Statement of Novelty

The biorefinery development in this study is based on the utilisation of winery wastes (grape pomaces and stalks) for the extraction of valued-added fractions and the production of succinic acid via fermentation. Deep eutectic solvents (DES) have been evaluated in the pretreatment of grape pomace and stalks to  improve the enzymatic hydrolysis of the lignocellulosic feedstock. The efficient recyclability and reusability of the selected solvent (choline chloride and lactic acid) were also evaluated. To the best of our knowledge, this is the first study evaluating the use of DES in the pretreatment of winery wastes for succinic acid production and the reutilisation of the solvent in repeated pretreatment cycles.

Introduction

Wine production generates large amounts of waste, including grape pomaces (20% of the grape weight) and grape stalks [1], the disposal of which poses environmental pollution problems [2, 3]. The exploitation of solid and liquid winery waste streams to produce numerous value-added products via extraction (e.g. polyphenols, grape seed oil, tannins), fermentation (e.g. microalgal biomass, succinic acid, polyhydroxyalkanoates) and chemical conversion (e.g. furfural) may restructure wineries into sustainable biorefineries [4,5,6,7,8,9]. Within this context, Filippi et al. [4] developed an integrated biorefinery process based on the exploitation of grape pomaces, stalks and wine lees for the extraction of phenolic-rich extract, grape seed oil, calcium tartrate and tannin-rich extracts, while the lignocellulosic fraction was pretreated with 1.19% NaOH at 70 °C and subsequent enzymatic hydrolysis to produce a sugar-rich hydrolysate for succinic acid production using the bacterial strain Actinobacillus succinogenes.

The development of cost-competitive biorefineries is dependent on the efficient pretreatment and subsequent enzymatic hydrolysis of lignocellulosic biomass into C5 and C6 sugars. Several physical and chemical lignocellulose pretreatment technologies, such as liquid hot water, alkaline, dilute acid and steam explosion, have been widely employed [10]. However, these methods lead to high energy requirements, operational costs, environmental concerns and equipment corrosion [11]. Ionic liquids have also been employed for lignocellulose pretreatment, but their industrial application is limited due to their high price, toxic nature and complex production [11, 12]. Deep Eutectic Solvents (DESs) are alternative solvents to ionic liquids with major advantages, such as lower cost, non-toxicity, easy preparation and biodegradability [13, 14].

DES are eutectic mixtures composed of two or three components at specific molar ratio with a freezing point lower than those of the individual components. The components required for DES formulation are a hydrogen bond donor (HBD) and a hydrogen bond acceptor (HBA). A commonly used HBA is choline chloride, which is inexpensive and biodegradable and is often used as animal feed, while carboxylic acids and polyols are mainly used as HBDs. Among HBDs, carboxylic acids are preferred for lignocellulose pretreatment [12, 15,16,17,18]. Lignocellulose deconstruction with DES is achieved via hydrogen bond formation, as they can donate and accept protons and electrons, while they can cleave the ether and hydrogen bonds contained in lignin. Despite the advantages of DES in lignocellulose pretreatment, their industrial implementation is hindered by the high cost of the pretreatment process mainly associated with the efficient recyclability and reusability of the solvents. For this reason, recent studies have focused on the efficient recycling of DES [19].

In 2021, the global succinic acid market was ca. $1,028 million with a projected growth to reach ca. $3,613 million by 2030 [20]. The expected succinic acid market growth is attributed to the increasing demand for bio-based chemicals, its increasing use in food, pharma and plastic industries and the need to reduce environmental issues. However, the succinic acid derived from fossil resources is cheaper than the bio-based counterpart. This problem could be alleviated through technological advances in fermentation and purification stages as well as the utilisation of crude renewable resources, such as sugarcane bagasse, winery wastes, spent sulphite liquor, textile waste and the organic fraction of municipal solid waste, integrated in multi-product biorefineries [4, 6, 21,22,23,24,25,26]. The methodology used for lignocellulose pretreatment affects significantly the processing costs and the environmental impact of succinic acid production. For this reason, sustainable pretreatment technologies should be developed based on solvent recycling with minimal environmental impact and low processing costs.

This study focusses on the evaluation of pretreatment of grape pomaces and stalks using DES for the production of a sugar-rich hydrolysate. Grape pomaces are initially processed for the extraction of sugars, polyphenols and grape seed oil as described by Filippi et al. [4]. The lignocellulose-rich mixtures of remaining grape pomaces and stalks were initially treated with four different DES to select the most promising one for further optimization. The efficient recyclability and reusability of the solvents was also evaluated. The hydrolysate was used in A. succinogenes cultures for succinic acid production and the fermentation efficiency was compared with the efficiency achieved with hydrolysates produced via alkaline pretreatment of grape pomaces and stalks. The main aim of this study is the integration of DES pretreatment of grape pomaces and stalks in the biorefinery concept developed by Filippi et al. [4] towards sustainable succinic acid production.

Materials and Methods

Raw Materials

Grape pomace and stalks were obtained from the winemaking process of four different Greek varieties (Agiorgitiko from Nemea, Muscat from Tyrnavos, Assyrtiko from Santorini and Savatiano from Attiki). The biorefinery concept presented by Filippi et al. [4] was employed to produce the lignocellulose-rich fraction used in this study for DES pretreatment (Fig. 1). The pomace was pretreated via water extraction of free sugars followed by solvent extraction of lipids and phenolic compounds. The lipids were extracted with ethyl lactate (2 h, 1:10 solid-to-solvent ratio, ambient temperature), while the phenolics were extracted with acidified aqueous ethanol (70%, 1:20 solid-to-solvent ratio) in an ultrasonic water bath for 20 min. The remaining pomace solids were mixed with grape stalks at 1:1 (w/w) ratio and treated with various DES.

