Biofuel, Bioenergy and Feed Valorization of By-Products and Residues from Hevea brasiliensis Cultivation to Enhance Sustainability

: In the latex production chain, rubber tree seeds ( Hevea brasiliensis ) represent an underutilized fraction with high potentialities, which can increase the sustainability of the whole process if rightly valorized. In the present study, the quality of all the fractions obtained from the rubber fruit were evaluated, with the aim to identify possible applications for their valorization with a circular economy perspective. Seeds from ﬁve di ﬀ erent varieties of rubber tree were analyzed. Furthermore, a whole mass and energy balance was deﬁned, which has allowed us to deﬁne hypothetical production scenarios. The obtained results show negligible di ﬀ erences among varieties. Shells and capsules have shown a composition similar to woody biomass, with high heating values (more than 16.5 MJ kg − 1 ), low nitrogen content (below 0.5% on weight basis ( w / w )) and reduced ash content (0.51% w / w and 1.90% w / w , respectively). Kernels were chemically extracted comparing two di ﬀ erent solvents: n-hexane and ethanol. Both solvents showed similar extraction yields, i.e., 49% w / w and 46% w / w for n-hexane and ethanol, respectively. The resulting extraction ﬂour was characterized by a high protein content (around 40% w / w ) making it suitable for animal feeding. The rubber seed oil could be used in blends of di ﬀ erent vegetable oils for biodiesel production. All this information is useful for improving the sustainability of the latex production chain and to assess the sustainability of possible bioenergy value chains. The results were statistically evaluated with the Tukey–Kramer test at the 0.05 level of signiﬁcance.

This comprehensive approach focused on this subject, also considering different varieties of Hevea brasiliensis, is missing in the scientific literature and could represent the basis for performing sustainability assessment of these productions based on primary data.

Introduction
Samples of rubber seed coming from Africa were supplied by Green Dragon Environmental Ltd. (Chichester, United Kingdom), a UK company providing professional engineering services to renewable energy companies. Biomass Lab of Università Politecnica delle Marche carried out several analyses on the rubber seed samples with the aim to verify mass balances of the first processing steps and the quality of products and residues. Different varieties of rubber seed were tested for solvent extraction to understand the variability of the extracted rubber seed oil (RSO) in terms of quantity and quality considering a possible application for biodiesel production, due to the interest of different stakeholders in this possibility. A test of mechanical extraction was carried out to assess the differences in terms of oil yield and have a comparison with data reported in scientific literature. The by-products of cultivation, seed preprocessing and solvent extraction were analyzed as well to understand possible valorization options with the aim to increase the sustainability of the biodiesel production chain and indirectly to increase also the sustainability of the latex production chain, currently the main driver of rubber tree cultivation. Materials, treatments and analyses are detailed in the following paragraphs.

Materials
Rubber seeds were handpicked in Alépé, Ivory Coast, during September 2019 and shipped to Biomass Lab in Italy. Seed samples from five varieties of rubber tree were evaluated, in particular: All samples, around 5-10 kg for each variety, were shipped and stored in a plastic bag until analysis. An additional sample of empty capsules was sent for evaluation.

Pre-Treatment
The samples received in the Biomass Lab were sealed in plastic bag and there was no evidence of impurities such as dirt, mud, sand and stones. For each sample of rubber seed, kernel and shell were split by manual dehulling. Rotten rubber seeds were removed from the not degraded one. The share of rotten seeds was always below 5% w/w. Masses of all the fractions were recorded after each step to define an overall mass balance. No drying step was needed as a pretreatment because of the relatively low moisture content of the received samples.

