixture (CO2+MeOH), a type II mixture (CO2+n-heptane), three type III mixtures (CO2+N2, CO2+H2, CO2+1-octanol) and a type IV mixture (CO2+2-octanol), according to the classification by van Konynenburg and Scott (1980). Our results demonstrate that the quartz sensor method is applicable for studying fluid phase equilibria of highly asymmetric binary systems, which do not necessarily have a continuous critical line between both critical points of pure components (e.g. type III/IV mixtures). Moreover, for the mixtures of CO2+N2 and CO2+H2, the phenomenon of retrograde condensation can be identified by the two ‘jumps’ along the isotherm at a given temperature between Tc and maxcondentherm. We have also shown that the liquid–liquid–vapour equilibrium can be studied using a single sensor. exhibits two inflexion points with increasing the total volume of the system, corresponding to the formations of the second liquid phase and the vapour phase (the third phase; Ke et al. 2007). The existence of a three-phase equilibrium, inferred from our measurements with a single sensor, was subsequently confirmed by using a conventional view cell.
This method is not limited to binary systems. Our recent work (Ke et al. 2006) has been focused on a ternary system of 2-propanol (IPA) + N2+CO2, which is a model system for understanding the phase behaviour of oxidations using O2 in CO2 (Bourne et al. 2009; Chapman et al. 2010). N2 has been used to replace O2 to avoid the possible oxidation of IPA during the measurements. Our results show clearly that the quartz sensor method can be extended for studying multi-component reaction mixtures.
All our measurements have been made in a large variable-volume cell (maximum volume 320 ml) fitted with a high-speed, magnetic stirrer at the bottom. The quartz sensor is mounted just above the stirrer. These key features ensure that the sensor responds only to the liquid phase when both the liquid and gas phase coexist in the cell. The temperature stability of the variable-volume cell is better than ±10 mK, which allows us to perform relatively high precision measurements. The experimental error for Tc is about ±0.2 K for most of the mixtures in our studies. However, the time for acquiring a single –p trace is 8–16 h because it takes a considerable amount of time for a large cell to reach thermal equilibrium. Even if the measurements have been fully automated, this static method is more time-consuming than the dynamic methods discussed next, which use small tubular vessels.
5. The fibre-optic reflectometer
Away from the critical point, coexisting gas and liquid phases usually have different refractive indices, a difference that can be used to monitor phase transitions occurring on the vapour–liquid phase boundaries. In the fibre-optic reflectometer (FOR), light from a diode travels along a silica-based, single-mode optic fibre until it reaches the far end of the fibre (the fibre/fluid interface) where part of the light is reflected back (see the inset in figure 4). The reflective coefficient, R, defined as the ratio of the intensity of the reflected beam (IR) to the incident beam (I0) is related to the refractive index of the fibre (n0) and the fluid (nf) (equation (5.1)). R, directly measured by FOR, is high for gas and low for liquid because n0 is a constant (=1.45 for quartz), and ngas<nliquid<n0 for most of supercritical fluids (Avdeev et al. 2004) 5.1
A schematic diagram of the FOR is shown in figure 4. A fibre X-splitter divides the light into two beams: reference and probe. The reference beam is monitored by a photodiode for elimination of any low-frequency fluctuations from the light source. The probe beam is introduced to a high-pressure measuring chamber through the optical fibre, and the reflected beam is monitored by the second photodiode.
A fibre-optic probe is highly versatile for high-pressure measurements in SCFs because of its small diameter (less than 250 μm), high pliability and toughness and suitability for use in both static and dynamic systems. In particular, the fibre-optic probes can be inserted into the inlet and outlet pipes of tubular SCF reactors for monitoring the phase state of reactants and products with scCO2 and can even be positioned in the centre of a catalyst bed for monitoring phase changes during the course of reactions.
Since the fibre-optic probe is a surface-sensitive device, its behaviour very much mimics that of the shear-mode quartz sensors when they have been used in the static systems (Ke et al. 2004). However, when installing the probe in a dynamic system, the ‘noisy’ signals obtained in the biphasic region give a clear indication of phase separation (Wu et al. 2006a). Figure 5 shows examples of the signals obtained at the bubble, dew and critical points for a mixture of CO2+toluene. Clearly, the signal jumps up at the bubble point, whereas it jumps down at the dew point when the system pressure is decreased below the phase separation pressure at a constant temperature. The size of the jump is small near the critical point, reflecting the fact that the refractive indices of the gas and liquid phases are almost identical under the conditions. The accuracy of the FOR method depends largely on the temperature and pressure stability of the apparatus. It is between ±0.2 and ±1 K for temperature, and ±1 bar for pressure, respectively, when a dynamic system with a BPR was used. It is worth pointing out that it takes about 20–40 min to measure each point on the phase boundary, which is far quicker than most conventional methods, especially for measuring data at high temperatures.
