Improving the Fractionated Catalytic Oxidation of Lignocellulosic

Research Article Cite This: ACS Sustainable Chem. Eng. 2019, 7, 9754−9762

pubs.acs.org/journal/ascecg

Improving the Fractionated Catalytic Oxidation of Lignocellulosic Biomass to Formic Acid and Cellulose by Using Design of Experiments Dorothea Voß, Heike Pickel, and Jakob Albert* Lehrstuhl für Chemische Reaktionstechnik, Friedrich-Alexander-Universität Erlangen-Nürnberg, Egerlandstraße 3, 91058 Erlangen, Germany Downloaded via GUILFORD COLG on July 26, 2019 at 09:37:24 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: The main objective of this contribution was to optimize the process conditions for a fractionated oxidation of lignocellulosic biomass (e.g., wood) to produce formic acid and cellulose for further applications. Using the Lindqvist-type polyoxometalate K5V3W3O19 as a homogeneous catalyst in aqueous media, we were able to selectively oxidize the hemicellulose and lignin fractions. While hemicellulose and lignin are converted to formic acid, the cellulose fraction remains untapped. Furthermore, we optimized the process conditions with the help of a Box−Behnken design of experiments. Thereby improved conditions showed an output of formic acid from lignin three times higher compared with the initial parameters. Furthermore, recycling of the catalyst has been carried out successfully without observing significant changes in vanadium species. Finally, we investigated the conversion of different real biomasses (hardwood, softwood, and algae) leading to new structure−performance relationships. KEYWORDS: Biomass fractionization, Selective catalytic oxidation, Formic acid, Lignocellulose, Design of experiments

■

INTRODUCTION Using fossil resources as a starting point for the production of fuels and platform chemicals is not very sustainable due to two reasons. First, the limited availability of fossil resources could become problematic, and second, there is a potential climate changing effect of burning fuels and at the end of the life cycle of the platform chemicals, too.1 As sustainability is a key market desire, biomass is gaining economic significance for the production of platform chemicals.2 The industry supplying this market, which has so far been based on fossil resources, is now changing toward a biobased economy.3 Biomass is defined as material of organic origin containing carbon. In order to avoid any dinner plate or fuel tank discussion about using food crops for producing chemicals, lignocellulose seems to be a promising starting material for the production of sustainable platform chemicals.4 The main part of plant material is lignocellulose, which consists mainly of cellulose, hemicellulose, and lignin.5,6 Starting from this raw material, a selective depolymerization of the biopolymers followed by further chemical transformation of the mono- or oligomers has to be carried out.7 Our suggested way to proceed is the selective catalytic oxidation of biomass. Hereby, the first step is an acid-catalyzed depolymerization followed by an oxidative cleavage of the carbon−carbon bonds in the biomolecule as a second step.8 In this way, we are able to convert biomass selectively into formic acid (FA) and CO2 by a catalytic partial oxidation, the so-called “OxFA process”.9,10 This slightly exothermic oxidation process operates at temperatures below 373 K and uses molecular oxygen or © 2019 American Chemical Society

synthetic air as oxidizing agent and polyoxometalates as homogeneous catalysts in aqueous solution.10 Applying these reaction conditions, it is of critical importance to avoid further oxidation of FA to CO2.11 As FA is the only detected product in the aqueous phase, we do not need any additional separation process. CO2 is the only side product at full conversion present in the gas phase. The interesting difference of the OxFA route compared to reductive biomass valorization approaches is that all thermally induced formation of solid or gluey byproducts is completely avoided.11,16 Moreover, applying the OxFA process allows for conversion of a wide range of biomasses into only two C1 products, FA and CO2. Furthermore, these two products are separated in the gas and in the liquid phase during the process. Therefore, the OxFA process is simpler and more robust compared to other biomass valorization technologies.10 Recent works on biomass oxidation focus as well on the formation of formic acid from mechanically activated microcrystalline cellulose in the presence of polyoxometalate catalysts.26 Polyoxometalates (POMs) are complexes of oxygen and light transition metals like vanadium, molybdenum, or tungsten at their highest oxidation state. POMs have a wide structural and chemical variability leading to different catalytic properties.12−14 Recent studies on polyoxometalate catalysts enabled a system for the selective conversion of lignin and Received: October 8, 2018 Revised: April 15, 2019 Published: May 9, 2019 9754

