University of Birmingham Ultrasound-induced emulsification of subcritical carbon dioxide/water with and without surfactant as a strategy for enhanced mass transport

Pulsed ultrasound was used to disperse a biphasic mixture of CO 2 /H 2 O in a 1 dm 3 high-pressure reactor at 30 (cid:2) C/80 bar. A view cell positioned in-line with the sonic vessel allowed observation of a turbid emulsion which lasted approximately 30 min after ceasing sonication. Within the ultrasound reactor, simultaneous CO 2 -continuous and H 2 O-continuous environments were identiﬁed. The hydrolysis of benzoyl chloride was employed to show that at similar power intensities, comparable initial rates (1.6 ± 0.3 (cid:2) 10 –3 s –1 at 95 W cm –2 ) were obtained with those reported for a 87 cm 3 reactor (1.8 ± 0.2 (cid:2) 10 –3 s –1 at 105 W cm –2 ), demonstrating the conservation of the physical effects of ultrasound in high-pressure systems (emulsiﬁcation induced by the action of acoustic forces near an interface). A comparison of benzoyl chloride hydrolysis rates and benzaldehyde mass transport relative to the non-sonicated, ‘silent’ cases conﬁrmed that the application of ultrasound achieved reaction rates which were over 200 times faster, by reducing the mass transport resistance between CO 2 and H 2 O. The versatility of the system was further demonstrated by ultrasound-induced hydrolysis in the presence of the polysorbate surfactant, Tween, which formed a more uniform CO 2 /H 2 O emulsion that signiﬁcantly increased benzoyl chloride hydrolysis rates. Finally, pulse rate was employed as a means of slowing down the rate of hydrolysis, further illustrating how ultrasound can be used as a valuable tool for controlling reactions in CO 2 /H 2 O solvent mixtures.


Introduction
One way of improving the separation between immiscible species following synthetic procedures, is to use a biphasic system involving an organic solvent and an aqueous phase [1,2].Owing to its low critical point (31.1 °C/73.8bar) and ease of separation upon depressurisation, subcritical or supercritical CO 2 may be employed usefully as a replacement for the organic phase [3,4].Moreover, as the solvating power of CO 2 is confined to small, relatively non-polar compounds, the presence of a second aqueous phase can greatly increase the accessibility of hydrophilic species (e.g.salts), that are commonly employed in many organic chemistry reactions.Additionally, the reaction of CO 2 with water leads to the formation of carbonic acid, providing an intrinsic proton source (pH 3-4) [5,6], which may be advantageous for several processes [7].
Due to the presence of a liquid-liquid interfacial barrier, mass transfer rates in biphasic systems may be improved by increasing the interfacial area through the formation of an emulsion [8].Emulsions of CO 2 /H 2 O (C/W) have the distinct advantage over conventional organic-phase emulsions of allowing the complete reversal to the biphase upon depressurisation of the system.Furthermore, the droplet size and stability of the emulsion may be controlled by specially-designed surfactant molecules allowing greater fine-tuning of the reaction system [9].Dispersions of CO 2 and water, as emulsions or microemulsions [10], have been developed for a range of synthetic processes, including hydrogenations [11][12][13][14], hydroformylations [15], the synthesis of metal nanoparticles [16], and polymerisations [17].Due to its favourable properties, pulsed ultrasound may also be used to generate emulsions and therefore increase the rate of reactions.Ultrasound may also be considered as a tool for green chemistry, by allowing the use of benign conditions and enhancing energy efficiency [18].The use of ultrasound in high-pressure systems has been reported for several extraction processes in CO 2 [19][20][21].
A previous study by Timko et al. investigated the hydrolysis of benzoyl chloride in a CO 2 /H 2 O biphase under ultrasonication in a reactor of 87.2 cm 3 volume [22].In the present study, we have designed a 1 dm 3 high-pressure ultrasound reactor and used this Contents lists available at SciVerse ScienceDirect Ultrasonics Sonochemistry j o u r n a l h o m e p a g e : w w w .e l s e v i e r .c o m / l o c a t e / u l t s o n to emulsify a dense CO 2 /H 2 O solvent system.Scaling up the hydrolysis of benzoyl chloride, we show that the efficiency is not lost at this scale-up factor, highlighting the conservation of the positive physical effects of ultrasound.Emulsion formation and subsequent breakdown was observed by passing the emulsions through a 30 cm 3 view cell and monitoring the change in turbidity over time.The advantages of ultrasound were further demonstrated for the hydrolysis reaction in the presence of Tween 80, a bio-compatible surfactant, able to stabilise a uniform C/W emulsion with estimated surface areas greater than 480 m 2 kg -1 .The combined effects of ultrasound and Tween 80 resulted in faster reaction rates than those reported for a pre-formed H 2 O in CO 2 (W/C) microemulsion [23], with higher energy efficiency than without the use of surfactant.Overall, the results demonstrate the potential utility of this reactor methodology for industrial applications.

