Recent Advances in Supercritical CO2 Extraction of Pigments, Lipids and Bioactive Compounds from Microalgae

Supercritical CO2 extraction is a green method that combines economic and environmental benefits. Microalgae, on the other hand, is a biomass in abundance, capable of providing a vast variety of valuable compounds, finding applications in the food industry, cosmetics, pharmaceuticals and biofuels. An extensive study on the existing literature concerning supercritical fluid extraction (SFE) of microalgae has been carried out focusing on carotenoids, chlorophylls, lipids and fatty acids recovery, as well as the bioactivity of the extracts. Moreover, kinetic models used to describe SFE process and experimental design are included. Finally, biomass pretreatment processes applied prior to SFE are mentioned, and other extraction methods used as benchmarks are also presented.


Introduction
In the last few years, the need for naturally derived products with a low environmental footprint is steadily emerging [1]. For this purpose, not only green processes need to be applied, but also, feedstock that can be obtained with a neutral impact on the ecosystem is desired [2]. Biomass, such as microalgae, seems to have many advantages, mainly due to its ease of availability, either from controlled cultures, where no arable land is required, or from natural sources, for instance fresh water, marine environments and wastewater [2][3][4].
Microalgae are a diverse group of eukaryotic organisms or prokaryotic cyanobacteria, which can be cultivated autotrophically, heterotrophically or mixotrophically [5]. They can be reproduced rapidly, where, under the appropriate conditions an exponential production rate can be reached [3,5]. Also, thanks to the wide diversity of species and different cultivation protocols, it is possible to recover various components, namely, pigments, lipids, proteins and fatty acids [6][7][8]. Those ingredients find application in the pharmaceutical and food industry, as well as in the production of biofuels. Consequently, microalgae species are studied and recorded with ever-increasing interest [9].
Concurrently, green extraction methods have also gained research interest. New extraction protocols focus on minimizing the energy demands and the use of solvents. Preferably, non-toxic and non-flammable solvents derived from biomass are used [10]. Plenty of novel extraction processes can be used for this objective, such as microwave (MAE), ultrasound (UAE) and UV light assisted extraction. These techniques apply energy to the system enabling shorter extraction times and lower solvent consumption, while achieving high recovery rates [11][12][13].
C. vulgaris

Microalgal Products
Microalgae is a rich source of bioactive compounds, for instance, chlorophylls, carotenoids, tocopherols and phenolics [8,[126][127][128] (Figure 1). These high-added value pigments are commercially exploited to produce food supplements, pharmaceuticals and cosmetics, thanks to their antioxidant, anti-inflammatory and anti-microbial properties, among others [3,128,129]. Depending on the species and the cultivation conditions, the variety and the amount of bioactive compounds in the cells may differ [8].
The data of these columns are analytically presented in Table 2, 2 Total Phenolic Content (TPC), Antioxidant Activity (AO) or Antimicrobial Activity (AM)., 3 The data of these columns are analytically presented in Table 3 2

. Microalgal Products
Microalgae is a rich source of bioactive compounds, for instance, chlorophylls, carotenoids, tocopherols and phenolics [8,[126][127][128] (Figure 1). These high-added value pigments are commercially exploited to produce food supplements, pharmaceuticals and cosmetics, thanks to their antioxidant, anti-inflammatory and anti-microbial properties, among others [3,128,129]. Depending on the species and the cultivation conditions, the variety and the amount of bioactive compounds in the cells may differ [8].  Carotenoids are tetraterpenoids, soluble in lipids and responsible for the photoprotection of the microalgal cell [8,130]. They present coloring and antioxidant activities, and as a result, they are commonly used in the food industry [129][130][131]. Furthermore, carotenoids can be divided into two categories depending on the presence of oxygen in their structure [130,131]. Xanthophylls, which contain oxygen, have gained significant industrial interest for having antioxidant and conservative properties [132]. In this group, astaxanthin, lutein and fucoxanthin are included [130][131][132]. Non-oxygen containing carotenoids are called carotenes (e.g., β-carotene) [130][131][132]. Carotenoids can be categorized into primary and secondary depending on their synthesis process [130,133]. Primary carotenoids are produced during photosynthesis and are crucial for the cell's viability, while secondary ones are produced when the cell is subdued due to stress, leading to carotenogenesis [132,133]. Factors, such as temperature, pH, salinity, light, nutrients, and the presence of oxidizing substances during cultivation may lead to an enhanced production of primary and secondary carotenoids [8].
In addition, chlorophylls are an extractable compound from microalgae [127,134]. Their role is to absorb solar energy, ensuring that the organism can photosynthesize [127,134]. Chlorophylls in nature may appear with plenty of isomers. The most common among microalgae is chlorophyll a, which is present in all species, while chlorophyll b is found in green algae [134,135]. Chlorophyll extracts are known for their antioxidant and antibacterial activity [127,134]. Consequently, they are widely used in pharmaceutical applications, but also, as a natural pigment due to their intense green color [127,134,135]. Their main disadvantage is that they need stabilization in order to be used as food additives, which can increase the cost and alter their beneficial properties [127].
Apart from pigments, microalgal strains also contain a significant number of fatty acids. They are carboxylic acids with compositions depending on the function they have in the cell [22]. Fatty acids can be categorized by the length of their hydrocarbon chain as short-, medium-, long-and very long-chain and by their structure as saturated (SFA), monounsaturated (MUFA) or polyunsaturated (PUFA) [22,136]. Commonly, PUFAs, such as docosahexaenoic acid (DHA), eicosapentaenoic acid (EPA) and γ-linolenic acid (GLA), are present in microalgae and find application in the food industry [6].

