Ultrasound-Assisted Extraction of Lavender ( Lavandula angustifolia Miller, Cultivar Rosa) Solid By-Products Remaining after the Distillation of the Essential Oil

: FINNOVER is an EU Interreg-Alcotra project that aims to bring new perspectives to ﬂoriculture enterprises by recovering useful bioproducts from the waste produced during processing of several aromatic species. In this study, a new operation strategy to recover lavender ( Lavandula angustifolia Mill.) solid by-products remaining after the extraction of the essential oil was developed. Pulsed ultrasound-assisted extraction was employed as a sustainable and eco-compatible technology to extract, in a very short time (10 min), this agricultural waste using a food-grade solvent (a mixture of ethanol/water). All the extracts obtained from both ﬂower and leaf waste and ﬂower-only residues, exhibit a promising total phenolic content (38–40 mg gallic acid/g of dry waste), radical scavenging activity (107–110 mg Trolox/g of dry waste) and total ﬂavonoid content (0.11–0.13 mg quercetin/g of dry waste). Moreover, the chromatographic analysis of these extracts has shown that this overlooked agriculture waste can represent a valuable source of multifunctional compounds. Particularly, they exhibit a content of polyphenols and ﬂavonoids up to 200 times higher than the corresponding leachate, and they are a valuable source of gentisic acid (1.4–13 mg/g dry waste) representing a new low-cost ingredient usable in different ﬁelds (i.e., cosmetic).


Introduction
In this research, the solid by-products remaining after the distillation of lavender, which are still little studied in scientific literature, are taken into account as a potential and still underutilized source of bioactive compounds to be valorized. These solid wastes, sometimes called "distilled straws of lavender" are lignocellulosic-rich materials, which have been proposed as an interesting source to be converted into biofuel [1]. However, this biomass is still rich in bioactive compounds, mainly antioxidants, whose extraction This technique has recently emerged among green technologies due to their advantages shown in several processing steps such as degassing, filtration, preservation (i.e., inactivation of microorganism and enzymes), and extraction of natural products [18]. According to the six principles of green extraction [19], ultrasound-assisted extraction (UAE) allows one to extract, in very short time, with high reproducibility, and with increased extraction efficiency (extraction yield) compared to the conventional extraction methods, several classes of bioactive compounds from different plants and foodstuffs [20][21][22]. In pulsed ultrasound-assisted extraction (PUAE), ultrasounds are applied in pulsed mode, i.e., the ultrasound processor is switched on and off intermittently during the entire extraction process. This extraction modality, compared to the continuous sonication, leads to less heat generation allowing a better control of the temperature rise. For this reason, it is the most suitable method for the extraction of thermolabile biomolecules (e.g., polyphenols) [23]. Moreover, as recently reported by Saratale et al. [24], ultrasounds are a very useful method for the treatment of biomass.
In this research, a strategy to recycle and valorize the solid waste remaining after LEO distillation, both the total waste (TW, distilled floral spikes) and the only-flower waste (FW, flower material detached from the distilled spikes), has been proposed. Particularly, PUAE was employed for the first time, according to our knowledge, to extract polyphenols and flavonoids from this natural matrix. The obtained extracts, using a foodgrade solvent (a mixture of ethanol and water) named respectively total waste extract (TWE) and flower waste extract (FWE), were spectrophotometrically and chromatographically analyzed and compared to the corresponding liquid aqueous by-products (hydrolat and leachate) of the distillation process.
The goal of this study has been the development and the characterization of a new natural extract starting from an agro-industrial waste according to eco-sustainable and circular economy criteria. Figure 2 shows the global scheme of the present work. This technique has recently emerged among green technologies due to their advantages shown in several processing steps such as degassing, filtration, preservation (i.e., inactivation of microorganism and enzymes), and extraction of natural products [18]. According to the six principles of green extraction [19], ultrasound-assisted extraction (UAE) allows one to extract, in very short time, with high reproducibility, and with increased extraction efficiency (extraction yield) compared to the conventional extraction methods, several classes of bioactive compounds from different plants and foodstuffs [20][21][22]. In pulsed ultrasound-assisted extraction (PUAE), ultrasounds are applied in pulsed mode, i.e., the ultrasound processor is switched on and off intermittently during the entire extraction process. This extraction modality, compared to the continuous sonication, leads to less heat generation allowing a better control of the temperature rise. For this reason, it is the most suitable method for the extraction of thermolabile biomolecules (e.g., polyphenols) [23]. Moreover, as recently reported by Saratale et al. [24], ultrasounds are a very useful method for the treatment of biomass.
In this research, a strategy to recycle and valorize the solid waste remaining after LEO distillation, both the total waste (TW, distilled floral spikes) and the only-flower waste (FW, flower material detached from the distilled spikes), has been proposed. Particularly, PUAE was employed for the first time, according to our knowledge, to extract polyphenols and flavonoids from this natural matrix. The obtained extracts, using a food-grade solvent (a mixture of ethanol and water) named respectively total waste extract (TWE) and flower waste extract (FWE), were spectrophotometrically and chromatographically analyzed and compared to the corresponding liquid aqueous by-products (hydrolat and leachate) of the distillation process.
The goal of this study has been the development and the characterization of a new natural extract starting from an agro-industrial waste according to eco-sustainable and circular economy criteria. Figure 2 shows the global scheme of the present work.

