Valorization of secondary feedstocks from the agroindustry by selective catalytic oxidation to formic and acetic acid using the OxFA process

The selective oxidative conversion of seven representative fully characterized biomasses recovered as secondary feedstocks from the agroindustry is reported. The reaction system, known as the “OxFA process,” involves a homogeneous polyoxometalate catalyst (H8PV5Mo7O40), gaseous oxygen, p-toluene sulfonic acid, and water as solvent. It took place at 20 bar and 90 °C and transformed agro-industrial wastes, such as coffee husks, cocoa husks, palm rachis, fiber and nuts, sugarcane bagasse, and rice husks into biogenic formic acid, acetic acid, and CO2 as sole products. Even though all samples were transformed; remarkably, the reaction obtains up to 64, and 55% combined yield of formic and acetic acid for coffee and cocoa husks as raw material within 24 h, respectively. In addition to the role of the catalysts and additive for promoting the reaction, the influence of biomass components (hemicellulose, cellulose and lignin) into biogenic formic acid formation has been also demonstrated. Thus, these results are of major interest for the application of novel oxidation techniques under real recovered biomass for producing value-added products.


Introduction
Biomass plays a key role towards a future green economy for the production of renewables, heat, biofuels, pharmaceuticals, and green chemical feedstocks [1][2][3]. However, bioenergy at industrial production might trigger possible water privation, can limit land availability, and certainly, will increase the food prices [4][5][6][7][8][9]. Second generation feedstocks, which include biomass residues from the agroindustry, appear as a stop-arguing-solution. Though, most of them remain untreated and underutilized. For instance, in emerging nations, which usually generate big amounts of biomass, half of them end up in open-air dumps or unplanned landfills [10,11]. Therefore, implementation of novel alternatives of biomass valorization is highly desirable.
Many emerging countries with a traditionally agricultural based-economy have a great potential for recovering and converting biomass residues into energy and more valuable green chemicals. Being palm heart, sugarcane, cocoa, palm oil, rice, maize, banana, and pineapple crop the most representative [12,13]. Huge efforts have been made in order to show its potential applying common biomass conversion processes in the last years. Producing pyrolysis gas, biooil, and solid biochar via thermochemical methods (e.g., fast-pyrolysis and combustion) has been studied [14][15][16]. Biochemical processes, like anaerobic digestion, have been also applied for biogas generation [17,18]. Moreover, alcoholic fermentation and submerged fermentation techniques have been studied for bioethanol and organic acid production, respectively [19]. Nevertheless, the use of alternative ways to convert agricultural biomass into products with more value-added applications (e.g., formic acid (FA), acetic acid (AA)) is still an unchartered territory. Formic acid, for example, has found applications in many industries, which include textile, agriculture, and pharma, among others. Lately, it has been also studied as energy carrier and ideal fuel for fuel cells [20,21]. In this sense, Albert et al. have shown the possibility of transforming biomass into green fuels and chemical via bio-FA decomposition to syngas (H 2 and CO), and subsequently, Fisher-Tropsch synthesis [22], showing its importance as a future energy vector.
In this context, one attractive approach for valorization of organic residues is the transformation of lignocellulosic biomass into formic acid and acetic acid via metal catalysts (see Fig. 1a and b) [23][24][25][26][27][28], the so-called OxFA process [29]. The OxFA process runs under moderate temperature conditions (< 373 K) with oxidants such as air or oxygen. In the last years, a number of water-soluble and non-soluble substrates have been successfully converted (glucose, cellobiose, sucrose, beech, wood, among many others) [23,26]. Polyoxometalates (POMs) are the chosen catalysts for this type of homogeneous catalytic reaction. Advantageous are their strong Brönsted acidity and stability in aqueous solution. In particular, POMs from the Keggintype [H 3+n PV n Mo 12 O 40 ] (also known as HPA-n) with different amounts of vanadium-substitution have been the most studied catalysts. High amounts of vanadium substitution in the catalysts have widely proven that both the reduction and reoxidation steps (see redox cycle in Fig. 1b) become faster. Being HPA-5 the fastest and with highest selectivity towards FA formation [30,31], due to active VO + 2 species dissolved in the reaction media. Concerning water-insoluble biomass, p-toluenesulfonic acid (TSA) showed to be an efficient additive as hydrolyzing and phase transfer agent promoting higher amounts of FA [29]. In addition to the selection of the right catalysts and assuring a fully homogenized reaction media, the biomass composition has shown to play a major role during the reaction. Previous studies have shown that POMs catalyze first the decomposition of hemicellulose, then oxidize the phenolic lignin. Thus, different biomass residues would find distinct alternative valorization pathways [27].
Herein, a complete screening for the oxidative conversion applying the OxFA process under seven representative fully characterized biomasses recovered the agroindustry is presented. In this first approach, the standard method of the OxFA process presented in Preuster and Albert [29] was applied using a HPA-5 polyoxometalate (H 8 PV 5 Mo 7 O 40 ) catalyst at 20 bar and 90 °C in aqueous solution. The yield and selectivity of the products are analyzed as function of the biomass composition. These results are of major interest for all geographies and countries with similar agro-industrial sectors and biomass composition, in order to valorize residues and producing value-added products for upcoming needs.
All chemicals were purchase and used as received without additional purification.