Apart from measuring phase boundaries, we have also demonstrated that the FOR method can be used for monitoring the volumetric expansion of organic solvents with scCO2 by immersing the end of the optic fibre into liquid (Avdeev et al. 2004). Recent work by Dudukovic and his co-workers (Mueller et al. 2007) has also reported a fibre-optic approach for measuring volumetric expansion, in which a multi-mode optical fibre is employed as a position sensor for monitoring the position of the meniscus between the liquid and the CO2 phase.
The FOR method has been applied to several systems of CO2+organics (e.g. MeOH, toluene, γ-valerolactone, etc.; Licence et al. 2005; Wu et al. 2006b; Hou et al. 2009) over a broad temperature, pressure and composition range, demonstrating that the FOR is a simple, rapid and versatile method for studying the phase behaviour of reaction mixtures in scCO2. However, the germanium-doped core of standard optic fibres was rapidly corroded by sub- and supercritical water, making measurements impossible in such mixtures. This problem can be partially overcome by using a fluorine-doped fibre, the corrosion rates of which are sufficiently slow for it to be usable for phase measurements (Bagratashvili et al. 2009).
6. The ‘holey’ fibre
Like the FOR, the holey fibre method exploits the small diameter of optic fibres and is targeted at mixtures containing high-temperature high-pressure (HTHP) water where the FOR is thwarted by corrosion (see above). Instead of optical properties, this method exploits the very narrow hollow channels that run along the length of so-called holey fibres (Monro & Ebendorff-Heidepriem 2006). These channels allow the fibre to be used as a device for withdrawing minute samples from the fluid mixture.
The method can only be used in a dynamic regime with a mixture flowing continuously through a system. The measuring chamber (figure 6) is very similar to one in the FOR. The position of the holey fibre end is very important and has to be 1–1.5 cm below the ‘nozzle’ in the measuring chamber. This design ensures that a fresh portion of mixture directly contacts the holey fibre and is sampled. The free end of the fibre is attached to a standard flame ionization detector (FID) as commonly used in gas chromatographs. The FID signal responds to the amount of organic compounds in the injected stream. Unlike the other methods described above, the holey fibre method is entirely empirical—it does not rely on any fundamental property of the fluid, apart from the flammability of at least one organic component in the mixture. The key point is that, because of the very small amount of fluid drawn into the fibre, the FID signal is much noisier when the fibre samples a multi-phase system than when it is fed with a homogeneous gas or liquid phase. Thus, an abrupt increase or decrease in noise in the FID signal can be used to locate a phase transition (figure 7; Novitskiy et al. 2009).
The method had been validated with CO2+acetone and with H2O + EtOH (figure 7); the results correlate well with literature data. The narrowness of the channels, typically 5–20 μm, means that the phase equilibrium is undisturbed because a negligible amount of material is withdrawn. However, the narrowness does render the channels liable to blockage, particularly if the mixture contains components such as sugars that can decompose into insoluble char.
7. ATR infrared spectroscopy
As explained above, there is an inherent difficulty in studying the phase behaviour of reaction mixtures containing near- and supercritical water because of the high temperature, high pressure and corrosive nature of water under these conditions. Attenuated total reflection infrared (ATR-IR) may provide an excellent solution to this problem. Unlike transmission IR, which requires two large windows placed in opposite sides of an IR cell, ATR-IR needs only a single window, which can be made of diamond with its exceptional physical and chemical stability. We use a commercial type IIa natural diamond bonded into a tungsten carbide mount to ensure the high reliability and safety.
Our high-pressure ATR-IR cell is based on the Specac Golden Gate SCFs analyser, but with modifications to the geometry of the top plate and the heating system to give the uniformity and stability needed for phase equilibrium measurements up to 650 K and 250 bar. The ATR-IR cell is incorporated into a dynamic system operated in a similar manner to the FOR and holey fibre methods.
With ATR, the IR beam penetrates the medium in contact with the surface of the diamond to a depth equal to half the wavelength of the IR beam. Hence, like the shear-mode sensor or the FOR, the ATR crystal interrogates the sample in immediate contact with its surface, but the penetration depth increases at lower wavenumbers, the so-called fingerprint region of the spectrum (Harrick 1967). Liquid samples give relatively strong absorptions because they are dense while gas samples have very weak absorptions. Therefore, when a gas → liquid phase transition occurs, the IR peak intensities rise sharply.