Fig. 1
figure 1

Process flow diagram of the proposed biorefinery based on the 1st pretreatment cycle of grape pomace and stalks mixture with ChCl:LA 1:10 at 120 °C for 1 h

DES Preparation

The different DES were formulated with choline chloride (ChCl) as HBA and various HBDs, namely formic acid (FA), acetic acid (AA), oxalic acid (OA) and lactic acid (LA), in different molar ratios. Choline chloride was dried under vacuum for 4 h before use. For the preparation of DESs, choline chloride was mixed with formic acid and acetic acid at 1:2 molar ratio [27, 28], while it was mixed with oxalic acid at 1:1 molar ratio [18]. In the case of lactic acid, four molar ChCl:LA ratios (1:2, 1:5, 1:10, 1:15) were evaluated. The different mixtures were heated to 60°C under constant stirring until a homogeneous colourless solution was formed. The solutions were cooled to ambient temperature in a desiccator before use.

Pretreatment of Lignocellulosic Biomass

The mixture of pretreated pomace solids and stalks (10 g on dry basis) was initially suspended in the four different DES at 1:10 solid-to-liquid ratio [29] under continuous stirring using a hotplate at 120°C and 2 h pretreatment duration. The molar ratios of ChCl to organic acids used in the initial set of experiments (ChCl:FA – 1:2, ChCl:AA – 1:2, ChCl:OA – 1:1, ChCl:LA – 1:10) that was carried out for screening purposes, was based on molar ratios reported in literature-cited publications for lignocellulose pretreatment [18, 27, 28, 30]. The pretreatment duration was initiated when the processing temperature (120°C) was reached. After the pretreatment duration was completed, temperature and stirring were switched off and the mixture was allowed to reach ambient temperature.

A second set of experiments was carried out with ChCl:FA at 1:2 molar ratio, 1:10 solid-to-liquid ratio (10 g solids on dry basis), 120°C and three different pretreatment durations (1, 2 and 3 h) in order to identify the suitability of this DES in lignocellulose pretreatment.

A third set of experiments was carried out ChCl:LA at four molar ratios (1:2, 1:5, 1:10, 1:15) 1:10 solid-to-liquid ratio (10 g solids on dry basis), 120°C and 2 h pretreatment duration in order to identify the optimal molar ratio for lignocellulose pretreatment. The selection of ChCl:LA molar ratio was based on literature-cited publications for lignocellulose pretreatment [18, 27, 30,31,32,33].

Based on the obtained results in the preliminary sets of experiments presented above, ChCl:LA was selected for further optimisation based on maximum lignin removal, enzymatic hydrolysis efficiency and solvent recyclability and reusability. Pretreated pomace solids and stalks were mixed with ChCl:LA at 1:10 solid-to-liquid ratio (10 g solids on dry basis) using different temperatures (80, 100, 120, 140°C) and reaction durations (1 and 2 h) in order to identify the optimal temperature and pretreatment duration for lignocellulose pretreatment.

Lignin Separation and DES Recovery

After pretreatment, the mixture was allowed to reach room temperature. The pretreated solids were separated from the liquid fraction by vacuum filtration using a No. 2 Whatman membrane filter. The solid fraction was thoroughly washed with hot water to achieve a neutral pH, indicating complete removal of DES and then air dried at 60°C to reach constant weight. The filtrate was collected by vacuum filtration. Except for DES removal from the solid fraction via water washing, lignin precipitation is also achieved. The filtrate containing water, DES, lignin and impurities was maintained at 4°C for 2 h before lignin separation from the solution by centrifugation at 10,000 × g for 10 min. The lignin-free filtrate that contained water, DES and impurities was subjected to vacuum evaporation at 60°C to remove water [31]. The recovered DES was reused without further purification for the pretreatment of mixed pretreated pomace and stalk solids using the same procedure as described above. Five pretreatment cycles were carried out with the same DES at two operating conditions (100°C for 2 h and 120°C for 1 h).

Stiasny Number Analysis

The lignin separated from the solution was freeze-dried and the solid obtained was subjected to Stiasny number analysis according to a method described by Ping et al. [34]. More specifically, 0.2 g of the freeze-dried sample was mixed with 5 mL of 37% aqueous formaldehyde and 5 mL of 10 M hydrochloric acid. The mixture was refluxed for 30 min and filtered through a sintered glass filter immediately before reaching ambient temperature. The precipitate was air dried until it reached a constant weight. The determination of the Stiasny number was based on the following equation:

$$Stiasny\;number\; = \;\frac{Dry \;weight \;of\; the \;precipitate}{{Dry\;weight\;of\;the \;initial\; lignin\; - \;tannins \;sample}}\; \times \;100$$
(1)

Enzymatic Hydrolysis of Glucan and Hemicellulose

Enzymatic hydrolysis of the mixture of pretreated pomaces and stalks was performed by utilising commercial enzyme preparations containing cellulases (20 FPU/g solids), β-glucosidase (80 U/g solids), xylanase (14 U/g solids) and BSA (8 mg/g solid) [4]. All hydrolysis experiments were conducted at 50°C for 48 h at a 1:10 solid-to-water ratio under mechanical stirring. In all cases, the pH of the hydrolysis was adjusted to 5 [6]. The determination of enzyme activity in all cases was carried out at 50°C in test tubes containing the substrate, 0.1 M acetate buffer and the enzyme preparation used. Cellulase (filter paper units) and β-glucosidase activity were assayed by measuring the glucose produced during hydrolysis of 50 mg Whatman No. 1 filter paper strip (1.0 × 6.0 cm) within 60 min at pH 4.8 and 0.5% cellobiose within 30 min at pH 5, respectively. The xylanase activity was assayed by measuring the xylose concentration produced after the hydrolysis of 0.25% birchwood xylan within 60 min at pH 5. One unit (U) of cellulase or β-glucosidase activity was defined as the amount of enzyme that releases 1 mg of glucose per minute while one unit (U) of xylanase was defined as the amount of enzyme that release 1 mg of xylose per minute.