Classification of Methods
The existence of thick cell wall structures in microalgae affects the efficiency of the extraction methods. Thus, in many species, the weakening of the cell wall is necessary in order to minimize the cost of the extraction process and to enhance the recovery of target compounds [137]. There is a wide range of methods that can be used and most of them affect the microalgae cells in different ways, while aiming at extracting different compounds. Additionally, it should be mentioned that not all the pretreatments are appropriate for every process, because they alter the cell in distinct ways. Thus, the technique applied should be taken under consideration [138].
There are two main reasons for the necessity of a pretreatment process prior to extraction. The first one is that, in many algal species, the target compound is part of the cell wall, so by decomposing the structure, the extraction is conducted more effectively. Secondly, when extracting intracellular ingredients, the weakening of the cell wall enhances their accessibility by facilitating their transport through the cell wall [139]. The algal cell walls have tensile strength around 9.5 MPa, a fact that makes pretreatment inevitable in some cases [140].
The techniques can be briefly divided into two categories: The latter includes two disruption methods: The mechanical techniques affect the cell by using shear forces, electrical pulses, waves or heat. Although they provide high recovery yields, they are not recommended for sensitive compounds due to high shear stress or temperature increases, unless a cooling mechanism is used. Their combination with other pretreatment methods may result in better recovery rates [138].
Bead milling is a commonly used process not only for algal biomasses, but also for grinding minerals and manufacturing paints. During this procedure, a given amount of energy is applied to the cell wall, causing the release of intracellular products. The results of pretreatment depend on bead size and type, as well as agitation speed, bead filling, chamber size and geometry, biomass concentration, and suspension flow rate [141,142].
Ultrasonication may also be performed as a biomass pretreatment technique. It can be described as a series of acoustic waves with frequencies that vary from 20 kHz up to some GHz. The waves transfer through the medium and create points with higher or lower pressure (compression or rarefaction, respectively). Those local changes, if they are intense enough, create bubbles which grow and undergo implosive collapse. This cavitation phenomenon is responsible for the ruptures caused in the cell wall surface. The energy is applied to the cell either by using an ultrasonic horn or by using an ultrasonic bath [11,143].
Microwaves are electromagnetic waves with frequencies between 0.3-300 GHz that generate heat depending on the polarity of the compounds. Those waves create electromagnetic fields causing the rotation of the polar compounds according to the direction of the field (dipole rotation). Respectively, ions in the medium tend to migrate with the field alternation (ionic conduction). The movement of the ions and rotation of the dipoles result in heat production by friction [11,144]. The intracellular water under microwaves evaporates, leading to an increase in pressure inside the cell and the expansion of the cell wall which causes its rupture [11].
High-pressure homogenization is a process used for sterilization and recovery of intracellular products [145]. The biomass is pumped through an orifice leading to a valve under high pressure and then expands in a lower pressure chamber. The disruption occurs because of the pressure drop which creates cavitation and shear stress on the cell wall surface [146]. The advantages of the specific method are the low heat formation, which lowers the risk of thermal degradation, and the ease of scale-up. On the other hand, for sufficient cell wall damage, a lot of circles of homogenization are required resulting in cost increases [145,147].
As in the case of ultrasonication, in hydrodynamic cavitation the cell wall ruptures because of cavitation. A Venturi valve is used in order to create a pressure drop and, therefore, cavitation bubbles which, by collapsing violently, cause damage to the cell. A major advantage of this technique is that the temperature does not increase [148,149].
The pretreatment in the case of pulsed electric field (PEF) concerns a mild disruption method because it forms pores on the cell wall for a short period of time without a significant increase in the temperature [138,150]. Specifically, when an external electric field is applied to the cell, it is believed that the lipids on the surface rearrange, enhancing the permeability of the compounds. PEF pretreatment has better results in higher cell densities, lower liquid content, and liquid systems with low cell density [151]. The conditions that affect the efficiency of the method are solvent type and dosage, temperature, and conductivity [138].
Steam explosion is a batch process where the biomass is treated under high pressure (1-3.5 MPa) and high temperature (160-260 • C). The cells are placed in a closed chamber and then the temperature and pressure are increased until the system equilibrates for 5-10 min. Afterwards, the vessel is depressurized rapidly. The sudden expansion causes the disruption of the cell wall [152]. The method is mostly used in lignocellulosic biomasses [153]. High operation temperatures might degrade thermolabile compounds, thus, lower temperatures are preferable [152].
The freeze-drying process is commonly used for drying thermolabile products in the food industry in order to maintain their quality. The procedure consists of two steps, the first is the freezing of the biomass and the second is the subjection of the sample to low pressure (approximately 1 kPa). By freeze-drying, ice crystals are formed from intracellular water, which makes the cells expand. The slower the freezing is, the larger are the crystals and the effect of the pretreatment on the biomass. A major disadvantage of the method is its high cost along with high residence time [140]. The results of this treatment are enhanced when used in combination with other methods (e.g., microwaves) [154].

Chemical
A lot of materials have been used for the disruption of the cell wall. The method and the compounds used depend on the cell wall structure, its composition and the suitability with the extraction technique applied. Commonly, the substances are: • Acids • Solvents (organic, ionic liquids, etc.) • Salts (e.g., osmotic shock with NaCl)

Enzymatic
Frequently, enzymatic lysis is used as a cell disruption method. Enzymes, as cellulose, break the linkage between sugars in a cellulosic chain [147]. This facilitates the extraction of the intracellular products due to their ease of accessibility through the disrupted cell wall. The method targets specific compounds depending on the enzyme used. The most used enzymes, apart from cellulose, are amylase, amyloglucosidases, lipases, and proteases [138,155]. Occasionally, a combination of enzymes in a single treatment can achieve better recovery yields [138]. Although it is an environmentally friendly procedure requiring low temperatures, the cost of the enzymes, the difficulty in scaling-up, and the slow reaction times make the method hard to apply in every case [138].

