Carbon supported Ru clusters prepared by pyrolysis of Ru precursor-impregnated biopolymer fibers

Ru clusters deposited on pyrolyzed bacterial nanocellulose (Ru/p-BNC) were prepared in a single step by controlled pyrolysis at 1250 °C (under Ar gas), starting from bacterial nanocellulose (BNC) fibers impregnated with [RuCl2(DMSO)4], which serves as a Ru precursor. Pyrolyzed BNC fibers, featuring a high specific surface area (>600 m2 g−1), are structurally and chemically robust supports for Ru nanoparticles, as demonstrated by temperature programmed pulse chemisorption experiments. The carbon fibers, as well as the carbon supported Ru clusters, were characterized by a range of complementary analytical techniques including high-resolution transmission electron microscopy (HRTEM), selected area electron diffraction (SAED) and Raman spectroscopy.


Introduction
The deposition of Ru containing particles on carbon materials, either as pure Ru clusters, as RuO 2 particles, or as a bimetallic phase (e.g. Pd/Ru particles) has been of scientic interest for decades. These materials are widely used as catalysts in a variety of reactions, employing a broad range of different carbon support materials, i.e. carbon nanobers (CNFs), carbon nanotubes (CNTs), activated carbon bers (ACFs), activated carbon cloth (ACC), or just amorphous carbon. [1][2][3][4][5][6][7][8] However, the production of nanostructured carbon supports oen requires a number of preparation steps, such as template impregnation, pyrolysis and/or template removal procedures, including the usage of aggressive chemicals (e.g. hydrogen uoride) 9,10 and thus environmentally benign approaches towards structurally and chemically well-dened nanocarbon materials should represent a sustainable and economically attractive alternative. The use of cellulosic precursors for the production of carbon bers is long since known and dates back to the late 19th century. 11 The graphitization of cellulose from different sources (such as plant and bacterial nanocellulose (BNC)), the resulting carbon morphologies and their potential use as catalyst supports have been studied extensively. [12][13][14][15][16][17] A review of the structural evolution of the resulting carbon bers and the operational parameters is given in ref. 11. While cellulose commonly is not as easily graphitizable as the more frequently used polyacrylonitrile (PAN), the crystallinity of bacterial cellulose is higher than that of cellulose from other biological resources, which possibly enhances its intrinsic propensity for graphitization. 11,17 The small sized microbrils which cellulose-producing bacteria assemble form bands of nanosized brils that feature a width of only $500Å, which renders these a technologically attractive biological source for brous carbon materials. 18 This microstructure leads to highly accessible surface areas of the pyrolyzed carbon bers, which in turn is advantageous for use as supports for metal catalysts. 19 We here report for the rst time a highly efficient one-step synthesis of carbon supported ruthenium clusters by controlled pyrolysis of microbial nanocellulose (Ru/p-BNC) impregnated with a Ru precursor, namely cis-[Ru II Cl 2 (DMSO) 4 ]. The morphology and structure of the thus-prepared materials were investigated by a range of complementary analytical techniques and preliminary tests of the catalytic properties of Ru/p-BNC were conducted through hydrogenation (methanation) of carbon monoxide by means of temperature controlled pulse chemisorption experiments.

p-BNC
Pyrolyzed bacterial nanocellulose (p-BNC) was analyzed by IR, TGA, ESEM, HRTEM, temperature controlled pulse chemisorption experiments and argon sorption measurements. The morphologies of the native and the pyrolyzed bers were investigated by ESEM. Fig. 1a shows the image of an untreated ber, while Fig. 1b shows the image of a sample pyrolyzed at 1000 C under a N 2 atmosphere. The adsorption isotherms of argon sorption measurements for two different BNC samples treated at 800 C and 1500 C under a N 2 atmosphere and another sample treated at 1250 C under an Ar atmosphere were recorded (see Fig. A in the ESI †). According to the BET-plot, the highest surface area value (624 m 2 g À1 ) was obtained for a sample pyrolyzed at 800 C while samples treated at 1250 C (in an Ar atmosphere for better comparison with Ru/p-BNC samples) and 1500 C resulted in a decrease of the specic surface area (307 m 2 g À1 and 287 m 2 g À1 , respectively). The decrease of specic surface area for cellulose treated at very high temperature is a well-known phenomenon, which was reported earlier by Harris. 20 Fourier-transform infrared measurements of the untreated bacterial nanocellulose and samples pyrolyzed at 1000 C and 1500 C under N 2 are shown in Fig. 2.