The efficiency of polysaccharide to sugar conversion yield was designated individually for glucan and hemicellulose as the ratio of sugars released to the sugar content in the glucan and hemicellulose contained both in the initial solid stream (Overall conversion yield) and the residual stream after pretreatment (Hydrolysis yield). The hydrolysis efficiency yield represents the sugars produced during enzymatic hydrolysis alone, while the overall conversion yield represents the conversion of the initial carbohydrate content in grape pomace and stalks mixture into sugars considering the effect of the isolation of soluble sugars, phenolics and lipids, the pretreatment and the enzymatic hydrolysis stages. These two yields were estimated considering the sugars produced during hydrolysis of glucan and hemicellulose fractions as described in equations (2) and (3).

$$\it {\text{Glucan}}\;{\text{conversion}}\;{\text{yield}} \left( \% \right)\; = \;\frac{{{\text{glucose}}\;{\text{released}} \left( g \right) \; \times \;0.9\; \times \;100}}{{{\text{glucan}}\;{\text{content}}\;{\text{in}}\;{\text{the}}\;{\text{substrate}} \left( g \right)}}$$
(2)
$$Hemicellulose\;conversion\;yield \left( \% \right)\; = \;\frac{xylose, arabinose, \;galactose\;and\;mannose\;released \left( g \right) \times 0.88 \times 100}{{hemi\, cellulose \,content \,in \,the \,substrate \left( g \right)}}$$
(3)

Microorganism and Fermentation Medium

The bacterial strain Actinobacillus succinogenes 130 Z (DSM-22257) was used in this study for succinic acid production in batch fermentation mode. Pre-culture preparation was carried out as previously described [6]. The fermentation was carried out in a bench-top bioreactor (Labfors 4, InforsHT) with a working volume of 0.5 L using the enzymatic hydrolysate produced after pretreatment of the mixture of grape pomace and stalks with ChCl:LA at 120°C for 1 h. The initial total sugar concentration of the medium was almost 60 g/L, while it was supplemented with 5 g/L yeast extract, 5 g/L MgCO3 and the following mineral medium: 1.16 g/L NaH2PO4·H2O, 0.31 g/L Na2HPO4, 1 g/L NaCl, 0.2 g/L MgCl2·6H2O, 0.2 g/L CaCl2·2H2O. Fermentation was conducted at 37°C under anaerobic conditions by supplying 0.1 vvm CO2 and continuous agitation at 200 rpm. The size of the inoculum was 10% (v/v), while the pH of the fermentation was controlled to 6.7 with 5 M NaOH.

The effect of the hydrolysate produced as a fermentation broth was evaluated by performing a batch fermentation with simulated commercial sugars with that obtained after the hydrolysis, while all other conditions were the same as described above.

Analytical Methods

The compositional analysis of lignin, glucan, xylan, galactan and mannan was performed using NREL protocol [35, 36] as described by Filippi et al. [6]. Sugars were analysed by High Performance Liquid Chromatography (Prominence, Shimadzu, Kyoto, Japan) equipped with a Shodex SP0810 column coupled to a differential refractometer (RID-10A, Shimadzu, Kyoto, Japan). The analysis was conducted at 65°C and 0.6 mL/min flow rate of pure water as the mobile phase. Organic acids and sugar alcohols were determined with a Rezex ROA-organic acid H+ column (300 mm length × 7.8 mm internal diameter, Phenomenex), with a mobile phase of 10 mM H2SO4 at 0.6 mL/min flow rate and 65°C. 5-hydroxymethylfurfural (HMF) and furfural were analysed using a HPLC–DAD equipped with a RP-C18 column (Waters), 20% (v/v) aqueous solution of methanol as mobile phase, 0.5 mL/min flow rate and ambient temperature. HMF was detected at 272 nm and furfural at 280 nm wavelength. Free amino nitrogen (FAN) content was measured according to the ninhydrin method [37]. In this work, hemicellulose was considered as the sum of xylan, arabinan, galactan and mannan as it was assumed that hemicellulose does not contain any glucan. The lignin, hemicellulose and glucan removal were calculated based on the following equations:

$${\text{Lignin}}\;{\text{removal}}\;\left( \% \right)\; = \;\frac{Mass\;of\;the\;removed\; lignin}{{Mass \;of \;the\; lignin\;in\;the\;initial\;sample}} \times 100$$
(4)
$${\text{Hemicellulose}}\;{\text{removal}}\;\left( \% \right)\; = \;\frac{Mass\;of\;the\;removed\;hemicellulose}{{Mass\;of\;the\;hemicellulose \;in\;the\;initial\;sample}}\; \times \;100$$
(5)
$${\text{Glucan}}\;{\text{removal}}\;\left( \% \right)\; = \;\frac{Mass\;of\;the\;removed\;glucan}{{Mass\;of\;the\;glucan\;in\;the\;initial \;sample}}\; \times \;100$$
(6)

Results and Discussion

Biorefinery Development

Grape pomaces of the four different Greek varieties were initially mixed and the mixture obtained consisted mainly of free sugars (6.7%), glucan (19.3%), hemicellulose (13.4%), lignin (31.2%), protein (8.7%), lipids (6.8%) and ash (7.2%). The free sugars, glucan and hemicellulose fraction can be used as carbon sources for the production of succinic acid [4]. Therefore, the free sugars contained in 1 kg of grape pomaces were initially extracted with water. After the removal of free sugars from the grape pomace mixture, grape-seed oil was extracted (24.3 g) using ethyl lactate as a greener alternative to hexane, as it is a non-corrosive, non-toxic (non-carcinogenic and non-teratogenic), non-ozone depleting, biodegradable and generally considered as GRAS solvent [43, 44]. Ethyl lactate has been employed for lipids extraction from oleaginous yeast [39] and microalgae [40] resulting in high recovery yields. Ethyl lactate resulted in 35.7% oil recovery that is higher than the one (31.3%) achieved when hexane was employed, indicating that ethyl lactate could replace hexane for grape seed oil extraction.

