Pressurized Liquid Extraction of Polyphenols and Anthocyanins from Saffron Processing Waste with Aqueous Organic Acid Solutions: Comparison with Stirred-Tank and Ultrasound-Assisted Techniques

of Polyphenols Anthocyanins from Saffron Processing Waste Aqueous Organic Acid Abstract: Follow up with our previous study on the extraction of saffron processing waste polyphenols using deep eutectic solvents, the objective of this examination was a comparative evaluation of pressurized liquid extraction (PLE), stirred-tank extraction (STE) and stirred-tank extraction with ultrasonication pretreatment (STE/UP) with respect to the recovery of pigments and antioxidant polyphenols from saffron processing waste. Aqueous solutions of citric and lactic acids at two different concentrations were used as green solvents. The extracts obtained under the speciﬁed conditions were analyzed for total pigment and total polyphenol yields as well as for their ferric-reducing power and antiradical activity. Furthermore, each produced extract was analyzed with liquid chromatography–mass spectrometry to proﬁle its analytical polyphenolic composition. In all cases, PLE provided inferior results compared to the two other techniques, producing extracts with lower polyphenolic concentration and weaker antioxidant properties. On the other hand, no speciﬁc pattern was detected concerning the effect of ultrasonication, acid type and acid concentration. Hierarchical cluster analysis indicated that stirred-tank extraction with 1% ( w / v ) lactic acid and ultrasonication pretreatment might be the highest-performing combination, providing extracts with increased polyphenol and pigment concentration; however, it also enhanced antioxidant activity. It was also concluded that the signiﬁcantly shorter extraction time when using PLE might be an important element in further optimizing the process, buttressing the use of this technique for the establishment of innovative and sustainable-by-design extraction


Introduction
As the world's population and industrial activity are rapidly expanding, resource depletion and environmental aggravation are challenges which need to be imminently addressed [1]. The agricultural and food industries are responsible for a large share of waste and byproducts generated as a result of farming practices and the harvesting and processing of raw materials. These side-streams are particularly rich in organic substances, and their uncontrolled dumping results in environmental pollution with detrimental consequences to the neighboring eco-systems and public health [2]. On the other hand, agri-food waste represents a vast pool of materials which can be used for the production of bio-based chemicals, bio-fuels, and high value-added substances [3]. Thus, in the framework of the circular economy, the rational utilization of agri-food waste biomass within a biorefinery concept may contribute to a fully sustainable agri-food sector.

Plant Material
Saffron (Crocus sativus L.) processing waste (SPW) consisting of saffron tepals was collected from a processing plant in Kozani (West Macedonia, Greece) immediately after manual processing of the saffron flowers. The material was transferred to the laboratory within 24 h and freeze-dried in a Biobase BK-FD10P freeze-drier (Shandong, China) for 24 h. The dried tissue was then ground in a domestic blender to give a powder with an average particle diameter of 0.637 mm and stored in air-tight tubes at −18 • C until used.

Pressurized-Liquid Extraction (PLE)
The equipment used was a PLE system (Fluid Management Systems, Inc., Watertown, MA, USA). An analytical description of the device is given in Figure 1. For the purposes of this study, static extraction was performed using a 100-mL stainless steel cell. An amount of SPW (2 g) was loaded onto the extraction cell and capped with two filtration end fittings. The PLE system then automatically pressurized and heated the sample while pumping solvent into the chamber for a predetermined resident time and solvent flow. The specific settings used were: filling (with solvent) time, 1.3 min; pressurization, 0.5 min; preheating (at 120 • C) time, 5 min; extraction (at 120 • C) time, 10 min; cooling (at T < 50 • C), 7 min; depressurization time, 0.02 min; solvent flush, 1.3 min; nitrogen flush, 1 min. Under these conditions, the liquid-to-solid ratio was 40 mL g −1 .
During heating, a temperature rise (overshooting) was observed which in no case exceeded 4 • C, while the temperature set was restored within 1-2 min. After the completion of the extraction, the extract was drained and transferred to a suitable vial for further processing or analysis. The choice of the extraction temperature and time was based on previous data [20][21][22], as well as preliminary experimentation. The solvents used were distilled water (pH = 5.61), 1% (w/v) citric acid (pH = 2.88), 5% (w/v) citric acid (pH = 2.42), 1% (w/v) lactic acid (pH = 2.71), and 5% (w/v) lactic acid (pH = 2.48).