The infrared spectrum of BNC in Fig. 2a shows the typical bands for the cellulose material. 21 The bands in the region of 500-700 cm À1 can therefore be assigned to the n O-H vibrations (out of plane bending) of the hydroxyl groups. The bands in the region of 1000-1200 cm À1 are mainly due to the n C-O stretching vibrations of hydroxyl and acetal groups. Bands in the region of 1200-1450 cm À1 can be assigned to the n O-H in-plane bending vibrations of the hydroxyl groups, while bands around 2900 cm À1 are due to the n C-H stretching vibrations of the cellulose structure. Finally, the broad bands in the region from 3000-3600 cm À1 can be assigned to the n O-H stretching modes of hydroxyl groups. 21 For the samples heated to 1000 C under N 2 (and 1500 C respectively, cf. Fig. 2b and c), the characteristic bands for cellulose cannot be observed, which is expected to be due to the carbonization process. The resulting feature-less IR spectra are typical for carbon materials/graphite. 22 HRTEM images of BNC bers pyrolyzed at 1500 C under N 2 in Fig. 3a and b show graphitic domains from which electron density proles can be extracted (see Fig. 3c). The layer distance was determined to be d 002 ¼ $0.347 nm, which is close to the d value of turbostratic graphite (random stacking of graphene layers, d ¼ 0.344 nm). 20,23 The SAED pattern of the sample in the inset in Fig. 3a shows diffraction rings which can be assigned to the (100) and (110) planes of a graphitic structure. The diffraction ring corresponding to the (002) plane is not clearly visible from the SAED pattern. This is probably due to the small domains of graphitic layers and the resulting very broad ring, which is overlaid by the central spot. This could be revealed by using a different camera length which also resulted in weaker overall intensity. The rotational average of this SAED pattern shows a broad peak for (002) planes which is centred at $3.60Å (see Fig. 4). The peaks for (100) and (110) planes are centred at 2.07Å and 1.21Å respectively.   Thermogravimetric analyses for samples of BNC have been conducted under Ar (purity 99.999%) atmosphere in the temperature range starting from room temperature up to a nal temperature of 1100-1400 C (see Fig. B in the ESI †). The mass losses of four independent samples are almost identical with values ranging from 89.0% loss for the sample heated up to 1100 C, 93.3% (1200 C), 91.8% (1300 C), and 92.7% (1400 C). The weight loss takes place in one big step with an onset at 311-315 C for the samples heated to 1200, 1300 and 1400 C, respectively. For the sample treated at 1100 C the onset is located at a slightly higher temperature (324 C). The loss of mass is mainly due to release of water, CO and CO 2 (determined by mass spectrometric measurements, see Fig. D in the ESI †).

Ru/p-BNC
HRTEM images of a sample of the Ru precursor impregnated nanocellulose pyrolyzed under Ar at 1400 C (Ru/p-BNC_28.4%) shows Ru clusters as dark spots conned to the surface of carbon bers ( Fig. 5a and b). The average size of $6.5-13 nm (depending on the ruthenium concentration) of the ruthenium clusters was calculated by a method described elsewhere. 24 Higher ruthenium-to-cellulose ratios led to larger ruthenium clusters (see Table 1).
The diffraction rings of the SAED pattern of the area in Fig. 5c can all be assigned to crystalline ruthenium with hexagonal close-packed (hcp) structure. The indexed peaks of the extracted rotational average of the SAED pattern at 2.336Å, 2.055Å, 1.576Å, 1.349Å and 1.212Å can be assigned to the (100), (101), (102), (110) and (103) planes of Ru-hcp (Fig. 5d). The peak at 1.137Å could either be assigned to the (112) or the (201) plane of Ru-hcp with theoretical values of 1.143Å and 1.129Å, respectively. 25 Broad diffraction rings of the carbon support (compare Fig. 3) can also be observed under the diffraction rings of the Ru-hcp structure at $2.05Å and $1.21Å. These values can be assigned to the (100) and (110) planes of the graphite structure, respectively. 26 Thermogravimetric measurements in conjunction with mass spectrometric sampling of the evolved gases of the Ru precursor and the Ru precursor-impregnated nanocellulose were performed under Ar (purity 99.999) up to 1400 C. The mass ratios of Ru precursor to nanocellulose are listed in Table 2. The decomposition of the Ru precursor (i.e. cis-[Ru II Cl 2 (DMSO) 4 ]) takes place in two major mass steps. The temperature onset of the rst mass-loss step is $187 C. This rst step can be separated into three sub-steps. The rst sub-step shows a gas evolution of DMSO (dimethylsulfoxide) and the beginning of a partial decomposition of DMSO in H 2 O and CO ($15.2% massloss). The second (21.0% mass-loss) and third sub-step (33.5% mass-loss) are much more complicated and not fully understood as yet. As main compounds, gaseous CO and H 2 O are detected. Apart from these gases chloromethane, DMS (dime-thylsulde) and CH 2 O (formaldehyde) are also formed. The temperature onset of the second major (7.9% mass-loss) massloss step is $977 C; at this temperature SO 2 was detected by    4 ], the Ru precursorimpregnated nanocellulose as well shows two major decomposition steps. The temperature onset of the rst mass-loss step varies between $150 C and $240 C, depending on the loading of the nanocellulose ber. A loading of the cellulose bers with the Ru precursor up to 1 : 1 ratio leads to a shi of the decomposition temperature towards a lower onset value. Ratios higher than 1 : 1 let the decomposition temperature ascend again (see Fig. C in the ESI †). The decomposition of the impregnated nanocellulose also shows two main mass steps. During the rst main mass step H 2 O, CO 2 , CO, DMSO, DMS, CH 2 O and chloromethane were detected by mass spectrometry. The temperature onset of the second main mass-loss step is $981 C; at this temperature, SO 2 was released. In addition, CO 2 was formed due the reduction of RuO 2 (see Fig. F in the ESI †).