Arthrospira
Despite the recalcitrant cell wall of these species, pretreatments before SFE were reported only in a few studies. All the methods applied were mechanical and the majority of them involved grinding [25][26][27]39]. The rest of the pretreatments mentioned were crushing with cutting mills [28] and milling with mortar and pestle [30].

Chlorella
Chlorella is known to have a thick cell wall, consequently, disruption methods are necessary in most cases. Frequently, milling or grinding were applied before extraction. In particular, it has been reported that disk milling increases the extraction yield from 0.076% to 0.299% in comparison with manual grinding, respectively. Adding dry ice to the manual grinding results in an extraction yield of 0.161% [58]. Also, another publication demonstrates the effect that the crushing has on the extraction yield, leading to a more than 100% increase in the yield [61]. Finally, it has been shown that by cell wall disruption with lyophilization and bead milling, a yield of 10.64% was achieved, compared to 9.25% without pretreatment [52]. Microwave pretreatment was also tested. In detail, when freezedried biomass was subjected to microwaves, the extraction yield increased from 3.90% to 4.86% for supercritical extraction at 28 MPa and 70 • C. More significantly though, was the effect of microwave pretreatment at lower extraction temperature, where the yield obtained was 4.73% compared to 1.81% without pretreatment [57].

Haematococcus
Haematococcus cells, due to their rigid cell wall, when in the red non-motile stage, need to undergo pretreatment in order for carotenoids to be extracted more effectively [156]. Aravena and del Valle have studied the effect of cells homogenization with water on astaxanthin recovery [84]. Compared to powdered biomass, the homogenization leads to worse results; in particular, for extraction at 40 • C and 75 MPa, a recovery of 58% was achieved with powdered Haematococcus, while with homogenized cells the recovery was approximately 49% in addition to a longer extraction period. Almost the same results have been derived at 70 • C, with 61% recovery for powdered biomass and 48.5% with a water homogenized one. Nobre et al. examined the effect that the duration of the crushing has on the extracts. Under the same extraction conditions, total carotenoid recovery has been increased from 59% to 92% by doubling the crushing time [86]. Valderrama et al. achieved a yield of 0.86% at 60 • C and 30 MPa by using crushed by cutting mills biomass, while the yield reached 1.26% when crushed and manually ground with ice biomass, was extracted under the same conditions [28].

Nannochloropsis
Nannochloropsis consists of a double layered cell wall; an external algaenan-based and an internal cellulose-based [157]. The thickness of the cell wall leads to different disruption attempts to maximize the effectiveness of the extraction method. Regarding SFE, homogenization [75,96,103] and grinding [104] have been applied to cultures. Moreover, high pressure homogenization has been tested [97]. Molino et al., have studied the outcome that accelerated solvent extraction (ASE) with n-hexane as pretreatment at 50 • C and 100 bar for 20 min [98]. Experimental design in bead-milling conditions was performed by Leone et al., focusing on the increase in extraction of lipid and total yield [106]. Microwaves seem to have a negative effect on the total recovery for the same extraction conditions since, according to Hernández et al., pretreatment for 5 minutes resulted in 8.2% yield and for 1 min in 11.9%, while the extraction yield was 12.9% when crude biomass was used [93]. Lipid yield showed different behavior, with optimum results, namely 10.8%, achieved when 1 min of microwave pretreatment was employed, while the yield was 6.9% in the case of 5 min pretreatment and 7.9% without any pretreatment. Also, water content remaining in biomass after different drying methods have been tested by Crampon et al. [102]. For freeze-drying, more humid cells resulted in higher extraction yields (same extraction conditions). Specifically, 18.4% water content resulted in 18.7% yield, while 8.5% and 4.3% water content led to 8.9% and 5.2% yield, respectively. Air dried Nannochloropsis with 20.4% water, yielded 22.6% and with 9.6% water content, 15.0%. Furthermore, the use of a more finely crushed biomass (<16 µm) led to a lower yield (10.3%) than that obtained with larger particles [102].

Scenedesmus
In the case of Scenedesmus, all of the investigated methods were mechanical, namely microwave, ultrasonication, homogenization, bead milling and grinding. The strains were lyophilized before being subjected to cell wall disruption and/or SFE. Unfortunately, even though pretreatment is commonly applied before SFE, there are very few publications investigating its impact on the extracts. For the recovery of carotenoids and other pigments, bead-milling of the Scenedesmus sample before extraction resulted in significantly higher yields [111].
Regarding lipid extraction, microwave pretreatment positively affects the yield, in particular, it has been noted an almost double lipid yield [113]. Nevertheless, the duration of the pretreatment with microwaves seems to reduce its effect, as shown by Hernández et al. [93]. Thus, 1 minute microwave pretreatment prior to SFE resulted in a higher yield than crude biomass, while 5 minutes pretreatment led to worse results compared to non-pretreated biomass.
Additionally, it was indicated that lyophilization as a pretreatment method does not affect FAME yields compared to fresh Scenedesmus samples [113]. However, it is mentioned that freeze-drying could possibly enhance the cell wall disruption in combination with other pretreatment techniques because of the increased specific area and the reduced diffusion gradient [154].

