Sequential Carotenoids Extraction and Biodiesel Production from Rhodosporidium toruloides NCYC 921 Biomass

A new process for co-extraction and separation of fatty acids and carotenoids from Rhodosporidium toruloides NCYC 921 biomass in order to achieve full exploitation of the yeast lipidic fraction is described. A saponification of the wet yeast biomass was performed using a potassium hydroxide solution (1.1 M) in ethanol 96%, at 65 °C for 180 min. In the carotenoid extraction step, a biphasic system with an organic:aqueous phases ratio of 0.49 mL/mL and a water content of 18.9% (w/w) was used. In the presence of an acid catalyst, the fatty acid fraction was esterified into fatty acids ethyl esters. The yeast biomass downstream processing allowed reaching a fatty acid and total carotenoids recovery yields of 91.0% and 85.2%, respectively. The process reported here takes advantage of various components of the yeast biomass, therefore maximizing the value derived from the biomass feedstock, with a minimal environmental impact within the frame of circular bioeconomy.


Introduction
Due to the increasing difficulty in accessing and extracting fossil fuels, their expected medium and long-term scarcity, and the environmental problems caused by their use (increase of greenhouse gases in the atmosphere resulting in climate changes), the scientific community has been searching for renewable energy sources whose production and use are economically and environmentally sustainable.
Biodiesel is a renewable, non-toxic biofuel, which, when used in the transport sector, emits less pollutant gases than fossil diesel. Transesterification is the conventional process for biodiesel production from oils and fats resulting in general in methyl fatty acid esters and glycerol. No other high value added lipidic products are produced in this process.
Also, the use of arable soils for vegetable oleaginous crops (usually corn, soybean, rape, sunflower) that are processed into biodiesel, instead of being used in the food industry, has resulted in higher prices for these products, raising public controversy all over the world [1]. Therefore, nowadays, new sources of oils and fats for the production of biodiesel are being searched and oleaginous microorganisms are a very good feedstock alternative. Autotrophic oleaginous microalgae have been suggested as possible sources of oil for biodiesel production [2]. However, the use of 1 3 heterotrophic oleaginous yeasts shows advantages over oleaginous microalgae, since the former present higher growth rates and biomass and oil productivities. Besides, they are grown in controlled closed systems, which reduces the risk of contamination, unlike microalgae cultures usually grown in low-cost outdoor raceways, with little or no control. In addition, yeasts can be grown in any region of the globe, and at any time of the year, which is not the case of autotrophic microalgae, as the outdoor cultivation depends strongly on sunlight availability and adequate outdoors temperature.
Nevertheless, the biodiesel microbial production is still far from being economically sustainable, since the production costs are still higher than fossil fuel costs and new approaches aiming at the microbial biodiesel production costs reduction must be looked for and studied.
One way to reduce these costs consists of taking integrated advantage of the various constituents/fractions of the microbial biomass, in order to obtain several products with commercial added-value, similarly to the oil refinery approach [3].
Besides intracellular triacylglycerols, some yeasts also produce significant amounts of carotenoids, which are pigments that are known to improve the human immune system response, to induce vitamin A synthesis, and to reduce the presence of free radicals. The human body does not synthesize carotenoids, so these compounds must be present in the human diet, either through the consumption of fruits and vegetables or through the ingestion of other products rich in these pigments. Carotenoids thus represent a group of compounds of commercial interest for the chemical, food and pharmaceutical industries. However, these compounds, when chemically produced, raise some objections in consumers who prefer to consume natural carotenoids, whose sources include microorganisms [4].
The yeast Rhodosporidium toruloides NCYC 921 has been widely reported as a potential oil producer yeast [5]. In addition, this species, often called "pink yeast", has also been reported as a source of carotenoids of high commercial interest which are used as natural food colorants and feed additives in aquaculture [6].
Most of the works reporting carotenoids and lipids coproduction from yeasts describe separate procedures to extract these products from different yeast biomass samples, as the lipid yeast fraction contains, in the same phase, saponifiable (fatty acids) and unsaponifiable compounds (carotenoids) [7][8][9][10]. Such approach is highly time-consuming and requires huge amounts of microbial biomass, reagents and consumables.
Also, in a recent study Kuan et al. [11] converted R. glutinis biomass into fatty acids methyl esters (FAME) by direct transesterification, claiming that this procedure decreases the energy expenditure and the amount of solvents, when compared with the conventional transesterification method for biodiesel production. However, FAME was the only final product obtained from this process.
The present study describes an innovative process involving the simultaneous extraction of carotenoids and saponifiable lipids from the same R. toruloides NCYC 921 biomass sample using a sequential process of saponification followed by liquid-liquid extraction and esterification of the fatty acids, and resulting in the co-production of fatty acids ethyl esters (FAEE; biodiesel) and carotenoids. The process here discussed takes advantage of the two different yeast biomass components, leading to the main production of biofuel and carotenoids and generating other byproducts potentially valuable along the process, therefore maximizing the value derived from the whole yeast biomass feedstock. Even the residues and effluents generated in the whole process may be used to produce biogas as previously reported [12]. The inclusion of all these operations will greatly improve the economics of the overall process, as the high-value added products obtained in the process will economically sustain the microbial biodiesel production.