Stirred-Tank Extraction in Batch Mode
An exact mass of 0.250 g of SPW was mixed with 10 mL of solvent (liquid-to-solid ratio 40 mL g −1 ) in a 25-mL Duran glass vial and stirred continuously at 500 rpm for 180 min at 80 • C into an oil bath heated by a thermostat-equipped hotplate (Witeg, Wertheim, Germany). After the extraction, all samples were centrifuged at 10,000× g for 10 min.

Batch Stirred-Tank Extraction with Ultrasonication Pretreatment
Prior to batch stirred-tank extraction samples were subjected to ultrasonication in pulse mode for 15 min in an Elma S 100 (H) heated ultrasonic bath (Elma Schmidbauer GmbH, Singen, Germany) at a frequency of 37 Hz and nominal power of 550 W. During this resident time, increases in the initial temperature (31 • C) in no case exceeded 6 • C. The actual ultrasonication power (P) dissipated into the system, as well as the acoustic energy density (AED), were determined using the following equations: where m corresponds to the mass of the coupling liquid (water) contained in the ultrasonication bath (in g), C p corresponds to the specific heat capacity of water (4.2 J g −1 K −1 ), and dT dt represents the temperature rise per s, which was calculated by fitting temperature change (dT) as measured by a thermocouple as a function of time [23]. P and AED were determined to be 159.6 W and 39.9 W L −1 , respectively.

Stirred-Tank Extraction in Batch Mode
An exact mass of 0.250 g of SPW was mixed with 10 mL of solvent (liquid-to-solid ratio 40 mL g −1 ) in a 25-mL Duran glass vial and stirred continuously at 500 rpm for 180 min at 80 °C into an oil bath heated by a thermostat-equipped hotplate (Witeg, Wertheim, Germany). After the extraction, all samples were centrifuged at 10,000× g for 10 min.

Batch Stirred-Tank Extraction with Ultrasonication Pretreatment
Prior to batch stirred-tank extraction samples were subjected to ultrasonication in pulse mode for 15 min in an Elma S 100 (H) heated ultrasonic bath (Elma Schmidbauer GmbH, Singen, Germany) at a frequency of 37 Hz and nominal power of 550 W. During this resident time, increases in the initial temperature (31 °C) in no case exceeded 6 °C. The actual ultrasonication power (P) dissipated into the system, as well as the acoustic energy density (AED), were determined using the following equations:

Determinations
Determination of total polyphenols was performed with Folin-Ciocalteu reagent and results were given as mg gallic acid equivalents (GAE) per g of dry mass (dm). The antiradical activity (A AR ) was estimated with a DPPH radical probe using a stoichiometric assay, and results were given as µmol DPPH per g dm. Ferric-reducing power (P R ) was determined using the TPTZ chromophore probe; results were expressed as µmol ascorbic acid equivalents (AAE) per g dm. The analytical protocols for all these methodologies have been previously reported in detail [24]. Total pigments were likewise determined with a protocol reported elsewhere [25].

Liquid Chromatography-Diode Array-Mass Spectrometry (LC-DAD-MS)
A published methodology was employed [24] using a Finnigan AQA mass spectrometer coupled to a Finnigan (San Jose, CA, USA) MAT Spectra System P4000 pump and a UV6000LP diode array detector. Chromatography was performed on a Fortis RP-18 column, 150 mm × 2.1 mm, 3 µm, at 40 • C, with a 10-µL injection loop. Mass spectra were acquired at 20 and 70 eV with electrospray ionization (ESI) in positive ion mode. Mass acquisition settings were as follows: temperature 250 • C, probe source voltage 25 V, detector voltage 450 V, and capillary voltage 4 kV. Elution was carried out with (A) 2% acetic acid and (B) methanol at a flow rate of 0.3 mL min −1 , as follows: 0-30 min, 0-100% methanol; 30-40 min, 100% methanol.