Other Cultures
Mechanical disruption methods as a pretreatment for enhanced extraction are also applied in other species. For instance, Halim et al. have extracted Chlorococcum, achieving 5.8% lipid yield with dried, and then ground in ring mill biomass, compared to 7.1% with wet biomass [68]. The effect of bead-milling prior to SFE has been tested in Pavlova cultures resulting in 17.9% lipid yield and 15.7% FAME yield for pretreated biomass, instead of 10.4% and 5.4% for crude biomass, respectively. Furthermore, grinding has been reported by Grierson et al. for Tetraselmis biomass [124]. Homogenization before extraction has also been used for Tetraselmis by Bong and Loh [103] and for Synechococcus by Cardoso et al. [17] and Macías-Sánchez et al. [75]. Hernández et al. have studied the effect of microwaves as a disruption method on the extraction yield of Tetraselmis [93]. For crude biomass, 14.8% yield has been achieved, while for 1-and 5-min pretreatment time the extraction yield was 4.7% and 5.2%, respectively. Microwaves combined with DES in Phaedactylum strains have increased lipid yield from 1% without pretreatment and 5.8% when only mixed with DES, to 6.6% for 30 min at 150 • C and 7.1% for 60 min at 100 • C. Finally, Montero et al. have attempted cell wall disruption by ultrasonication, but the method did not affect the extraction efficiency [122].

Principles and Process
Supercritical Fluid Extraction (SFE) is a green process for the recovery of compounds from a solid matrix using supercritical fluids as solvents. Fluids are in supercritical state when their temperature and pressure are above critical point (T c , P c ). They demonstrate properties such as low viscosity, density comparable to that of liquids, gas-like diffusion and near zero surface tension. Under these conditions, the extraction capacity of many compounds increases, therefore, supercritical fluids become a suitable solvent for a variety of applications [14]. The most commonly used solvent for SFE is supercritical CO 2 thanks to its low critical temperature (31.1 • C) and lack of toxicity, which allows the extraction of thermolabile compounds. Moreover, Sc-CO 2 is non-flammable, readily available, cost-effective and can be removed from the extracts by expansion to ambient conditions without any further processing, due to its gaseous state under atmospheric temperature and pressure [9,11]. Apart from that, in the supercritical region, solubility increases with the increase in density, which allows the regulation of selectivity by adjusting extraction conditions, such as temperature and pressure. For highly polar compounds, modifiers, such as alcohols, can be used in order to enhance the solubility. Furthermore, the yield and the selectivity of the process can be improved by the use of co-solvents. The above properties generate a highly selective extraction technique, resulting in extracts with high purity [11].

Arthrospira
Apart from γ-linolenic acid, which is the compound extracted in the majority of SFE applications, Arthrospira (Spirulina) can also provide extracts with high concentrations of carotenoids. Specifically, Canela et al. have recovered 2.27 mg/0.8 kg algae per extraction bead, at the optimal extraction conditions, namely a temperature of 30 • C, 18 MPa pressure and 11 hours extraction time [27]. Temperature, in that study, varied from 20 to 70 • C and pressure from 15 to 18 MPa. Valderrama et al. have achieved 3% phycocyanine yield and more than 97% astaxanthin recovery by extracting A. maxima strains at 60 • C and 30 MPa, both with and without the use of 10% w/w ethanol [28]. Similarly, experiments at 40-80 • C, 15-35 MPa and 5-15% v/v ethanol led to 48 mg/100 g biomass zeaxanthin, 7.5 mg/100 g biomass cryptoxanthin and 118 mg/100 g biomass β-carotene yield at 35 MPa and 15% v/v ethanol [29]. Also, in another study, the maximum amount of 283 µg/g biomass total carotenoids and 5.01 µg/g biomass total tocopherols have been recovered from A. platensis at 60 • C and 450 bar with 53.22% v/v ethanol [30]. SFE on pretreated A. platensis, also, resulted in extract composed of approximately 290 ppm zeaxanthin, 73 ppm myxoxanthophyl fucoside, 55 ppm β-carotene and 535 ppm chlorophyll a with antioxidant activity close to 70 µg/mL (EC 50 ) [34]. Additionally, Wang et al. have extracted at 48 • C, 20 MPa using ethanol as entrainer, 77.8 g β-carotene/kg biomass , 113.2 g vitamin a /kg biomass , 3.4 g α-tocopherol /kg biomass and 85.1 g flavonoids /kg biomass [35]. Finally, 6.84 mg/g biomass chlorophyll a was recovered from A. platensis at 53.4 • C and 48.7 MPa with 40% aq. ethanol [37].