In contrast to the other methods described here, ATR-IR also gives compositional information of the sample. This has an unexpected consequence for phase behaviour because, using H2O + EtOH, we have demonstrated that this information can be used to locate bubble points (figure 8), which shows three spectra corresponding respectively to the gas, biphasic and liquid regions of phase space. When the pressure is low, the peak heights are close to zero after the background correction, spectrum 1. As the pressure rises, the peak heights start to increase, and when system reaches the dew point, peaks rise dramatically (spectrum 2) indicating that a liquid phase has formed on the surface of the ATR crystal. Increasing pressure above the dew point does not cause any further dramatic changes in the spectra because the surface of the crystal is always in contact with liquid, whether there is a homogenous liquid or a biphasic fluid in the ATR cell. However, the composition of the liquid changes as the two phase region is traversed and these changes are registered in the relative intensities of the IR bands of H2O and EtOH. Once the system crosses the bubble-point curve, the composition stops changing (e.g. spectrum 3 in figure 8a) because the system is a single liquid phase (see the inset in figure 8a).
The ATR-IR approach has been applied to obtain whole p–x phase diagrams of the H2O + EtOH system. It shows good agreement with literature values by using the isothermal method to measure a series of the mixtures with different compositions (figure 8b). Thus, we have the surprising result that one can detect both the dew and bubble points with a single ATR crystal positioned at the bottom of the cell even though the vapour phase forms at the top of the cell.
8. Pressure drop method
All of the methods described so far have involved synthetic mixtures which, even if multi-component, are less complicated than real reaction mixtures that change their composition as the reaction proceeds. If phase behaviour measurements are needed for a specific application, e.g. a reaction, or for particle formation, then the most efficient way to measure the phase behaviour would be to do so in situ so that no additional time would be required to determine the phase behaviour. Furthermore, if one can make the detection method objective, it could be automated, thereby allowing many monotonous measurements to be carried out unattended. The pressure drop method satisfies the criteria both of in operando detection and of objectivity.
When a fluid flows through a packed bed of spheres, energy is dissipated due to friction, generating a pressure gradient along the length of the packed bed. The pressure drop method is conceptually very simple. CO2 and another substrate are combined and mixed, and the pressure is measured both before and after a packed column. By monitoring the pressure drop as a function of either temperature, pressure or composition, it is possible to observe distinct changes in the pressure drop (e.g. figure 9; Akien et al. 2010). These changes identify a phase transition introduced by alteration of temperature, pressure and composition of mixtures which changes the superficial velocities and viscosities of fluids, and hence the flow patterns, pressure drops, liquid holdups, etc. The pressure drops in the two-phase flow regions shown in figure 9 are essentially larger than those in the single-phase flow region under similar conditions because of the interactions between the gas and liquid phases, e.g. surface tension.
Since the changes in the pressure drop at phase transitions are quite well-defined, they can be determined computationally, and so the measurement process can be automated, since the pressure, temperature and flow rates of both CO2 and substrate can all be controlled via a computer. Although the sensitivity towards dew points is lower at high temperatures, bubble points are very easily determined, complementing the abilities of the FOR method well. In fact, the two methods can be combined in a single experiment.
Almost uniquely, it is also possible to measure the phase behaviour of a reaction in operando, by measuring the pressure drop across the heterogeneous catalyst itself, as shown in figure 10. Thus, our initial studies demonstrate that the pressure drop method is a convenient method for determining phase transitions in continuous flow, either for automated measurements of non-reacting systems or for measurements of reacting mixtures.
Table 1 summarizes the key features of the different methods described in this review for locating phase boundaries. These methods are intended to complement more traditional methods rather than to supplant them for the study of reaction mixtures. The range of such mixtures is so broad that no single method is suitable for all of them. The common features of our methods are that all of them are objective in the sense that they can be used in opaque apparatus and that the phase transitions are located from the output of sensors rather than visually. Most of the methods are rapid compared with more conventional techniques.
Although the two-phase flow through packed beds is a widely studied area of chemical engineering (Kleinstreuer 2003), studies on reaction mixtures in the near- and supercritical conditions have remained unexplored until recently (Akien et al. 2010), and many fundamental aspects, such as the relationship between viscosity, density and flow patterns, require further detailed studies. By linking studies of reactions and phase behaviour, we believe that one will be able to increase an understanding of both the science and engineering that will underpin the more sustainable production of chemicals and materials.
We thank the EPSRC, TSB, the Russian Foundation for Basic Research, The Royal Society, The Paul Instrument Fund, Thomas Swan & Co Ltd, INVISTA Performance Technologies, AstraZeneca and the University of Nottingham for support. M.W.G. gratefully acknowledges receipt of a Royal Society Wolfson Merit award. We thank Professors B. Han, H. Yan, P. J. King and A. Aguiar-Ricardo, Drs A. Kordikowski, R. Oag, V. K. Popov, M. Sokolova, M. V. Avdeev, W. Wu and all our colleagues who have assisted in the development of these new techniques, and M. Guyler, R. Wilson, P. Fields, D. Litchfield and J. Warren for technical support.
One contribution to the 2010 Anniversary Series ‘A collection of reviews celebrating the Royal Society’s 350th Anniversary’.
- Received May 24, 2010.
- Accepted June 30, 2010.
- © 2010 The Royal Society