DOI: 10.1021/acssuschemeng.8b05095 ACS Sustainable Chem. Eng. 2019, 7, 9754−9762

Research Article

ACS Sustainable Chemistry & Engineering Scheme 1. Main Reaction Steps for POM-Catalyzed Xylose Oxidation in Aqueous Solution

hemicellulose without converting the high value cellulose.15,16 Using this catalyst system, it was possible for the first time to produce FA and high-grade cellulose in only one reactor setup, avoiding any pretreatment or further downstream processing of the biomass.17 In a previous study,17 we found out that increasing the degree of vanadium substitution within the Lindqvist POM structure K2+x[VxW6−xO19] also increases the oxidation activity. While compounds with zero to two vanadium substitutions showed only very little catalytic activity, the three-vanadium-substituted structure K5[V3W3O19] combined high activity with the goal of a fractionated biomass conversion. While the cellulose model compounds glucose and cellobiose were untapped, all tested model compounds for both hemicellulose and lignin showed selective oxidation to FA and CO2. Obviously, the active sites in the Lindqvist-POM structure are not strong enough to oxidize the linear β-1,4glycosidic bonds in the cellobiose structure. However, C6 sugars like glucose are known to have an inhibiting effect on the active species.15 Moreover, these results could be confirmed using the main components cellulose, hemicellulose, and lignin. This is due to the crystalline nature of the cellulose where the functionalized carbon sites are not accessible for the catalyst due to the highly interlinked molecular strains stabilized by hydrogen bonds.25 However, the conversion rates were very still far too low for an industrial application and therefore have to be improved. Details about the catalytic reaction pathways are summarized elsewhere.11 In this contribution, we investigated the applicability of the Lindqvist-type polyoxometalate catalyst K5V3W3O19 for fractionated conversion of lignocellulosic biomass. Moreover, we studied the transformation of the main components of lignocellulosic biomass and the influence of an upscaling of the process using more efficient gas entrainment. The reaction conditions were optimized for xylose and lignin by means of a Box−Behnken design of experiments.18,19 Moreover, we investigated the recycling of the applied catalyst and studied structural changes in the catalyst species by applying 51V NMR spectroscopy. Finally, the conversion of three different types of real biomass using our improved reaction system has been studied.

The catalyst K5[V3W3O19] (IPA-3) selectively converts model compounds for hemicellulose (xylose, arabinose, mannose, and galactose), as well as model compounds for lignin (phenol, 4-hydroxy-3-methylbenzoic acid (4H3MB)), to FA. Model substrates for cellulose (glucose and cellobiose) were not converted using this catalyst under mild oxidation conditions.17 In this study, we started with some control experiments using different analytical tools than in the previous study having higher resolution (IC) allowing us also to detect traces of potential side products. Therefore, we used watersoluble, well-defined model substrates, namely, cellobiose, xylose, and 4H3MB in order to confirm results of the previous studies and to set a benchmark for the applied design of experiments (DoE) study. The experiments shown in Table 1 Table 1. Transformation of Model Substrates for All Main Components of Lignocellulosic Biomass Using Lindqvisttype POM K5V3W3O19 entry

substrate

molecular compositionb

1 2 3

cellobiose xylose 4H3MBd

C12H22O11 C5H10O5 C8H8O4

a

FA yieldc (%)

AA yieldc (%)

total carbon yieldc (%)

0.0 10.0 1.4

0.0 5.6 0.5

1.3 23.5 3.1

a

Reaction conditions: 10-fold, 2 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed, 6 h reaction time. bDetermined via C,H,N,S elemental analysis. cYields determined as described in “Determination of Yield and Selectivity”. d4-Hydroxy-3-methylbenzoic acid.

were performed at 90 °C, 30 bar oxygen pressure, and 1100 rpm stirrer speed using 2 mmol of substrate and 0.1 mmol of IPA-3 catalyst in 10 g of water as the solvent for 6 h reaction time. The analysis of liquid products with IC indicates the formation of FA and AA using xylose and 4H3MB as substrate (Table 1, entries 2 and 3). Using 4H3MB as a substrate leads also to the formation of oxalic acid (OA). The blank experiments without any catalyst under the same reaction conditions resulted in only formation of small amounts of CO2 due to the thermally induced total oxidation of the carbon framework. Experiments using the Lindqvist-type POM catalyst K4[V2W4O19] (IPA-2) showed the same qualitative results, as the model substrates xylose and 4H3MB were also converted but in lower yields, whereas cellobiose was untapped again. The IPA-3 catalyst showed the desired behavior for the model substrates and consequently was tested in further experiments with the main components of lignocellulosic biomass (Table 2). Herein, we wanted to investigate the

■

RESULTS AND DISCUSSION Fractionated Transformation of Lignocellulosic Biomass. Previous findings using Lindqvist-type polyoxometalates showed promising results for the fractionated biomass conversion in aqueous media.17,20 Scheme 1 shows the main reaction pathways exemplarily for xylose as a substrate based on former work and literature data.10,27 9755


Research Article

ACS Sustainable Chemistry & Engineering

the amount of substrate between 2 and 8 mmol, respectively. These three factors were optimized regarding FA yield, FA selectivity, and total carbon yield. Figure 1 shows the proclaimed experiments for the optimization as points on the center of the cube edges.