Chlorella
Chlorella cultures can be used as a source of carotenoids, such as astaxanthin, canthaxanthin, lutein and β-carotene, chlorophylls and phenolic compounds. The extraction conditions, along with the use of co-solvent, can alter the extract's composition of bioactive compounds and, thus, their antioxidant activity.
Kitada et al. have studied the effect of pressure, temperature and co-solvent on the carotenoid extraction from C. vulgaris [59]. Specifically, at 70 • C, 2.5 mL/min flow rate and 300 min extraction time, the lutein extracted was 0.13, 0.46, 0.40 and 0.61 mg/g biomass at 20, 30, 40 and 50 MPa, respectively. The increase in temperature at a constant pressure of 30 MPa, increased the recovered lutein from 0.46 at 60 • C to 0.57 mg/g at 80 • C. The use of ethanol as co-solvent presented generally better results compared to acetone under the same conditions. Namely, 1.54 mg/g biomass lutein, 0.13 mg/g biomass β-carotene, 11.43 mg/g biomass α-chlorophyll and 3.90 mg/g biomass β-chlorophyll were recovered with ethanol and 0.94 mg/g biomass lutein, 0.01 mg/g biomass β-carotene, 3.30 mg/g biomass αchlorophyll and 0.59 mg/g biomass β-chlorophyll were recovered with acetone. Similarly, another study indicated that the increase in pressure at 40 • C led to higher lutein recoveries. More explicitly, at 20 MPa, 1.34% lutein recovery was achieved, at 30 MPa 1.64% and at 40 MPa 1.78% [64]. Temperature increase seemed to present the opposite effect at 40 MPa, by decreasing lutein recovery to 0.67% at 80 • C [64]. The flow rate of ethanol as entrainer resulted in 1.78% lutein recovery at 0.3 mL/min, in 1.80% at 0.4 mL/min and in 1.68% at 0.5 mL/min [64]. Gouveia et al. using extraction conditions of 40 • C, 30.0 MPa and 0.0397 kg/h Sc-CO 2 , have reported maximum total carotenoid recovery of 69.1% for completely crushed C. vulgaris cells without the use of co-solvent, while when mixed with oil and with double the flow rate the recovery obtained was 16.6% [58]. Fairly crushed and slightly crushed cells without the use of entrainers led to a recovery of 37.3% and 17.4%, respectively. Different co-solvents showed little impact on the carotenoid recovery since 19.7% was achieved with oil and 20.2% with ethanol. Safi et al. accomplished better results in overall extract characterization for bead milled C. vulgaris biomass by increasing pressure from 35 MPa to 60 MPa [52]. In terms of total mass recovered, at 60 MPa pressure 10.64% yield was achieved, in contrast to 9.7% at 35 MPa. Total carotenoids and total chlorophylls reached 60 MPa 1.72 mg/g dry biomass and 1.61 mg/g dry biomass , respectively.
Mendes et al. have investigated the effect of three operational conditions (temperature, pressure and pretreatment) on the carotenoid recovery [24]. The optimum carotenoid recovery for crude C. vulgaris, almost 500 mg/kg dry algae , was achieved at maximum temperature and pressure, i.e., 55 • C and 35 MPa. From the three degrees of crushing, whole, slightly, and well crushed, the second presented analogous results with the third, approximately 40% total carotenoids yield, but with larger requirements of Sc-CO 2 . In a similar study, under the same extraction conditions, best results were derived for the most intense extraction conditions for both crude and pretreated biomass, i.e., 171.1 mg carotenoids per 100 g oil and 0.05% w/w carotenoid yield [61,62]. Hu et al. have carried out an orthogonal experimental design that consisted of 16 experiments, where each factor consisted of four levels, in order to examine the effect of five factors (temperature, pressure, duration, Sc-CO 2 flow rate and co-solvent quantity) on extraction yield and antioxidant capacity [46]. Yield reached its maximum value, 7.78%, at 32 • C, 40 MPa, 20 kg/h Sc-CO 2 flow rate, 180 min and 1 mL ethanol per gram of C. pyrenoidosa. The inhibition at those conditions was 42.03%, while the optimum was 54.16% with 3.50% yield at 40 • C, 35 MPa, 20 kg/h Sc-CO 2 flow rate, 150 min and 1.5 mL/g ethanol. Consequently, the most effective parameters were pressure for yield and modifier for antioxidant activity. Georgiopoulou et al. studied the SFE of C. vulgaris and specifically the effect of temperature, pressure and solvent flow rate on total extraction yield, antioxidant activity, total phenolic content and target carotenoid compounds, by applying experimental design [66]. The experiment under the optimum conditions (60 • C, 250 bar and 40 g Sc-CO 2 /min) resulted in 3.37% yield, 44.35 mg extr /mg DPPH antioxidant activity using an IC 50 assay, total phenolic content equal to 18.29 mg gallic acid/g extract , 35.55 mg/g extract total chlorophyll content, 21.14 and 10.00 mg/g extract total and selected carotenoid content, respectively. Furthermore, the addition of 10% w/w ethanol as entrainer enhanced antioxidant activity and yield. Wang et al. investigated the properties of the extract obtained by the SFE of Chlorella at 50 • C, 31 MPa, 6 Nl/min and the use of 50% aqueous ethanol [65]. The total polyphenol content of the extract was 13.40 mg GAE /g extract , while the total flavonoid content was 3.18 mg QE /g extract . The inhibition value in the DPPH assay was 47.24% compared to gallic acid's 100% inhibition. In other research, in which experimental design was employed, the recovery of lutein from superfine pulverized C. pyrenoidosa with the use of ethanol as entrainer, reached its maximum value, 87.0% extraction yield. The conditions of that experiment were 50 • C, 25 MPa, 240 min duration and 50% w/v ethanol [47].