System configuration (Fig. 1)
A high-pressure, cylindrical 1 dm 3 stainless-steel vessel (Parr, USA I.D. = 9.5 cm, H = 16 cm) was designed to incorporate an ultrasound probe and manufactured for use in the present study.CO 2 was delivered to the vessel via a high-pressure pump (Thar Instruments), electronically connected to a PC.The amount of CO 2 entering the reactor was recorded via a mass flow meter positioned inline before the pump.The vessel was insulated by an electric heating jacket, and the temperature was controlled by a temperature controller (Parr 4838) coupled to a J type thermocouple, positioned in an immersed thermowell.The pressure was recorded to within ±0.1 bar using a transducer (Druck PTX 1400; Druck Ltd, Leicester, UK) with dedicated display.Ultrasound was introduced from a 20 kHz VCX-1500W processor (Sonics & Materials, CT, USA), coupled with a titanium alloy horn of tip diameter 2 cm.Tip amplitude (20-100%) was manually selected and related to the energy transferred to the fluid, while the pulse rate could define the ratio of ultrasound ON/ultrasound OFF based on a sonication cycle (e.g.25%, 1 s ON/3 s OFF).A calibration between delivered and dissipated power was undertaken using the calorimetric method [24].Power density values reported here (W cm -3 ) were obtained by dividing the total energy per pulse time (J s -1 ) by the vessel volume.When comparing vessels of different sizes, power intensity (W cm -2 ) may also be encountered.Here power intensity values were calculated by dividing the power by the irradiating surface area of the tip.As a consequence of high amplitude settings resulting in a rapid $3 °C rise in the temperature of the vessel, water cooling was employed to keep the vessel temperature constant during experiments.A dual piston HPLC pump (Jasco PU-2087) with a maximum flow rate of 20 cm 3 min -1 was used to recirculate the vessel contents.Connections to the vessel were made by 1/4 00 and 1/8 00 stainless steel Swagelok tubing, while connections to the pump were made in 1/16 00 tubing.

General method for sonication of the CO 2 /H 2 O biphase
The vessel was initially heated until the desired temperature (30 °C) was reached.Water was then added to give a volume be-tween 400 and 650 mL.CO 2 was supplied to the sealed vessel at a flow rate of $40 g min -1 , until the desired pressure was reached.Following 30 min of equilibration/recirculation, sonication was initiated and continued for a known time.

Observation of emulsion stability
A 30 cm 3 high-pressure stainless-steel view cell I.D. = 2.5 cm, H = 5.7 cm (Parr Instruments, Moline, IL, USA) fitted with two inline sapphire windows (D = 1.5 cm), was used to visualise the dispersion generated in the ultrasound reactor (Fig. 1F), at 30 °C/ 80 bar.Heating was provided by two electric heating plates connected to a dedicated controller, and monitored by an immersed J type thermocouple.Emulsion stability was assessed by measuring the increase in light intensity (Visilight I-Led) over time, following termination of a 10-min sonication/circulation period.Time-lapse recordings were carried out using a Photron high-speed camera at 10-s time intervals, and the digital images transferred to a PC, using a dedicated interface.The images were analysed with ImageJ for the sum of the greyscale pixel values (integrated density) inside a defined area, and each time-point value divided by the maximum density recorded in the absence of turbidity.Experiments were also performed under conditions above the CO 2 critical point in order to minimise any effect associated with phase transition.