Chemical Oil Extraction
Two solvents were tested for the chemical extractions: hexane and ethanol, which represent the common extraction solvent used in industry and a possible alternative, respectively. Although ethanol is not so common as seed oil extraction solvent, it can be biogenic and can be easily obtained from a large variety of biological materials [28] also in developing countries (e.g., by fermentation) and could be employed for rubber seed oil extraction at local level to further improve sustainability. Furthermore, hexane shows toxic properties and originates from non-renewable sources [29].
For each extraction test, 250 g of kernels were ground in a bench-top mill (mod. A11 basic analytical mill, IKA, Staufen im Breisgau, Germany) just before extraction. Afterwards, a representative portion of 25 g was extracted with 400 mL of n-hexane (Sigma Aldrich, purity > 95%) or ethanol (Baker, purity 96%) in a standard Soxhlet apparatus for 24 h.
The extraction solvent was removed from the rubber seed (RSO) oil at 70 • C or 85 • C depending on the solvent. The solvents were recovered by condensation. A following treatment at about 100 • C under vacuum in a rotavapor apparatus (mod. Laborota4000, Heidolph, Schwabach, Germany) was performed to remove water from the oil sample. The RSO obtained was stored in the dark at 4 • C until analysis.
The solid residue (flour) was dried at 85 • C to remove residual solvents before storing it in plastic bag until analysis. The mass balance of the whole extraction process was evaluated. Chemical extraction tests were performed in triplicate for each solvent. The results were statistically evaluated with the Tukey-Kramer test at the 0.05 level of significance.

Mechanical Oil Extraction
Preliminary tests were carried out to tune the mechanical extraction of the rubber seeds on a single screw expeller for vegetable oil extraction (mod. 205 S, COTER, Bracco Srl, Bagnatica, Italy). Dried seed results are unsuitable for mechanical extraction with the used device, causing the block of the system. The successful tests were conducted by using not dried and not dehulled seed. A test was carried out with the kernel but the dehulled material was not able to reach the high pressures required for mechanical extraction (data not shown), so the extraction test was conducted on the whole seed, in line with literature [13].
Mechanical extraction yield was assessed on GT1 variety due to the mass of sample required. A portion of around 5 kg of whole seeds was extracted with the expeller at 80 • C and with an 8 mm diameter compression ring with three extraction cycles.
The raw RSO was clarified by decantation and centrifugation and the sludge removed. The solid residue (oil cake) was stored in plastic bag until analysis.

Chemical-Physical Analysis of Rubber Seed Oil and By-Products/Residues
A list of the analytical parameters evaluated is reported in Table 1, describing references of standard methods employed, principle of analysis and fractions analyzed (see standard test methods for further details).
Except for RSO, all fractions were initially characterized by a proximate analysis (i.e., moisture content, higher and lower heating values, ash content) necessary to define mass balances and to understand which are the most promising applications for such materials.
Shells, kernels and capsules were further characterized as required or evaluating a possible application in the solid biofuel sector, i.e., chlorine and sulfur contents, elemental analysis (CHNO).
The RSO extracted was analyzed for the most important quality parameters: acid number, iodine number, saponification number. Moreover, RSO was investigated analyzing the fatty acid methyl ester (FAME) composition, important for a possible application for biodiesel production. Kinematic viscosity and density were determined only on the oil extracted mechanically due to the significant amount needed for the analysis. The flours were also analyzed for cyanide content and other parameters useful for a possible use for animal feeding. Analyses were performed in triplicate except for RSO analyses. The results were statistically evaluated with the Tukey-Kramer test at the 0.05 level of significance.

Mass and Energy Balances and Possible Production Scenarios
Taking into account all results obtained in this study (primary results) and some others taken from the literature (secondary results), it was possible to define a production scenario for latex by-products, in order to appreciate the positive contribution to the latex production chain.
The following data were taken from literature or estimated (secondary data): • Mass yield of rubber seed referred to 1 hectare equal to 150 kg ha −1 [2,4,12,13]. Higher yields are reported by other authors [27] but in the present study the lower yield was used to take into account that part of the seeds could be difficult to harvest and should be used to help preserve the ecological balance in the soil.

•
According to the authors' knowledge, there are no data in literature related to capsule yield. The value was estimated from the small quantity obtained and it should be considered as a preliminary indication. This value was also employed to estimate the rubber tree fruit yield. • Energy content of rubber seed was calculated by energy contents and mass balances of both shell and kernel fractions.

•
In the chemical extraction step ethanol was considered as solvent since it represents a more environmentally sustainable scenario.
Mass and energy balances for each residue were evaluated taking into account some considerations: • Masses were expressed on dry matter basis since moisture content is a parameter strongly dependent by logistic variables.