Raw Materials
Lavender of Lavandula angustifolia cultivar Rosa was collected in August 2017 from plants growing in the outdoor facilities of Franco Stalla at Albenga, Savona Province (Italy) and was immediately authenticated by an agronomist at Istituto Regionale per la Floricoltura (IRF). The conventional steam distillation technique was immediately applied for the extraction of the LEO. The total solid waste (TW) composed from flower and leaf residues remaining after the distillation of essential oil, was supplied to the food chemistry laboratory of the University of Genoa Department of Pharmacy for their further extraction. Moreover, the only-flower residues (FW) were manually separated from TW to make a comparison. The distillation wastewaters namely hydrolat (HYDR) and leachate (LEACH), which represent the liquid aqueous by-products of the LEO distillation, were also provided for further investigation as reported in Figure 2. Leachate concentration was calculated considering an average production of 1400 L of leachate from 35000 kg of fresh plant material and a 77% moisture content for lavender spikes, resulting in a dry weight/volume ratio of 5.75 g/mL.

Equipment
A traditional Binder FED53 heating oven (Goettingen, Germany) and a Retsch Grindomix 200 M (Haan, Germany) were used for preliminary drying and grinding of the samples, respectively. A MA40 infrared moisture analyzer (Sartorius, Goettingen, Germany) was used to detect the residual moisture content of the solid wastes before their

Raw Materials
Lavender of Lavandula angustifolia cultivar Rosa was collected in August 2017 from plants growing in the outdoor facilities of Franco Stalla at Albenga, Savona Province (Italy) and was immediately authenticated by an agronomist at Istituto Regionale per la Floricoltura (IRF). The conventional steam distillation technique was immediately applied for the extraction of the LEO. The total solid waste (TW) composed from flower and leaf residues remaining after the distillation of essential oil, was supplied to the food chemistry laboratory of the University of Genoa Department of Pharmacy for their further extraction. Moreover, the only-flower residues (FW) were manually separated from TW to make a comparison. The distillation wastewaters namely hydrolat (HYDR) and leachate (LEACH), which represent the liquid aqueous by-products of the LEO distillation, were also provided for further investigation as reported in Figure 2. Leachate concentration was calculated considering an average production of 1400 L of leachate from 35,000 kg of fresh plant material and a 77% moisture content for lavender spikes, resulting in a dry weight/volume ratio of 5.75 g/mL.

Equipment
A traditional Binder FED53 heating oven (Goettingen, Germany) and a Retsch Grindomix 200 M (Haan, Germany) were used for preliminary drying and grinding of the samples, respectively. A MA40 infrared moisture analyzer (Sartorius, Goettingen, Germany) was used to detect the residual moisture content of the solid wastes before their drying. UAE was performed directly by a Hielscher UP200St sonicator (Teltow, Germany). An UV-Vis Agilent 8453 spectrophotometer (Waldbronn, Germany) was used for the determinations of total phenolic content (TPC) and radical scavenging activity (RSA). Total flavonoids were determined using a T60 spectrophotometer (PG Instruments Ltd., Leichestershire, UK). HPLC analyses were performed using a Perkin-Elmer (Waltham, MA, USA) Series 200 LC pump coupled with a 785A UV/VIS detector.