Compound injection, sampling and analysis
For greater precision, hydrophobic compounds were added to the CO 2 phase via a 6-port valve positioned in-line with the CO 2 feed, and connected by 1/16 00 Swagelok tubing.A standard 1 mL bypass loop was used to inject a known amount of each compound.Following equilibration of the reactor contents at 30 °C/70 bar, either benzaldehyde or benzoyl chloride was injected into the flowing CO 2 feed, while final pressurisation to 80 bar was achieved.Aliquots (2 mL) were taken from the recirculating H 2 O line at the specified time intervals and analysed immediately by gas chromatography (Shimadzu GC-2010, combined with autosampler and injector AOC-20, and equipped with a DB-5 column and an FID).Four or five aliquots were taken over the course of each experiment at different time points.For the same set of conditions, experiments were repeated three times, generating aggregated data of non-overlapping time points.The final concentration was obtained from a standard calibration line.The H 2 O recirculating line originated from a 14-cm 1/4 00 stainless steel dip tube immersed in the reactor.The sampling point was therefore positioned 8 cm below the probe tip.Aliquots from the CO 2 phase on the other hand were taken from a line originating from a port at the top of the reactor, and were collected by slowly bubbling into cold acetone.In order to verify that all of the compounds tested were dissolved in the CO 2 phase prior to initiating sonication, visual experiments were conducted by injecting 100 lL of compound into the view cell charged with just CO 2 .At the pressure and temperature described in this paper, only a single phase was observed.

Benzaldehyde mass transport
Mass transport experiments were conducted in order to assess the dispersive capacity of ultrasound for this vessel geometry (1 dm 3 ) using a hydrophobic compound, and to potentially compare mass transport rates with other systems reported in the literature.Benzaldehyde was chosen for the mass transport experiments as its solubility in CO 2 has been studied previously [25,26].Moreover, it contains an important functional group for organic chemistry, namely the carbonyl group.The mass transport of benzaldehyde from CO 2 to H 2 O was assessed by the two-film theory, previously described for similar systems by Timko [27], and Tai et al. [28] 1 .
Due to its influence on the mass transport, the partition coefficient was determined by measuring the concentrations of benzaldehyde partitioned in both phases, following 4 h of equilibration, and the resulting value was found to match to within 5% that reported in the literature (K CW = 12) [29].For comparison with values reported in the literature, the effective mass transport coefficient K mtC a⁄ (s -1 ) was calculated, where 'a⁄' is the surface area per unit volume (cm -1 ).

Kinetics of benzoyl chloride hydrolysis
The hydrolysis of benzoyl chloride represents a useful reaction for modelling kinetics in which a hydrophobic species (the acid chloride) must cross the interface, meet the hydrophilic species (H 2 O), and react (Scheme 1).In the present system, the reverse case is also true, whereby H 2 O travels to the CO 2 side and reacts with the acid chloride.
Initial rates of benzoyl chloride hydrolysis were analysed under amplitude settings of 20-70%, and were found to follow first-order kinetics, as previously reported [23,30]. 1 The reaction was also studied in the presence of the surfactant polysorbate, Tween 80.This surfactant was chosen due to an existing in-depth knowledge of its C/W emulsion properties from the literature [31].To this end, the same procedure was followed as that described above other than 0.5 or 1.0 wt.% of surfactant (with respect to water) was dissolved in the water prior to charging the vessel.Samples were taken at various time points, and the reaction was quenched by trapping the carboxylic acid hydrolysis product in acetone on ice (1:6 dilution by volume) before analysing by GC.As sonication in the presence of Tween resulted in a greater fraction of CO 2 being dispersed in H 2 O compared to when sonication was carried out without Tween, a few considerations had to be made.To investigate whether or not the benzoic acid hydrolysis product had formed in situ as a result of ultrasound mixing, or by mixing during sample depressurisation, the same experimental protocol was carried out with an additional visual aid using the view cell.Following 1 min of sonication, ultrasound was stopped and the contents were recirculated until the emulsion reverted to the biphase.After 1 h, when only a clear solution was observed to pass through the view cell, samples were collected in the usual manner.The sample concentration closely paralleled that obtained from 1 min ongoing sonication, indicating that the benzoic acid had not formed as a consequence of the benzoyl chloride-containing emulsion rapidly breaking down upon depressurisation into the sample vial.The yields were plotted as a function of time at different ultrasound power settings.