Catalyst synthesis and characterization
According to literature [25,32], the heteropolyacid catalyst (H 8 [PV 5 Mo 7 O 40 ]) (HPA-5) was synthesized. A P:V:Mo (1:5:7) ratio by a ICP-OES (Perkin-Elmer, Plasma 400) analysis was obtained. As depicted in Fig. S1 in Supporting Information (SI), the infrared spectra showed a characteristic stretching vibration for the kegging-oxoanion. Spectra was recorded in FT-IR spectrophotometer (Shimadzu, Prestige-21) in combination with a Golden Gate single reflection diamond ATR crystal in the range of 4000 to 400 cm −1 . As shown in Fig. S2 in SI, a two-step loss of water typical for heteropolyacids of the Keggin type was identified. Eleven and 5 wt% loss of crystallization and constitutional water, respectively, led to in 14 mol of hydrated H 2 O per mole of heteropolyacid catalyst. Moreover, a catalyst molar mass of 1857 g mol −1 was calculated. Samples were tested in a thermogravimetric analyzer (SETARAM Instruments, SET-SYS-1750 CS Evolution).

Biomass selection and characterization
Seven different residual biomasses were selected (please refer to table S1 in SI) and characterized as follows. All biomasses were dried, cut, grounded, and sieved (sieve opening: 250 µm). One and 2 g of dried biomass were used for the determination of fats and proteins content, and percentage of ash, lignin, and cellulose, respectively.
Ash content was determined applying the AOAC 942.5 standard gravimetric method [33]. Biomass fat content was analyzed by a modified version of the Randall method according to the official standards AOAC 2003.05 and AOAC 2003.06 [34] with solvent extractor equipment VELP Scientifica SER 148.
The amount of protein was quantified as nitrogen content based on the Kjeldahl method and the official method AOAC 960.52 [33] in a digester DK6 VELP Scientifica and a semiautomatic distillation equipment UDK 139. This method allowed obtaining the amount of nitrogen by the titration volume and then calculating the percentage of proteins using a nitrogen to protein conversion factor of 6.25 for all biomasses except rice husks that has a protein conversion factor of 5.95 according to the table of specific factors of food protein content [35]. Total carbohydrates content was obtained mathematically by subtracting the results of fat, protein, and ash from the total dry sample weight. Cellulose content was determined using the protocol described by Kürschner-Hanack using acetic and nitric acid [36]. The lignin content was obtained applying the AOAC 973.18 method [33], and hemi-cellulose was calculated by subtracting cellulose and lignin from the total carbohydrate content (see Table 1 and Table S2 in SI). Finally, the morphology and the surface of biomass samples were analyzed in a scanning electron microscope JEOL JSM-IT300LA (15 kV, 50 Pa). Images of raw biomasses can be seen in Fig. S3 in SI.
For molar calculations, elemental analysis was conducted to measure C, H, N, S, and O contents in the recovered biomass. The Flash 2000 Thermo Scientific elemental analyzer was operated with constant helium and/or oxygen flow (250 and 300 mL min −1 , respectively) at 950 °C. For each measurement, approximately 2 mg of biomass sample were used (see Table S3 in SI).

Biomass oxidation reactions
A 600-mL high pressure reactor (Hastelloy, C276) equipped with a gas entrainment impeller was used. For working with oxygen in a safe environment, all fittings, valves, and pipes were made of stainless Steel-Austenitic-1.4571. No leakage was ensured with a PTFE sealing. A safety released valve adjusted to 100 bar was also installed.
For each oxidation reaction, 40 mmol of biomass and 0.4 mmol of HPA-5 catalyst (substrate to catalyst ratio = 100:1) as well as 4 mmol of the additive p-toluenesulfonic acid (substrate to additive ratio = 10:1) were added to the reaction system. Then, 100 mL of water were added. Resulting in a turnover number of the catalysts of 100 which lies within an acceptable value for batch processes at laboratory scale. The autoclave was purged at least three times with oxygen. The heating was started on and the stirrer speed set to 300 rpm. Soon after the reactor reached the set temperature (90 °C), the oxygen pressure was adjusted to 20 bar. Then, the stirrer speed was increased to 1000 rpm. These conditions were kept for 24 h, while pressure and temperature were constantly monitored.