Catalytic activity
Two samples of Ru/p-BNC were used for the preliminary catalytic tests, namely Ru/p-BNC_5.0% and Ru/p-BNC_3.8%. The results obtained with the sample Ru/p-BNC_5.0% (sample mass: 8.0 mg) indicated promising catalytic activity of the material. To verify these results a second independent sample was prepared under the same conditions. The amount of ruthenium in this sample was 3.8 wt%. This sample (45.4 mg) was used to analyze the amount of active sites of the ruthenium clusters by employing carbon monoxide pulse chemisorption. Prior to this, the sample was reduced with 10% hydrogen in argon at an oven temperature of 550 C (sample: 510 C). Aerwards, pulse chemisorption measurements were performed at room temperature, injecting pulses of CO/He (5 vol%/95 vol%) into a carrier gas stream of helium. The amount of carbon monoxide adsorbed on the ruthenium particles was 0.24 mmol. Assuming a Ru/CO-ratio of 1 : 1, ca. 1.3% of the ruthenium atoms are covered with CO. The particle size of this sample was determined to be 6.5 AE 2.9 nm. Thus the amount of titrated surface atoms was calculated to be 6.3% for the average particle size of 6.5 nm. Ruthenium particles are the most active particles for the Fischer-Tropsch reaction. 27 Carbon monoxide is reduced by hydrogen to hydrocarbons and water in the gas phase at higher temperatures and ambient pressure. 28 Hence, the carbon supported ruthenium clusters were used in another pulse experiment, applying a temperature ramp up to 400 C. Hydrogen (10 vol% in Ar) was used as the carrier gas, while carbon monoxide (5 vol% in He) was used as the pulse gas. The reaction products were analyzed by mass spectrometry. At 135 C the rst release of methane was detected by means of mass spectrometry (Fig. 6). Upon reaching 400 C (sample temp.: 360 C) the oven temperature was kept constant, showing a constant production of methane (1.18 mL per pulse of carbon monoxide) and water between 250 C and 360 C. Another hint for the formation of methane is the delay of the liberation of the pulses with the m/z value of 15, depicted in Fig. H of the ESI, † in comparison to the methane calibration pulses. Typically, the reaction products of a Fischer-Tropsch reaction are higher chains of hydrocarbons and water, which could not be detected during this experiment. Thus the catalytic reaction of Ru/p-BNC can be described as methanation.
A third pulse experiment was conducted with hydrogen (10 vol% in Ar) and nitrogen (100%) as pulse gas. A temperature ramp up to 800 C was applied, revealing that the carbon supported ruthenium clusters reduce the support material at temperatures above 525 C (see Fig. J in the ESI †), as reported elsewhere. 29 A comparison of the adsorption isotherms of argon sorption measurements of Ru/p-BNC_5.0% samples showed an increase according to the BET-plot before (288 m 2 g À1 ) and aer catalytic tests (384 m 2 g À1 ) including several reduction steps under hydrogen above 550 C (see Fig. A in the ESI †). The reason for this might be a reaction of the carbon supporting material with hydrogen and the ruthenium clusters at temperatures above $525 C (see Fig. J in the ESI †) or a deposition of carbon according to the Boudouard reaction (2CO 4 C + CO 2 ). Both effects might lead to the formation of further micropores within the carbon supporting material; hence a higher surface area was generated.
