In [22] we conducted supercritical CO₂ extraction of microalgae (nannochloropsis salina) and analyzed the extracts. Atmospheric and vacuum drying was conducted on the setup described in [23] during 14 - 16 hours which resulted in humidity decrease from 68.63% to 3.88%. The microalgae sheets obtained were ground to 0.5 - 1 mm and loaded to the extraction cell of the experimental setup (Figure 2). Extraction was conducted at 313 K, 35 MPa. Extract samples weighing 0.14 - 0.62 g. underwent chromatography-mass-spectrometry analysis at DFS mass- spectrometer [17] . Analysis parameters were as follows: injector temperature 553 K (exposure time 60 min), thermostat initial temperature 393 K, heating rate 6 K/min; carrier gas-helium, flow rate 1 ml/min, flow splitting 1:10; capillary column with immobilized nonpolar phase type DB-5.30 m/0.25 mm/0.25 μ. Electron ionization mass-spectra were registered in the 100 - 1000 amu mass range. Spectrometer mass adjustment was carried out using Autotune OEM software. Component identification was done using NIST-05 specialized mass-spectra library.

Vegetable oils mainly consist of triglycerides of saturated and unsaturated fatty acids. In terms of process modelling, certain stages of proposed biodiesel production flow require reliable phase equilibrium data of oil in the media for supercritical extraction (CO₂) and supercritical transesterification (methanol).

To study phase equilibria of the triolein-methanol and ethyl-EPA-CO₂ binary systems, following the description provided in [24] we built the experimental setup described in Figure 3 equipping it with the P-50 metering pumps supplied by Thar SFC.

The phase equilibrium cell was evacuated, then filled with the solute (triolein or ethyl-EPA). After that the solvent (methanol or CO₂) was fed with themetering pump to the experimental pressure and the equilibrium cell was heated to the experimental temperature. After 60 min stirring the equilibrium cell was put to rest vertically for 90 min.

The upper phase of triolein-methanol system was collected using syringe. The sample volume V_sam. and the volume of the triolein V_tr. separated to the bottom phase of the sample were measured with a 5-ml volumetric tube.

Triolein volume concentration:

Triolein mass concentration:

where ρ_tr. and ρ_met. are the triolein and methanol densities.

Triolein molar concentration:

where μ_tr. and μ_met. are the triolein and methanol molar masses.

Triolein, essentially thealmond oil supplied by Aspera Ltd. (Moscow) had the purity of 90%, the rest falling to trilinolenin and negligible amount of vitamins and proteins. 99% pure methanol was supplied by Kazanorgsintez JSC.

In the ethyl-EPA-CO₂ system the upper phase was collected to a preliminarily weighed balloon submersed in water. Sample volume was defined as the volume of displaced water. After phase separation was complete inside the balloon, it was wiped dry and weighed. The mass of ethyl-EPA dissolved in CO₂ was defined as the mass difference of empty balloon and balloon with ethyl-EPA.

CO₂ mass concentration in the upper phase is defined as:

where is the carbon dioxide density inside the balloon, V_atm is the volume of carbon dioxide under pressure inside the balloon [25] .

Molar concentration of CO₂ in the upper phase is defined as:

where and are the molar masses of ethyl EPA and CO₂.

Ethyl EPA supplied by takeomega3.com was 90% pure, the remaining 10% falling to ethyl docosahexaenoate and tocopherols. 99.5% pure carbon dioxide was supplied by OOO TehGaz.

32-ml batch-type experimental setup was created [26] designed for 60 MPa pressure and 800 K temperature (Figure 4).

The transesterification reaction is conducted in autoclave 1 that is isochorically heated by a muffle furnace 2 to the experimental temperature and pressure. Mounted on a hinge, the furnace can be rotated about horizontal axis to provide stirring inside the high-pressure autoclave. Temperature is measured with chromel-alumel thermocouple whose thermojunction is located centrally inside the autoclave. The thermocouple is preliminarily calibrated with the PTS-10 platinum resistance thermometer. Measurement accuracy is ±5 K. Pressure is measured with Stanley-Korund DI-001 pressure sensor 3 powered by the direct current power supply 5. Pressure and temperature measurements are displayed on TRM-101 indicating gauges 6. Process pressure can be adjusted with the VK-97 high pressure needle valve 4.

The reaction autoclave is 500 mm long, cylinder-shaped, with the external diameter of 26 mm, internal diameter of 12 mm (Figure 5).

The setup embodies the transesterification process designed by Saka and Kusdiana [27] . Transesterification experiments were done with the non-refined palm oil.

Reaction products separation and composition analysis are described in [26] .

In this work the emulsion of initial reagents was obtained using 24 kHz, 200 W Hielscher Ultrasonic UIP200 hd ultrasonic emulsifier [28] .

Emulsion stability by time was assessed visually, by the amount of oil separating from the emulsion inside measuring tank. We called the share of oil that separated from the emulsion as emulsion layering coefficient.

