#### Abstract

Absorption of pure oxygen into aqueous emulsions of n-heptane, n-dodecane, and n-hexadecane, respectively, has been studied at 0 to 100% oil volume fraction in a stirred tank at the stirring speed of 1000βmin^{β1}. The volumetric mass transfer coefficient, , was evaluated from the pressure decrease under isochoric and isothermal (298.2βK) conditions. The O/W emulsions of both n-dodecane and n-hexadecane show a maximum at 1-2% oil fraction as reported in several previous studies. Much stronger effects never reported before were observed at high oil fractions. Particularly, all n-heptane emulsions showed higher mass-transfer coefficients than both of the pure phases. The increase is by upto a factor of 38 as compared to pure water at 50% n-heptane. The effect is tentatively interpreted by oil spreading on the bubble surface enabled by a high spreading coefficient. In W/O emulsions of n-heptane and n-dodecane increases with the dispersed water volume fraction; the reason for this surprising trend is not clear.

#### 1. Introduction

Oxygen absorption into emulsions is encountered, for example, in fermentations with an oil as the carbon source. Most studies have been carried out at low oil volume fractions typical for this application [1, 2]. Literature data on the effect of n-alkane addition on the volumetric mass transfer coefficient for oxygen are illustrated in Figure 1. The value has been reported to increase [3, 4], decrease [5, 6], or remain unaffected [7] compared to the one in water. Da Silva et al*.* [8] reported that 1% n-hexadecane or n-dodecane increased in a stirred tank by factors of 1.68 and 1.36, respectively; Kundu et al*. *[3] found that addition of 1% n-dodecane or n-heptane could enhance oxygen transfer in a bubble column up to fourfold; Jia et al*.* [9] also found a fourfold increase by 2% soybean oil in an air-lift reactor. Among other factors, the oil spreading coefficient *S* (1) has been used to explain the differences between the oils as follows:
The values of spreading coefficients *S* reported in the literature for the three n-alkanes used in this study, n-heptane, n-dodecane, and n-hexadecane, differ considerably (Table 1). In the present work, absorption of pure oxygen into aqueous emulsions of these n-alkanes has been studied in the full range of oil volume fraction (0 to 100%) with a barometric technique. The main advantage of the pressure technique is that it can give information on both and oxygen solubility. At high-oil volume fractions, not considered in the previous studies, the high oxygen solubilities in the oils (high driving force) should have a strong effect on the mass transfer characteristics.

#### 2. Experimental

##### 2.1. Chemicals

The oxygen gas had a purity of 99.999% (Westfalen AG, Germany). The emulsion (liquid phase) was composed of double-distilled water and n-heptane, n-dodecane, or n-hexadecane, respectively, with purities β₯99% (Merck, Germany).

##### 2.2. Spreading Coefficient

The spreading coefficient* S* was calculated based on the values of the surface and interfacial tensions measured at 298.2βK with a tensiometer (K11, KrΓΌss, Germany) using the ring method [10]. The measurements were carried out with mutually saturated liquid phases. The reported surface tensions are mean values of 3 measurements (maximum relative standard deviation 0.25%) and the interfacial tensions are mean values of 4 measurements (maximum relative standard deviation 4%). It should be noted that the aim was to understand the behavior of the oils used in this study; therefore, the oils were used without purification.

The evaluated spreading coefficients *S* are listed in Table 2. Different from the literature data (Table 1), all three oils have positive *S* values, but the value for n-heptane is noticeably higher than those for n-dodecane and n-hexadecane.

##### 2.3. Volumetric Mass Transfer Coefficient and Oxygen Solubility

The mass transfer experiments were carried out in a stirred glass vessel with a total volume of 1830βmL and an inner diameter of 100βmm. A schematic diagram of the experimental setup is shown in Figure 2. The stirrer shaft was equipped with two pitched-blade PTFE stirrers (50βmm diameter, 4 blades) mounted at 20βmm (liquid phase) and 140βmm height (gas phase). Four baffles with a width of 10βmm were mounted to the walls. The reactor was jacketed and the temperature was controlled to βK with a thermostat. All experiments were carried out under surface aeration with gas entrainment into bubbles at the same high stirring speed (1000βmin^{β1}).

The volumetric mass transfer coefficient () was evaluated from the pressure decrease measured during batchwise saturation of the initially gas-free emulsion under isothermal and isochoric conditions. To this end, the desired mixture of oil and water (total volume 700βmL) was placed in the reactor and degassed by applying vacuum under stirring. The liquid losses during degassing were determined using a cold trap with liquid nitrogen as the coolant. About 1.3βmL n-heptane loss was compensated by adding the respective volume in excess, whereas the losses were negligible for n-dodecane and n-hexadecane.

After degassing, the stirrer was stopped, and the reactor was pressurized in about 10βs to approximately 0.15βMPa with pure oxygen. After about 20βs, when the pressure reading had become constant, the absorption experiment was started by switching on the agitator. The time course of the pressure decrease in the head space was measured with a pressure transducer (0β0.5βMPa 0.05%). A typical experimental record is shown in Figure 3.