After oil extraction, the remaining pomace solids were treated with acidified aqueous ethanol to extract a phenolic-rich extract (40.3 g) containing 8.3 g gallic acid equivalents with a strong antioxidant activity index (AAI) of 1.41. Grape-seed oil and the phenolic rich extract could be used in cosmetics and the pharmaceutical industry [4]. The grape pomace solids that remained after the extraction of free sugars, lipids and phenolics was mixed with grape stalks in 1:1 ratio. This mixture, containing mainly lignin (35.0%), hemicellulose (15.3%), glucan (22.1%), protein (11.0%) and extractives (16.2%) (Fig. 1), was subjected to DES pretreatment.

As shown in the biorefinery concept presented by Filippi et al. [4], grape stalks contain mainly lignocellulose. Thus, value-added components were not extracted from grape stalks. The process developed in this study could be applied to either pretreated grape pomace and grape stalks alone or the combined streams.

Effect of Pretreatment Using Different DES

Among the carboxylic acids tested in this work, low lignin removal was achieved with ChCl:OA (7.8%) and ChCl:AA (22.5%), while ChCl:AA pretreatment resulted in high glucan removal (41.4%) (Table 1). ChCl:FA and ChCl:LA (1:10 molar ratio) pretreatment of grape pomace and stalks led to lignin removal of 47.9% and 47.1%, respectively. It is also observed that the removal of hemicellulose (higher than 58.7%) is higher than that of glucan (higher than 23.0%) when ChCl:FA and ChCl:LA pretreatment was employed. This is consistent with previous studies stating that the solubility of cellulose in DES is generally low, while the solubility of hemicellulose is higher [31]. This could be attributed to the shorter chain length of hemicellulose, its amorphous nature, its lower degree of polymerization and its branched structure [13].

Table 1 Composition of remaining solids after pretreatment of grape stalks and pomace with DES at 120°C for 2 h. The DES are composed of choline chloride and different organic acids at various molar ratios

The selected DES have been employed in different lignocellulosic biomass sources showing high selectivity to extract lignin and enhance cellulose digestibility [42]. Okuofu et al. [43] evaluated various DES that were composed of different organic acids (lactic acid, formic acid, acetic acid) as HBD and choline chloride as HBA for the pretreatment of Bambara groundnut haulm in order to enhance its saccharification for bioethanol production. According to their results, ChCl:LA pretreatment at 100°C for 1 h was selected as the most efficient conditions in terms of hemicellulose (54.5%) and lignin (60.7%) removal, while a high sugar recovery yield of 94.8% was obtained after enzymatic hydrolysis of the solids. Xu et al. [15] evaluated the effect of DESs synthesised by choline chloride and seven different HBDs (urea, glycerol, formic acid, acetic acid, oxalic acid, malonic acid and citric acid) on improving the cellulose hydrolysis of corn stover. The best results were obtained with ChCl:FA resulting in 99% glucan hydrolysis yield.

Based on the results presented in Table 1, the DESs composed of choline chloride and either lactic acid or formic acid were selected in order to evaluate their effect on enzymatic hydrolysis of the pretreated biomass as well as their recyclability and reusability. This decision was based on the high lignin removal achieved with these two DESs.

Effect of Pretreatment Duration Using ChCl:FA

Figure 2 shows the remaining solids, lignin, hemicellulose and glucan after pretreatment of grape pomace and stalks with ChCl:FA at 120°C and three different pretreatment durations (1, 2 and 3 h). Around 31.9%, 47.9% and 48% of lignin was removed at pretreatment durations of 1, 2 and 3 h, respectively (Fig. 2b). Figure 2 shows that increasing pretreatment duration from 1 to 2 h leads to increasing removal of all components. Increasing the pretreatment duration to 3 h has no significant effect on the removal of solids and lignin, but increases the removal of hemicellulose and glucan to 68.5% and 34.9%, respectively (Figs. 2c and 2d). Xu et al. [15] reported that the glucose concentration obtained after enzymatic hydrolysis of corn stover pretreated with ChCl:FA at 130°C was increased from 7.83 to 17 g/L when the pretreatment duration was increased from 0.5 to 3 h.

Fig. 2
figure 2

Pretreatment of grape pomaces and stalks using ChCl:FA at different temperatures and pretreatment durations. The dark coloured bar section shows the remaining component after pretreatment of 100 g, while the light coloured bar section shows the removal of each component. The components evaluated were (a) solids, (b) lignin, (c) hemicellulose and (d) glucan

An important criterion for the identification of the optimal DES for biomass pretreatment is its efficient recyclability and reusability. For this reason, it was evaluated whether the ChCl:FA could be efficiently recycled and reused after pretreatment of grape pomaces and stalks mixture. In the first pretreatment cycle, the grape pomace and stalks mixture was pretreated with ChCl:FA at 120°C and 2 h pretreatment duration. After recycling of ChCl:FA, the solvent was reused in the second pretreatment cycle resulting in only 2.5% lignin removal (data not shown) demonstrating that ChCl:FA presents inefficient recyclability. Kohli et al. [12] reported that the toxic and caustic properties of ChCl:FA, as compared to that of ChCl:LA, hinders the use of formic acid in DES usage in lignocellulose pretreatment. For the above reasons, ChCl:LA was selected for further experiments on the pretreatment of grape pomaces and stalks.