Emulsion appearance and stability
Emulsions originating in the H 2 O-continuous phase were assessed first.Approximately 10 min after commencing sonication, turbidity was observed to increase inside the view cell (Fig. 2a  and b).Following the ceasing of sonication, this turbidity remained for approximately 30 min, gradually reverting to the biphase with a characteristic demulsification front (Fig. 2c).Droplets could be observed to collect on the upper part of the inside wall of the window, which increased in size and number over time, indicating that the emulsion was breaking down by CO 2 droplet coalescence.
Turbid emulsions generated by ultrasound were reported as being indicative of droplets with Sauter mean diameters of 9 lm for H 2 O droplets in CO 2 -continuous (W/C), and 15 lm for CO 2 droplets in H 2 O-continuous (C/W) emulsions [22].In our system, turbidity was observed regardless of whether the H 2 O-or CO 2 -continuous phase was recirculated through the view cell during sonication (Fig. 3).This allowed measurement of emulsion stability by assessing the decrease in turbidity in the view cell.The data points for the fastest increase observed in light intensity were plotted and found to fit a second-order polynomial function (r 2 > 0.99, Fig. 4).This was found to closely match the observed time course for emulsion breakdown visualised by eye.Volume fractions of CO 2 in H 2 O were evaluated by the tracer technique, 1 and were found to be between 1% and 6%, depending on the standoff distance of the biphase line from the probe (Fig. 5).Optimal dispersion of CO 2 in H 2 O was achieved when the biphase line was positioned between 1 cm and 2 cm below the probe surface.This observation was consistent with previous studies in liquids at ambient conditions that showed cavitation bubble velocity and bubble flow rate decrease dramatically in intensity at distances of more than 4 cm below the probe [32].Emulsification may also result in H 2 O droplets dispersing in the CO 2 phase.As a consequence, water collected from the CO 2 -continuous phase revealed the H 2 O volume fraction to be 6.5% (± 1) following sonication.

Mass transport
Following injection of 1 mL benzaldehyde, sample analysis revealed that its concentration measured in the upper part of the reactor during sonication was roughly ten times higher than that in the lower part.Together with volume fraction measurements from the previous section, the results indicated that two distinct environments existed simultaneously during sonication, one CO 2 -continuous and one H 2 O-continuous in the upper and lower parts of the reactor, respectively.The change of benzaldehyde concentration over time under only circulation ('silent') and under sonication conditions, is presented in Fig. 6.The equilibrium for benzaldehyde mass transport was reached within approximately 2 h under only circulation conditions, whereas it was reached within 1 min following the start of sonication.Analysis of the log plot yielded slopes of 2.7 and 0.023 Â 10 -2 s -1 for the sonicated and 'silent' case, respectively.Insertion of known parameters (V W = 650 cm 3 , V C = 350 cm 3 and K CW = 12), yielded respective K mtC a ⁄ values of 1.9 Â 10 -3 s -1 and %0.016 Â 10 -3 s -1 , and calculation of their ratio afforded a sonic enhancement factor of $120.The value for the sonicated case was within the range for the transfer of most species in liquid-liquid mixtures.Tai et al. studied the impeller-induced mass transport of zinc(II) ions in a 1.3 dm 3 batch with a CO 2 /H 2 O biphase, reporting values of 0.2-2.6Â 10 -3 s -1 at impeller speeds of 400-1000 rpm at 40 °C/83 bar [28].Mass transport is furthermore highly dependent on the type of contactor and agitation employed.For example, the effective mass transport coefficient ranged between 0.016 and 1.6 Â 10 -3 s -1 for a stirred tank, or as high as 30.6 s -1 in a microchannel [33].In a different study, the effective gas-liquid mass transfer coefficient increased from $0.1 to 10 Â 10 -3 s -1 when 500 rpm stirring was complemented with ultrasound in a 1 L vessel [34].Separation of K mtC into its H 2 O and CO 2 components 2 (K W and K C respectively) revealed an estimated K W %1.5 ± 0.5 Â 10 -4 cm s -1 .This value was found to be approximately half that reported for a 87.2 cm 3 vessel with a higher interfacial area to volume ratio, under gentle circulation (a/V w = 3.7 higher [22]), and was within an order of magnitude for the transfer of most organic species in water.Hydrophobic compounds such as benzaldehyde are always expected to transfer through CO 2 faster than through H 2 O, highlighting how pulsed ultrasound may be employed to enhance mass transport rates of such species through the interface, by almost two orders of magnitude.  Once K mtc was obtained from K mtc a ⁄ (by inserting 'a' = 71 cm 2 and V W = 650 cm 3 ), kw and kc were derived from the relation: , and solved for K C / K W P 9.