Microorganism
The yeast R. toruloides NCYC 921 was purchased to the National Collection of Yeast Cultures (Norwich, UK). The strain was stored on slants of Malt Extract Agar, at + 4 °C.

Biomass Production
The growth conditions and the analytical methods (High Performance Liquid Chromatography-HPLC and Flow Cytometry-FC) used for the biomass production process characterisation were previously described [5], and so, only an outline will be given here.

Inoculum Preparation
Rhodosporidium toruloides yeast cells from one slant grown for 72 h at 30 °C were transferred to 1 L baffled Erlenmeyer flask containing 150 mL growth medium with the following composition (g/L) [13] 4 , 0.5. Glucose was added to the culture at a final concentration of 35 g/L. The medium pH was adjusted to 5.5 before inoculation.

Fermentation Conditions
After 1 day growth (exponential growth phase) the innoculum culture was used to inoculate a 7 L bioreactor (5% (v/v); 5 L working volume) (FerMac 360 bioreactor, Electrolab Biotech, UK) equipped with two Rushton impellers and 4 baffles, and with an initial volume of 2850 mL of the medium described above. All the cultivations were conducted at 30 °C. The feeding strategy was fed-batch and followed the procedure described by Dias et al. [5].

Biomass Harvesting and Characterization
At the end of the production process, biomass was harvested by centrifugation (Sigma 2-16K, Sartorius, Germany) at + 4 °C and 10,000 rpm for 10 min and stored in the dark at − 20 °C until needed. The biomass lot recovered was characterised regarding moisture, carotenoid and lipid contents. The first value was obtained by sample drying, and the latter by performing the laboratory traditional extraction methods described by Freitas et al. [14]. Figure 1 outlines the main steps involved in the carotenoids and FAEE co-production and fractionation from the R. toruloides biomass. A direct saponification on the yeast biomass, followed by the addition of an organic solvent, allowed the separation of the unsaponifiable fraction containing the carotenoids (which remained in the organic phase) from the saponifiable fraction containing the fatty acid soaps (which remained in the aqueous phase). The neutralisation of the aqueous phase followed by a separation step resulted in an ethanolic fraction containing free fatty acids (FFA) which were further converted into FAEE biodiesel under acidic catalysis. Ethanol was used in the esters production as it was present in the extract and also because a major concern of this process is the use of environmentally friendly chemicals that may be produced from other sources rather than fossil fuels.

Carotenoids and FAEE Reaction, Extraction and Separation
The main saponification and the carotenoids extraction steps were optimised in order to select the best operational conditions that led to the highest product recovery yields.