Ultrasonic-Assisted Extraction of the Solid By-Products
The solid wastes (TW and FW) were dried at 40 • C for 48 h until the residual water amounts, evaluated by a moister analyzer, were lower than 11%. After drying, wastes were finely ground by a Grindomix (company, city, state abbrev if USA, country; process conditions: 20 s at 5000 rpm) and then sieved through a 150 µm sieve. 1 g of ground solid waste (TW and FW, respectively) were immersed in 40 mL of ethanol:water mixture (70:30) in a polypropylene 50 mL centrifuge tube with 30 mm of internal diameter. Then, this mixture was submitted to ultrasounds applied in pulsed modality, using an ultrasonic titanium probe (7 mm diameter), at a 26 kHz frequency and an effective power of 200 W, and keeping temperature under control during all the extraction time (<60 • C). Process conditions of UAE were set up based on published results and several preliminary trials were made to adapt them [25,26]. The experimental details are reported in Table 1. The obtained extracts, named TWE and FWE, were centrifuged at 3500 rpm for 10 min (RCF or G-force equal to 2016) and then they were filtered through a Büchner funnel, using Whatman no. 1 paper. The final extracts were stored frozen at −20 • C until their analysis. The 1,1-diphenyl-2-picrylhydrazyl (DPPH) radical scavenging activity was determined according to the in vitro assay described by Brand-Williams et al. [27]. 0.250 mL of the extracts were transferred into a 10 mL volumetric flask and were mixed with a daily prepared DPPH solution (10 −4 M in methanol). After 30 min incubation in the dark and at room temperature, the residual absorbance was read at 515 nm against a blank (solution without radical). The initial DPPH concentration was measured by control samples (without extracts) obtained by diluting 0.250 mL of methanol with the DPPH solution in a 10 mL volumetric flask. The absorbance measurements were transformed in antioxidant activity using Trolox and ascorbic acid as reference standards. A multi-point calibration curves in Trolox and ascorbic acid (R 2 = 0.9976 and R 2 = 0.9975, respectively) were obtained. The RSA was expressed as Trolox equivalent antioxidant capacity (TEAC, mg TE/gof dry waste) and ascorbic acid equivalent antioxidant capacity (AEAC, mg AAE/g of dry waste), respectively. Two replicated determinations were performed for each extract and the averaged results were reported.

Determination of Total Phenolic Content (TPC)
The total phenolic content (TPC) of the extracts was determined by the Folin-Ciocalteu method previously reported by Singleton et al. [28]. Briefly, 200 µL of suitably diluted extracts or standard, 1 mL of Folin-Ciocalteu reagent (previously diluted 1:10 with deionized water) and 800 µL of aqueous sodium carbonate 7.5% w/v were mixed in a test tube and vortexed. The mixture was equilibrated in the dark at room temperature for 30 min and then the absorbance was read spectrophotometrically at 760 nm using quartz cuvettes with 1 cm of path length. TPC was calculated using gallic acid as reference standard for the calibration curve (R 2 = 0.9887) and results were expressed as milligrams of gallic acid in one gram of dry waste (mg GA/gdry waste). Determinations were replicated for each sample and the averaged results were reported.

Determination of Total Flavonoid Content (TFC)
The total flavonoid content (TFC) was estimated following the aluminium chloride colorimetric method [29]. Aliquots of 0.5 mL of lavender extracts or diluted standards (quercetin) were mixed with 0.1 mL of AlCl 3 , 0.1 mL of CH 3 COONa and 2.8 mL of deionized water. Absorbance of the samples was measured at 415 nm wavelength after 40 min of incubation in the dark at room temperature. Diluted samples of commercial quercetin in a range between 5 and 150 µg/mL were used as standards for the calibration curve (R 2 = 0.9987). Flavonoid concentrations were expressed as mg/g of quercetin equivalents (QE).