Each of the samples was found out to be a multicomponent mixture consisting primarily of triglycerides and esters of saturated and unsaturated fatty acids. Composition of one of the samples is show in the chromatogram (Figure 6) and in Table 1.

Experimental phase equilibrium data for the triolein-methanol system at 413 K and various pressures are presented in Figure 7.

The phase equilibrium diagram is in satisfactory agreement with the literature data [29] . There seems to be discrepancy at 10 MPa, however this and neighboring isotherms of the same literature work altogether fail to represent a data array without unexplained anomalies which can mean significant measurement error in this measurement scale. Anyway the research has proved very low solubility in sub-critical area of triolein, one of the main components of vegetable oils, in methanol, the other component of the transesterification reaction, which is a significant limiting factor for the reaction.

Phase equilibrium was also studied for the ethyl eicosapentaenoate (ethyl-EPA)-CO₂ system at 313, 323 and 333 K in the 10 - 21 MPa pressure range. Experimental data are presented in Figures 8-10.

Experimental data agree satisfactorily with the literature data, while discrepancies may be explained by differences in the composition of substances studied. In all figures, the experimental curve tends to horizontal, as pressure grows, which can mean full solubility of the components.

Maximum error in measuring phase equilibrium characteristics was calculated as 9.7%.

Experiments on obtaining biodiesel fuel in batch-type setup were conducted in the 563 - 793 K temperature range,

Ethyl 5, 8, 11, 14, 17-eicosapentaenoate, an omega-3 component was found among extracted substances in quantity of 1.75%.

19.7 - 44 MPa pressure range, with methanol-to-oil molar ratios of 304:1 to 6:1.

Non-refined palm oil produced by Seloga Petroleum (Malaysia) from elaeis guineensis crop in 2007 had the following characteristics: humidity and impurities content < 0.1%, free fatty acids < 0.1%, iodine number 50, cloud point 283 K, peroxide number < 1. Refined palm oil branded Zlata Palma had the following characteristics: free fatty acids content < 0.1%, humidity < 0.1%, iodine number 64, cloud point 277 K, saturated fatty acid components 39.7%, monounsaturated fatty acid components 46%, polyunsaturated fatty acids 14.2%. 99% methanol was supplied by Kazanorgsintez JSC.

Reaction yield was defined as:

where V_rme and V_met are the volumes of the raw methyl esters phase and the methanol phase in the sample after reaction.

Figure 11 and Figure 12 compare the experimental data for FAME yield at 10 min reaction time at various temperatures and methanol-to-oil molar ratios to our previously published results [26] and literature data [11] [12]

[31] - [33] .

Figure 13 compares experimental data on the FAME yield of refined palm oil at 10 min reaction time at various temperatures and methanol-to-oil molar ratios with our previously published results [26] and literature data [34] .

Figure 14 features the pressure dependence of reaction yield at constant temperature and molar ratio.

Dependences presented are in qualitative agreement with those in [3] [16] [35] and [36] . At pressures above 20 MPa the influence of temperature on the reaction yield was shown to be more significant than that of pressure which is congruent with the literature results [36] - [38] .

Calculated maximum error in measuring FAME yield is 2.27%.

Experimental data are presented on phase separation dynamics of ultrasonicated emulsions of ethanol and rapeseed oil. Rapeseed oil provided by OAO Nefis-Cosmetics was produced from brassica napus (Tatarstan, harvested in 2008) by GOST 1129-93, acid number < 3, humidity and volatile components content < 0.1%, negligible amount of FAME present. 99% pure ethanol was supplied by Kazanorgsintez JSC.

Emulsion examined had the ethanol-to-oil molar ratio of 150:1 to 7:1 (oil volume concentrations of 0.1 - 0.7). Ultrasonic treatment time was conducted for 30 - 120 s, the treatment power being 100 - 200 W. Emulsions with the ratio 7:1 (volume concentration 0.7) proved to essentially separate very fast, within 1 - 5 minutes with further slowing of separation while emulsions with the ratio of 17:1 (volume concentration 0.5) and higher remained relatively stable for longer time range and separated slower (Figure 15).

Emulsions obtained were examined under microscope to learn the emulsion grain size. Samples with 2.26 - 8.6 μm grain size (Figure 16) were used in transesterification experiments which proved that fatty acid alkyl ester yield is highest when the grain size is smallest (Figure 17).

Experimental data are presented on temperature dependence of reaction yield for non-refined rapeseed oil and methanol mixtures. Molar ratio was 153:1, reaction time, 10 min. Transesterification reactions were carried out in

the same experimental setup (Figure 4). FAME yields for reaction mixtures with and without preliminary ultrasonication are compared in Figure 18.

Ultrasonication proved to significantly increase FAME yield at relatively low reaction temperatures (563 K) compared to reaction mixture without preliminary ultrasonication. This difference tends to vanish at higher temperatures (633 K).

No clear correlation impact was found of ultrasonication time (30 - 120 s) and power (100 - 200 W) to emulsion properties.

Maximum error in measuring emulsion layering coefficient was calculated as 1.86%.