Yeast Biomass Saponification Optimisation
The saponification process was previously optimised, testing different biomass conditions (pre-treatments), assay atmospheres and temperatures. For these studies, assays were performed, in duplicate, in 100 mL closed Erlenmeyer flasks containing 15.2 mL/g wet biomass (wb) of alcoholic alkaline solution (KOH 0.34 M in ethanol 96% (v/v)) and around 2.5 g of yeast wet biomass harvested at the end of the fermentation process. After incubation for 1 h in a rotary shaker (3527, Labline) at 200 rpm, the mixture was filtered in a G3 Gooch crucible and the pellet washed with an extra volume of 10 mL of ethanol. The final liquid fraction was recovered and analysed for carotenoid content and for the remaining alkaline reagent quantification as an indirect measure of the saponified compounds release. Whenever needed, fatty acids (FA) were also recovered from the extracts and their contents characterised by gas chromatography (GC) after derivatization with boron trifluoride.
The saponification process was optimised regarding: 1. Biomass disruption technique and assay' atmosphere: Biomass (2.5 g; wet) was pre-treated according to the following techniques: • Bead-milling-biomass was milled in a MM 400, Retsch apparatus at 25 Hz, with 4 stainless steel beads (10 mm diameter) for 5 min; • Bead-beating-1 g of glass beads (diameter of 425-600 µm or 2 mm) was added to the wet biomass and the mixture was stirred in a vortex for 1 min, followed by 1 min in an ice bath, being this procedure repeated 5 times; • Homogenization-biomass was disrupted in a Heidolph DIAX homogenizer at 8000 rpm for 5 min; • Drying-biomass was dried in a forced air oven at 60 °C until constant weight; • Lyophilization-biomass was frozen and lyophilised at − 54 °C in a Heto PowerDry LL3000 from Ther-moFisher Scientific device. Different types of atmosphere inside the Erlenmeyer flasks where the saponification reaction occurred at 55 °C were also tested: laboratory atmosphere, nitrogen and argon.

Temperature:
Regarding the study of the influence of the temperature, assyas were performed in the range of 35-80 °C with wet biomass.

Alkali and ethanol concentrations and reaction time:
The remaining operational factors influencing the saponification reaction were the alkali (KOH) and ethanol concentrations and the reaction time, whose values were optimised using a central composite design (CCD) planned for three variables at two levels (2 3 ) and expanded for the levels + 1.682/− 1.682 (Table 1). 17 experiments were carried out, using 38 mL of ethanolic KOH, at 65 °C and under laboratory atmosphere for 1 h. This number included 8 factorial points (levels + 1 and − 1), 6 expansion points and 3 replicates of the centre point (0).

Carotenoid Extraction Optimisation
The experiments were performed in a separatory funnel containing 5 mL of saponified alcoholic extract (8.3% (w/w) of water). For the optimisation of carotenoid extraction from the alkaline alcoholic extracts, a two variables at two levels (2 2 ) expanded factorial design was planned ( Table 2). The parameters to be optimised were the proportion of water in the liquid-liquid system (8.3-20% w/w) and the phases ratio regarding the volume of hexane (0.2-0.6 mL/mL). After the extraction step, the organic phase was characterized in terms of total carotenoids content by HPLC. Eleven experiments

Residue
Alcoholic Extract

FAEE Production
After saponification and carotenoid extraction in the biphasic system, an adequate volume of concentrated acid (37% hydrochloric or 95-98% sulphuric purchased chemicals) was added to the alcoholic phase (containing soaps; V = 60 mL) to decrease the pH to around 8 ( Fig. 1). Afterwards, samples were centrifuged at + 4 °C and 10 000 rpm (Multifuge 3SR+, Thermo Scientific) to remove the salt precipitate (KCl/K 2 SO 4 ) from the supernatant containing the soaps. To the remaining liquid phase was added more concentrated HCl until reaching pH 3, to promote FFA release.
To produce FAEE, to the extract treated with hydrochloric acid (HCl) was then added an extra amount of 25 µL HCl/ mL to promote FFA esterification and the mixture was incubated at 55 °C and 150 rpm for 4 h (Orbit Environ-Shaker, Lab-Line). Ethanol was then removed in a rotavapor (R-200 with V-800 vacuum control and B-490 heating bath, Buchi) at 40 °C and 175 mPa allowing obtaining a FAEE phase.

Optimised Carotenoids and FAEE Co-production Overall Process
In order to evaluate the final co-extraction process yield, the overall process was performed sequentially (Fig. 1). Saponification and extraction steps were carried out using the operational conditions previously optimised (Table 6) and the carotenoid recovery yield (w/w) was calculated as the ratio between the carotenoid amount in the initial yeast biomass and the carotenoid amount in the hexane phase. FA recovery yield (w/w) was calculated as the ratio between the FA amount in the yeast biomass before the saponification, extraction and separation steps and the produced FAEE amount. The global procedure was carried out in duplicate for 25 g of wet biomass (humidity − 65%). The carotenoid amount was evaluated by HPLC and the quantification of the amount of FAEE was determined by GC.