HPLC Analysis
Reversed phase (RP) chromatography of the studied extracts was performed by using a Kinetex EVO C18 column 25 cm × 4.6 mm, 4.7 µm particle size (Phenomenex Inc., Torrance, CA, USA). The solvent gradient method used 0.1% H 3 PO 4 in water (A) and acetonitrile (B). Gradient (% in A): 0-30 min from 10% to 58% B, 1 min from 58% to 90% B, 5 min isocratic 90% B, 1 min from 90% to 10% B, and 25 min isocratic 10% B [30]. Injection volume was 50 µL; solvent flow rate was set at 1 mL/min. Detection was performed at 280 and 350 nm wavelength to discriminate between phenols and flavonoids. Peak identification and quantification, when possible, was made using external standards. When identification was no possible, peak concentration was calculated as gallic acid equivalents (mg/g) or as quercetin equivalents (mg/g) for unknown phenols and flavonoids, respectively.

Statistical Analysis
Data were analysed using the R environment [31]. Differences in RSA, TPC, and TFC among the studied extracts (TWE, FWE, LEACH) were analysed via analysis of variance using the aov function. Multiple comparisons between extracts were carried out using Tukey's Honest Significance tests with the TukeyHSD function. The same approach was used to analyse differences among lavender extracts in the relative composition of single phenolic compounds and flavonoids as observed through HPLC. Results throughout the text were expressed as means ±95% confidence intervals.

