Improving the Fractionated Catalytic Oxidation of Lignocellulosic

Subscriber access provided by Bibliothèque de l'Université Paris-Sud

Article

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 ACS Sustainable Chem. Eng., Just Accepted Manuscript • DOI: 10.1021/ acssuschemeng.8b05095 • Publication Date (Web): 09 May 2019 Downloaded from http://pubs.acs.org on May 10, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Improving the fractionated catalytic oxidation of lignocellulosic biomass to formic acid and cellulose by using design of experiments Dorothea Voßa, Heike Pickela and Jakob Alberta* Lehrstuhl für Chemische Reaktionstechnik, Friedrich-Alexander-Universität Erlangen-Nürnberg, Egerlandstraße 3, 91058 Erlangen, Germany * E-Mail: [email protected] Contact details of the corresponding author: Phone: (+49)-(0)9131-85-67417 Fax: (+49)-(0)9131-85-27421; e-mail: [email protected]

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 is so far based on fossil resources, is now changing towards 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 ACS Paragon Plus Environment

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

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 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 by-products 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 more simple and 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 hemicellulose without converting the high value cellulose. 15,16 Using this catalyst system, it was for the first time possible 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 threevanadium 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-1,4 glycosidic 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 crystallinic 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, ACS Paragon Plus Environment

Page 2 of 20

Page 3 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


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.

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

Scheme 1: Main reaction steps for POM-catalyzed xylose oxidation in aqueous solution.

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-3methylbenzoic 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 water-soluble, 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 were performed at 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed using 2 mmol substrate and 0.1 mmol IPA-3 catalyst in 10 g water as the solvent for 6 hours 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 2-3). 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 thermal 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.

ACS Paragon Plus Environment


Page 4 of 20

Table 1: Transformation of model substrates for all main components of lignocellulosic biomass using Lindqvist-type POM K5V3W3O19. Entrya

Substrate

Molecular compositionb

FA-Yieldc (%)

AA-Yieldc (%)

Total Carbon Yieldc (%)

1 2 3

Cellobiose Xylose 4H3MBd

C12H22O11 C5H10O5 C8H8O4

0.0 10.0 1.4

0.0 5.6 0.5

1.3 23.5 3.1

Reaction conditions: a) 10-fold, 2 mmol substrate, 0.1 mmol K5V3W3O19 catalyst, 10 g water as solvent, 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed, 6 hours reaction time, b) determined via C,H,N,S elemental analysis, c) Yields determined as described in the corresponding section of the experimental part, d) 4-Hydroxy-3-methylbenzoic acid.

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 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 substrate and 0.1 mmol IPA-3 catalyst in 10 g water as solvent at 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed and a reaction time of 24 hours (Table 2, entries 1, 3 and 5) in the 20 ml tenfold screening apparatus. Second, the process was scaled ten times up (Table 2, entries 2, 4 and 6) in the 600 ml batch autoclave. For this purpose, the oxidation experiments were performed with 100 mmol substrate and 1 mmol IPA-3 catalyst in 100 g water as a solvent at 90 °C, 30 bar oxygen pressure, 1000 rpm stirrer speed and a reaction time of 24 hours. As expected, the real lignocellulosic compounds hemicellulose (xylan) and lignin were oxidized under these conditions and FA as well as 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 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 upscaling the process by the factor of ten, 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.


Page 5 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 2: Transformation of all main components of lignocellulosic biomass using Lindqvist-type POM K5V3W3O19. Entry 1a 2b 3a 4b 5a 6b

Substrate

Molecular compositionc

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

Reaction conditions: a) 10-fold, 10 mmol substrate, 0.1 mmol K5V3W3O19 catalyst, 10 g water as solvent, 90 °C, 30 bar oxygen pressure, 1100 rpm stirrer speed, 24 hours reaction time, b) 600mL-batch, 100 mmol substrate, 1 mmol K5V3W3O19 catalyst, 100 g water as solvent, 90 °C, 30 bar oxygen pressure, 1000 rpm stirrer speed, 24 hours reaction time, c) determined via C,H,N,S elemental analysis, d) Yields determined as described in the corresponding section of the experimental part.