Carotenoid Content Quantification by HPLC
The quantification of pigments in the samples was performed by HPLC in a Hewlett-Packard 1100 series device equipped with a µ-Bondapak C18 (250 mm × 4.0 mm) reversed phase column from VDS Optilab, a UV detector set at 450 nm and an injection loop of 20 µL. Samples were eluted at room temperature with a mobile phase of acetonitrile, methanol (with 0.1% (v/v) of triethylamine) and ethyl acetate (75%:15%:10%) at a flow-rate of 0.8 mL/min. Calibration curves were built for β-carotene in hexane, and torulene and torularhodin amounts were reported as β-carotene equivalent concentrations due to the lack of standards with known concentration.
Nevertheless, in strong alkaline ethanolic solutions (saponification extracts) it was very difficult to create reliable calibration curves for β-carotene and so the quantification is reported as peak area units (AU) being the results merely comparative.
Peaks in the HPLC chromatograms were assigned to β-carotene, torulene and torularhodin based on the analysis of separate solutions of the compounds and by comparison with published results from Park et al. [15]. Although torulene and torularhodin standards were not available in the market, separate extracts of these compounds were obtained after extraction from the biomass with DMSO by a conventional method [14] followed by thin-layer chromatographic analysis using silica gel 60 plates and an eluent mixture of petroleum ether (65-90 °C)/acetone/diethylamine (100/40/10), band scrapping and dissolution in hexane, allowing to confirm peak assignment.

Saponified Compounds Content Quantification
The saponified fraction evaluation was performed indirectly by titration of the unreacted alkaline reagent with hydrochloric acid (0.1 M) in the presence of phenolphthalein using 20 mL aliquots. Data are reported as meq of saponified compounds per litre of alcoholic alkaline solution.

Biomass Moisture Determination
Biomass portions (5 g) were dried at 80 °C in a forced air oven (ULE 508, Memmert) until constant weight to moisture content determination.

Fatty Acid FAEE Quantification by GC
Fatty acid containing samples were previously derivatized into FAME with boron trifluoride according to the EN ISO 5509 standard. Derivatized samples and samples containing FAEE were then analysed on a CP-3800 equipment from Varian (USA) with a flame ionization detector. The separation of the compounds was carried out on a capillary column (Supelcowax 10) with 30 m, internal diameter 0.32 mm and 0.25 µm of film thickness. Helium was used as carrier gas at a flow rate of 1.4 mL/min and the injector and detector temperatures were 250 °C. The column temperature rose from 60 °C (2 min) to 200 °C at 10 °C/min and then to 230 °C at 5 °C/ min. Methyl heptadecanoate was used as internal standard and the esters were identified by comparison with a chromatogram from a standard mixture (SUPELCO 37).

FAEE Characterization as Potential Biodiesel
Using the FAEE fatty acid composition and some empirical correlations reported in Nascimento et al. [16], the values for the cetane number (CN), the iodine value (IV), the saponification value (SV) and the cold filter plugging point (CFPP) of the ester fraction were estimated.

Biomass Pre-treatment
Several physical methods can be used to promote microbial cell membrane disruption and to enhance the solvent penetration, before saponification. Li et al. [17] for example, used freeze dried microalgal biomass in the saponification reaction. However, since these type of technologies are energy demanding, in this work, different biomass disruption techniques (lyophilisation, drying, bead-milling, bead-beating and homogenisation) were tested before the saponification step, and the yields were compared to those obtained with the original wet biomass harvested from the bioreactor and concentrated by centrifugation. In addition, as it is known that carotenoids are easily oxidized [18], different types of atmosphere (air = laboratory atmosphere, nitrogen and argon) in the Erlenmeyer flasks where the saponification reaction was conducted were also tested for their influence on the recovery efficiency.
The results obtained are shown in Fig. 2 and allow concluding that, in tight Erlenmeyer flasks, the reaction atmosphere composition does not have a significant impact on the saponified compounds and carotenoid extraction efficiency, being observed only slight differences among the experiments carried out under the different atmospheres. Although the presence of argon may in some cases improve carotenoid recovery, its effect on the lipid extraction is detrimental mainly in assays performed with biomass disrupted by contact methods (homogenisation and beating).
Regarding the need for a cell pre-treatment to improve the carotenoid and saponified compounds recovery yield, it was observed that no technique was effective. Saponified fraction amount is optimised by disruption with lighter contact methods (beating). However, carotenoids are compounds very sensitive to environmental factors such as light and temperature and, from results shown in Fig. 2, it can be inferred that pre-treatment methods that increase sample temperature degrade yeast carotenoids. Indeed, homogenization, beadbeating and bead milling promote sample heating. On the other hand, longer treatment processes such as lyophilisation and dryness are the ones with more negative impact mainly over the extraction of saponified compounds probably due changes in the inner cell environment and lipid layers. In general, subjecting biomass to any manipulation seems to always be detrimental for carotenoids condition. So, the best is then to use wet biomass after harvesting avoiding the use of energy demanding biomass pre-treatment steps and expensive inert gas reaction atmospheres.