Effect of Pretreatment Using Different ChCl:LA Molar Ratios

An important parameter affecting the pretreatment of lignocellulosic biomass with DES is the molar ratio of HBA to HBD. On this basis, the effect of the molar ratio of ChCl:LA on the pretreatment of grape pomaces and stalks was investigated (Table 1) under the same operating conditions (2 h pretreatment duration at 120°C). The results obtained with 1:2 and 1:5 ChCl:LA molar ratios were similar in terms of solids, lignin, hemicellulose and glucan removal corresponding to approximately 39.0%, 21.0%, 44.0% and 25.0%, respectively. It is also observed that increasing the lactic acid content in the ChCl:LA (1:10 molar ratio), the removal of solids, lignin and hemicellulose increased rapidly to 49.1%, 47.1% and 58.7% respectively, while glucan removal was slightly increased to 26.5%. Further increase in lactic acid content (ChCl:LA 1:15 molar ratio) resulted in higher removal of solids (55.8%), hemicellulose (72.1%) and glucan (43.6%) (Table 1). The effect of different molar ratios of ChCl:LA in the pretreatment stage has been studied in literature-cited publications using different raw materials, such as corncob [18], rice straw [32] and wheat straw [30]. Su et al. [33] evaluated the effect of different molar ratios of ChCl:LA (1:2, 1:4, 1:6, 1:8, 1:10) at different temperatures (110 and 130°C) on the pretreatment of poplar sawdust reporting that increasing lactic acid content led to increasing removal of solids, lignin and xylan. Increasing the ratio of ChCl:LA from 1:2 to 1:10 at 110°C increased the glucan to glucose conversion yield from 28.2% to 37.7% [43]. Further increasing the temperature to 130°C led to the highest glucan to glucose conversion yield of 75.8% at a molar ratio of 1:2 [33]. Based on the lignin removal achieved with the different ChCl:LA molar ratios, the 1:10 molar ratio was selected for subsequent experiments.

Effect of Temperature and Pretreatment Duration

Besides the molar ratio, the temperature and pretreatment duration are important operating conditions, which influence pretreatment efficiency and processing costs [12, 33]. The main aim is to maximise the degradation of lignin without significant loss of solids and polysaccharides, which will be subsequently used for the production of succinic acid through microbial fermentation.

Pretreatment using ChCl:LA at a molar ratio of 1:10 was evaluated at different temperatures and pretreatment durations (Fig. 3). The results presented in Fig. 3 generally show that the solids and individual component removal is increased with increasing temperature or pretreatment duration. The pretreatment carried out at 80°C resulted in low removal of solids (32.2%), lignin (15.7%), hemicellulose (30.2%) and glucan (9.4%). Conducting the pretreatment at 100°C for 1 h leads to higher removal of solids (41.0%), hemicellulose (44.8%) and glucan (19.4%), while the removal of lignin (ca. 20%) was slightly higher to that obtained at 80°C. Increasing the pretreatment duration from 1 to 2 h at 100°C increase lignin removal to 38.8%, while hemicellulose and glucan removal remain constant. When the temperature is increased to 120°C, the removal of lignin and hemicellulose is increased as compared to the same pretreatment duration carried out at 100°C, while glucan removal is slightly increased. In the pretreatment carried out at 140°C for 1 h the removal of solids (55.8%), hemicellulose (72%) and glucan (41.5%) reach the highest values indicating that this condition is not favourable for enzymatic hydrolysis of the resulting solids, as less glucan and hemicellulose is available for enzymatic hydrolysis.

Fig. 3
figure 3

Solids (a) lignin (b), hemicellulose (c) and glucan (d) remaining after grape pomace and stalks pretreatment with ChCl:LA at different temperatures and pretreatment durations. The dark coloured bar section shows the remaining component after pretreatment of 100 g, while the light coloured bar section shows the removal of each component

Shen et al. [44] evaluated the effect of temperature (90, 100, 110, 120, 130°C) using ChCl:LA (1:10 molar ratio) on Eucalyptus pretreatment leading to decreased remaining solids from 78.3% at 90°C to 39.7% at 130°C and decreased remaining lignin from 60% at 90°C to less than 10% at 120°C, while the remaining cellulose (82–85%) was not significantly affected [44].

The criteria for the selection of the operating conditions of ChCl:LA pretreatment for further evaluation were based on the lignin removal and the remaining polysaccharides. Thus, the two combinations of 100°C and 2 h pretreatment duration as well as 120°C and 1 h pretreatment duration were chosen in order to assess their effect on enzymatic hydrolysis at sequential pretreatment cycles.

Effect of DES Pretreatment on Enzymatic Hydrolysis of the Remaining Solids

The effect of DES pretreatment on the enzymatic hydrolysis of the remaining grape pomaces and stalks was initially evaluated after pretreatment with ChCl:LA (1:10 molar ratio) at 100°C for 2 h (Fig. 4). Under these pretreatment conditions, the composition of the pretreated solids was 31.3% glucan, 14.7% hemicellulose, 39.2% lignin and 7.3% protein. After enzymatic hydrolysis, the glucan and hemicellulose hydrolysis yield obtained in the 1st pretreatment cycle were 77.4% and 22.7% (Fig. 4e) respectively, while the overall conversion yields were 53.6% and 7.4% (Fig. 4f). The concentration of total sugars after the 1st pretreatment cycle carried out with ChCl:LA (1:10 molar ratio) at 100°C for 2 h was 29.6 g/L (Fig. 5a). After the recycling of ChCl:LA, it was reused in the 2nd pretreatment cycle under the same operating conditions where the removal of solids, lignin, hemicellulose, and glucan was decreased to 41.1%, 31.9%, 51.5% and 15.6%, respectively (Fig. 4). The hydrolysis yield after enzymatic hydrolysis was similar (78.2% for glucan and 25.3% for hemicellulose) to the 1st pretreatment cycle, while the overall conversion yield was increased to 60% for glucan and 11.7% for hemicellulose. The increase in overall conversion yield was attributed to the lower removal of glucan and hemicellulose in the 2nd cycle as compared to the 1st cycle (Figs. 4c and 4d). After the 4th cycle, the efficiency of lignin removal was significantly decreased to only 2.8% (Fig. 4b), while the removal of glucan and hemicellulose were only 5.4% (Fig. 4d) and 28.6% (Fig. 4c), respectively. The inefficient pretreatment with ChCl:LA in the 4th cycle was also indicated by the low total sugar production (20.1 g/L) via enzymatic hydrolysis (Fig. 5a), resulting in glucan and hemicellulose overall conversion yields of 48.0% and 9.2%, respectively.