Table 2. Transformation of All Main Components of Lignocellulosic Biomass Using Lindqvist-type POM K5V3W3O19 entry

substrate

molecular compositionc

1a 2b 3a 4b 5a 6b

Cellulose

C1.07H1.88O1

Xylan

C0.99H1.76O1

Lignin

C2.61H3.01O1

FA yieldd (%)

AA yieldd (%)

total carbon yieldd (%)

0.0 0.0 17.7 23.7 1.7 6.6

0.0 0.0 3.7 3.8 0.8 2.1

0.2 0.5 65.7 82.6 23.9 41.5

a

Reaction conditions: 10-fold, 10 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed, 24 h reaction time. bReaction conditions: 600 mL batch, 100 mmol of substrate, 1 mmol of K5V3W3O19 catalyst, 100 g of water as solvent, 90 °C, 30 bar oxygen pressure, 1000 rpm stirrer speed, 24 h reaction time. cDetermined via C,H,N,S elemental analysis. dYields determined as described in “Determination of Yield and Selectivity”.

influence of different reaction scales as well as different gas entrainment into the liquid reaction phase. First, the experiments were performed as described above with 10 mmol of substrate and 0.1 mmol of IPA-3 catalyst in 10 g of water as solvent at 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed, and a reaction time of 24 h (Table 2, entries 1, 3, and 5) in the 20 mL 10-fold screening apparatus. Second, the process was scaled up ten times (Table 2, entries 2, 4, and 6) in the 600 mL batch autoclave. For this purpose, the oxidation experiments were performed with 100 mmol of substrate and 1 mmol of IPA-3 catalyst in 100 g of water as a solvent at 90 °C, 30 bar oxygen pressure, 1000 rpm stirrer speed, and a reaction time of 24 h. As expected, the real lignocellulosic compounds hemicellulose (xylan) and lignin were oxidized under these conditions, and FA and acetic acid (AA) were formed as liquid phase reaction products in quantitative yields. Using lignin as a substrate leads again to the formation of small amounts of OA (yields up to 0.5%). Cellulose was not converted within the investigated reaction time. Surprisingly, the yields of FA and AA using the main components were higher than those using the model substrates. This is well in line with previous findings reported elsewhere, indicating that the linkage of the different main components inside the lignocellulosic matrix influences their catalytic behavior. Consequently, all subsequent experiments were done directly with lignin instead of using the model substrate 4H3MB. After scaling up the process by the factor of 10, the yields of the produced acids increased (Table 2, entries 2, 4, and 6). This is caused by the more effective gas entrainment into the system leading to lower influence of mass transfer limitations. Sensitivity Analysis and Parameter Optimization Using Box−Behnken Design of Experiments. A detailed sensitivity analysis was done using the statistical method of Box and Behnken for the design of experiments.18 The experiments were performed in a 10-fold screening plant using xylose and lignin as the substrates. The influence of the three parameters reaction temperature, oxygen partial pressure, and the substrate/catalyst ratio were investigated. The temperature was varied between 90 and 140 °C. An oxygen partial pressure range between 10 and 50 bar was investigated, and the ratio between substrate and catalyst loading was varied by changing

Figure 1. Experimental points using the Box−Behnken design of experiments.

The experiment at 115 °C, 30 bar oxygen partial pressure, and substrate/catalyst ratio of 50 (orange point in Figure 1) was executed three times to determine the quality of the experiments. Table 3 shows a list of all experiments for the sensitivity analysis. Entries 1−3 represent the center point experiments. The parameter optimization experiments were done for xylose (as model substrate for xylan) and for lignin separately. These experiments always used an amount of 0.1 mmol of IPA3 catalyst, 10 g of water as a solvent, a stirrer speed of 1100 rpm, and a reaction time of 6 h. Table 3. Experiments of the Sensitivity Analysis Using Box− Behnken Design of Experimentsa entry

temperature (°C)

pressure (bar)

substrate/catalyst ratio

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

115 115 115 115 115 115 115 90 90 90 90 140 140 140 140

30 30 30 10 50 10 50 10 50 30 30 10 50 30 30

50 50 50 20 20 80 80 50 50 20 80 50 50 20 80

a

Reaction conditions: 10-fold, 2−8 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 90−140 °C, 10−50 bar oxygen pressure, 1100 rpm stirrer speed, 6 h reaction time. 9756


Research Article


Figure 2. Effect of the interaction of temperature and pressure on the yield and selectivity of FA and the overall yield.

The Design-Expert Software Version 10 of Stat-Ease was used to evaluate the results. From the center experiments (Table 3, entries 1−3), first the variance and standard deviation were calculated in order to assess whether the differences in the results from the experiments 4−15 (see Table 3) are significant. To validate the results, a variance analysis was done and the software fitted the results of the experiments with a quadratic term of the form

other hand, we focused on the highest yield of FA. Therefore, for the following experiments with all main components of lignocellulosic biomass, the conditions of 115 °C reaction temperature, 50 bar oxygen partial pressure, and a substrate/ catalyst ratio of 40 were chosen. The experiments were performed in the 10-fold reactor setup with an amount of 0.1 mmol of IPA-3 catalyst, 10 g of water as a solvent, a stirrer speed of 1100 rpm, and a reaction time of 24 h. The received yields are summarized in Table 4.