Temperature
Once the biomass pre-treatment need was discarded and reaction atmosphere selected (air), the effect of temperature on the extraction efficiency was studied (Fig. 3). It was observed that, although higher temperatures increase the extract saponified content, they had a harmful effect over carotenoids. The extraction capability of carotenoids attained its maximum at 65 °C. So, as for the saponified amount 65 °C was nearly as good as 80 °C, 65 °C was the selected temperature for the saponification step. These results are in accordance to those reported by Binnal and Babu [19] who studied the effect of temperature on the direct saponification yield of wet Chlorella protothecoides biomass and concluded that the yield of soaps increased with increasing temperature up to 60 °C at which the maximum yield was observed. Also González et al. [20] carried out optimised saponification reactions to purify polyunsaturated fatty acids at 60 °C for 60 min. Binnal and Babu [19] noticed a decrease in the yield of soaps for temperatures above 65 °C which was attributed to a decrease in the available quantity of ethanol due to vaporization. In the present work, the tightly closed Erlenmeyer flasks wherein the saponification reaction took place might have avoided ethanol evaporation phenomena, which may explain the high saponification yields observed at temperatures higher than 65 °C as a result of higher ethanol availability.
Regarding the detrimental temperature effect over carotenoids, results in Fig. 3 are consistent with the results described by Bhosale et al. [18], who reported a decrease in the yeast carotenoid recovery with the increase in spray drying inlet temperature .

KOH and Ethanol Concentrations and Reaction Time
The previous studies allowed selecting the best operational conditions for the carotenoid and saponified compounds extraction from R. toruloides biomass: untreated wet biomass, air as reaction atmosphere and 65 °C of temperature.
The saponification efficiency is also influenced by other factors such as the alkaline catalyst (KOH) and ethanol concentrations, and the reaction (saponfication) time duration. Those were optimised using a central composite design (CCD) planned for three variables at two levels (2 3 ). The response variables taken into consideration were the amount of extracted saponified compounds and carotenoids. The first one was reported as meq of saponified compounds per g of wet biomass and the other one as area units accounted for in HPLC analysis per g of wet biomass. Results from the 17 experiments performed were analysed using the Statistica 10 program to determine the effect of each factor and the main interactions between them. In Table 3 can be observed that the duration of the process and ethanol concentration  seemed to be the factors with higher impact on the extraction efficiency regarding soaps formation (lowest values of p). However, values of 0.05 < p < 0.1 are considered as a mere suggestion lacking of significant statistical relevance. On the contrary, ethanol linear dependence is the only factor with a strong statistical impact on carotenoids release and it has a positive effect (higher ethanol purity promotes higher extraction). The polynomial equations obtained by the model were as follows: From Eqs. (1) and (2) were generated the response surfaces presented in Figs. 4 and 5, showing the effect of the three variables respectively on the saponified compounds and carotenoid extraction efficiencies. Regarding saponifiable compounds recovery (Fig. 4), it can be seen that the best conditions should include a high KOH concentration and the presence of water in the ethanol solution. However, the reaction time would be a compromise choice taking into consideration the process cost as for high concentrations of KOH the reaction time may be lower while with lower concentrations of catalyst the efficiency increases in time.
The analysis of the results in terms of carotenoids extraction efficiency (Fig. 5) shows that the best conditions include a low water proportion (as expected, since those compounds are lipophilic), a high reaction time and a moderate KOH concentration.
Having low coincidence of results for the co-extration of carotenoids and lipids, the authors final choice was towards the best conditions that led to the maximum carotenoid extraction efficiency as carotenoids are the products with the highest added-value in the process.
So, partial differentiation of the polynomial equation was used to find the optimum point: KOH concentration: 1.  AU/g wb (3.5% below the predicted maximum value). This represents a minimal reduction in the carotenoids content retrieval, when compared to the impact of ethanol 100% (v/v) prices in the process costs. Gonzalez et al. [20] have also optimized the choice of the extraction solvents for algae biomass saponification and find out that the best was ethanol 96% while using a lower KOH concentration (0.32 M).
Having chosen the most suitable operational conditions taking into consideration the main goals of this work, the next step was to validate the model results for the chosen reaction conditions (KOH = 1.1 M; Ethanol = 96% (v/v); reaction time-180 min).
In this work, the validation assay generated saponified compounds and carotenoids extraction levels of 2.17 ± 0.09 meq/g wb and 1187.17 ± 76.28 AU/g wb respectively, showing a deviation from the model of 0.6%.
However, after performing the saponification step under the optimised conditions it was observed that the final biomass residue remained coloured indicating that the carotenoid extraction was not complete. Due to the inability of increasing ethanol concentration and to the fact that the model showed that no beneficial effects were obtained from increasing KOH concentration, the effect of sequential saponification cicles was evaluated to compare with reaction time increase. Results showed that increasing the reaction time up to 180 min promotes an increase in the amount of carotenoid (the most valuable compounds) recovery of 44% while performing 3 cycles of 1 h allowed to retrieve 70% more. The higher amount of reagents used in the reaction increases process cost; however those that are not consummed in the reaction may be recovered and reintroduced in the process. By the contrary, the energy cost increase is the factor that will need to be taking into consideration versus the increase in the extraction yield of the most valuable compounds.