Fig. 4
figure 4

Solids (a) lignin (b) hemicellulose (c) and cellulose (d) removal achieved after sequential pretreatment cycles of grape pomaces and stalks using ChCl:LA (1:10 molar ratio) at 100°C for 2 h. The hydrolysis yield (e) and the overall conversion yield (f) obtained after enzymatic hydrolysis of the remaining solids after each sequential pretreatment cycle are presented. The dark coloured bar section shows the remaining component after pretreatment of 100 g, while the light coloured bar section shows the removal of each component. The glucan to glucose yields are presented with black coloured bars (e, f) and the hemicellulose to sugar yields are presented with grey coloured bars (e, f)

Fig. 5
figure 5

Total sugars produced during enzymatic hydrolysis of grape pomaces and stalks pretreated at sequential cycles with ChCl:LA (1:10 molar ratio) at (a) 100°C for 2 h and (b) 120°C for 1 h. (open circle) 1st cycle, (filled circle) 2nd cycle, (open square) 3rd cycle, (filled square) 4th cycle, (filled triangle) 5th cycle

The enzymatic hydrolysis yield, recyclability and reusability of the DES were also evaluated after the pretreatment with ChCl:LA at 120°C for 1 h and the results obtained are presented in Fig. 6. The removal of solids (50.0%), lignin (40.0%) and hemicellulose (52.4%) in the 1st and 2nd pretreatment cycles were similar, while glucan removal was decreased from 29.5% in the 1st cycle to 22.4% in the 2nd cycle. In the subsequent 3rd and 4th pretreatment cycles, the removal of all components is decreased. In the 5th pretreatment cycle, the removal of glucan was almost constant at 13.0% (Fig. 6d), but the removal of lignin was decreased to 4.7% (Fig. 6b) depicting the inefficient pretreatment with ChCl:LA in the 5th cycle.

Fig. 6
figure 6

Solids (a) lignin (b) hemicellulose (c) and glucan (d) removal achieved after sequential pretreatment cycles of grape pomaces and stalks using ChCl:LA (1:10 molar ratio) at 120°C for 1 h. The hydrolysis yield (e) and the overall conversion yield (f) obtained after enzymatic hydrolysis of the remaining solids after each sequential pretreatment cycle are presented. The dark coloured bar section shows the remaining component after pretreatment of 100 g, while the light coloured bar section shows the removal of each component. The glucan to glucose yields are presented with black coloured bars (e, f) and the hemicellulose to sugar yields are presented with grey coloured bars (e, f)

The efficiency of ChCl:LA pretreatment in the 5 sequential pretreatment cycles is also evident in the hydrolysis yield (Fig. 6e) and the overall conversion yield (Fig. 6f). The glucan hydrolysis yield was ca. 90% in the 1st and 2nd cycles, which was gradually decreased to 85.4% in the 3rd cycle, 81% in the 4th cycle and 59.9% in the 5th cycle (Fig. 6e). The overall conversion yield of glucan was gradually increased to 65% in the 3rd cycle followed by a reduction to 47.9% in the 5th cycle (Fig. 6f). The total sugar production during enzymatic hydrolysis was 38.9 g/L using the remaining solids after the 1st pretreatment cycle that was reduced to 32.6 g/L in the 4th cycle and 25.6 g/L in the 5th cycle (Fig. 5b). These results confirm that the efficiency of enzymatic hydrolysis correlates strongly with the delignification achieved during ChCl:LA pretreatment.

The recyclability of DES is one of the most important parameters for the economic efficiency of the proposed process, since the recovery of the solvent reduces operating costs and the overall process sustainability [16, 44]. For this reason, literature-cited publication focus not only on the maximization of delignification and yield of the upcoming hydrolysis, but also on the efficient reuse of DES. Shen et al. [44] investigated the efficiency of recycling of ChCl:LA (1:10 molar ratio) in the pretreatment of Eucalyptus solids at 110°C for 6 h followed by enzymatic hydrolysis. A similar trend was reported by Shen et al. [44] as the one presented in the present study regarding the recovery yield of solids, lignin, hemicellulose and cellulose at sequential ChCl:LA pretreatment cycles. The glucan hydrolysis yield was 94.3% in the 1st pretreatment cycle followed by a reduction to 73.8% after 3 reuses of the solvent [44]. Chen et al. [16] evaluated the pretreatment performance of choline chloride mixture with glycerol for fractionation of switchgrass, resulting in a glucose yield of 89%, while the solvent was recycled and reused successfully for at least four pretreatment cycles [16]. Ci et al. [45] reported that pretreatment of wheat straw with PBDES (choline chloride, polyethylene glycol-200, boric acid) at 120°C for 4 h resulted in 59.3% glucan hydrolysis yield, while the solvent could be efficiently recycled for 3 cycles [45]. Guo et al. [46] reported the pretreatment of corncob with benzyl-trimethyl-ammonium chloride and lactic acid at 120°C for 2 h. The enzymatic hydrolysis yield obtained was 94% in the 1st pretreatment cycle that was decreased to 83% in the 5th cycle. Guo et al. [46] reported that the accumulation of carboxylic acids, derived from the decomposition of cellulose and hemicellulose, and phenolic compounds derived from lignin were the main reasons for the decrease in pretreatment efficiency.