Statistical Analyses
At least two individual extractions were accomplished for each treatment, and all determinations were performed in triplicate. The values reported are the average ± standard deviation (SD). Linear regressions were established using SigmaPlot™ 12.5 (Systat Software Inc., San Jose, CA, USA), and distribution analyses with JMP™ Pro 13 (SAS, Cary, NC, USA). All statistical analyses were performed with at least at a 95% significance level. Figure 2A illustrates the yield in total polyphenols (Y TP ) achieved with the various solvents tested through deployment of pressurized liquid extraction (PLE), stirred-tank extraction (STE) and stirred-tank extraction with ultrasonication pretreatment (STE/UP). The highest Y TP (38.24 mg GAE g −1 dm) was recorded for the STE/UP performed with 1% (w/v) lactic acid (LA). Extractions carried out with 5% (w/v) citric acid (CA) and 5% LA under the same conditions had comparable performance, since there was no statistical difference amongst the Y TP obtained (p < 0.05). The same held true for the extractions carried out with water, 1% CA, 5% CA and 5% LA in stirred-tank mode without ultrasonication pretreatment. These findings show that there was no clear effect of either the type of the acid or its concentration on extraction efficiency. Furthermore, ultrasonication pretreatment offered virtually no advantage over simple stirred-tank extraction. On the other hand, all samples generated with PLE displayed lower Y TP , which reached statistical significance (p < 0.05) irrespective of the acid used or its concentration.

Yield in Total Polyphenols and Total Pigments
The results on the yield in total pigments (Y TPm ) provided a different image ( Figure 2B). The extractions carried out with 1% CA, 1% LA and 5% LA in stirred-tank mode, and those with 1% CA and 1% LA in stirred-tank mode with ultrasonication pretreatment exhibited the highest and most statistically significant Y TPm (p < 0.05). On the contrary, PLE performed with any of the extraction media tested showed significantly lower Y TPm (p < 0.05). In this case as well, a distinction between CA and LA or between STE and STE/UP was not apparent. Taking into consideration both Y TP and Y TPm , the samples prepared with 1% CA-STE, 5% CA-STE, 5% LA-STE, and 1% LA STE/UP were those with the richest composition.
hibited the highest and most statistically significant YTPm (p < 0.05). On the contrary, PLE performed with any of the extraction media tested showed significantly lower YTPm (p < 0.05). In this case as well, a distinction between CA and LA or between STE and STE/UP was not apparent. Taking into consideration both YTP and YTPm, the samples prepared with 1% CA-STE, 5% CA-STE, 5% LA-STE, and 1% LA STE/UP were those with the richest composition.

Effect on the Antioxidant Properties
The results on the determination of reducing power (PR) and antiradical activity (AAR) are depicted in Figure 3. Extracts produced with water, 5% CA and 5% LA in stirred-tank mode had significantly higher AAR (p < 0.05). The same was observed for the extract generated with 5% CA in stirred-tank mode after ultrasonication pretreatment. By contrast, all extracts produced with PLE showed lower AAR (p < 0.05) ( Figure 3A). The pattern concerning PR was essentially similar, with the samples prepared with 1% CA-STE, 5% CA-STE, 5% LA-STE and 5% CA STE/UP showing the highest values. On the other hand, all PLE samples as well as 1% LA-STE/UP had significantly lower PR ( Figure 3B). Considering both AAR and PR, the samples prepared with 5% CA-STE, 5% LA-STE and 5% CA-STE/UP were the most efficacious in terms of expressing antioxidant activity.