Scale-up feasibility by kinetics of benzoyl chloride hydrolysis
Hydrolysis represents an important reaction and is employed in the preparation of many pharmaceutical compounds [35].A range of amplitude settings was used to study the hydrolysis of benzoyl chloride and for the calculation of the respective rate constants (Fig. 7).
Comparison with a smaller, 87.2 cm 3 reactor reported in the literature revealed that initial rates were similar for both systems at approximately equal power intensities (power intensity here was 95.5 W cm -2 with the 1500 W sonifier and a probe tip surface area of 3.14 cm 2 , whereas for the smaller vessel the intensity was 104 W cm -2 with a 400 W sonifier and a probe tip surface area of 0.5 cm 2 , Table 1, Entries 1, 2).In the study by Timko et al., no sonochemical effects by ultrasound were observed, inferring that faster hydrolysis rates were due to increased interfacial areas between CO 2 and H 2 O.Further demonstration of the conservation of efficiency, was provided by observing an almost equivalent rise in the rate constant over the same range of power densities, for both vessel dimensions (rate constant increased roughly 3 times in the range 0.1-0.3W cm -3 , Fig. 8) [30].
Scale-up of sonochemical processes is usually problematic due to their dependence on localised cavitation events (immediately below the irradiating probe), and wide variation in the energy dissipation in the bulk volume [24].Scale-up operations may also be hindered by the fact that the progressive erosion and shortening of the probe may eventually lead to reduced efficiency and increased process costs [36].However, ultrasound in high-pressure, and especially near-critical, fluids, displays distinct advantages compared to sonication of fluids under ambient conditions, including 'soft' cavitation (therefore limiting erosion of the probe tip), and increased biphasic interfacial disruption (better dispersive and micro-mixing capacity) due to the low surface tension, high vapour density and low liquid density of CO 2 [37,38].Furthermore, in the vicinity of a heterogeneous surface, the acoustic wave leads to the generation of physical and mechanical phenomena such as acoustic streaming and microjetting, which may be beneficial regardless of any sonochemical effect taking place [39,40].As a consequence, two immiscible liquids will generate an emulsion, which may not be solely localised beneath the irradiating surface once formed, but as in the present case, will occupy larger fractions of the vessel volume.Factors governing ultrasound efficiency for heterogeneous systems include distance of the biphase line from the probe (here with 650 cm 3 H 2 O 75 wt.%,this was 1 cm below the ultrasound probe), probe surface area, and vessel geometry.An aluminium foil test [41] revealed that significant erosion took place at 1 cm below the probe, indicating the presence of cavitation events with a circular pattern and diameter ($1.2 cm) approximately equal to that of the probe (Fig. 9).
Ultrasound-induced cavitation in liquid CO 2 has proven to be beneficial in a number of applications, including the preparation of polymers, drug nanoparticles, and ceramic materials [42][43][44].In this system, hydrolysis occurred in two distinct environments, a H 2 O-continuous phase existing below the probe, where benzoyl chloride-containing CO 2 droplets were constantly being formed and reacted with surrounding H 2 O, and a CO 2 -continuous phase in which H 2 O was constantly fluxing to react with benzoyl chloride (Scheme 2).