•
Energies were expressed as lower heating value (LHV) to obtain a more realistic estimation of the energetic performances, keeping the dry basis as mentioned above. • Mass and energy balance were referred to 1 ha of rubber tree cultivation • Mass percent values were referred on whole seed basis since represents the main residue of the latex chain. Table 2 lists seed moisture contents and mass balances of kernel and shell. The average seed moisture value is about 16.1% w/w, ranging from 11.9% w/w (GT1) to 20.5% w/w (PB 260). Literature reports several values for seed moisture content from around 3% w/w [7,30] up to around 25-30% w/w for collected seed [13]. Since rubber tree capsules explode when ripe, releasing the seeds on the ground, moisture content is strictly related to the collection methods employed, i.e., collection time after the seed dispersal, the amount of rain form the dispersal, storage conditions [2], more than the variety typology. Note: values in the same column that do not share a letter are significantly different at p-value 0.05.

Results and Discussion
High fungal contamination could be probably related to the high seed moisture content, which promotes the mold formation during the storage as observed by Widyarani et al. [2].
Shell and kernel represent 47.5% w/w db and 52.3% w/w db of the whole seed, respectively, showing a low data variation among the tested varieties. The obtained results are in line with literature [2,13,31]. Table 3 reports characterization results for kernel, shell and capsule.  Note: values in the same column that do not share a letter are significantly different at p-value 0.05 (e.g., "fg" and "ef" are not significantly different because sharing "f". In addition, "fg" is also not significantly different from "g", while "ef" is significantly different from "g").
It is clear that kernel and shell show a very different composition. Due to the higher protein content, kernel contains much more nitrogen (3.3% w/w) and sulfur (0.12% w/w) than shell (0.50% w/w of ash and 0.01% w/w of sulfur). For the same reason, kernel exhibits a higher ash content than shell, 3.25% and 0.51%, respectively.
Ishak et al. [26] found similar results for ultimate analysis, highlighting higher values for C (64.45%), N (3.63%) and S (0.32%) in kernel with respect to shell. However, shell shows a higher N content (1.5%) with respect to this work. Rubber seed shell characterized by Reshad et al. [10] showed an N content of 3.13%.
Although both fractions are characterized by high heating values, the RSO contained in kernel gives to it a higher energy content with respect to shell, 24 MJ/kg against 16.5 MJ/kg of LHV, respectively. The same trend is evident for the respective carbon contents, since vegetable oil is mainly composed of carbon (above 70% w/w). Chlorine contents are low for both fractions, i.e., less than 0.02% w/w.
Similar results for heating values can be found in literature [10,26].
From the results reported in Table 3, it can be noticed that shell has a composition similar to woody biomass, with some characteristics that comply even with solid biofuel standard limits for high quality pellet classes, i.e., ash content, heating value and Cl/S contents for class A1 and nitrogen content for class A2 [32]. Similar considerations can be applied to capsule fraction, which presents a lower N content (<0.1% w/w) than shell but a higher ash (1.90% w/w) and S (0.03% w/w) contents, making it a material suitable for energy applications complying with the recent hogfuel requirements [33].
Given the small quantity of capsule sample analyzed, the results should be considered just as a preliminary indication, needing a further in-depth analysis for understanding the possible quality variability.
Different authors have found low ash and moisture contents for the shell fraction [10,26,31]. The results obtained from characterization analysis show negligible differences among varieties, suggesting a quite standardized feedstock.
High protein and oil contents make kernel an unsuitable material for energy valorization, from both environmental and economic point of views. In fact, a high amount of pollutants can be emitted during combustion of kernel, conversely more valuable products can be obtained after oil extraction, i.e., biodiesel from RSO and feed from extracted flour. Table 4 reports results for the chemical oil extraction. Hexane extracts more RSO than ethanol, until 8.0% w/w more, although differences in yields on an average basis are limited, i.e., 49.0% w/w against 46.2% w/w, respectively. Most of the papers found in literature about the oil chemical extraction of rubber seed use hexane as a solvent; few evaluate other solvents with a different dielectric constant [8,9,18,[34][35][36]. To the authors' knowledge, papers focused on employing ethanol are limited [34,36].