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

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



Page 6 of 20

The experiment at 115 °C, 30 bar oxygen partial pressure and a 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 represents the center point experiments. Table 3: Experiments of the sensitivity analysis using Box-Behnken design of experiments. 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

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

The parameter optimization experiments were done for xylose (as model substrate for xylan) and for lignin separately. It was always used an amount of 0.1 mmol IPA-3 catalyst, 10 g of water as a solvent, a stirrer speed of 1100 rpm and a reaction time of 6 hours. 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 to 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 𝑦 = 𝑎 + 𝑏 ∙ 𝑥1 +𝑐 ∙ 𝑥2 +𝑑 ∙ 𝑥3 +𝑒 ∙ 𝑥1,2 +𝑓 ∙ 𝑥1,3 +𝑔 ∙ 𝑥2,3 +ℎ ∙ 𝑥21 +𝑖 ∙ 𝑥22 +𝑗 ∙ 𝑥23

(1)

(with the constants a-j, the varied factors x1-x3 and the 2-factor 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 the 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 ACS Paragon Plus Environment

Page 7 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


FA selectivity played only a minor role. As an additional condition, a complete re-oxidation 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.

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

For lignin, the optimal parameters were determined with the same weighting of the three target values to 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 Due to the fact that 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 re-oxidation of the used catalyst had to be ensured. On the 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 tenfold reactor setup with an amount of 0.1 mmol IPA-3 catalyst, 10 g water as a solvent, a stirrer speed of 1100 rpm and a reaction time of 24 hours. The received yields are summarized in Table 4. Table 4: Transformation of all main components of lignocellulosic biomass using Lindqvist-type POM K5V3W3O19 at optimized conditions. 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

Reaction conditions: a) 10-fold, 4 mmol substrate, 0.1 mmol K5V3W3O19 catalyst, 10 g water as solvent, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 24 hours reaction time, b) Yields determined as described in the corresponding section of the experimental part.



The experiments using the main components of lignocellulosic biomass once more confirmed our expectations. Cellulose was not converted within the investigated reaction time, neither under optimized conditions, while xylan and lignin were transformed to FA and AA (Table 4). In comparison to the experiments under non-optimized conditions (Table 2, entries 1, 3 and 5), the yield of FA using lignin as substrate could be increased by the factor of three 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 both degradation pathways to FA as well as 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. 24,7

25 20 Yield /%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 20

Lignin (non optimized) Lignin (optimized) Xylan (non optimized) Xylan (optimized)

17.7

15 10 5.6

5

3.8 1.7

3.7 2,0

0.8

0 Formic acid

Acetic acid

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

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 were converted at 115 °C under 50 bar oxygen pressure for 6 hours reaction time. The experiment was performed in the tenfold reactor setup with an amount of 0.1 mmol IPA-3 catalyst, 10 g of water as a solvent and a stirrer speed of 1100 rpm. The substrate/catalyst ratio was chosen as low to ensure a high conversion of the substrate so that an accumulation of xylose should 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 were 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).


Page 9 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


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

Reaction conditions: a) 10-fold, 2 mmol xylose, 0.1 mmol K5V3W3O19 catalyst, 10 g water as solvent, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 6 hours reaction time, b) Yields determined as described in the corresponding section of the experimental part.

The chosen reaction conditions led to a complete re-oxidation 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 non-converted and new-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. 51V-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 VO3species (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 VO3clusters 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 ppm 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.



after 3rd recycle after 2nd recycle after 1st recycle

after reaction

before reaction

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 substrate, 0.1 mmol K5V3W3O19 catalyst, 10 g water, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 6 hours reaction time).

Application of the optimized conditions to real lignocellulosic biomass To round up 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 3rd 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 with a substrate/catalyst ratio of 40. The tenfold reactor setup was used with an amount of 0.1 mmol IPA-3 catalyst and 10 g water as solvent in each reactor. The stirrer speed was set to 1100 rpm and a reaction time of 72 hours 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. 51V-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 indicate 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 ppm and -560 ppm) has been formed.