Haematococcus
Haematococcus pluvialis has gained significant research interest due its high content of natural astaxanthin [158]. Yothipitak et al. have estimated that the recovery of astaxanthin could reach 22.66 mg/g biomass by SFE at high pressure and temperature (64 MPa and 90 • C) [80]. SFE, with or without the use of co-solvent, appears to be an adequate technique for astaxanthin extraction, reaching, in certain cases, more than 80% recovery. Extraction of lyophilized H. pluvialis at 45 • C, 48.3 MPa and 2.7 mL/min Sc-CO 2 flow rate, led to almost 85% astaxanthin recovery [85]. Likewise, 83% recovery, equal to 22.84 mg/g biomass , was achieved at slightly higher pressure and flow rate (50 MPa and 3 mL/min) and 80 • C [81]. Moreover, ethanol as co-solvent has been widely investigated. Bustamante et al. recovered 84% of biomass astaxanthin at 40 • C and 55 MPa with the addition of 4.5 v/v ethanol [82] and, correspondingly, Pan et al. recovered 73.9% by using 9.23 mL/g biomass of aqueous ethanol under moderate conditions [83]. Similar studies of SFE at 70 • C and 40 MPa with 5% v/v ethanol led to 80.6% astaxanthin recovery [87], while at 65 • C, 43.5 MPa with 2.3 mL/g ethanol and at 55 • C, 20 MPa with 13% w/w ethanol, the recovery obtained was 87.4% and 82.3%, respectively [89,90]. SFE of powdered biomass resulted in 61% astaxanthin recovery at 70 • C and 55 MPa [84], while SFE of lyophilized and crushed H. pluvialis with 9.4% w/w ethanol as co-solvent led to a recovery of 92% of total carotenoids, 76% of βcarotene and 90% of astaxanthin [86]. Dried H. pluvialis extraction with 10% v/v olive oil as co-solvent under optimum conditions (70 • C, 40 MPa) resulted in 51% recovery of available astaxanthin [88]. Finally, extraction of red phase Haematococcus at 65 • C and 55 MPa resulted in high astaxanthin and lutein recoveries, 92-98.6% and 52.3-93%, respectively [91,92]

Scenedesmus
Scenedesmus cells contain both carotenoids and chlorophylls that can be recovered by SFE with or without the use of co-solvent [159]. A lutein recovery has been reported for S. almeriansis of 0.0466 mg/g biomass at 60 • C, 400 bar and extraction duration of 300 min [111]. Also, for the same species, another study reports a recovery of 2.97 mg/ g biomass of lutein for a shorter extraction time, but increased temperature and pressure, i.e., 65 • C and 550 bar [112]. The addition of a polar co-solvent in the SFE could affect the extraction of the target compounds by increasing the solvent's polarity, and therefore, their solubility in the medium [160]. Indeed, the lutein yield seemed to have been augmented from 0.206 mg/g biomass to 2.210 mg/g biomass by adding 30% v/v ethanol maintaining the same temperature, pressure and time [118]. Similarly, the yield increased from 0.2105 mg/g biomass lutein to 0.4361 mg/g biomass with the addition of 10% v/v ethanol [119]. Remarkably, the extraction conditions which lead to the maximization of the lutein yield does not always match with the most intense ones. The same phenomenon is observed for β-carotene and lutein extraction, which both reach their maximum recovery (1.5 mg/g biomass and 0.047 mg/g biomass, respectively) at 60 • C, 400 bar and 300 minutes total extraction [111]. In this case, co-solvent contribution seems to be not so intense, since the use of 10% v/v ethanol led to the increase in the extracted β-carotene from 0.0547 mg to 0.0599 mg per dry biomass [119]. As a result, the best total carotenoid recovery does not occur under very intense extraction conditions. For example, SFEat 40 • C, 400 bar, and 2 h duration resulted in a recovery equal to 48.39 mg/g extract and 0.303 mg/g biomass at 250 bar, the same temperature and double duration [114,131]. Additionally, more carotenoids were detected, such as astaxanthin, neoxanthin, violaxanthin and zeaxanthin, and the recovery of all of them, except for violaxanthin appeared to increase with the use of co-solvent [119].
In terms of chlorophylls, they seem to have similar behavior to carotenoids. At 50 • C, 250 bar, and extraction time equal to 120 min, 15.68 mg/g extract of chlorophylls were recovered [114]. Chlorophyll a is extracted in larger quantities in contrast to chlorophyll c. For example, Guedes et al. extracted 0.848 mg/g biomass of chlorophyll a while chlorophyll b and c quantities obtained were 0.356 mg/g biomass and 0.018 mg/g biomass , respectively [131].
The extraction yields reported in the various studies show significant diversity, possibly due to different species, different cultivation and different SFE conditions. The species obliquus presents the lowest yields among them all. The highest cited is 8.3% at 20 • C, 120 bar and 540 min total extraction time [2]. Also, SFE at 40 • C, 400 bar for 120 min resulted in 1.15% yield as reported by Gilbert-López et al. [114], while Choi et al. obtained a yield of 4.20% under almost the same conditions [115]. By the addition of 15% v/v ethanol as co-solvent, the latter yield was increased to 14.51% [115]. However, other research presented a 0.247% yield with 5% v/v ethanol at 65 • C, 300 bar and for 90 min, which deviates significantly from the results of the other researchers [44].
The SFE of the species almeriensis at 60 • C, 400 bar and 120 min total extraction time, led to 1.50% yield [112]. Similarly, SFE at 45 • C, 300 bar and 90 min with the addition of 5% v/v ethanol resulted in 19.4% yield [93]. The extraction of species of obtusiusculus at 20 • C, 120 bar and 540 min resulted in a yield of 6.4% [2]. Ultimately, SFE of unspecified Scenedesmus species led to yields up to 6.81% [120].