The oil content yields obtained are in line with literature, the values of which range from around 33% to 55% for hexane extraction [26,31,34,37].
Mohd-Setapar et al. [34] have extracted around 45% of RSO, employing ethanol against around 55% obtained with hexane. Wildan et al. [36] found that hexane extracts around 12% more RSO from waste rubber seed with respect to ethanol.
After the solvent removal step from the ethanol/oil mixture, a white solid residue remains in the flask which accounts for around 10.3% on kernel basis.
Other authors [34] have observed white particles in both water and ethanol/water (70:30 v/v) after rubber seed extraction. Same authors describe the oily residues obtained as too viscous and not transparent as hexane extracted one. In the present work RSO-ethanol was easily separated from the solid residue after a mild warming of the mixture at 40 • C. At these conditions the RSO has become more fluid and it was separated by pouring it off from the flask while the residue remained stuck on flask wall.
Since the residue was highly water soluble, it could be composed mainly by sugars as it is known that ethanol exhibits a good ability to extract these compounds [38].
Differences in extraction yields and extract compositions are linked to solvent and compounds polarities. The higher ethanol polarity with respect to hexane allows for a co-extraction of compounds other than triglycerides, which are insoluble in non-polar solvent [34]. For the same reason hexane left more exhaust kernel (flour) at the end of the extraction process than ethanol, with a 51% and 43.4% mass balance, respectively.
Visually, the RSO extracted with ethanol was orange and turbid with respect to the hexane extracted RSO, yellowish and more transparent.
In Figure 1 are plotted the results of mechanical oil extraction for GT1 variety. Mechanical extraction produces 15.9% w/w of RSO on whole seed and as received basis, equal to 35.4% w/w on kernel dry basis, which accounts for 73% w/w of the total RSO contained (referred to hexane extraction).
Resources 2020, 9, x FOR PEER REVIEW 10 of 18 Mohd-Setapar et al. [34] have extracted around 45% of RSO, employing ethanol against around 55% obtained with hexane. Wildan et al. [36] found that hexane extracts around 12% more RSO from waste rubber seed with respect to ethanol.
After the solvent removal step from the ethanol/oil mixture, a white solid residue remains in the flask which accounts for around 10.3% on kernel basis.
Other authors [34] have observed white particles in both water and ethanol/water (70:30 v/v) after rubber seed extraction. Same authors describe the oily residues obtained as too viscous and not transparent as hexane extracted one. In the present work RSO-ethanol was easily separated from the solid residue after a mild warming of the mixture at 40 °C. At these conditions the RSO has become more fluid and it was separated by pouring it off from the flask while the residue remained stuck on flask wall.
Since the residue was highly water soluble, it could be composed mainly by sugars as it is known that ethanol exhibits a good ability to extract these compounds [38].
Differences in extraction yields and extract compositions are linked to solvent and compounds polarities. The higher ethanol polarity with respect to hexane allows for a co-extraction of compounds other than triglycerides, which are insoluble in non-polar solvent [34]. For the same reason hexane left more exhaust kernel (flour) at the end of the extraction process than ethanol, with a 51% and 43.4% mass balance, respectively.
Visually, the RSO extracted with ethanol was orange and turbid with respect to the hexane extracted RSO, yellowish and more transparent.
In Figure 1 are plotted the results of mechanical oil extraction for GT1 variety. Mechanical extraction produces 15.9% w/w of RSO on whole seed and as received basis, equal to 35.4% w/w on kernel dry basis, which accounts for 73% w/w of the total RSO contained (referred to hexane extraction). Besides extraction cake, a solid residue (sludge) was obtained after the RSO clarification process (i.e., decantation and centrifugation), which has retained part of the oil. Since in the industrial Besides extraction cake, a solid residue (sludge) was obtained after the RSO clarification process (i.e., decantation and centrifugation), which has retained part of the oil. Since in the industrial extraction process sludge is re-extracted by addition to the initial feedstock (i.e., rubber seed), the oil contribution associated with this portion is not considered in the overall mass balance. Similar results were reported also by other authors. Ebewele et al. [37] have extracted RSO mechanically using a hydraulic press obtaining a yield of 28.5% w/w of kernel weight at 8 MPa and 70 • C. Lower pressure yields less RSO, as reported by Sabarish [39], which has obtained a yield of RSO about 16.5% w/w db of kernel weight with 3.5 MPa pressure. Aigbodion et al. [40] have extracted 23% w/w RSO of whole seed using a pilot mill. Solid fraction produced after mechanical extraction (extraction cake) was higher than the one obtained from solvent extraction (flour) due to the presence of residual oil and the shell fraction.