Product analysis
After the reaction took place, the system was cooled down. Gaseous products were collected in gas bags and its composition were analyzed by a gas chromatograph (Varian 450-GC (TCD/FID)) equipped with a Shin Carbon ST column (2 m × 0.75 mm ID) was used. Moreover, liquid-phase products were analyzed with a HPLC (Jasco Instrument) equipped with a Shodex SUGAR SH1011 column (300 mm × 8 mm). Prior to injection, the sample was filtered with a one-time Micropur PET filter (Altmann Analytik) with a pore width of 0.2 µm to remove the remaining catalysts. The liquid additive toluene sulfonic acid was not removed from the mixture prior to analysis. Though for scale up processes, its separation has been studied in previous studies [37]. A Jasco HPLC with Shodex SH1011 separation column (300 mm × 8 mm) was used to analyze the liquid samples. The outer material of the column is SUS-316 steel, and the column material is styrene, divinylbenzene, and copolymer. The samples were injected through a 45-µm syringe filter into a GC vial for preparation. For measurement, 10 µl of sample is injected and measured at a flow rate of 1 ml/min for 45 min. The eluent is 0.005 M sulfuric acid. The temperature of the column is kept constant at 50 °C using an oven (Jeatstram 2-Plus from Jasco). As detector served a RI detector (200 series) from Perkin Elmer.

Determination of yield and selectivity
The yields and selectivity of the gaseous and liquid products (FA, AA, CO 2 , and CO) were calculated by Eq. (1).
where Y i is the molar carbon yield of each product (determined either by HPLC (liquid phase) or GC (gas phase), n Product,I is the molar carbon amount of each product, and n C-Atoms substrate is the molar amount of carbon initially present in each substrate determined by CHNS analysis. The total carbon yield (TOC) was calculated by Eq. (2). This value was used to represent biomass conversion, as no other products were detected via the analytical methods.
The selectivity of the products was calculated by the product yield divided by the TOC.