Effect on the Antioxidant Properties
The results on the determination of reducing power (P R ) and antiradical activity (A AR ) are depicted in Figure 3. Extracts produced with water, 5% CA and 5% LA in stirred-tank mode had significantly higher A AR (p < 0.05). The same was observed for the extract generated with 5% CA in stirred-tank mode after ultrasonication pretreatment. By contrast, all extracts produced with PLE showed lower A AR (p < 0.05) ( Figure 3A). The pattern concerning P R was essentially similar, with the samples prepared with 1% CA-STE, 5% CA-STE, 5% LA-STE and 5% CA STE/UP showing the highest values. On the other hand, all PLE samples as well as 1% LA-STE/UP had significantly lower P R ( Figure 3B). Considering both A AR and P R , the samples prepared with 5% CA-STE, 5% LA-STE and 5% CA-STE/UP were the most efficacious in terms of expressing antioxidant activity.

Effect on the Flavonol and Anthocyanin Profile
The extract generated from each assay was subjected to HPLC analysis in order to portray the analytical polyphenolic composition and detect differences attributed to different extraction modes. Chromatograms were traced at both 360 and 520 nm; in all cases, seven principal compounds could be tentatively identified and quantified: four flavonol glycosides and three anthocyanin pigments (Figure 4). The tentative identification of these substances was based on mass spectral data, as previously described [24]; the quantitative data concerning flavonol and anthocyanin composition are given in Table 1 and Table 2, respectively.