Mechanical extraction with continuous screw press produces RSO without solvent employment and with relatively low operating costs; although, this solution does not allow one to separate and valorize the shell fraction before the extraction process. Moreover, from the animal nutrition point of view, the presence of residual oil in the extraction cake could make it difficult to manage fat balance in animal feeding applications and can significantly reduce its storage time (i.e., oil rancidification). Table 5 reports characterization results for the chemical extraction flours. The comparison between flours obtained with two extraction solvents does not highlight important differences in composition. Flour from ethanol extraction shows a higher protein content than hexane, 43.7% w/w against 37.4% w/w, respectively. The difference could be attributable to the solid residue extracted by ethanol which returns a flour richer in protein. To the author knowledge in literature there are no papers on characterization of ethanol extracted flours. The obtained results are in line with Oluodo [12], which reviewed different papers on nutritional composition of rubber seed founding a content of moisture 3.99%, crude protein 23%, crude fat 68.5% and ash 4.3%.
Cyanide is recognized as a possible drawback of using this material for feeding and a specific analysis was carried out. Cyanide content was tested in GT1 flour extracted with hexane and ethanol and resulted 36 and 19 ppm, respectively. In both cases the cyanide amount was less than 500 ppm, which is considered the general safe level for animal forages [41]. Differences in cyanide contents of meals, higher for hexane extraction than ethanol, could be due to a partial removal of cyanogenic glycoside (called linamarin). Since linamarin is a water soluble compound [42], it could be partly removed by a polar solvent like ethanol during RSO extraction.
The results of oil characterization are shown in Table 6. High and variable values of acid number were obtained for RSO but are in line with literature [8,18,35]. According to several authors, acid number can also be significantly affected by logistics factors (i.e., collection time and method, storage) [11,31,43,44]. Abdulkadir [7] found an increase in RSO acid number from an initial value of 34 mg KOH/g for fresh seeds up to around 80 mg KOH/g for RSO extracted (with hexane) after two months of storage due to enzyme activities [31]. The obtained results agree with literature since around three months passed from seeds collection to laboratory analyses. The results show a higher acid number for RSO extracted with hexane than ethanol, as found by Attah [35], which compared FFA extraction for different solvents with similarly different dielectric constants. Thus, ethanol seems to extract fewer acid compounds than hexane in aged seeds. Iodine number and saponification number are in line with literature [17,22] and not affected by solvent polarity [35].  −1 a.r. ) (% w w −1 a.r. ) Note: values in the same column that do not share a letter are significantly different at p-value 0.05 (e.g., "cd" and "de" are not significantly different because sharing "d". In addition, "cd" is also not significantly different from "bc", while "cd" is significantly different from "e"). The results of fatty acid composition are in line with the range reported in scientific literature [14,18,22,45]. The different solvents used for extraction do not seem to influence the fatty acid composition. With respect to a standard vegetable oil used for biodiesel production (EN 14214) [46], it seems there is a higher amount of polyunsaturated fatty acids in RSO, which could make the biodiesel produced unstable, suggesting the use in blends of different vegetable oils.
The kinematic viscosity (40 • C) of RSO obtained by mechanical extraction was 36.5 cSt for GT1. The density of oil (at 30 • C) was about 0.88-0.89 g/cm 3 , in line with scientific literature [8,9,19,31]. Figure 2 shows an overall mass and energy balance reported on a hectare basis for by-product and residues obtained from the latex production chain. The results of fatty acid composition are in line with the range reported in scientific literature [14,18,22,45]. The different solvents used for extraction do not seem to influence the fatty acid composition. With respect to a standard vegetable oil used for biodiesel production (EN 14214) [46], it seems there is a higher amount of polyunsaturated fatty acids in RSO, which could make the biodiesel produced unstable, suggesting the use in blends of different vegetable oils.