A reference sample of p-BNC (4.0 mg) showed signs of a reaction with hydrogen gas above 450 C, evolving traces of methane, according to the results of mass spectrometry (m/z 15 and m/z 16). 28

Raman spectroscopy
According to the literature the graphitization degree of carbonaceous samples can be determined by different techniques, i.e. X-ray diffraction, 30 electron microscopy 31 or Raman spectroscopy. 32 We chose the latter technique and followed a procedure described by Endo et al. which shows the so-called Dand G-bands of the carbonaceous sample, in which the D-band (centered at approx. 1350 cm À1 ) is related to disordered graphene layers while the G-band (centered at approx. 1592 cm À1 ) accounts for the E 2g mode at the G-point of the graphite crystal lattice. 33 Thus, determining the ratio between the peak areas (A D-band / A G-band ) of a sample provides a means to evaluate its graphitization degree, which can also be employed as a qualitative measure for comparing different samples (sharing similar precursor chemistry and pre-treatment conditions, however). For comparison purposes, the D-($1350 cm À1 ) and G-band ($1592 cm À1 ) peak positions, the full width at half maximum (FWHM) and the relative peak area ratio of the two bands (A D-band /A G-band ) were tted using a Gaussian/Lorentzian mixed shape t function (see Table A in the ESI †). 34 BNC pyrolyzed at 1400 C shows the highest graphitization degree, related to the FWHM from D-and G-band and the relative peak area ratio (A D-band /A G-band ). Also the Ru/p-BNC_3.8 samples show an increase in graphitization at higher temperatures. However in direct comparison, the samples including ruthenium have a broader FWHM and a higher relative peak area ratio (A D-band /A G-band ) than the p-BNC. This indicates a lower graphitization degree, irrespective of the gas atmosphere and the sample.
Comparing different pyrolysis gases, argon seems to be better suited than nitrogen for the pure BNC, in order to obtain a high graphitization degree of the samples (A D-band /A G-band -ratio and the FWHM of G-bands), which contrasts with the ndings of Fey et al. reported in 2002. 35 However, the degree of graphitization of Ru/p-BNC_3.8 is higher for samples treated with N 2 .

Conclusions
By pyrolyzation of bacterial nanocellulose at temperatures ranging from 800-1500 C it was possible to synthesize a carbon material with a brous structure and a moderately high specic surface area of up to 624 m 2 g À1 . A high surface area represents a crucial feature for carbon materials being used as supports for metal catalysts or as supercapacitors in ionic liquids. 19,36,37 The carbon bers produced from bacterial nanocellulose are partly graphitized, which should give rise to good electrical conductivity, which in turn could also be advantageous for use as supercaps or for electrocatalytic applications. 15,17,38 A promising method for generating even higher surface area values within the nanocellulose-derived carbon bers might be a chemical (e.g. KOH) or physical (e.g. CO 2 ) treatment as described elsewhere. 17,39 The use of the pyrolyzed bacterial nanocellulose as a catalyst support was tested by in situ preparation of nanosized Ru clusters on p-BNC. The untreated bers were immersed in [RuCl 2 (DMSO) 4 ] solution and then heated to 1250 or 1400 C under an Ar atmosphere. The Ru clusters were identied as ruthenium showing a hexagonal close-packed lattice structure with average metal cluster sizes ranging from $6.5 nm to 13 nm, depending on the concentration of the impregnating precursor solution (see Table 1).
Raman measurements show that the graphitization degree of Ru/p-BNC_3.8 is lower than of p-BPF. This is shown for two different pyrolysis temperatures and two different reaction gases. According to the results argon as the inert gas seems to support the graphitization of pure BNC bers better than nitrogen, whereas nitrogen seems to support the graphitization of Ru/p-BNC_3.8 better than argon.
The efficient generation of carbon nanober surface-conned Ru clusters in a single preparation step renders this procedure an economically attractive, sustainable approach toward large-scale production of Ru-based supported catalysts. The brous structure of the carbon provides open access for reagents (either in condensed or gas phase) to the catalytically active Ru centers.
As a preliminary test for the catalytic properties of the material the low-pressure (partial pressure of CO: <0.05 bar) reduction of carbon monoxide by hydrogen gas (at temperatures starting at 135 C) was demonstrated successfully. TEM investigations of a Ru/p-BNC_5.0% sample aer several catalytic tests did not show any discernible changes with respect to the carbon ber structure. The average ruthenium cluster size slightly  6 Mass spectrum of CO pulse chemisorption measurements in 10% hydrogen in Ar. A temperature ramp (red) was conducted from room temperature to 400 C. m/z 28 (purple) represents the carbon monoxide pulses, m/z 16 (blue) represents oxygen from carbon monoxide, as well as the methane formed at temperatures above 135 C (growth of the peaks). The ratio of m/z 15 (green): m/z 16 is used to prove the formation of methane.
increased from 8.6 nm to 8.9 nm. These promising results might point to a highly desirable long term stability of the catalyst in gas phase hydrogenation processes.

