Olive stones (Olea europaea L. of the variety ‘Picual’) were provided by a local mill industry from Jaén (Spain). The raw material was dried at room temperature under shade for three weeks and then stored in sealed plastic bags until use.

Experiments were conducted in a 2 L Parr batch reactor, Series 4522 (Moline, IL, USA). For the preliminary assessment of the hydrolytic power of sulphuric acid and nitric acid, 50 g of dry olive stones were loaded into the reactor along with 300 mL (solid/liquid ratio=1/6) of distilled water (autohydrolysis) or the desired acid solution (dilute acid hydrolysis). For the optimization of the dilute sulphuric acid hydrolysis, different olive stone loadings (29.3 g, 50.0 g, 100 g, 150.0 g and 170.7 g) were placed into the reactor along with 300 mL of the pertinent sulphuric acid solution (0.006 M, 0.010 M, 0.020 M, 0.030 M or 0.034 M), i.e. solid /liquid ratios between 1/10.23 and 1/1.76 (g·mL⁻¹). The mixture was stirred at 250 rpm and heated until the desired maximum temperature was reached (200, 205, 210 and 220 °C for preliminary trials, 200 °C for dilute acid hydrolysis optimization, respectively). Finally, the reactor was cooled to room temperature in less than 10 min (residence time at the maximum temperature=0). After hydrolysis, the content of the reactor was separated by filtration into solid residue and hydrolysate. The former was water-washed to remove any remaining sugars, dried at room temperature and characterized by the methods described in Section 2.4. The liquid fraction was analyzed in D-glucose, D-xylose, L-arabinose, D-galactose, 5-hydroxymethyl-furfural (5-HMF) and acetic acid.

The severity parameter of autohydrolyses and acid hydrolyses, log R₀, was calculated according to Overend and Chornet (1987). The parameter R₀ combines the effect of reaction time and temperature on the polysaccharide-degradation process as follows: R 0 = ∫ o t e t ( t ) - T r / 14.75 d t

where T(t) is a function that relates the reaction temperature with reaction time (t), and Tr is the reference temperature (100 °C). More details about the procedure can be found elsewhere since the severity parameter has been widely used in the autohydrolysis of lignocellulose biomasses such as maize, poplar, birch, olive pruning and olive stones (Tortosa et al., 1995; Fernández-Bolaños et al., 2001; Cuevas et al., 2009; García et al., 2010; García Martín et al., 2010).

Response surface methodology (RSM) was employed for statistical data treatment and the optimization of sulphuric acid hydrolysis conditions by multiple regression analysis, using the software Model 7.0 software (Umetrics AB, Umeå, Sweden). A 2² central composite circumscribed design (CCCD) with two independent variables (sulphuric acid concentration and olive stone loading) at two different levels (from 0.006 M to 0.034 M, and from 29.3 g to 170.7 g, respectively). Four star (axial) points and three central points (Fig. 1) were adopted to find linear, quadratic and interaction effects of independent process variables on experimental responses.

A second-order polynomial model was fitted to the experimental data, Y = b 0 + ∑ i = 1 2 b i ⋅ X i + ∑ i = 1 , j = 1 2 b ij ⋅ X i ⋅ X j + ∑ i = 1 2 b ii ⋅ X i 2 ( i < J )

where Y is the response variable, X_i and X_j are the experimental factors with coded units, and b₀, b_i, b_ii and b_ij are the constant, linear, quadratic and cross product coefficients, respectively. The statistical validation was performed by one-way ANOVA test with a 95% confidence, and the optimal condition values were determined based on the response surface methodology.

The raw material as well as the solids resulting from autohydrolyses and acid hydrolyses were characterized according to moisture (TAPPI T257), insoluble acid lignin (IAL; TAPPI T222 os-74), and ash (TAPPI T211). Neutral detergent fiber (NDF) and acid detergent fiber (ADF) were quantified according to Van Soest and Wine (1967). The percentages of hemicellulose (HEM) and cellulose (CEL) were calculated according to the expressions:

HEM=NDF–ADF

CEL=ADF–IAL

The percentage of xylans (20.6%) and acetyl groups (4.0%) in the raw material was determined by a modification of the method proposed by Puls et al. (1985). The potential yield of D-xylose is defined as 0.88 g xylan per g pentose. This number results from the quantitative conversion of D-xylose to xylan using the stoichiometric factor 150/132 (ratio of the molecular weight of pentose per molecular weight of anhydro pentoses that makes up xylans). The maximum D-xylose extraction attainable was then 23.4%. Similarly, the conversion factor of acetic acid to the acetyl group is 0.72 and thereby the potential yield of acetic acid was 5.6%. The concentration of D-glucose, D-xylose, L-arabinose, D-galactose and 5-hydroxymethyl-furfural in the hydrolysates were analyzed by high-performance liquid ionic chromatography (HPLIC) using amperometric detection. The HPLIC (Dionex ICS 3000, Sunnyvale, CA, USA) was equipped with a CARBOPAD PA20 analytical column (3×150 mm) in combination with a CARBOPAD PA20 guard column (3×30 mm). Elution took place at 30 °C. The eluent was 0.002 M NaOH, at a flow rate of 1 mL·min⁻¹. After dilution, the samples were filtered through a 0.22 µm nylon membrane (Millipore). Acetic acid was quantified following the enzymatic method described by Bergmeyer and Möllering (1974). All determinations were performed in duplicate.