y = a + bx1 + cx 2 + dx3 + ex1,2 + fx1,3 + gx 2,3 + hx12 + ix 2 2 + jx32

Table 4. Transformation of All Main Components of Lignocellulosic Biomass Using Lindqvist-type POM K5V3W3O19 at Optimized Conditions

(1)

with the constants a−j, the varied factors x1−x3, and the 2factor interactions x1,2, x1,3, and x2,3. From the statistically determined figures for temperature, pressure, and substrate/ catalyst ratio, the optima regarding FA yield, FA selectivity, and the total carbon yield were calculated. For xylose, the optimal parameters were determined to 115 °C reaction temperature, 50 bar oxygen partial pressure, and a substrate/catalyst ratio of 40 (see Figure 2). These optimized parameters represent a compromise of the three target values (FA yield, selectivity, and total yield). In this compromise, the three target values were weighted differently. The yield of FA was considered to be the most important target. This was followed by the total carbon yield, while the FA selectivity played only a minor role. As an additional condition, a complete reoxidation of the catalyst was required, which could be visually assessed by the color of the catalyst solution. This experiment resulted in a FA yield of 40.0% with a selectivity of 49.3% and a total carbon yield of 81.1%. The results of the DoE experiments listed in Table 3 with xylose as substrate are summarized in Table S1 in the Supporting Information. For lignin, the optimal parameters were determined with the same weighting of the three target values to be 140 °C reaction temperature, 50 bar oxygen partial pressure, and a substrate/ catalyst ratio of 40 (see Figure S3 in the Supporting Information). This results in significantly lower yields of FA (5.7%), AA (4.2%), and CO2 (47.1%). Nevertheless, a definite maximum at maximal temperature and pressure could be found for lignin. All results of the DoE experiments listed in Table 3 with lignin as a substrate are summarized in Table S2 in the Supporting Information. Application of the Optimized Conditions to the Main Components of Lignocellulosic Biomass. Because real lignocellulosic biomass contains a mixture of cellulose, hemicellulose, and lignin, a compromise for the optimized parameters had to be found. On the one hand, a complete reoxidation of the used catalyst had to be ensured. On the

entrya

substrate

FA yieldb (%)

AA yieldb (%)

total carbon yieldb (%)

1 2 3

cellulose xylan lignin

0.0 24.7 5.6

0.0 2.0 3.8

2.8 88.3 57.0

a

Reaction conditions: 10-fold, 4 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 24 h reaction time. bYields determined as described in “Determination of Yield and Selectivity”.

The experiments using the main components of lignocellulosic biomass once more confirmed our expectations. Cellulose was not converted within the investigated reaction time under optimized conditions, while xylan and lignin were transformed to FA and AA (Table 4). In comparison to the experiments under nonoptimized conditions (Table 2, entries 1, 3, and 5), the yield of FA using lignin as substrate could be increased by a factor of 3 from 1.7% to 5.6%. This is also visualized in Figure 3. The yield of AA was more than quadruplicated due to the optimization. This shows that optimizing the process conditions accelerates degradation pathways both to FA and to AA. Nevertheless, the improvement of FA yield using xylan as substrate is lower; the yield of AA decreased by optimizing the process. This shows a preference of the catalytic degradation pathway and leads into a higher selectivity regarding FA. Recycling of the IPA-3 Catalyst over Four Batches. To investigate the catalyst recyclability, the liquid phase of an experiment with xylose as a substrate was recycled three times. First, 2 mmol of xylose was converted at 115 °C under 50 bar oxygen pressure for 6 h reaction time. The experiment was performed in the 10-fold reactor setup with an amount of 0.1 mmol of IPA-3 catalyst, 10 g of water as a solvent, and a stirrer speed of 1100 rpm. The substrate/catalyst ratio was chosen as 9757


Research Article


Figure 3. Comparison of experiments under nonoptimized (10-fold, 10 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed, 24 h reaction time) and optimized reaction conditions (10-fold, 4 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 24 h reaction time).

low to ensure a high conversion of the substrate so that an accumulation of xylose could be avoided (Table 5, entry 1).

After the reaction, a sample for NMR and IC analysis was taken and the gas phase was analyzed by GC. For the first recycling experiment (Table 5, entry 2), 2 mmol of xylose was added to the reaction mixture of the first experiment. Furthermore, the amount of liquid samples taken was replaced by water. This recycling experiment was repeated twice (Table 5, entries 3 and 4). The chosen reaction conditions led to a complete reoxidation of the catalyst in all recycling experiments, which was visible by a yellow/orange solution after reaction and further confirmed by 51V NMR spectroscopy. Exclusively FA and CO2 could be detected within the detection limits of the applied analysis. The product FA was not removed after each recycling step so that it accumulated during the experiments. The calculated FA yields in Table 5 include only the newly formed FA in each experiment considering the nonconverted

Table 5. Recycling of the Lindqvist-type POM K5V3W3O19 at Optimized Conditions over Four Batches entrya

FA yieldb (%)


FA selectivityb (%)

1 2 3 4

39.1 34.3 34.8 35.5

80.1 74.1 74.9 72.4

48.9 46.3 46.5 49.0

a

Reaction conditions: 10-fold, 2 mmol of xylose, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 6 h reaction time. bYields determined as described in “Determination of Yield and Selectivity”.