Mass and Energy Balances and Possible Production Scenarios
The kinematic viscosity (40 °C) of RSO obtained by mechanical extraction was 36.5 cSt for GT1. The density of oil (at 30 °C) was about 0.88-0.89 g/cm 3 , in line with scientific literature [8,9,19,31]. Figure 2 shows an overall mass and energy balance reported on a hectare basis for by-product and residues obtained from the latex production chain.

Mass and Energy Balances and Possible Production Scenarios
Since there were no substantial differences among tested varieties, averaged data were employed to make calculations. The percentages represent the mass fraction with respect to the rubber seed mass (100%). Consequently, the rubber tree fruits are reported as 200% because the capsules were also included.
It can be noticed that a preliminary separation of shell and kernel, followed by a chemical extraction, enables a recovery of 74% of the initial energy content (seed) only considering an energy valorization of shell and RSO. Extraction meal instead, due to the high protein content, can be addressed to animal feeding, adding value to this fraction. On the other hand, a mechanical extraction needs the whole seed (according to results of this study), which entails a lower RSO extraction and a solid residue (oil cake) with high nitrogen content and low storage stability (oil rancidification). Since there were no substantial differences among tested varieties, averaged data were employed to make calculations.
The percentages represent the mass fraction with respect to the rubber seed mass (100%). Consequently, the rubber tree fruits are reported as 200% because the capsules were also included.
It can be noticed that a preliminary separation of shell and kernel, followed by a chemical extraction, enables a recovery of 74% of the initial energy content (seed) only considering an energy valorization of shell and RSO. Extraction meal instead, due to the high protein content, can be addressed to animal feeding, adding value to this fraction. On the other hand, a mechanical extraction needs the whole seed (according to results of this study), which entails a lower RSO extraction and a solid residue (oil cake) with high nitrogen content and low storage stability (oil rancidification).
From the results reported in Figure 2 and in previous tables, it is possible to define production scenarios to valorize residues and products. All the values reported below are referred to 1 ha cultivation of rubber seed.
Capsules can be burnt in specific devices to produce the thermal energy needed to dry about 5 ton of rubber seed (from 25 to 5% moisture content [13]) to improve seed stability. Shell can be used as well to dry about 4.5 ton of rubber seed. Considering these numbers, capsules and shells can widely satisfy the need of thermal energy for drying rubber seeds and also a significant quantity of other products.
As an alternative, capsules and shells could be used to produce electricity in biomass power plants. As an indication, a production of about 300 kWh e can be estimated.
Another possibility for shell could be the use as feedstock for the production of pellet to be combusted in small domestic stoves for heating. It can be estimated that shell pellet can be used in an 11 kW stove for about 24 h at maximum power.
According to the RSO production calculated in this study, it could be possible to produce more than 30 kg of biodiesel, the same amount needed for driving a car for more than 450 km.
A last consideration can be made on the extraction meal. As an indication, it could be exploited to substitute feeding materials equivalent to about 30 kg of soybean meal. This possibility can be achieved by carefully including a limited amount of rubber seed meal in the animal diet. The manure can be then employed to return nutrients and organic matter to the soil closing the cycle. According to some authors, the extraction meal itself could be used as a weed controller or as a biofertilizer [47].

Conclusions
The study evaluated the quantity of residues of rubber tree cultivation and their quality to give useful indications for a possible valorization. From an energy point of view, there is a significant amount of biomass material linked to shell and capsules that could be exploited to provide heat for many applications also within the possible "RSO to biodiesel" chain. Capsules and shells can be used to produce enough thermal energy for drying rubber seeds and also a significant amount of other products.
From each hectare cultivated with rubber seed, about 30 kg of RSO biodiesel could be produced, substituting about 26 kg of fossil fuels.
The extraction meal could be used as a biofertilizer or for feeding purposes to further improve the sustainability. For each hectare of rubber seed cultivation, about 30 kg of soybean meal can be substituted with rubber seed meal.
This information is useful to improve the sustainability of the latex production chain and to assess the sustainability of the possible bioenergy value chains. The quantity and quality of the studied biomass materials are interesting, but the practical feasibility of these win-win solutions should be defined case by case considering logistical and technical issues.