Filippi et al. [4] optimized via experimental design the alkaline pretreatment of the same mixture of grape pomaces and stalks. The optimal pretreatment conditions (1.19% NaOH, 70°C, 30 min pretreatment duration) resulted in the removal of 50.8% lignin, 29.9% hemicellulose and 23.6% glucan. The glucan hydrolysis yield of the pretreated solids was 81.5%, which was lower than the one obtained in this study, while the overall glucan conversion yield (56.9%) was also lower than the one obtained in the present study at 120°C and 1 h pretreatment duration (Fig. 6e and 6f). The overall hemicellulose conversion yield (18.0%) [4] achieved via alkaline pretreatment was similar to the one observed in this study (Fig. 6f). The efficient recycling of DES for 4 pretreatment cycles is an additional advantage that cannot be achieved in alkaline pretreatment. Consequently, the use of DES for the pretreatment of grape pomaces and stalks should be further optimized in order to broaden its sustainability potential.

Succinic Acid Production

The enzymatic hydrolysate produced from the solids pretreated with ChCl:LA was used as substrate for succinic acid production in batch fermentation mode. The consumption of total sugars and FAN along with the production of organic acids are depicted in Fig. 7. The fermentation was initiated with 62.9 g/L total sugars and 259 mg/L FAN. The succinic acid concentration reached 20.9 g/L after 23 h with simultaneous production of 2.2 g/L formic acid and 4.5 g/L acetic acid. The production of succinic acid reached the highest concentration (36.0 g/L) at 55 h resulting in a yield of 0.62 gSA per g total sugars and a productivity of 0.65 g/(L·h). At the same time, acetic acid reached 5.8 g/L, while the formic acid concentration remained almost constant (2.6 g/L). The sugar to succinic acid conversion yield was estimated considering the sugars and succinic acid contained in the samples taken during fermentation. Around 1.45 g/L of total sugars were left unconsumed, while the FAN concentration remained stable at around 50 mg/L after 23 h.

Fig. 7
figure 7

Batch bioreactor fermentation with A. succinogenes using the hydrolysate produced after DES pretreatment and enzymatic hydrolysis of grape pomaces and stalks. Total sugars (filled square), succinic acid (filled circle), formic acid (times symbol), acetic acid (open triangle), FAN (open square), glucose (open diamond), xylose (filled diamond) and arabinose (open circle)

A batch fermentation using commercial sugars at the same composition as in the hydrolysate was also carried out aiming to compare the fermentation efficiency achieved with commercial sugars and the crude hydrolysate derived after ChCl:LA treatment (Fig. 8). Slightly higher succinic acid concentration (38.7 g/L at 50 h), yield (0.66 gSA per g of total sugars) and productivity [0.77 g/(L·h)] were achieved when commercial sugars were used. The total sugar to succinic acid conversion yield was estimated considering both the sugars and succinic acid contained in the samples taken during fermentations. In the case of by-product formation, the final formic acid and acetic acid concentrations were 1.9 g/L and 5.9 g/L, respectively. These results depict that ChCl:LA pretreatment could be employed for succinic acid production as there is only a slight reduction in fermentation efficiency as compared to the utilisation of commercial sugars.

Fig. 8
figure 8

Batch bioreactor fermentation with A. succinogenes using commercial sugars with the same composition as the one contained in the winery waste hydrolysate. Total sugars (filled square), succinic acid (filled circle), formic acid (times symbol), acetic acid (open triangle), FAN (open square), glucose (open diamond), xylose (filled diamond) and arabinose (open circle)

The presence of HMF and furfural in the hydrolysate produced was also investigated, as these compounds could inhibit succinic acid production. According to the results, 186.8 mg/L furfural was identifying (184 mg furfural per 100 g dry sample), while no HMF was detected. The slight difference in succinic acid production efficiency using the produced hydrolysate (Fig. 7) and the commercial sugars (Fig. 8) could be attributed to the presence of furfural, since it has been demonstrated that A. succinogenes can be inhibited by furfural even when it is present at low concentrations [6].

Filippi et al. [4, 6] has carried out fermentations with A. succinogenes for succinic acid production using crude hydrolysates derived from winery waste streams that were processed with different pretreatment methods. More specifically, Filippi et al. [6] pretreated a mixture of grape pomaces and stalks with 1% (w/v) NaOH at 100°C for 3 h to remove lignin, while the resulting solids were subjected to acid pretreatment with 3% (v/v) H2SO4. The remaining solids were subjected to enzymatic hydrolysis and the hydrolysate was used in A. succinogenes cultures leading to 24.6 g/L succinic acid production with a productivity of 0.75 g/(L·h) and a yield of 0.47 gSA per g consumed sugars [6]. Filippi et al. [4] produced a sugar-rich hydrolysate after pretreatment of a mixture of grape pomaces and stalks with 1.19% NaOH at 70°C for 30 min followed by enzymatic hydrolysis. The sugar-rich hydrolysate was supplemented with a nitrogen-rich hydrolysate produced after hydrolysis of wine lees with crude enzymes produced via solid state fermentation. A batch A. succinogenes fermentation carried out with these two hydrolysates resulted in the production of 24.9 g/L succinic acid with a yield of 0.54 gSA per g total sugar and a productivity of 0.62 g/(L·h). In addition, a fed batch experiment was also conducted using the same hydrolysates, but the feeding solution consisted of concentrated free sugars obtained after aqueous extraction of grape pomaces. In this case, the succinic acid concentration of 37.2 g/L was achieved with a productivity of 0.79 g/(L·h) and a yield of 0.64 gSA per g consumed sugars [4]. Alexandri et al. [47] also investigated the production of succinic acid with A. succinogenes using sugar beet pulp pretreated with acid prior to enzymatic hydrolysis. The succinic acid concentration achieved, using 32.3 g/L initial sugars, was 20.0 g/L with a yield of 0.62 gSA per g consumed sugars and a productivity of 0.69 g/(L·h) [47]. As compared to the results presented above, the use of ChCl:LA pretreatment can efficiently replace conventional pretreatment methods leading to the production of crude hydrolysates that can sustain sufficient A. succinogenes growth and succinic acid production.