Carotenoid Recovery by Liquid-Liquid Extraction
After the saponification reaction step where the compounds were extracted from the intact cells, it was necessary to separate the carotenoids from the alcoholic phase containing the saponified matter, which was strongly alkaline and contained ethanol, glycerol (from lipid hydrolysis) and other yeast compounds/residues. The first step was the recovery of the carotenoid unsaponified fraction, containning pigments such as β-carotene, torulene and torularhodin, by a liquid-liquid extraction procedure. Carotenoids, being liposoluble compounds, are easily extracted to a non-polar solvent such as hexane which was used due to its lower toxicity when compared to other suitable solvents such as chloroform [20].
To optimise the carotenoid recovery from the ethanolic basic extract, it was planned a set of 11 experiments using a central composite design for two variables. The variables taken into account were the water content in the liquid-liquid system (8.3-20% w/w) and the organic:aqueous phases ratio (0.2-0.6 mL hexane/mL extract + water) and the efficiency of the process was defined in terms of the amount of extracted unsaponified carotenoids, now expressed as (µg). Besides hexane quantity, water content was also chosen as an important parameter affecting the recovery process. Because ethanol and hexane are miscible substances it was indispensable to improve the polarity of the aqueous extract to promote a good phase separation and carotenoid distribution as carotenoids are apolar or low polarity compounds. The lower limit for the water amount was defined as 8.3% (w/w) meaning that no extra water was added to the system as this was the natural concentration of water in the extract resulting from the biomass moisture (65.3%) and the ethanol solution purity (96% v/v). The effect of each parameter and the main interactions between them are shown in Table 4 and the response surfaces in Fig. 6. Both parameters tested showed statistically relevant influence (p < 0.05) in the system being the parameter that revealed the strongest effect on the carotenoids extraction efficiency the water content (Fig. 6). This observation was expected since for some of the experiments a null amount for the carotenoid extraction was attributed because the system was totally miscible.
The equation describing the modelled system is: The partial differentiation of the polynomial equation generated an optimum set of conditions which included: 0.49 mL hexane/mL aqueous phase and 18.9% (w/w) of water in the liquid-liquid system. The conclusions were validated in a new experiment performed in the optimal conditions: 14.96 ± 0.68 µg [deviation of 1.6% from the predicted value (15.21 µg)].