Figure 4. 51V NMR spectra of the liquid phase before and after reaction as well as after the three recycling experiments (conditions: 10-fold, 2 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 6 h reaction time). 9758


Research Article


Figure 5. 51V NMR spectra of the liquid phase after reaction with beech, spruce, and chlorella as a substrate compared to the reference catalyst before reaction (conditions: 10-fold, 4 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 72 h reaction time). 51

and newly added amount of xylose. However, no significant decrease in the catalyst activity was observed during recycling. The FA yield and selectivity remain almost constant over all four batches. 51 V NMR spectra of the liquid phase before and after the reaction as well as after the three recycles are shown in Figure 4. Before reaction, the significant peaks for [V2W4O19]4− (−511.3 ppm and −518.4 ppm) and [V3W3O19]5− (−498.9 ppm and −505.2 ppm) as well as the presence of different VO3− species (around −575 ppm) are visible.21,22 After reaction, the peaks are shifted to smaller chemical shifts due to the decrease in pH of the reaction mixtures. In addition, no peaks of VO3− clusters are visible. The ratios of the different catalyst isomers are changed by the reaction as well. Within each recycling step, the decreased pH leads to an increased shift. The new peaks between −530 and −570 ppm after reaction and the recycling experiments indicate a change of the catalyst structure. However, this change has no effect on the catalytic activity. Application of the Optimized Conditions to Real Lignocellulosic Biomass. To round out our study, we investigated the conversion of different real biomasses. Beech wood was used to represent hardwoods, spruce for softwoods, and chlorella (a green algae) for a third generation biomass containing a high nitrogen amount. The experiments were performed as described above at conditions of 115 °C reaction temperature, 50 bar oxygen partial pressure, and substrate/ catalyst ratio of 40. The 10-fold reactor setup was used with an amount of 0.1 mmol of IPA-3 catalyst and 10 g of water as solvent in each reactor. The stirrer speed was set to 1100 rpm, and a reaction time of 72 h was chosen. Beech and spruce wood were purchased from JBACH GmbH. Elementary (C, H, N and S) analysis gave a molecular composition of C1.23H1.85O1 for beech and C1.30H1.98O1 for spruce wood. Chlorella is commercially produced by Algomed. The molecular composition was determined to C7.03H12.18O4.17N1.

V NMR spectra of the liquid phase after each experiment are shown in Figure 5. All reaction solutions showed a yellow to orange color after the experiments, which indicated a complete reoxidation of the catalyst. 51V NMR spectra show only slight changes in the catalyst structure after the experiments with beech and spruce as substrates. The spectra are comparable to the reference before reaction with only a shift due to a pH change. The more intense shift after the experiment with beech indicates that more acids were produced in this reaction compared to the experiment with spruce as a substrate. After oxidation of chlorella, the catalyst shows a greatly changed structure. Although the characteristic peaks of [V2W4O19]4− and [V3W3O19]5− can be recognized, a major amount of VO3− species (between −550 and −560 ppm) has been formed. As expected, all real lignocellulosic biomass substrates were oxidized, and FA and AA were formed (see Table 6). Using beech as a substrate leads to FA yield of 12.5% and selectivity regarding FA of 25.1% (Table 6, entry 1). Spruce wood was converted mainly into CO2 due to total oxidation (Table 6, entry 2). This is in contrast to former observations using different kinds of wood and is subject of further studies. Using Table 6. Catalytic Conversion of Real Lignocellulosic Biomass Using Lindqvist-type POM K5V3W3O19 at Optimized Conditions entry

substrate

FA yieldb (%)

1 2 3

beech spruce chlorella

12.5 1.6 0.2

a

AA yieldb (%)


FA selectivityb (%)

7.7 6.2 18.7

49.9 55.1 68.5

25.1 2.8 0.3

a

Reaction conditions: 10-fold, 4 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water as solvent, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 72 h reaction time. bYields determined as described in “Determination of Yield and Selectivity”. 9759


Research Article


Figure 6. FTIR spectra of the substrates beech, spruce, and chlorella before and after reaction (reaction conditions: 10-fold, 4 mmol of substrate, 0.1 mmol of K5V3W3O19 catalyst, 10 g of water, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 72 h reaction time).

biomass, the FA yield and selectivity were satisfactory. Finally chlorella, representing a third generation biomass as substrate, processed under the same conditions gave noticeable higher AA yields. We hope that our investigations help to further push this technology forward to industrial implementation and shed more light into selective biomass conversion processes using polyoxometalate catalysts.