Table 1
Comparison of benzoyl chloride hydrolysis rates between various emulsion systems.As expected, the reaction proceeded faster under sonication compared to silent conditions, by effectively increasing the CO 2 / H 2 O interfacial area.Sonication with 70% amplitude afforded the maximum yield in 30 min (Fig. 11, dashed lines) 3 .In the absence of sonication, the hydrolysis of benzoyl chloride was extremely slow (t 1/2 % 27 h).The sonic enhancement factor for hydrolysis rates relative to the silent case was therefore almost 230.When assessing reaction rates in binary solvent systems, hydrophobic interactions between the solvent and the reagent must also be taken into account; in this respect, ultrasound would alter the solvent structure and solvation [45], thus affecting the activation energy and entropy values for the reaction.Using different alcohol-water compositions, Salmar et al. showed that the solvent affected the rate constants depending on the alcohol/water fraction used, by hydrophobic stabilisation of either the transition complex or the ground state of the reagent [46].
The results from this part of the study highlight how the physical effects of ultrasound may be applied to include the efficient processing of high-pressure fluids on the litre scale, and may provide scope for identifying specific reaction pathways that may benefit from emulsification of H 2 O-containing heterogeneous systems [47].Furthermore, in this system, the reaction between CO 2 and H 2 O formed a spontaneous acidic environment (pH 6 4), potentially acting as an intrinsic catalyst that tends to neutral pH following CO 2 venting.

Emulsion formation using Tween 80
Sonication in the presence of 1.0 wt.% Tween 80 generated a uniform, white emulsion (Fig. 10).Analysis of samples taken from the upper, CO 2 -continuous section of the reactor, following injection of 1 mL octane (a hydrophobic tracer that preferentially partitions in CO 2 [29]), showed a 2.5-fold increase in the concentration of octane during sonication that reverted to the biphase when sonication was ceased.
Additionally, the quantity of water present in these samples disappeared gradually following termination of ultrasound.Quantification of the water mass collected, in combination with octane concentrations, suggested that the emulsion consisted of 35 ± 2% CO 2 and 70 ± 5% H 2 O by volume, which gradually reverted to a separated biphase approximately 1 h after ceasing sonication.Comparison of samples from the upper and lower sections of the reactor showed similar compositions, indicating uniformity throughout the reactor.
The hydrolysis reaction was next assessed in the presence of Tween.
Studies revealed that a combination of ultrasound and Tween was an effective environment for producing fast conversions of benzoyl chloride to benzoic acid (Fig. 11, solid lines).
Tween possesses an amphiphilic structure, and has been shown to stabilise C/W emulsions exclusively (even at 90 wt.%CO 2 for concentrated emulsions), with mean droplet diameters of 0.5-5.4lm depending on the shear rates through a sand pack [31].
Confirmation of the C/W nature of the emulsion was obtained by injecting benzoyl chloride into the pre-formed emulsion, without further sonication, and analysing samples from both the upper and lower sections of the reactor.If the emulsion proved to be CO 2 -continuous (W/C), substantial dissolution of either benzoyl chloride, benzoic acid, or both, would have occurred.The samples, however, contained neither species implying that following injection, benzoyl chloride encountered a H 2 O-continuous phase, and due to its density (1.21 g cm -3 ), resided at the bottom of the vessel without reacting.Tween-stabilised hydrolysis rates under ongoing sonication showed two distinct stages, the first achieving $80% yield almost instantly, and a second showing slower conversion, reaching 100% yield after approximately 30 min.The rate of the Tween-stabilised hydrolysis could be controlled by manipulating the pulse rate.Pulsed ultrasound has distinct advantages over continuous ultrasound, such as an increase in the number of spatial reactive sites and residual pressure amplitude during the OFF stage of the cycle, and enhanced energy efficiency by decreased power draw [48,49].Pulse rate may therefore be regarded as a valuable variable for process optimisation.Using 10% pulse rate and 20% amplitude, 100% yield of benzoic acid was reached within $120 min, significantly slower than when higher pulse rates were   employed.Observation through the view cell revealed that turbidity was of a lower intensity and lagged by several minutes in reaching a maximum, suggesting that the emulsion consisted of a lower surface area during low pulse rate.At steady-state, assuming the highest mean CO 2 droplet diameter of 5 lm (as reported under conditions of shearing), and 35% volume of the dispersed phase, the total emulsion interfacial area was estimated4 at 420 m 2 or 480 m 2 kg -1 .Under the same conditions without Tween, surface areas for ultrasound-generated C/W and W/C emulsions were reported as 20 and 100 m 2 kg -1 , respectively, making the Tween-stabilised emulsion surface area at least four times higher [22].However, surface area may not independently account for the large increase observed in hydrolysis rates.Additional ultrasound-induced effects such as convective forces on existing droplets, may define mass transport and reaction rates, with droplet size playing an important factor in droplet dynamics.In support of this, benzoyl chloride hydrolysis in a pre-formed W/C emulsion was much slower (Table 1, Entry 4).
Another advantage of using a surfactant is improved energy efficiency, as demonstrated by experiments using different pulse rates.After 5 min of sonication at 20% amplitude, inclusion of surfactant yielded a fivefold higher concentration in benzoic acid with a 2.5-times lower energy requirement (2700 J with 10% pulse rate compared to 6750 J with 25% pulse rate for sonication with and without surfactant, respectively).A combination of low pulse rate and surfactant may therefore offer a useful processing strategy for minimising energy expenditure without compromising the yield.