Page 10 of 20

Page 11 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


after reaction with chlorella

after reaction with spruce

after reaction with beech

reference

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 substrate, 0.1 mmol K5V3W3O19 catalyst, 10 g water, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 72 hours reaction time).

As expected, all real lignocellulosic biomass substrates were oxidized and FA as well as AA were formed (see Table 6). Using beech as a substrate leads to a FA yield of 12.5 % and a 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 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 as well 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 as well converted during the reaction. Table 6: Catalytic conversion of real lignocellulosic biomass using Lindqvist-type POM K5V3W3O19 at optimized conditions. Entrya

Substrate

FA-Yieldb (%)

1

Beech Spruce

12.5 1.6

Chlorella

0.2

2 3

AA-Yieldb (%)


FA-Selectivityb (%)

7.7

49.9

25.1

6.2 18.7

55.1 68.5

2.8 0.3

Reaction conditions: a) 10-fold, 4 mmol substrate, 0.1 mmol K5V3W3O19 catalyst, 10 g water as solvent, 115 °C, 50 bar oxygen pressure, 1100 rpm stirrer speed, 72 hours reaction time, b) Yields determined as described in the corresponding section of the experimental part.



To determine if the cellulose part out 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 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. Therefor numerous new intensive maxima were formed. Apparently, these bands are caused by intermediates of the oxidative degradation of chlorella.

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


Page 12 of 20

Page 13 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


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 Lindqvist-type 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 loosing performance as well as significant changes in catalyst structure indicated by 51V-NMR spectroscopy. The optimized conditions were further applied to the conversion of real biomass. When using hardwood (beech) as an example for real biomass, the FA-yield and selectivity were satisfactory. Finally Chlorella, representing a third generation biomass as substrate processed under the same conditions gave a 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.



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. [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 tenfold 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 work-up procedure For the oxidation reactions, each autoclave was filled with substrate, catalyst (0.1 mol L-1) and water (10 g respectively 100 g) as the solvent. The substrate concentration varies between 0.2 mol L-1 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 pre-pressurized 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 °C respectively 115 °C) was reached, the oxygen pressure was increased to the required pressure (30 bar respectively 50 bar) and the stirring speed was set to 1000 rpm in order to 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 Perkin Elmer 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 ACS Paragon Plus Environment

Page 14 of 20

Page 15 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


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 ICS-3000 equipped with a Dionex™ IonPac™ AS11-HC-2µm (250 mm x 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 x 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(Catoms 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.



Supporting Information Electronic Supplementary Information (ESI) available: 51V-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 lignin, HPLC chromatogram of an exemplarily reaction solution, IC chromatogram of reaction solution from Table 1, entry 2.

Acknowledgements 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 gratefully acknowledged. We thank Julian Mehler supporting the catalyst synthesis, Nicola Taccardi for performing ICPOES and Benjamin Bertleff for performing IC measurements. The authors declare no conflicts of interest.


Page 16 of 20

Page 17 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


References [1]

Cherubini, F. The biorefinery concept: Using biomass instead of oil for producing energy and chemicals. Energy Conversion and Management 2010, 51, 1412-1421. DOI: 10.1016/j.enconman.2010.01.015

[2]

Covestro AG. https://www.covestro.com/en/sustainability/service-downloads/policiescommitments [2018-08-22].