chlorella as a substrate leads predominantly to the formation of AA (Table 6, entry 3). Additionally, small amounts of OA (yield of 1.5%) as well as traces of other acids and CO (yield of 0.9%) could be detected. Since the algae has been shown to contain nitrogen (N) and also traces of the element sulfur (S), compounds with these elements, such as sulfuric acid or nitric acid, were certainly also formed. The orange color of the solid obtained after the reaction lead to the assumption that chlorophyll was also converted during the reaction. To determine if the cellulose part of the real lignocellulosic biomass substrates was converted or structurally changed during the reaction, the solid reaction residues were dried and analyzed by FTIR (Figure 6). The recorded spectra for beech, spruce, and chlorella before reaction are comparable to those in the literature.23,24 To evaluate the results, we focused on the characteristic bands for xylose at 1740 cm−1 (representing the hemicellulose part) and for lignin at 1510 cm−1. These characteristic bands disappeared almost completely after reaction with beech and spruce. This leads to the assumption that the hemicellulose and lignin fraction decreased during the conversion. In case of chlorella as a substrate, the spectrum changed almost completely. The characteristic band for lignin cannot be clearly identified as it overlays with additional N−H vibrations. The absorption peak at 1740 cm−1 is not visible anymore after reaction. Therefore numerous new intensive maxima were formed. Apparently, these bands are caused by intermediates of the oxidative degradation of chlorella.

■

EXPERIMENTAL DETAILS

Chemicals. All chemicals were obtained commercially and used as received without further purification. The Lindqvist-type polyoxometalate catalyst K5V3W3O19 was synthesized according to the literature procedure.20 Oxygen (4.5 GA 201) was bought from Linde AG and used as the oxidizing agent. Demineralized water was used as a solvent. Cellulose was purchased from JBACH GmbH. Elementary (C, H, N, S) analysis gave a molecular composition of C1.07H1.88O1. Hemicellulose (xylan from beech wood) is commercially available from Carl Roth GmbH. The molecular composition was determined to C0.99H1.76O1. Lignin was purchased from Fraunhofer ICT Pfinztal. The molecular composition via elemental analysis gave C2.61H3.01O1. Experimental Setup. The oxidation reactions were carried out in a 10-fold screening plant with a batch mode reactor setup. It consists of ten 20 mL autoclaves made of Hastelloy C276 (10-fold). All pipes, valves, and fittings were made of stainless steel 1.4571. The gaskets used were made of Teflon. The autoclaves were connected in parallel to a single oxygen supply line via individual couplings and placed inside a heating plate in order to adjust the required temperature. The heating plate was equipped with a magnetic stirrer whereby magnetic stir bars could be used for stirring. Additionally, each reactor was connected to a rupture disk with a burst pressure maximum of 90 bar. The scale-up experiments with all main components of lignocellulosic biomass were carried out in a 600 mL Hastelloy C276 fed-batch autoclave equipped with a gas entrainment impeller. A flat Teflon sealing located at the reactor head was used. All pipes, valves and fittings were made of stainless steel 1.4571. The reactor was equipped with a cooling coil made of Hastelloy C276 and a heating jacket to control the reactor temperature. The oxygen was fed into the system by a mass flow controller to keep up the required pressure. A pressure control valve was installed to prevent higher pressures than 65 bar. Additionally, the reactor was connected to a rupture disk with a burst pressure maximum of 100 bar. Typical Workup Procedure. For the oxidation reactions, each autoclave was filled with substrate, catalyst (0.1 mol L−1), and water (10 or 100 g) as the solvent. The substrate concentration varied between 0.2 and 1 mol L−1. The system was purged with 10 bar oxygen pressure to remove the residual air out of the reactors. In the following, all reactors were prepressurized with about 25 bar (40 bar) oxygen pressure, the stirrer was set to 300 rpm, and the heating was switched on. When the desired temperature (90 or 115 °C) was reached, the oxygen pressure was increased to the required pressure (30 or 50 bar) and the stirring speed was set to 1000 rpm in order to

■

CONCLUSION The main objective of this project was to optimize the process conditions for the fractionated oxidation of lignocellulosic biomass (e.g., wood) to produce formic acid and high-grade cellulose for further processing in one single reaction step. Experiments with model substrates as well as with the main components of lignocellulosic biomass using the Lindqvisttype polyoxometalate K5V3W3O19 showed the desired results that the cellulose fraction is not converted. Using the Box− Behnken design of experiments with a number of 15 experiments, we obtained optimized reaction conditions for xylan as well as for lignin. The optimized conditions led to an output of formic acid (FA) from lignin three times higher compared with the initial conditions. Recycling of the catalyst could be demonstrated for four batches without losing performance, as well as significant changes in catalyst structure indicated by 51 V NMR spectroscopy. The optimized conditions were further applied to the conversion of real biomass. When using hardwood (beech) as an example for real 9760