Neutralization for FFA Release
After carotenoid separation/recovery, the remaining alkaline extract containing glycerol, ethanol and fatty acid soaps was neutralized to obtain the corresponding FFA which could eventually be transformed into biodiesel.
Among the most used inorganic acids, hydrochloric, sulphuric and phosphoric acids are the ones whose use could be envisaged at high concentration levels taking into account a possible reuse of the generated salts in the R. toruloides growth culture medium or even as supplements in anaerobic experiments degradation. Nevertheless, phosphate salts have higher solubility products and consequently would be more difficult to separate from the extract. So, sulphuric and hydrochloric acids were chosen to test the efficiency for FFA neutralisation.
In both cases, a first neutralization step to pH 8 was performed to allow the removal from the extract of the precipitated salts, so that the solid powder would not interfere negatively in the biodiesel production process. Then, more acid was added until pH 3 to obtain the corresponding FFA. Probably due to some adsoption of the soap molecules to the precipitate, the FFA release using sulphuric acid was 25% lower than with HCl (46.21 ± 1.26 mg/g wb versus 61.60 ± 4.91 mg/g wb).

Esterification of FFA
After neutralizing the alcoholic extract, and as fatty acid ethyl esters (FAEE) may be used as biodiesel rather than methyl esters, an esterification step was performed directly in the extract without precipitate using the ethanol already present and adding an extra amount of hydrochloric acid to act as catalyst. After 4 h at 55 °C, with an ethanol amount of 280 mol/   Fig. 6 Effect of phases ratio and water amount on the carotenoid recovery efficiency from alcoholic extracts of R. toruloides in a liquid-liquid extraction system (V extract = 5 mL) mol FFA and an acid catalyst proportion of 53% (w/w FFA), the FFA in the extract were converted into FAEE. The FAEE obtained represented a recovery of 85% (w/w) regarding R. toruloides original biomass.
Biodiesel quality is strongly defined by the ratio between saturated and unsaturated fatty acids. Increased levels for calorific value, cetane number and oxidation stability are associated to the presence of saturated long fatty acids while unsaturated chains increase flow characteristics at low temperatures [21]. The composition of the mixture of FAEE obtained in this work was rich in palmitic, oleic and linoleic acids reflecting the yeast lipids composition as shown in Table 5. Based on this observation, some properties of a biodiesel of similar composition may be predicted, and the values obtained can be compared to those defined in the biodiesel quality standard EN 14214:2012 (Table 5).
The proportion of linolenic acid (5.6% w/w) falls within the range defined in the biodiesel standard (< 12%) and no fatty acids with 4 double bonds were detected in the mixture. Cetane number and iodine value estimates also comply with the biodiesel standard. CFPP is the only parameter whose estimated value does not comply with the standard values for B100. Nevertheless, this is not impeditive for use as a biodiesel if R. toruloides FAEE are mixed with other types of biodiesel or diesel, providing the resulting mixture meets all the EN 14214:2012 requirements.

Optimised Overall Process for Carotenoid and FAEE Co-production
After selecting the optimal operational conditions that led to the highest product yields (Table 6), a new portion of yeast biomass was processed according to the protocol depicted in Fig. 1. The initial biomass before fractionation contained 36.8 ± 0.1% FFA (w/w dry biomass) and 0.023 ± 0.002% total carotenoids (w/w dry biomass). At end of the overall process, FFA (in the form of FAEE) and total carotenoids recovered amounts were 31.7 ± 0.17% (w/w dry biomass) and 0.021 ± 0.001% (w/w dry biomass) respectively, which correspond to 85.3% and 91.0% recovery yields relatively to the initial yeast biomass. González et al. [20] also used a saponification reaction to separate P. tricornutum polyunsaturated fatty acids fraction from the unsaponifiable fraction and obtained similar yields (86.0-90.0%) using wet microalgae biomass.

Conclusions
The approach here described allowed R. toruloides yeast biomass fractionation, taking advantage of the intracellular lipidic fractions (saponifiable and unsaponifiable), which resulted in the co-production of biofuel and carotenoids-rich extracts with increasing value, using low-priced and environmentally friendly chemicals. Several process steps were optimised which allowed selecting the best operational conditions that led to the highest carotenoid content. This process shows many advantages over the traditional methods for lipid and carotenoid separated extractions such as time and chemicals savings, and can also be applied to other microbial resources such as microalgae and bacteria. AMB/116594/2010 entitled "CAROFUEL -New process for a sustainable microbial biodiesel production: The yeast Rhodotorula glutinis biorefinery as a source of biodiesel, biogas and carotenoids" supported by FCT (Fundação para a Ciência e a Tecnologia-Portugal) (also supported by FEDER funding through COMPETE -Programa Operacional Factores de Competitividade). The authors acknowledge