Conclusions
This study highlights two important aspects regarding the utilisation of ultrasound in high-pressure systems.First, processes that rely on the physical effects of sonication and lead to dispersion of heterogeneous phases may be safely scaled-up (over 11-fold in the present study) without loss of efficiency, due to less dependence on cavitation and the sonolytic effect which may quickly dissipate from the irradiating surface in larger vessels.Second, the effect of ultrasound may be complemented by including low quantities of a bio-compatible surfactant (as low as 0.5 wt.%), resulting in complete emulsification of the biphasic mixture, which maintains a defined C/W nature and uniformity throughout the reactor.It is therefore easy to envisage how the nature of the environment may be best suited to different applications either: (i) ongoing reaction/separation (without Tween two distinct environments co-exist, allowing preferential partitioning in the CO 2 and H 2 O phases in real time), or (ii) complete emulsification forming only one type of environment (with Tween, achieving greater uniformity and surface area).

Fig. 1 .
Fig. 1.Schematic showing the experimental setup utilised for sonication of CO 2 / H 2 O in the 1 dm 3 reactor A: CO 2 pump; B: high-pressure valve; C: 3-way sampling valve; D: circulation pump; E: 6-port valve for injecting compounds into the CO 2 stream; F: bypass of the H 2 O recirculating line for visualisation through the 30 cm 3 view cell.

Fig. 3 .Fig. 4 .
Fig. 3. Darkening of the recirculated CO 2 -continuous phase through the view cell, following sonication in the 1 dm 3 high-pressure vessel: (a) for image analysis experiments, a light source was located at the rear window to illuminate the contents of the cell; (b) approximately 3 min after the start of sonication, a turbid phase was seen in the cell; (c) after 10 min of sonication/circulation, complete opacity was observed; (d) 5 min after stopping sonication, light transmittance was partially restored due to reversal to the biphase.Conditions: 30 °C/80 bar, 70% amplitude, 25% pulse rate.

Fig. 5 .Fig. 6 .
Fig. 5. Volume fractions of CO 2 dispersed in H 2 O following sonication.The standoff distance signifies the position of the biphase line with respect to the probe.The negative values denote conditions in which the biphase line was above the probe surface (i.e.probe immersed in H 2 O).The position of the biphase line was adjusted by changing the initial H 2 O volume in the reactor, which was (from left to right): 800, 700, 650, 630, 600 and 580 mL H 2 O.The number of experimental points for each standoff distance was between 3 and 7.No tracer was detected in the H 2 O phase after 3 h of only recirculation.Conditions: 30°/80 bar, 70% amplitude, 25% pulse rate.

Scheme 2 .
Scheme 2. Simplified diagram showing the simultaneous formation of CO 2 in H 2 O and H 2 O in CO 2 environments by ultrasound (left).The reaction may proceed in both environments, by the hydrophobic compound encountering a water droplet in the bulk CO 2 phase, or by the compound-carrying CO 2 droplet interacting with the bulk water phase (right, BC = benzoyl chloride).