The volumetric mass transfer coefficient was evaluated based on a pseudohomogenous model, that is, the oil-water emulsion was treated as a pseudohomogenous phase with mean physicochemical properties, specifically, mean gas solubility. This allowed determining the -value from the decrease of pressure from initially to finally (see Figure 2) using (2) suggested by Albal et al*.* [16]:
Here and are the partial pressures of water and oil, respectively. The data at 15% to 90% saturation was used to evaluate in order to exclude the initial re-emulsification phase as well as the final phase with low driving force. The reported values are mean values of at least 5 measurements; the mean relative standard deviation was 8% for n-dodecane and n-heptane emulsions, and 7% for n-hexadecane emulsions.

From the total pressure decrease, *ΞP*, the overall oxygen solubility in the emulsion at a partial pressure of 101325βPa can be calculated as follows:
For the oxygen solubility, the mean relative standard deviation was always less than 5%.

#### 3. Results and Discussion

##### 3.1. Effect of Oil Volume Fraction on **O**_{2} Solubility

_{2}

The oxygen solubilities in the emulsions increases linearly with oil volume fraction (Figure 4). The oxygen solubility decreases with increasing chain length of the n-alkanes. The relative solubility (oxygen solubility in the pure oil compared to pure water) is 11.1, 7.7, and 6.5 for n-heptane, n-dodecane, and n-hexadecane, respectively.

##### 3.2. Phase Inversion

In the previous study on CO_{2} absorption [10], the electrical conductivity method showed the phase inversion taking place in the range of 60% to 65% oil volume fraction for n-dodecane as well as n-hexadecane. For n-heptane, the conductivity variation was rather continuous; observation of deemulsification suggested the inversion region to be at 50%β60% n-heptane.

##### 3.3. Effect of Oil Volume Fraction on

The volumetric mass transfer coefficient in O/W emulsions of n-dodecane and n-hexadecane show similar trends (Figures 5 and 6): As the oil volume fraction increases, first increases to a maximum at a small oil volume fraction and then decreases towards the phase inversion region. The increase in at 1% n-dodecane and at 2% n-hexadecane is by 44% and 26%, respectively. Note that the standard deviation, indicated by error bars, is higher in this range. An increase in at low oil concentrations (1-2%) was previously observed for oxygen by several authors. According to Clarke and Correia [1] hydrocarbons tend to decrease the surface tension of the water phase, resulting in smaller bubble size, therefore higher interfacial area (*a*), and hence higher . However, in this study the surface tension of water (72.0βmNβm^{β1}) and that of water saturated with n-heptane (71.8βmNβm^{β1}) or n-hexadecane (71.2βmNβm^{β1}) were practically the same (cf. Table 2); only saturation with n-dodecane slightly reduced the surface tension (68.2βmN m^{β1}). This means that the βsurface tension effectβ is negligible for the systems studied. Bruining et al. [17] and Cents et al*.* [18] proposed that an additional transport mechanism might be responsible for the maximum at low oil fraction. Small oil droplets with higher solubility for the gas might enter the water film at the G/L interface, absorb the gas, and then return to the bulk water phase where the absorbed gas is discharged. After the maximum, the strong decrease of is quite expected due to the increase of the emulsion viscosity. All trends are similar to those for CO_{2} absorption in the same emulsions [10] but the effects are much more pronounced for oxygen. The main difference between the gases is that the CO_{2} solubilities in the organic liquids are only slightly higher than in water.

The O/W emulsions of n-heptane show quite different trends with increasing oil fraction (Figure 7): strongly increases up to 15% oil fraction and then again from 40% oil towards the phase inversion region. All n-heptane emulsions have higher volumetric mass transfer coefficients than pure water.

The increase in at 15% n-heptane volume fraction (+336% as compared to water) can probably be explained by the bubble-covering mechanism proposed by Rols et al*.* [19]. This mechanism is linked to the spreading coefficient *S*. Pinho and Alves [11] recently concluded that there is direct contact between gas and oil when , and very high mass transfer coefficients can be expected. The high *S* value for n-heptane (Table 2) might enable n-heptane to spread as a film on the oxygen bubbles. When a bubble finally breaks at the top surface, small oxygen-rich droplets may leave the oil film and release oxygen to the water phase. It should be noted that n-heptane has the highest oxygen solubility of the three alkanes used and the value in pure n-heptane is remarkably higher than that in pure water (+358%), whereas the values in pure n-dodecane and in pure n-hexadecane are similar to the one in pure water. This means that if n-dodecane or n-hexadecane could also spread as a thin film on the oxygen bubbles, the mechanism would be less effective. The differences of the values in the pure oils mainly result from the viscosities [20] of only 0.39βmPa s for n-heptane and 1.38βmPa s and 3.03βmPa s for n-dodecane and n-hexadecane, respectively.