Mass Balances of the Biorefinery Concept

The development of novel biorefineries based on the conversion of lignocellulosic biomass into key platform chemicals and value-added products requires the implementation of efficient strategies aimed at cascade biomass utilisation [48]. Following this concept of cascade utilisation of lignocellulosic feedstocks using green solvents to replace the conventional pretreatment methods, this study demonstrated the development of a biorefinery concept where ChCl:LA could be employed for the pretreatment of grape pomaces and stalks. The material balances of the proposed biorefinery concept are presented in Fig. 1. Based on the experimental results utilising 1 kg of grape pomace, 130 g of free sugars, 24.3 g of grape seed oil and 40.3 g of phenolic-rich extract could be extracted. The remaining solids were mixed with 1 kg of grape stalks and pretreated with ChCl:LA 1:10 molar ratio at 120°C for 1 h. After pretreatment, two main streams were obtained. A solid stream (896.4 g) which was subsequently subjected to enzymatic hydrolysis and a liquid stream.

Around 251.4 g solids were obtained after centrifugation of the liquid fraction that resulted after pretreatment of grape pomaces and stalks with ChCl:LA. This solid fraction contains mainly lignin and tannins. Analysis of the Stiasny number indicates how reactive the tannins are towards formaldehyde and whether the solid fraction containing lignin and condensed tannins is suitable for the production of green adhesives or phenol–formaldehyde resins. The threshold of the Stiasny number for the production of phenol–formaldehyde type resins is 60% [49] and for the production of adhesives it is 65% [50]. In this study, a Stiasny number of 65% was obtained, which is lower than that (76%) obtained when NaOH was used for the pretreatment of the same mixture of grape pomace and stalks [4]. Apart from the Stiasny number, the ability of crude tannin extracts isolated from grape pomace to produce adhesives is strongly influenced by the method used for the isolation of lignin—crude condensed tannin mixtures. To this end, Ping et al. [34] reported that the crude tannin extract obtained after lyophilization of the liquid resulted in more promising results than precipitation of lignin with HCl, although the Stiasny number was much higher in the second case (89–98% in the case of HCl precipitation and 32–55% when lyophilization was used) [34]. These results suggest that the crude condensed tannin fraction should be further evaluated in future work for the production of formaldehyde resins and adhesives.

Lignin separation from lignocellulosic biomass via DES pretreatment has been described in literature-cited publication. Alvarez-Vasco et al. [28] reported the isolation of lignin, with purity up to 95%, from hardwood (up to 78% lignin extraction from poplar) and softwood (up to 58% lignin extraction from D. fir) via treatment with DES mixtures prepared from ChCl as HBA and four HBDs (i.e. acetic acid, lactic acid, levulinic acid, glycerol). Cronin et al. [51] reported the extraction of lignin at high purity (94.7%) and yield (75%) via corn stover treatment with ChCl:LA. Moreover, lignin derived via DES pretreatment could be used in the production of phenol–formaldehyde resins presenting better characteristics than the original resin in terms of bonding strength, thermal stability and sunshine gel time [52, 53]. Hong et al. [54] reported that DES composed of choline chloride and zinc chloride could efficiently modify chemically the wheat straw alkali lignin that could be used as phenol replacement in the synthesis of phenol–formaldehyde adhesives [54]. These findings suggest that DES pretreatment is a promising technology for lignin extraction along with enzymatic hydrolysis of the remaining solids, while further valorisation of the derived lignin is still a challenge.

The liquid fraction obtained by centrifugation after lignin isolation contained 112 g glucan, 118.6 g hemicellulose, 1.8 g lignin and 129.2 g protein (Fig. 1). The high amount of carbohydrates (230.6 g) contained in this fraction should be exploited in future studies. Moreover, the isolation of these compounds may also increase the pretreatment efficiency at sequential cycles.

Enzymatic hydrolysis of the solid stream remaining after the 1 pretreatment cycle resulted in the production of 288.9 g glucose, 39.2 g xylose, and 6.5 g arabinose (Fig. 1). Fermentation of these carbon sources with the bacterial strain A. succinogenes led to the production of 200.8 g succinic acid. After reuse of the solvent and considering the removal of solids, the glucan and hemicellulose content in the remaining solids and the yield of the enzymatic hydrolysis, the succinic acid produced was 208 g in the 2nd pretreatment cycle, 204.9 g in the 3rd cycle, 184.5 g in the 4th cycle and 94.3 g in the 5th cycle.

Conclusions

This study demonstrates that the integration of ChCl:LA pretreatment of grape pomaces and stalks within a biorefinery concept is feasible and highly efficient considering ChCl:LA recycling, hydrolysis efficiency of glucan and hemicellulose, and succinic acid production. The efficiency of succinic acid production in the crude hydrolysate was similar to the respective efficiency achieved when commercial sugars were used. The ChCl:LA could be efficiently recycled and reused for at least 4 pretreatment cycles showing that ChCl:LA is a promising solvent for the pretreatment of the lignocellulosic fraction of grape pomaces and stalks. This process could be applied in different wine producing regions after certain modifications depending on the composition of grape pomaces and stalks.