Results and discussion
The study shows the potential of residual biomass as a secondary feedstock for oxidative conversion into more valuable chemicals, such as formic acid and acetic acid, but also carbon dioxide formation. For this purpose, seven representative biomass residues from the agroindustry were selected, namely, coffee, cocoa and rice husks, sugarcane bagasse, and waste from a palm oil factory (e.g., palm fiber, palm rachis, and palm nuts) (see source and location of biomasses in Table S1 in SI). For understanding the efficiency of the (2) TOC = ∑ i n product,i n c−atoms,substrate oxidative reaction system, the seven substrates were fully characterized. Their cellulose, lignin, and hemicellulose composition were determined, as described in the experimental section and can be seen in Table 1 and Figure S4 in the supplementary Information. Moreover, the elemental analysis of each substrate allows the calculation of yields based on the carbon atoms present in the biomass (see results in Table S3 in SI). With these results, the molecular composition of each biomass was obtained.
All biomasses were processed with a Keggin-type polyoxometalate catalyst (HPA-5, [H 8 PV 5 Mo 7 O 40 ]), TSA as an additive, and water as a solvent for 24 h, at 20 bar oxygen pressure, and 90 °C. For HPA-5 catalyzed redox processes, it has been widely demonstrated that moderate oxygen pressure (20 bar) guarantees the substrate oxidation to be ratedetermining, as the reoxidation of higher substituted HPAs is favored by higher amounts of vanadium in the Keggin structure [25]. At 90 °C, the HPA-5 catalyst shows a maximum of activity, while temperatures over 100 °C are not favorable due to possible product decomposition. In addition, in a former publication [27], TSA showed to act as the most efficient solubility promoter for water-insoluble biomass. As already mentioned, p-toluenesulfonic acid speeds the biomass conversion up without being consumed or transformed during the reaction. The results of the recovered agro-industrial biomass experiments applying the above described parameters are shown in Tables 2 and 3, respectively.
The results in Table 2 clearly show that the used homogeneously catalyzed oxidation system is able to convert all applied biomasses like coffee husks (97% combined yield), cocoa husks (100% combined yield), palm rachis, fibers and nuts (92, 88, and 71% combined yield, respectively), sugarcane bagasse (79% combined yield), and rice husks (53% combined yield) into formic acid, acetic acid, and carbon dioxide. Under the reaction conditions, formation of carbon monoxide (≈ < 1.5% CO yield) due to FA decomposition can be neglected. These results were expected as the HPA-5 catalyst has proven to convert all three components of lignocellulosic biomass (cellulose, lignin, and hemicellulose) under the applied reaction conditions. However, HPA-5 catalyst performance towards FA production would differ depending on the representative group (in order of activity: cellulose ≈ hemicellulose ≫ lignin) [27]. Thus, yield to FA and AA strongly depends on the substrate structure and the number of oxygen-functionalities in the carbon-network.
Remarkably, between the seven fully characterized substrates, the best recovered biomass for converting into formic acid and acetic acid was found to be coffee husks with 64% FA + AA-combined yield (sample 1), followed by cocoa husks 55% FA + AA-combined yield (Sample 2), and palm rachis and sugarcane bagasse with ≈ 50% FA + AA-combined yield (Samples 3 and 6, respectively). Interestingly, for future applications, in all biomasses, selectivity towards FA is favorable, being almost twice as high as for AA (see Table 3). These values, as much as the authors know, are the highest obtained by the oxidative conversion of residual biomass. Note that all these latter four represent a high percentage of the total residual biomass produced in developing countries, making this technology an interesting opportunity for supporting the food and agro-industry.
Superior coffee and cocoa husk conversion into FA and AA can be correlated to their high amount of cellulose in the structure (≈ 40%, see example for coffee husks in Fig. 2a). On the other hand, the huge content of hemicellulose is responsible for positive, but lower results, on palm rachis and sugarcane bagasse wastes (≈ 58%, see example for sugarcane bagasse Fig. 2b). Finally, as mentioned above, the HPA-5 catalyst is less able to convert lignin during the oxidative process; therefore, lower conversions using palm nuts and palm fibers can be explained. If we compare them, as show in Table 2, a major production of CO and CO 2 in substrates with higher lignin content is observable. Thus, certainly, most lignin is transformed into CO and CO 2 (see example for palm nut in Fig. 2c; for other biomass compositions, please refer to Fig. S4 in SI). These results, are in accordance with former studies developed under biogenic substrates [26]. For rice husks, a proper correlation between its poor conversion (53%) and biomass composition could not be found. However, after reaction, a solid leftover, containing mainly ashes, was observed. This is in accordance with the high ash content (≈ 20%) compared to other biomasses (≈ < 5%) found in rice husks after characterization (see values in Table S2 in SI). It is well known that ashes contain a number of alkali and alkaline metals like Ca, Na, K, and Mg, but also Si and P, and transition metals such as Fe, Al, and Mn, which can affect positively or negatively the reaction [38,39]. Herewith, the high ash content has a negative effect on the conversion of the substrate (see Table 2, the higher the ash content, the lower is the conversion resp. total carbon yield). However, product selectivity (see Table 3) is not affected. Even though CO 2 is formed as main by-product in the gas-phase, the presented valorization method will certainly add on to the CO 2 abatement in the atmosphere compared to common FA production methods like the hydrolysis of methyl formate. As CO 2 contained in FA and CO 2 by-product is trapped from the atmosphere while plant growing. In fact, the gaseous carbon dioxide can be separated and used for other biomass productions, such as large-scale micro algae production [40].

Conclusion
This study gives good arguments for the scientific and economic usefulness of the OxFA process for converting recovered biomass from agro-industry into valuable chemicals. It was possible to expand the biomass scope for the reaction significantly, by transforming many complex water-insoluble feedstocks, which are highly attractive considering other geographies and countries with similar agro-industrial sectors in Latin America, Asia, and Africa.
Seven fully characterized residual biomasses recovered from agro-industrial wastes were successfully valorized for the production of important commodities, for present and future applications, such as formic acid, acetic acid, and CO 2 in high yields. The developed reaction works with an affordable POM catalyst (HPA-5), TSA as additive, and oxygen at relatively mild reaction conditions (90 °C, 20 bar O 2 , and 24 h) in aqueous phase. Remarkably, coffee husk and cocoa husk residues showed quite promising results with high yields towards formic and acetic acid up to 64 and 57%, respectively. With favorable selectivity towards FA formation (coffee husks: FA/AA selectivity [%]: 43/22), being almost twice as high as for AA. Higher amounts of cellulose and hemicellulose contents in the biomass structure could be successfully correlated to higher yields towards FA and AA during the oxidative process (coffee husks > cocoa husks > palm rachis ≈ sugarcane bagasse), while substrates with higher lignin concentrations showed the lowest FA yields (e.g. Palm fiber and nut). This complete screening of several representative residual biomasses obtained from food providing agro-industry opens new interesting perspectives towards biomass valorization. This will promote the use of the "OxFA process" in new hydrogen production routes. However, there are still some technical drawbacks that must be solved, e.g., the efficient catalyst recirculation for reducing production costs. Moreover, the results also suggest a need for further investigations concerning the application of different types of polyoxometalate catalysts, for not only producing FA, but also high-purity cellulose that can be used for bioethanol production, for instance.