Effect on the Flavonol and Anthocyanin Profile
The extract generated from each assay was subjected to HPLC analysis in order to portray the analytical polyphenolic composition and detect differences attributed to different extraction modes. Chromatograms were traced at both 360 and 520 nm; in all cases, seven principal compounds could be tentatively identified and quantified: four flavonol glycosides and three anthocyanin pigments (Figure 4). The tentative identification of these substances was based on mass spectral data, as previously described [24]; the quantitative data concerning flavonol and anthocyanin composition are given in Tables 1 and 2, respectively. sition in the samples produced with PLE was identical to those seen in samples generated with STE and STE/UP, which showing that there was no selectivity towards any SPW flavonol. The extraction with 1% LA-STE/UP was proven to be the most efficient for all flavonols, affording significantly higher yields (p < 0.05). On the contrary, extractions with 1% CA-PLE, 5% CA-PLE and 5% LA-PLE were the least efficient in this regard. Considering the total flavonol yield, the extractions with Water-STE, Water-STE/UP, 1% LA-STE/UP and 5% LA-STE/UP were of equivalent performance. These findings suggest that ultrasonication pretreatment enabled the extraction of increased flavonol amounts. On the other hand, the use of citric acid appeared to disfavor flavonol recovery.
With respect to anthocyanins, extraction with 1% LA-STE/UP was once again the most efficacious, while 1% CA-STE, 5% CA-STE, 5% LA-STE and 5% LA-STE/UP were of comparable efficiency.   The most abundant flavonol glycoside was kaempferol 3-O-sophoroside (K S ), followed by kaempferol 3-O-sophoroside 7-O-glucoside (K SG ), quercetin 3-O-sophoroside (Q S ) and kaempferol 3-O-glucoside (K G ). This finding was in accordance with recent findings on SPW extraction with a deep eutectic solvent [24]. The pattern of flavonol composition in the samples produced with PLE was identical to those seen in samples generated with STE and STE/UP, which showing that there was no selectivity towards any SPW flavonol. The extraction with 1% LA-STE/UP was proven to be the most efficient for all flavonols, affording significantly higher yields (p < 0.05). On the contrary, extractions with 1% CA-PLE, 5% CA-PLE and 5% LA-PLE were the least efficient in this regard. Considering the total flavonol yield, the extractions with Water-STE, Water-STE/UP, 1% LA-STE/UP and 5% LA-STE/UP were of equivalent performance.
These findings suggest that ultrasonication pretreatment enabled the extraction of increased flavonol amounts. On the other hand, the use of citric acid appeared to disfavor flavonol recovery.
With respect to anthocyanins, extraction with 1% LA-STE/UP was once again the most efficacious, while 1% CA-STE, 5% CA-STE, 5% LA-STE and 5% LA-STE/UP were of comparable efficiency.
However, unlike flavonols, PLE afforded significantly higher levels of delphinidin 3,5-di-O-glucoside (D DG ) compared to both STE and STE/UP. This outcome indicates that D DG recovery might be favored with PLE. On average, the most abundant anthocyanin was delphinidin 3-O-glucoside (D G ), followed by petunidin 3,5-di-O-glucoside (P DG ) and delphinidin 3,5-di-O-glucoside (D DG ). These findings contrast with a recent investigation in which it was demonstrated that D DG was the predominant anthocyanin [24,26].
Taking into account the total anthocyanin yield, it was evident that STE and STE/UP were of higher efficiency compared to PLE; however, the distinction between STE and STE/UP was unclear. By jointly considering the yield in total flavonols and total anthocyanins, the highest-performing system was 1% LA-STE/UP. This was corroborated by the data on Y TP , Y TPm (Figure 2), A AR and P R (Figure 3). To confirm these observations, a hierarchical cluster analysis was performed including the yield in all individual polyphenols as well as A AR and P R . As can be seen in Figure 5, 1% LA-STE/UP was clustered separately, which could be considered sound evidence of its supremacy over all other extracts. Apart from Water-PLE, all other PLE samples were grouped together, which clearly points out their similarity in extraction yield and antioxidant properties.
were of higher efficiency compared to PLE; however, the distinction between STE and STE/UP was unclear. By jointly considering the yield in total flavonols and total anthocyanins, the highest-performing system was 1% LA-STE/UP. This was corroborated by the data on YTP, YTPm (Figure 2), AAR and PR (Figure 3). To confirm these observations, a hierarchical cluster analysis was performed including the yield in all individual polyphenols as well as AAR and PR. As can be seen in Figure 5, 1% LA-STE/UP was clustered separately, which could be considered sound evidence of its supremacy over all other extracts. Apart from Water-PLE, all other PLE samples were grouped together, which clearly points out their similarity in extraction yield and antioxidant properties. This outcome suggests that under the PLE conditions employed, the addition of citric acid or lactic acid does not foster extraction efficiency or antioxidant activity. Likewise, 5% CA-STE, 5% LA-STE, 5% CA-STE/UP and 5% LA-STE/UP were on the same cluster, an indication of their comparable efficiency. Thus, it can be supported that extraction with aqueous solution containing 5% of either citric or lactic acid showed no significant differences, while ultrasonication pretreatment offered no detectable statistically significant advantage. It is also to be noted that the categorization of Water-STE and Water-STE/UP in the same cluster provides additional evidence that ultrasonication pretreatment of SPW might not always provide a significant benefit in terms of increasing polyphenol extraction yield and enhancing antioxidant activity.