The results obtained during the preliminary assessment of the hydrolytic power of sulphuric acid nitric acid illustrated that the use of dilute sulphuric acid led to the maximum D-xylose yield along with higher solid recoveries. This solid obtained after hydrolysis, composed mainly of cellulose and lignin, is a potential raw material for the manufacture of pellets which are used for heat and power production (García et al., 2012; Rafiqul et al., 2013). Therefore, sulphuric acid was chosen instead of nitric acid for optimizing the dilute acid hydrolysis. In spite of the fact that 210 °C was the temperature that maximized D-xylose production, 200 °C was the temperature selected to avoid sugar degradation as far as possible when using H₂SO₄ concentrations higher than 0.025 M.

D-xylose was found to be the main sugar in hydrolysates, as expected. D-xylose yield was much higher than those of other hemicellulosic sugars (L-arabinose, D-galactose and D-glucose) and was strongly dependent on experimental operating conditions (Table 4).

The resulting equation for the percentage of D-xylose yield (Y_XYL,%) based on actual values was as follow:

Y_XYL (%)=20.5959–0.0908 L+316.6140 A–0.0008 L²–25276.6 A²+9.775 L A

where L and A stand for the actual olive stone loadings and acid concentrations, respectively.

The D-xylose yield model obtained was statistically significant (p-value <0.001). The goodness of fit (R²) was 0.995 and thereby the model as fitted explained 99.5% of variability for D-xylose yield while the remaining 0.5% was explained by the residue (Table 3). According to RMS outputs, 20.4 g D-xylose can be obtained from 100 g olive stones (87.2% D-xylose extraction) under the most favorable conditions (50 g solid loading, 0.016 M H₂SO₄). This yield is comparable to those achieved by Saleh et al. (2014) and Cuevas et al. (2013). Saleh et al. (2014) extracted 89.7% of the potential D-xylose of olive stones (20.9 g D-xylose per 100 g olive stones) by dilute sulphuric acid hydrolysis with the same solid/liquid ratio (1/6) at a temperature of 195 °C, a reaction time of 5 min at this temperature, and a sulphuric acid concentration of 0.025 M. In the meantime, Cuevas et al. (2013) achieved full D-xylose recovery (23.3 g D-xylose per 100 g olive stones) by the synergistic action of hemicellulases and β-glucosidases on pre-treated olive stones by autohydrolysis. Perceptibly, dilute sulphuric hydrolysis can achieve the same yields than enzymatic hydrolysis (preceded by a pre-treatement) in one single stage.

Results illustrated that the higher the solid loading, the higher the required acid concentration to achieve the maximum D-xylose yield. From the response surface and mixture contour (Figure 6) plots it was concluded that the highest D- xylose production was achieved nearby the diagonal that pass through the points (−1, −1), (0, 0) and (+1, +1) of the factorial design (Figure 1), i.e. 50 g and 0.01 M, 100 g and 0.02 M, and 150 g and 0.03 M, respectively. The low sulphuric acid concentration was a hindrance to achieving high D-xylose yields for the conditions above this diagonal of the factorial design (high solid loadings). On the contrary, the excess of acid in the conditions below the diagonal of the factorial design (low solid loadings) probably led to sugar degradation to some extent which resulted in slightly lower D-xylose yields. Similarly, the maximum yields of the other hemicellulosic sugars (L-arabinose and D-galactose) were found at the diagonal points of the factorial design (Table 4). Unfortunately, the conditions which maximized solid recovery (Table 2) did not correspond to those that minimized hemicellulosic sugar yield (Table 4). Furthermore, solid solubilization was related to D-glucose production as maximum D-glucose yields were found at the conditions that minimized solid recovery, namely run numbers 1 and 3 (Tables 2 and 4). This fact is logical since the increase in cellulose hydrolysis, and therefore the increase in D-glucose yield, leads to lower solid recovery after dilute sulphuric acid hydrolysis.

Finally, no trend was found either for acetic acid or 5-hydroxymethyl-furfural. Complete release of acetic acid arising from acetyl groups, even exceeding the theoretical maximum attainable (5.6%), was found (Table 4). High concentrations of acetic acid can inhibit the microbial conversion of D-xylose to xylitol. Nevertheless, some xylose-fermenting yeasts, mainly from the Candida genus, have been found not to be inhibited by acetic acid. In this way, García et al. (2011) found a slow consumption of acetic acid by Candida tropicalis from the start of the fermentation process until depletion. These authors (García et al., 2011) emphasized that the acetic acid consumption was simultaneous with those of D-glucose and D-xylose. With regard to 5-hydroxymethyl-furfural yields, the values achieved were low and cannot be regarded as a potential source of inhibition in the subsequent fermentation of sugars.

On a large scale, xylitol is currently produced by the chemical reduction in D-xylose derived mainly from wood hydrolysates (Parajó et al., 1998; Winkelhausen and Kuzmanova, 1998). The xylitol production from olive stones, by acid hydrolysis and subsequent fermentation (using non-traditional yeasts) could be an interesting alternative to the current synthesis.