Overall Differences between Solid Waste Hydroalcoholic Extracts (TWE and FEW) Obtained by PUAE and Distillation Wastewaters (LEACH)
As reported in Table 2, the analyzed extracts significantly differed in their antioxidant activities (ANOVA F 2,3 = 365.4, p < 0.001 for TEAC; F 2,3 = 374.9, p < 0.001 for AEAC), and in their phenolic and flavonoid contents (ANOVA F 2,3 = 1100, p < 0.001, and F 2,3 = 177.1, p < 0.001, respectively). Solid wastes (TWE and FWE) showed very similar RSA (expressed as TEAC and AEAC), TPC (expressed as GAE) and TFC values (expressed as QE). In turn, when compared with solid wastes, distillation wastewaters showed a significantly lower antioxidant activity (more than 30 times lower, both for TEAC and AEAC) and phenolic and flavonoid concentration (ca. 52 times and 300 times, respectively). All results are reported as mean values ±95% confidence intervals (n = 2). In each column, means ± 95% confidence intervals followed by different letters (a, b) in superscript are significantly different after ANOVA and Tukey's Honest Significant Difference tests at α = 0.05.
Hydrolat (HYDR), generated during the extraction process of essential oils from aromatic plants, is made up of condensation water containing water-soluble, oxygenated, and volatile components in which very small amounts of essential oils remain dispersed [17]. Smigielski et al. [32] describing the composition of the hydrolat of L. angustifolia reported the following major chemical components: linalool (26.5%), borneol (9.0%), cis-linalool oxide (6.6%), and trans-linalool oxide (5.2%). These hydrosols showed antibacterial in vitro activity [33]. However, no study has been published on the antioxidant activity of lavender hydrolats. As reported in Table 2, HYDR was very poor in phenolic compounds (0.70 + 0.00 mg of gallic acid in 100 mL) while both the antiradical activity (RSA) and the flavonoid content (TFC) were not detectable.
Recently, the application of UAE to different plant matrices has produced very interesting results. Particularly, this extraction method was proposed as an innovative alternative to conventional processes to extract antioxidants, such as phenolic compounds, from several natural matrices [22][23][24][25][26]. The main advantages of UAE are the mass transfer intensification, the vegetal cell disruption, and a higher penetration of the extraction solvent to the plant tissues. Indeed, the high-frequency ultrasonic waves through the cavitation process break the vegetal cells releasing the cell contents into the extraction medium [20]. Several articles have been published on the application of UAE to extract phenolic compounds from plants such as myrtle (Myrtus communis L.) [34], pomegranate (Punica granatum L.) byproducts [22,25,26], maqui (Aristotelia chilensis) [35], black chokeberry (Aronia melanocarpa L.) [36], sloes (Prunus spinosa L.) [37] and buds (i.e., Ribes nigrum, Castanea sativa) [23,24]. All of these studies described ultrasound as an efficient and a more suitable method for the extraction of these degradable bioactive compounds, also in comparison to other green extractive technologies (e.g., microwave).
Moreover, several comparative studies on the extraction of phenolic compounds from the flowers and buds of different species were carried out using ultrasounds, traditional maceration, and extraction assisted by heat (i.e., decoction). These studies confirmed that UAE is a promising rapid and efficient green technology, which allows to significantly improve the extraction of phenolic compounds compared to other conventional methods [38,39].
In 2006, Proestos and Komaitis [40] demonstrated the superiority of UAE in extracting phenolics from aromatic plants (such as Rosmarinus officinalis, Vitex agnus-castus, Origanum dictamnus, O. majorana, Salvia officinalis and Teucrium polium) over conventional extraction techniques. The content of phenolic compounds extracted in an ultrasonic bath at the optimized conditions of 1 h at 60 • C using methanol as extraction solvent, are between 11 and 23 mg of gallic acid per g of dry sample depending on the species. In this work, using an eco-compatible food-grade solvent (a mixture of EtOH and water) and a direct UAE through a probe, which is much more powerful because the ultrasonic intensity is delivered on a small surface (only the tip of the probe) compared to the ultrasonic bath [41], about double extractive yields were obtained (38-40 mg GA/g of dry waste). As regards RSA, Chrysargyris et al. [42] reported the antioxidant activity of several medicinal and aromatic plants (i.e., Artemisia abrotanum L., Pelargonium roseum L., Laurus nobilis L., Rosmarinus officinalis L., Mentha spicata L., Lavandula angustifolia L., Aloysia triphylla L.) in the range between 6.1-47.7 mg of trolox/g of fresh weight. Particularly, the methanolic extract of L. angustifolia (stem and leaves) showed RSA equal to 16.06 mg of Trolox/g of fresh weight. In this work, both the extracts obtained from lavender waste (TWE and FW) showed a higher antioxidant activity (107-110 mg Trolox/g dry waste).
In the present study, we observed relatively low contents of flavonoid compounds per g dry weight as compared to TPC ( Table 2). TFC values were within the concentration range (between µg/g and mg/g) of those observed by previous authors in distilled straws [1,43]. Besides the expected variability in flavonoid concentrations between lavender cultivars from different geographic locations, differences in flavonoid quantification may also result from variability in distillation procedures (water and/or steam distillation, distillation time, temperature, etc.) affecting the quality of the remaining plant material. In addition, the use of a AlCl 3 chelation method for the quantification of total flavonoids (this study) may result in underestimation of flavonoid concentrations because glycoside sugar moieties can prevent chelation with AlCl 3 and the associated bathochromic shifts, rendering undetectable these molecules [44]. Since glycoside flavonoids are frequent in distilled straws [45], flavonoid quantification using colorimetric (through AlCl 3 chelation) or chromatographic (quantification of HPLC peaks) may result in significant discrepancies. In the present study, reported TFC values agreed with flavonoid concentrations estimated from the corresponding HPLC peak areas (data not shown).