Other Cultures
In addition to the species mentioned above, Dunaliella salina cultures are also a major carotenoid and chlorophyll source. Specifically, extraction carried out at 40 MPa and 60 • C recovered 12.17 µg/mg biomass carotenoids and 0.227 µg/mg biomass chlorophylls [74]. By using 5% mol ethanol as co-solvent, under the same conditions, the yield altered to 9.629 µg/mg biomass carotenoids and 0.700 µg/mg biomass chlorophylls [76]. Similarly, Pour Hosseini et al., at slightly lower temperature and without co-solvent, obtained 115.44 µg/g biomass total carotenoids and 32.68 µg/g biomass chlorophylls [77]. Under milder conditions, namely 45 • C and 20 MPa with 5% w/w ethanol, Molino et al. recovered 25.5% of β-carotene from D. salina [78]. Total carotenoid content was also determined at 27.5 • C, 44.2 MPa and 45 • C, 20 MPa and found to be equal to 7.2 mg/100 g extract and 25 g/kg biomass , respectively [72,79].
Lastly, carotenoids such as β-carotene, β-cryptoxanthin and zeaxanthin were recovered from Synechococcus sp. Explicitly, maximum recovery 71.6%, 90.3% and 36.4%, of β-carotene, β-cryptoxanthin and zeaxanthin, respectively, was achieved [122]. Additionally, the SFE at 40 • C,40 MPa and 5% mol ethanol performed by Cardoso et al., resulted in 20.35 mg/g extract β-carotene and 25.96 mg/g extract zeaxanthin [17]. The addition of ethanol as co-solvent appears to have a positive effect on the pigment extraction. Macías-Sánchez et al., by using 5% mol ethanol under the same extraction conditions, achieved an increase from 1.51 to 1.86 µg/g biomass in carotenoid recovery and from 0.078 to 0.286 µg/g biomass in chlorophyll recovery [76,123].

Arthrospira
The most common fatty acid extracted through SFE from Arthrospira cultures is GLA and, in general, an alcohol as co-solvent is used. GLA yield equal to 0.44% was achieved by conducting SFE of A. maxima at 60 • C, 35 MPa, 2 g/min solvent flow rate and 10% v/v ethanol [24][25][26]. Sajilata [57].
Lipid recovery from Chlorella by applying SFE was mainly conducted with the use of co-solvent. In detail, SFE of Chlorella sp. with 5% ethanol at 60 • C and 30 MPa led to 79.53% lipid yield [54]. Also, at lower pressure while using 0.4 mL/min hexane, lipid yield was determined as 63.78% [53]

Scenedesmus
The EFA with the highest concentration in the lipid extracts of Scenedesmus by SFE was found to be α-linolenic acid (ALA). Specifically, for the species obliquus, when extracted at 45 • C and 150 bar for 30 minutes, the percentage of ALA in the extracted lipids reached 21.47% [44], while in other research it was found to be equal to 28.44% by conducting extraction at 20 • C and 120 bar for 540 min total extraction time [2]. The concentration of LA in the aforementioned cases was 10.33% and 10.21%, respectively. It should be noted that the optimum extraction conditions, regarding the highest concentration of ALA and LA in the extracts, coincide. Contrariwise, an almost four times higher concentration of LA compared to ALA in S. obliquus extracts obtained by SFE at 40 • C and 379 bar is reported [115]. Moreover, for the species obstusiusculus, less ALA and LA were recovered in comparison with obliquus under the same conditions [2]. S. almeriensis extracts, in contrary to other species, contain 2.9% LA while no ALA was detected. However, these extracts contained more EPA (7.9%) compared to those of obliquus and obstusiusculus species which had less than 0.59% [

Kinetic Models
The mathematical modeling of SFE in solid matrixes provides valuable information about the course of extraction. Using as independent variables, the operational conditions, such models describe the progress of the extraction over time, making the optimization and the simulation of the process possible [161,162]. The solid particles are usually depicted as spheres or cylinders and the mass transfer phenomena occurring in the biomass can be described by linear driving force models, shrinking core models, broken plus intact cell models and the combination of the latter [162]. Some hypotheses can be made in order to simplify the kinetic models, such as immobilized cells with constant density and porosity and isothermal and isobaric conditions in the extractor [162].

Broken Plus Intact Cell Model
This model based on Lack's plug flow model was proposed by Sovová and coworkers [161,163], and assumes that cell walls function as an additional resistance to the extraction of the solute. Grinding of the biomass results in disrupted and intact cells where the solute transfers to the supercritical phase through convection and molecular diffusion, respectively [162]. The extract primarily gets exhausted from the broken cells and gradually from the intact, resulting in three mass transfer periods. Initially, the extraction rate increases constantly and then falls progressively, ending up in a diffusion controlled period [164]. Sovová [44]. Other studies involve Chlorella vulgaris [55,66], Haematococcus pluvialis [82] and Nannochloropsis gaditana [97].

Other Models
Apart from models such as the linear driving force model (LDF) and shrinking core model, desorption, solubility based on Fick's diffusion law models are often employed for the description of the SFE process on microalga. Examined species are A. maxima and A. platensis [25,27], C. protothecoides [43], Chlorococcum sp., Synechococcus sp., D. salina, N. gadiatana [75] and Nannochloropsis sp. [104].