Such an outcome apparently contradicts recent examinations in which ultrasonication pretreatment significantly boosted polyphenol extraction using various means, including β-cyclodextrin [27], deep eutectic solvents [8], and hydroethanolic solutions [9]. However, negative effects have also been reported [28]. Therefore, it would be reasonable to presume that different plant matrices may behave in a different manner as a response to ultrasonication prior to performing stirred-tank extraction. Furthermore, the role of extraction media should also be taken into account. Early investigations highlighted the importance of the type and concentration of acid on the aqueous extraction of anthocyanins from red grape pomace [29]. In the same line, a more recent study has suggested that aqueous solutions of acetic acid are more efficient for flavanol extraction from red grape pomace compared to citric acid solution [30]. In that study, a significant role in the extraction performance was also attributed to acid concentration. Such an approach was more This outcome suggests that under the PLE conditions employed, the addition of citric acid or lactic acid does not foster extraction efficiency or antioxidant activity. Likewise, 5% CA-STE, 5% LA-STE, 5% CA-STE/UP and 5% LA-STE/UP were on the same cluster, an indication of their comparable efficiency. Thus, it can be supported that extraction with aqueous solution containing 5% of either citric or lactic acid showed no significant differences, while ultrasonication pretreatment offered no detectable statistically significant advantage. It is also to be noted that the categorization of Water-STE and Water-STE/UP in the same cluster provides additional evidence that ultrasonication pretreatment of SPW might not always provide a significant benefit in terms of increasing polyphenol extraction yield and enhancing antioxidant activity.
Such an outcome apparently contradicts recent examinations in which ultrasonication pretreatment significantly boosted polyphenol extraction using various means, including β-cyclodextrin [27], deep eutectic solvents [8], and hydroethanolic solutions [9]. However, negative effects have also been reported [28]. Therefore, it would be reasonable to presume that different plant matrices may behave in a different manner as a response to ultrasonication prior to performing stirred-tank extraction. Furthermore, the role of extraction media should also be taken into account. Early investigations highlighted the importance of the type and concentration of acid on the aqueous extraction of anthocyanins from red grape pomace [29]. In the same line, a more recent study has suggested that aqueous solutions of acetic acid are more efficient for flavanol extraction from red grape pomace compared to citric acid solution [30]. In that study, a significant role in the extraction performance was also attributed to acid concentration. Such an approach was more thoroughly carried out by deploying response surface methodology, where it was demonstrated that acidification with lactic acid provided a more effective means of recovering flavonoids from red grape pomace compared to acetic, tartaric and citric acids [31].
The efficacy of PLE compared to both conventional STE and STE implemented after ultrasonication pretreatment was lower, as judged by the Y TP , A AR , P R , as well as the yield of major polyphenols. These results contrasted with previous studies where PLE outperformed both conventional and emerging techniques applied for anthocyanin extraction [32,33] and other polyphenols [22]. However, other investigations showed that these differences might be marginal [34]. Nevertheless, it should be emphasized that the yields attained using PLE cannot be overlooked considering the significantly shorter required extraction time. Furthermore, although PLE was carried out at 120 • C, considerably higher than the 80 • C used for STE and STE/UP, no alteration in the polyphenolic profile was observed. This may indicate that SPW polyphenols are rather stable under these conditions. Such evidence could be a key element in the future optimization of PLE methodology, which is anticipated to shed more light on the potential of PLE for obtaining high extraction yields and extracts with improved antioxidant properties from SPW. Generally, PLE processes have the advantage of providing important enhancements compared to conventional extraction procedures, including higher extraction yields and recoveries, faster extraction, and lower solvent volumes. The use of a high temperature results in an increase in the rate of mass transfer, enhancement of the solubility of the target compounds, and a decrease in solvent viscosity [7]. Moreover, the use of alternative solvents such as deep eutectic solvents should also be considered. The use of such a solvent composed of lactic acid and glycine has been demonstrated to significantly enhance the performance of polyphenol extraction from SPW compared to conventional solvents [24]. Thus, a combination of deep eutectic solvents with PLE might provide a highly effective means of polyphenol and pigment extraction from SPW.

Conclusions
Follow up with a previous study of ours on the use of deep eutectic solvents, in this study, pressurized-liquid extraction was compared to conventional stirred-tank extraction and stirred-tank extraction integrated by ultrasonication pretreatment, in order to obtain evidence regarding their suitability for antioxidant polyphenol and pigment extraction from saffron processing wastes (floral residues). The solvents used were green aqueous solutions of citric and lactic acids. The outcome of the investigation evidenced that stirred-tank extraction and stirred-tank extraction including ultrasonication pretreatment outperformed pressurized-liquid extraction under the conditions employed. This appraisal was based on the yields of total pigments and total polyphenols, as well as the antioxidant properties of the obtained extracts. Determination of the analytical polyphenolic composition also showed that the extracts generated with either stirred-tank extraction or stirred-tank extraction with ultrasonication pretreatment were richer in the major substances identified. On the other hand, the significantly shorter extraction time used for pressurized-liquid extraction should not be overlooked. It is suggested that optimization of the pressurizedliquid extraction process based on the data reported in this study might establish a green and efficient methodology for the recovery of polyphenols and pigments from saffron processing waste. Such an approach would contribute to more effective promotion of this precious residue.