Differences between Solid Waste Hydroalcoholic Extracts (TWE and FEW) and Distillation Wastewaters (LEACH) in the Relative Composition of Single Phenolic and Flavonoid Compounds
The analysis of the HPLC profiles for phenolic and flavonoid compounds revealed clear differences in the number and relative abundance of single compounds depending on the type of extract considered ( Figure 3A-C). Overall number of chromatographed compounds ranged between 29 and 30 (Table 3).
At the qualitative level, distillation wastewaters were characterized by the absence of p-coumaric acid and two non-polar unidentified flavonoids, likely aglycones. In turn, TWE samples were apparently lacking syringic acid, rutin and an unidentified polar flavonoid. These were the three most abundant compounds in LEACH samples, suggesting that steam distillation was a relatively efficient extraction method for some abundant compounds in the selected lavender cultivar. Syringic acid, rutin and the abovementioned flavonoid were present instead in the FEW samples, likely due to a concentration effect resulting from discarding the stem and the few leaves usually present in floral spikes used for essential oil distillation. The presence of flavonoid glycosides is very variable within the subgenus Lavandula, resulting in flavonoid profiles that may differ between groups of species [46]. There is little information of flavonoid variation at the cultivar level, although our results suggest that PUAE of distillation by-products may also contribute to lavender characterization at this level. The most abundant compound in TWE samples was the phenol gentisic acid (2,5-dihydroxybenzoic acid), accounting for more than 40% of the polyphenolic fraction obtained by PUAE (Table 3).  Table 3. The studied compounds differed in molecular weight and structure and not all of them showed maximum absorbance at 280 nm wavelength; therefore, observed peak areas and/or peak heights do not necessarily match calculated concentrations. Chromatograms were obtained at 280 nm wavelength after RP-HPLC. Note that retention times may vary for the same compounds depending on the specific laboratory conditions during the analysis. Numbers within square brackets indicate unidentified compounds. Peak numbers correspond to those reported in Table 3. The studied compounds differed in molecular weight and structure and not all of them showed maximum absorbance at 280 nm wavelength; therefore, observed peak areas and/or peak heights do not necessarily match calculated concentrations. Table 3. Observed differences in the relative composition of 32 phenolic compounds present in the distillation wastewaters (LEACH) and the corresponding distilled straws of inflorescences (TWE and FWE) from Lavandula angustifolia var. Rosa. In Table S1 the same data quantified as µg/g DW are reported. This is indicative of the fact that polar phenolic compounds were more abundant in TWEs than in FWEs, likely due to the relatively higher concentrations of phenolic compounds (about four times higher in the case of gentisic acid) in leaves and/or stem tissues. Other phenolic compounds, such as caffeic, chlorogenic and p-coumaric acids were also present in the studied extracts, although their relative abundance significantly depended on the type of extract (Table 3). In general, as in the case of gentisic acid, the proportion of these compounds was higher in TWE than in the other two extracts. Regarding the identity of the chromatographed phenolics, the occurrence of caffeic and p-coumaric acids was already reported in lavender distilled straws [1] as typical phenolic constituents of this kind of distillation by-product. Besides simple phenols, Torras-Claveria et al. [45] showed that lavender distilled straws are characterized by the presence of numerous glycosylated forms, which renders unlikely their identification using common external standards, as we tried to do in this study. In addition, distilled straws are a source of new polyphenolic compounds [47], which likely differ in composition and abundance between lavender species and cultivars.

No
The overall amount of flavonoids was much higher in PUAE samples than in distillation wastewaters (Table 2), and the relative proportion of these compounds significantly varied between extracts (Table 3). Distillation wastewaters (LEACH) and PUAE samples (TWE and FWE) showed clearly different chromatographic profiles ( Figure 3A-C), mostly due to differences in the relative amounts of flavonoids. As expected, hydroalcoholic extracts resulted in more concentrated flavonoid samples, likely due to the affinity of many of these compounds for solvents less polar than water [48]. Further processing of the distillation straws by collecting and extracting only flower tissues (FWE) resulted in a different chromatographic profile showing an even proportion of some of the compounds, mostly flavonoids, obtained in the TWE ( Figure 3B,C). Among the identified flavonoids, the aglycones luteolin and apigenin were already reported in flower distilled straws from L. angustifolia [1]. The presence of these two compounds agreed with the observed accumulation of flavone 7-O-glycosides derived from luteolin and apigenin, as a typical biochemical trait of L. angustifolia [49]. The ability of PUAE to extract a wide range of phenolic and flavonoid compounds that were just partially released from the plant material during the essential oil distillation process confirmed that distilled straws constitute a rich source of secondary metabolites. These compounds can be used, e.g., as flavour or as antioxidants [50], or even as a potential synergist of insecticides [51].
Since phenolic compounds and flavonoids show different biological activities depending on their structure [52], observed differences in the relative abundances of these compounds in the studied extracts likely resulted in overall differences in extract properties. By comparing the proportional abundance of the most abundant compound in each type of extract, it was possible to clearly characterize these hypothesized differences. As mentioned before, TWE extracts were distinguished by the abundance of gentisic acid (43.84% of the total concentration), while the most abundant compounds in LEACH and FWE samples were rutin (21.33%) and syringic acid (11.09%), respectively. Maximum percentage abundances are equivalent to the Berger-Parker index of dominance [53]. Therefore, an increase of the reciprocal form of this index, in this case 2.28 for TWE, 4.69 for LEACH and 9.02 for FWE indicated an increase in chemical diversity associated with a reduction in compound dominance [54]. These results reinforced the view that the two extracts from distilled straws (TW and FW) were significantly different between them ( Figure 3B,C), pointing to the potential importance of post-processing of distillation by-products.