Maceration
Maceration, is a commonly used method for microalgae extraction. Specifically, for A. maxima, maceration was conducted by using as solvent hexane, ethanol or acetone under ambient conditions in order to determine its lipid and GLA content [24][25][26]. Similarly, for A.pacifica, methanol with acetyl chloride as solvent was used for GLA recovery [32] and hexane for lipid yield [39]. Gouveia et al. used soy bean oil and acetone extraction for total lipid determination at 25 • C and 100 • C on C. vulgaris [58]. Also, the latter for the same species was determined with hexane and acetone maceration by Mendes et al. [61]. Lipid content of Chlorococcum sp. was specified by hexane and isopropanol/hexane extraction [68] while for P. tricornutum DMC was employed as solvent [109].
Hydrocarbon content of B. braunii was determined by using hexane [24,40]. Morcelli et al. by using ethyl acetate and methanol measured the concentration of violaxanthin, lutein and total carotenoids for C. sorokiana [49]. Total carotenoid content was determined by employing acetone for C. vulgaris [40], N. gaditana [97], Nannochloropsis sp. [105] and S. obliquus [116]. The latter study also estimated the extract's composition regarding chlorophyll α, b and c. Relatedly, maceration with acetone led to astaxanthin extraction from H. pluvialis [84,85]. Among others, acetone was also utilized to recover lutein from Scenedesmus sp. [118] and S. almeriansis [111], as well as for the determination of total extractable compounds for S. obliquus [114], and for β-carotene extraction from S. almeriansis [111]. Other solvents, such as alcohols, were additionally used for pigment extraction. Methanol maceration was employed for total carotenoid and chlorophyll content determination in the case of D. salina [77]. Similarly, ethanol extractions were performed on I. galbana for the determination of total extractable compounds [94], on Monoraphidium sp. for astaxanthin and total chlorophyll recovery [95] and on C. vulgaris for astaxanthin, lutein, β-carotene and total chlorophyll content determination [66]. Lutein recovery from Scenedesmus sp. was achieved by using various solvents, such as methanol, ethanol, propanol and butanol [118]. Finally, ethyl acetate maceration was used for total carotenoid extraction from Nannochloropsis sp. [105] and tetrahydrofuran with methanol for zeaxanthin, β-carotene and β-cryptoxanthin recovery [29].

Soxhlet
The Soxhlet technique is commonly used as a reference method for the determination of total extractable content of the solid matrix. Its application to microalgae can lead to the extraction of lipids, chlorophylls and bioactive compounds. By using this method with hexane, total lipid extraction was achieved for C. protothecoides [43], C. vulgaris [55,56], Chlorococcum sp. [68], N. granulata [99], N. oculata [101], Nannochloropsis sp. [104,105], Pavlova sp. [108] and Scenedesmus sp. [120]. Additionally, FAME recovery was performed for N. granulata [99] and Pavlova sp. [108], as well as, SFA and PUFA extraction from Nannochloropsis sp. s [104]. The mixture of methanol/chloroform is also widely used for lipid content determination of biomass. Soxhlet extraction using methanol/chloroform was performed in the case of C. vulgaris to recover neutral lipids, phospholipids and glycolipids [63].
Fatty acids were also recovered by using the Bligh and Dyer protocol. Indicatively, total FA content was determined for B. braunii [41] and free FA conversion for N. oculata [102]. For the latter, triglycerides and sterols were extracted similarly. Additionally, total FA, polyunsaturated FA and EPA content of Phaeodactylum tricornutum were determined [109]. γ-Linolenic acid was extracted from A. maxima [25,26] and from A. platensis, assisted by ultrasonication [32].

Ultrasound Assisted Extraction
The present method is suitable for the recovery of heat-sensitive substances due to low temperatures, even ambient ones, during the extraction. Also, it has a shorter duration than conventional extraction methods and generally presents a higher yield. The process is fairly simple and the equipment required is readily available and relatively inexpensive [11]. In literature, many solvents have been used for the UAE of bioactive compounds and lipids, most of them being alcohols. Namely, methanol was used to extract carotenoids and chlorophylls from D. salina, N. gadiatana and Synechococcus sp. [74,76,96,123], while mixed with ethyl acetate, it recovered FAME and lipids from Pavlova sp. [108] and commercial DHA algae [70], respectively. Aqueous ethanol was employed for quercetin extraction from C. vulgaris [65]. Carotenoids and fatty acids were extracted using DMF. Specifically, total chlorophyll and carotenoid contents of D. salina, N. gaditana and Synechococcus sp. were determined [74,76,122], as well as, myxoxanthophyl, β-carotene, β-cryptoxanthin, zeaxanthin, oleic, linoleic, palmitic and palmitoleic acid content of the latter species [17,122].

Microwave Assisted Extraction
Microwave assisted extraction (MAE) is a non-conventional method which uses electromagnetic waves, with frequencies of 2.45 GHz approximately, in order to recover analytes from solids [12,169]. The extraction process is a result of the synergistic combination of bipolar rotation and ionic conduction [169]. Bipolar rotation happens to solvent's and matrix's molecules that have a dipole moment when applying electric field, disrupting weak hydrogen bonds [169]. Those phenomena cause the release of thermal energy, increasing the temperature of the solution. Optimal results can be achieved using solvents with higher dielectric constants [169]. High extraction yields for natural matrices can be obtained due to the effect that an electric field has on cell structure [170]. Namely, the traces of water that exist inside the dried material evaporate, increasing intracellular pressure and, thus, creating ruptures in the cell wall [171]. Esquivel-Hernandez et al. extracted 2.46 µg/g tocopherols and 629 µg/g total carotenoids from A. platensis using a mixture of methanol, ethyl acetate and light petroleum (1:1:1 v/v/v) at 50 • C [30].

Conclusions
An in-depth investigation of the literature on the field of SFE application for the recovery of valuable extracts from microalgae has been performed and presented in comprehensive and easily read Tables. SFE using CO 2 as solvent is suitable for the extraction of solvent-free, high-quality products that, due to the low to moderate operating temperatures applied, maintain their bioactive properties.
A total of thirty-eight different microalgae species are included in this study, and SFE operating conditions are presented along with the extracts' yield, bioactive compounds content and properties. Modeling attempts of the extraction process are also reported as such information is important for the optimization and scale-up of the process. Finally, other extraction methods-if available-are briefly presented for comparison purposes.
Arthrospira (Spirulina), Chlorella, Dunaliella, Haematococcus and Nannochloropsis are the most investigated microalgae in the literature regarding SFE, which results in promising extracts for applications in either food and cosmetics or biofuels industries.

Conflicts of Interest:
The authors declare no conflict of interest.