[3]

Shafiee, S.; Topal, E. When will fossil fuel reserves be diminished? Energy Policy 2009, 37, 181-189. DOI:10.1016/j.enpol.2008.08.016

[4]

Palkovits, R. Cellulose und heterogene Katalyse – Eine Kombination mit Zukunft. Chemie Ingenieur Technik 2011, 83, 411-419. DOI: 10.1002/cite.201000194

[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: 978-1-4200-3082-2

[6]

Witzke, I. Alternative Kraftstoffe für die dieselmotorische Verbrennung aus kohlenhydrathaltigen Biomassen und basierend auf bio- und chemisch-katalytischen Herstellverfahren; Springer: Wiesbaden, 2017. 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. Angewandte Chemie 2007, 46, 7164-7183 . DOI: 10.1002/anie.200604274

[8]

Sutton, A. D.; Waldie, F. D.; Wu, R.; Schlaf, M.; Silks, L. A.; Gordon, J. C. The hydrodeoxygenation of bioderived furans into alkanes. Nature Chemistry 2013, 5, 428432. DOI: 10.1038/NCHEM.1609

[9]

Wölfel, R.; Taccardi, N.; Bösmann, A.; Wasserscheid, P. Selective catalytic conversion of biobased carbohydrates to formic acid using molecular oxygen. Green Chemistry 2011, 13, 2759-2763. DOI: 10.1039/c1gc15434f

[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. Chemical Society Reviews 2018, 47, 1351-1390. DOI: 10.1039/c7cs00213k

[12]

Hill, C. L.; Prosser-McCartha, C. M. Homogeneous catalysis by transition metal oxygen anion clusters. Coordination Chemistry Reviews 1995, 143, 407-455. DOI: 10.1016/0010-8545(95)01141-B

[13]

Pope, M. T.; Müller, A. Polyoxometalate Chemistry: An Old Field with New Dimensions in Several Disciplines. Angewandte Chemie 1991, 30, 34-48. DOI: 10.1002/anie.199100341



[14]

Hill, C. L. Progress and challenges in polyoxometalate-based catalysis and catalytic materials chemistry. Journal of Molecular Catalysis 2007, 262, 2-6. DOI:10.1016/j.molcata.2006.08.042

[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 Chemistry & Engineering 2017, 5, 7383–7392. DOI: 10.1021/acssuschemeng.7b01723

[16]

Albert, J.; Wasserscheid, P. Expanding the scope of biogenic substrates for the selective production of formic acid from water-insoluble and wet waste biomass. Green Chemistry 2015, 17, 5164-5171. DOI: 10.1039/c5gc01474c

[17]

Albert, J. Selective oxidation of lignocellulosic biomass to formic acid and high-grade cellulose using tailor-made polyoxometalate catalysts. Faraday Discussion 2017, 202, 99-109. DOI: 10.1039/c7fd00047b

[18]

Siebertz, K.; van Bebber, D.; Hochkirchen, T. Statistische Versuchsplanung: Design of Experiments; Springer Vieweg: Berlin, 2010. DOI: 10.1007/978-3-662-55743-3

[19]

Kleppmann, W. Versuchsplanung: Produkte und Prozesse optimieren; Carl Hanser Verlag: München, 2016. eBook ISBN: 978-3-446-44717-2

[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. DOI: 10.1002/slct.201600797

[21]

Andersson, I.; Hastings, J. J.; Howarth, O. W.; Pettersson, L. Aqueous tungstovanadate equilibria. Journal of the Chemical Society 1996, 2705-2711. DOI: 10.1039/DT9960002705

[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. Journal of the Chemical Society 1976, 2290-2297. DOI: 10.1039/DT9760002290

[23]

Böttcher, J. H. Quantitative Analyse von Holz und Holz-Komponenten mittels FTIRSpektroskopie 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. Journal of Environmental Monitoring 2011, 13, 879-885. DOI: 10.1039/c0em00593b

[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 and Environmental Science 2012, 5 (7), 7956-7962. DOI: 10.1039/c2ee21428h


Page 18 of 20

Page 19 of 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[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. Catalysis Today 2016, 278, 7481. DOI: 10.1016/j.cattod.2016.03.030

[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. DOI: 10.1039/C4GC00970C



For Table of Contents Use Only Design of experiments is used to improve the process conditions for the fractionated oxidation of lignocellulosic biomass to formic acid and high-grade cellulose.


Page 20 of 20

Improving the Fractionated Catalytic Oxidation of Lignocellulosic

Recommend Documents