Research Article

ACS Sustainable Chemistry & Engineering start the gas entrainment. This moment was set as starting time of the experiment. Analytics. The Lindqvist-type POM catalyst K5[V3W3O19] was characterized with ICP-OES measurements performed on a PerkinElmer Plasma 400 resulting in a v/w ratio of 3/3.12 (detection limit of the applied method is 1 ppm). The 51V NMR spectra showed the same chemical shifts according to Andersson et al. (see Supporting Information, Figure S1).21 Moreover, FTIR spectra confirmed the presence of Lindqvist species (see Supporting Information, Figure S2).20 NMR spectra were recorded on a JEOL ECX-400 MHz (detection limit of the applied method is 500 ppm). Liquid products were analyzed by means of IC analytics on an ICS3000 equipped with a Dionex IonPac AS11-HC-2 μm (250 mm × 2 mm) column (detection limit of the applied method is 1 ppm). Diluted sulfuric acid was used as an eluent. Moreover, HPLC measurements were performed using a HPLC from Jasco equipped with a Rezex RHM-Monosaccharide H+ (8%) column from Phenomenex Ltd. Gaseous product analysis was performed using a Varian GC 450 equipped with a 2 m × 0.75 mm ID ShinCarbon ST column (detection limit of the applied method is 100 ppm). Elementary (C, H, N, S) analysis of the complex biogenic substrates was carried out using a Euro Vector EA 3000. Determination of Yield and Selectivity. After all oxidation reactions, the detected carboxylic acids (namely, FA, acetic acid (AA), and oxalic acid (OA)) in the liquid phase were quantitatively determined by IC. Moreover, the gaseous products CO2 and CO were quantified with a TCD-GC-detector. The yields of FA, AA, and OA were determined by means of IC measurements and calculated as n(acid)/n(C atoms substrate). The yields of CO2 and CO were determined by means of GC analysis and calculated as n(CO2)/n(C atoms substrate) and n(CO)/n(C atoms substrate), respectively. The total carbon yield (FA + AA + OA + CO2 + CO) was calculated by (n(FA) + n(AA) + n(OA) + n(CO2) + n(CO))/n(C atoms substrate) and serves as a representative for the substrate conversion. Moreover, HPLC measurements of the reaction solution confirmed qualitatively the suggested reaction pathways in Scheme 1 and can be found in the Supporting Information, Figure S4. The selectivity of FA was calculated by the product yield divided by the total carbon yield.

■

fully acknowledged. We thank Julian Mehler for supporting the catalyst synthesis, Nicola Taccardi for performing ICP-OES, and Benjamin Bertleff for performing IC measurements.

■

(1) Cherubini, F. The biorefinery concept: Using biomass instead of oil for producing energy and chemicals. Energy Convers. Manage. 2010, 51, 1412−1421. (2) Covestro AG. https://www.covestro.com/en/sustainability/ service-downloads/policies-commitments [2018-08-22]. (3) Shafiee, S.; Topal, E. When will fossil fuel reserves be diminished? Energy Policy 2009, 37, 181−189. (4) Palkovits, R. Cellulose und heterogene Katalyse − Eine Kombination mit Zukunft. Chem. Ing. Tech. 2011, 83, 411−419. (5) Wyman, C. E.; Decker, S. R.; Himmel, M. E.; Brady, J. W.; Skopec, C. E.; Viikavi, L. Polysaccharides: Structural Diversity and Functional Versatility; CRC Press: New York, 2004. eBook ISBN: 9781-4200-3082-2. (6) Witzke, I. Alternative Kraftstoffe fü r die dieselmotorische Verbrennung aus kohlenhydrathaltigen Biomassen und basierend auf bio- und chemisch-katalytischen Herstellverfahren; Springer: Wiesbaden, 2017. DOI: DOI: 10.1007/978-3-658-17676-1. (7) Chheda, J. N.; Huber, G. W.; Dumesic, J. A. Liquid-Phase Catalytic Processing of Biomass-Derived Oxygenated Hydrocarbons to Fuels and Chemicals. Angew. Chem., Int. Ed. 2007, 46, 7164−7183. (8) Sutton, A. D.; Waldie, F. D.; Wu, R.; Schlaf, M.; Silks, L. A.; Gordon, J. C. The hydrodeoxygenation of bioderived furans into alkanes. Nat. Chem. 2013, 5, 428−432. (9) Wölfel, R.; Taccardi, N.; Bösmann, A.; Wasserscheid, P. Selective catalytic conversion of biobased carbohydrates to formic acid using molecular oxygen. Green Chem. 2011, 13, 2759−2763. (10) Albert, J. Der Erlanger OxFA-Prozess; SVH-Südwestdeutscher Verlag für Hochschulschriften: Erlangen, 2016. (11) Zhang, Z.; Huber, G. W. Catalytic oxidation of carbohydrates into organic acids and furan chemicals. Chem. Soc. Rev. 2018, 47, 1351−1390. (12) Hill, C. L.; Prosser-McCartha, C. M. Homogeneous catalysis by transition metal oxygen anion clusters. Coord. Chem. Rev. 1995, 143, 407−455. (13) Pope, M. T.; Müller, A. Polyoxometalate Chemistry: An Old Field with New Dimensions in Several Disciplines. Angew. Chem., Int. Ed. Engl. 1991, 30, 34−48. (14) Hill, C. L. Progress and challenges in polyoxometalate-based catalysis and catalytic materials chemistry. J. Mol. Catal. A: Chem. 2007, 262, 2−6. (15) Reichert, J.; Albert, J. Detailed Kinetic Investigations on the Selective Oxidation of Biomass to Formic Acid (OxFA Process) Using Model Substrates and Real Biomass. ACS Sustainable Chem. Eng. 2017, 5, 7383−7392. (16) Albert, J.; Wasserscheid, P. Expanding the scope of biogenic substrates for the selective production of formic acid from waterinsoluble and wet waste biomass. Green Chem. 2015, 17, 5164−5171. (17) Albert, J. Selective oxidation of lignocellulosic biomass to formic acid and high-grade cellulose using tailor-made polyoxometalate catalysts. Faraday Discuss. 2017, 202, 99−109. (18) Siebertz, K.; van Bebber, D.; Hochkirchen, T. Statistische Versuchsplanung: Design of Experiments; Springer Vieweg: Berlin, 2010. DOI: DOI: 10.1007/978-3-662-55743-3. (19) Kleppmann, W. Versuchsplanung: Produkte und Prozesse optimieren; Carl Hanser Verlag: München, 2016. eBook ISBN: 9783-446-44717-2. (20) Albert, J.; Mehler, J.; Tucher, J.; Kastner, K.; Streb, C. One-step Synthesizable Lindqvist-isopolyoxometalates as Promising New Catalysts for Selective Conversion of Glucose as a Model Substrate for Lignocellulosic Biomass to Formic Acid. Chemistry Select 2016, 1, 2889−2894.