Future Research Directions
Lavandula angustifolia is mainly grown for the purpose of obtaining essential oils by hydrodistillation and the oil yield is estimated to be 40 kg per hectare [55]. Lavender flowers have up to 3% essential oil, so the amount of residual vegetable material remaining after processing is copious. Even if some process wastes of essential oil, as hydrolat, have already found a reuse possibility, the solid straws remaining after the process still represent a natural source to be further exploited. The high-power ultrasound corresponding to frequencies of 20 or 25 kHz seem to offer a good and they could be applied using two types of devices, ultrasonic bath, or probe-type ultrasound equipment. The probe system is more powerful due to an ultrasonic intensity delivered through a smaller surface, when comparing to the ultrasonic bath [18]. Even though the quantity of product that can be treated by an ultrasonic probe is restricted to rather small volumes, nevertheless many lavender essential oil producers are small enterprises which could exploit this extraction system on pilot scale after reoptimising the process variables to adapt them on larger scale. Both the PUAE extracts obtained are rich in gentisic acid (2,5-dihydroxybenzoic acid) whose amounts are comparable with the richest natural sources of this phenolic acid already reported in literature [56,57]. There are several industrial uses of this compound and many published studies on its biological activities that make the natural extracts rich in gentisic acid very interesting to be exploited in numerous food, cosmetic and pharmaceutical applications.

Conclusions
The solid by-products remaining after the traditional distillation of lavender essential oil constitute an interesting source of antioxidant compounds such as polyphenols and flavonoids, whose recovery could be performed as a first step in their valorization before using the remaining lignocellulosic biomass for other purposes (e.g., biofuel production). The "environmentally friendly" PUAE demonstrated to be a promising strategy to valorize these wastes obtaining, using only food-grade solvent (a mixture of ethanol and water), new potential ingredients rich in polyphenolic compounds that can be used in the cosmetic and pharmaceutical industries. Gentisic acid (2,5-dihydroxybenzoic acid) which is particularly abundant in both the solid waste extracts (TWE, FEW) is described as an analgesic, anti-inflammatory, antirheumatic, antiarthritic, cytostatic agent, and a fibroblast growth factor (FGF) inhibitor and it seems able to inhibit low-density lipoprotein oxidation in human plasma [58]. Moreover, extracts rich in gentisic acid could be exploited as cosmetic ingredient for skin-lightening applications since this phenolic compound is largely used in cosmetics just as lightening and whitening skin ingredient [58,59]. Likewise, the relative abundance of quercetin derivatives might find application in skin care formulates for their anti-inflammatory properties [60]. As recently reported in the literature [61], the size of the market for natural phenolic acids will grow exponentially over the next five years. Also, as far as gentisic acid is concerned, plants are considered its best source, and lavender flowers (Lavandula officinalis L.) have been reported as one of the richest sources. However, to our knowledge, the highest amount reported in the literature (8.6 mg/g DW) was obtained from the flowers by maceration with methanol [62]. In this research the change in the extraction method, using PUAE with a food grade solvent, allows to reach a similar or even higher quantity of gentisic acid (TWE: 10-13 mg/g DW; FWE: 1.4-4 mg/g DW; Table S1) in the corresponding extract in an ecological way. This approach could represent an opportunity for lavender essential oil producers who could increase their production revenues by valorising solid by-products.