In W/O emulsions, from 100% n-hexadecane down to the phase inversion region, monotonously decreases with increasing dispersed water fraction. This trend is expected due to the increase in viscosity of the W/O emulsions with increasing water content [21]. Very surprisingly, for both n-heptane and n-dodecane, from 100% oil volume fraction down to the phase inversion region, does not decrease but increases substantially! The maximum increase in compared to the pure oil is found at the phase inversion region where it is +101% for n-heptane and even +189% for n-dodecane. (For n-dodecane the jump at the phase-inversion point is by a factor of 7.8). More extensive repetitions than usual have verified these results. To find an explanation, the structure of the W/O emulsion was studied with an endoscopic photographic technique [10, 22, 23]. The endoscopic photoprobe was installed at about half the emulsion height; after about 2.5βmin. of stirring at 1000βrpm, photos were taken at 10βfps (frames per second). Figure 8 is a typical photo taken at 70% n-dodecane volume fraction. In this W/O emulsion, the water drops have a surface-to-volume mean diameter of 1.16βmm (standard deviation: 0.15βmm; count: 1142 droplets). There are some very tiny droplets but there is no multiemulsion (e.g., O/W/O emulsion).

It may be argued that the driving force should be based on the solubility in the continuous phase rather than the mean solubility as assumed in (2). This will shift the values in O/W emulsions upwards (lower driving force) while those in W/O emulsions will be shifted downwards (higher driving force). Nielsen et al. [5] additionally suggested a comparison with the volumetric mass transfer coefficient in the respective continuous phase in the form of the mass transfer enhancement factor *E *as follows:
One might expect *E* to decrease from 1 on both sides to lower values at the phase inversion point due to the viscosity effect. Figures 9 and 10 present the effects of oil volume fraction on the mass transfer enhancement factor *E*. In O/W emulsions of both n-dodecane and n-hexadecane, as the oil fraction increases, at 25% oil fraction maxima are observed. There the mass transfer rate is enhanced by a factor of 2.3 for n-dodecane and a factor of 1.5 for n-hexadecane. At further increase in the oil fraction towards the phase-inversion region, the *E *value for n-dodecane decreases but still shows some mass transfer enhancement, whereas mass transfer retardation is observed above 40% n-hexadecane. This difference reflects the stronger decrease in the case of n-hexadecane (Figure 5) as compared to n-dodecane (Figure 6), for example, at 60% oil fraction is lower than in water by a factor of 0.12 for n-hexadecane and for n-dodecane the factor is only 0.42. In addition, the relative oxygen solubility in n-hexadecane as compared to water is 6.5 whereas it is 7.7 for n-dodecane. Both aspects lead to higher *E* values for n-dodecane (Figure 8).

In the case of O/W emulsions of n-heptane (Figure 10), as the oil volume fraction is increased towards the phase-inversion region, high mass transfer enhancement is always observed. This reflects the high values (Figure 7) as well as the high relative oxygen solubility of 11.1 in n-heptane as compared to water. Mass transfer is enhanced by *E* = 37.7 at the heptane volume fraction of 50%. The trend in O/W emulsions might be explained by the bubble-covering mechanism. However, further research is needed to explain the surprising increase in and *E* with increasing water content (from 100% oil down to the phase inversion region) for W/O emulsions of n-dodecane and n-heptane.

#### 4. Conclusions

Oxygen absorption into aqueous emulsions of n-heptane, n-dodecane, and n-hexadecane has been studied in the full range of oil volume fraction. The oxygen solubility in emulsions of all three n-alkanes increases linearly with the oil volume fraction. All three oils have positive spreading coefficients but a particularly high* S* value is found for n-heptane. In O/W emulsions, maxima at low oil fraction (1-2%) are found for n-dodecane and n-hexadecane. Emulsions of n-heptane always show higher than the pure liquids. This could be explained by a bubble-covering mechanism enabled by high spreading coefficient. In W/O emulsions of both n-heptane and n-dodecane, increases with increasing dispersed water content towards a maximum at the phase-inversion region. This is a quite unexpected trend that still needs to be explained.

#### Nomenclature

Specific interfacial area [] | |

: | Oxygen solubility [kmol ] |

: | Volumetric mass transfer coefficient [] |

: | Mass transfer coefficient [m] |

: | Initial (final) total pressure [Pa] |

: | Partial pressure of oil (water) [Pa] |

: | Gas constant [Pa β] |

: | Spreading coefficient [Nβ] |

: | Temperature [K] |

: | Time [s] |

: | Gas phase volume [] |

: | Emulsion volume [] |

(): | Surface tension of water (oil) [N ] |

: | Oil-water interfacial tension |

Ξ¦oil: | Oil volume fraction [β] |

: | Mass transfer enhancement factor [β]. |

#### Acknowledgment

The authors thank SOPATec (Berlin, Germany) for the chance to use its endoscopic photoprobe. This paper was presented as a poster at the 10th International Conference on Gas/Liquid and Gas/Liquid/Solid Reactor Engineering (GLS10), Braga, Portugal, 26β29 June 2011.