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acssuschemeng.8b05095. 51 V NMR spectrum of Lindqvist-type POM catalyst K5[V3W3O19], FTIR spectra of the Lindqvist-type POM catalysts K5[V3W3O19] and K2[W6O19] as a reference, Design of experiment results for xylose and lignin, HPLC chromatogram of an exemplarily reaction solution, IC chromatogram of reaction solution from Table 1, entry 2 (PDF)

■

REFERENCES

AUTHOR INFORMATION

Corresponding Author

*Phone: (+49)-(0)9131-85-67417. Fax: (+49)-(0)9131-8527421. E-mail: [email protected]. ORCID

Jakob Albert: 0000-0002-3923-2269 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Financial support by the Max-Buchner-Stiftung (www. dechema.de/mbf.html) and the Cluster of Excellence “Engineering of Advanced Materials” (www.eam.fau.de) is grate9761


Research Article

ACS Sustainable Chemistry & Engineering (21) Andersson, I.; Hastings, J. J.; Howarth, O. W.; Pettersson, L. Aqueous tungstovanadate equilibria. J. Chem. Soc., Dalton Trans. 1996, 2705−2711. (22) O’Donnell, S. E.; Pope, M- T. Applications of Vanadium-51 and Phosphorus-31 Nuclear Magnetic Resonance Spectroscopy to the Study of Iso- and Hetero-polyvanadates. J. Chem. Soc., Dalton Trans. 1976, 2290−2297. (23) Böttcher, J. H. Quantitative Analyse von Holz und HolzKomponenten mittels FTIR-Spektroskopie unter Anwendung multivariater statistischer Verfahren; Dissertation, Universität Hamburg, Hamburg, 1993. (24) Wen, Y.; Chen, H.; Yuan, Y.; Xu, D.; Kang, X. Enantioselective ecotoxicity of the herbicide dichlorprop and complexes formed with chitosan in two fresh water green algae. J. Environ. Monit. 2011, 13, 879−885. (25) Albert, J.; Wölfel, R.; Bösmann, A.; Wasserscheid, P. Selective oxidation of complex, water-insoluble biomass to formic acid using additives as reaction accelerators. Energy Environ. Sci. 2012, 5 (7), 7956−7962. (26) Gromov, N. V.; Taran, O. P.; Delidovich, I. V.; Pestunov, A. V.; Rodikova, Y. A.; Yatsenko, D. A.; Zhizhina, E. G.; Parmon, V. N. Hydrolytic oxidation of cellulose to formic acid in the presence of Mo-V-P heteropoly acid catalysts. Catal. Today 2016, 278, 74−81. (27) Niu, M.; Hou, Y.; Ren, S.; Wang, W.; Zheng, Q.; Wu, W. The relationship between oxidation and hydrolysis in the conversion of cellulose in NaVO3−H2SO4 aqueous solution with O2. Green Chem. 2015, 17, 335−342.

9762


Improving the Fractionated Catalytic Oxidation of Lignocellulosic

Recommend Documents