Experimental section
Preparation of pyrolyzed bacterial nanocellulose bers (p-BNC) Bacterial nanocellulose bers were obtained from Jenpolymer Materials Ltd & Co. KG in the block-shaped form. Pyrolysis was conducted in a Gero HTRH-70-300/16 tube furnace, equipped with an alumina tube 1000 mm in length and 70 mm in diameter (Alsint 99,7 TYP C 799 according to DIN EN 60672), under N 2 (purity 99.999%, gas ow 70 mL min À1 ) atmosphere and under Ar (99.999%, gas ow 70 mL min À1 ) atmosphere at 1250 C, respectively. The biopolymer was transferred to an alumina crucible and heated (heating rate 5 K min À1 ) to the desired temperature (see Table 2). Activated carbon was used in an extra alumina crucible as an oxygen getter and placed near the sample. At the outlet of the tube furnace the oxygen content was monitored during the whole process by means of a solid electrolyte sensor purchased from Zirox® (O2-DF-13.0). Aer holding the temperature for 3 h the pyrolyzed nanocellulose bers (p-BNC) were cooled down to room temperature. to room temperature. The ruthenium content in weight percent (wt%) was calculated from the value of the residual mass from TGA data, under the assumption of a complete conversion of the ruthenium(II)-complex into ruthenium(0) (see Table 3 and Characterization TEM investigations were performed using a JEOL 2100F microscope with a FEG electron source operated at 200 kV. For sample preparation, a small portion of material was dispersed in $1 mL of ethanol and sonicated for 2-3 minutes. Holey carbon-coated copper grids were then dipped into the dispersion and dried at RT. The soware package DiffTools was used for evaluation of SAED-patterns. 42 EDX measurements were conducted with an EDAX Genesis detector as part of the JEOL 2100F setup. Data analysis was performed with the corresponding soware package. Crystal phase determinations were performed with a Seifert 3003 TT diffractometer, collecting X-ray diffraction data in the range 10 # 2q $ 100 with a step width of 0.01 . Data collection was carried out by means of a Meteor 1D line detector and took 100 s per data point. (Cu-Ka radiation, Bragg-Brentano geometry). The X-ray tube was operated with 40 kV and 40 mA and a nickel lter was used to suppress K b radiation.
Fourier-transform infrared spectra of the samples were recorded using a Bruker Equinox 55 equipped with a Bruker Platinum ATR unit from 4000-400 cm À1 using 4 cm À1 resolution and 32 scans per sample.
Raman spectra were recorded with a DXR™ Raman microscope from Thermo Scientic with a 532 nm laser, a magnication of 10 with an estimated resolution from 5.5 to 8.3 cm À1 . The data were collected by applying a Laser power of 0.5 mW and an exposure time of 1 second for each spectrum. Raman spectra of each sample were recorded at seven different positions. Each spectrum was smoothened, base line corrected and tted with a two peak Gaussian/Lorentzian t function. For comparison purposes, the peak position, peak area and FHWM of the G-and D-band of the spectra were evaluated (see Table A in the ESI †).
ESEM investigations were performed using a FEI/Philips XL30 FEG ESEM. The samples were sputtered with an Au layer in order to avoid charging effects.
Argon sorption isotherms were measured with a Quantachrome Autosorb-1C instrument at 77 K up to p/p 0 ¼ 1. High purity gas was used for the adsorption experiments (argon 99.999%). Prior to measurements, the samples were heated at 300 C for 2 h in a vacuum.
Thermogravimetric analysis (TGA) was performed with a NETZSCH STA 409 coupled to a mass spectrometer via Skimmer®, and a NETZSCH STA 409 PC Luxx®.
Temperature programmed pulse chemisorption measurements were performed using a BelCat-B catalyst analyzer (Bel Japan, Inc.) equipped with a thermal-conductivity detector and a coupled mass spectrometer (OmniStar GSD 320, Pfeiffer Vacuum). The volume of the pulse loops was 0.920 mL and 0.326 mL, the gas ow rates were set to 30 mL min À1 . Two different samples were used for the catalytic tests: Ru/p-BNC_5.0% (8.0 mg) and Ru/p-BNC_3.8% (45.4 mg). Each sample was placed between two plugs of quartz wool in a quartz glass reactor. Prior to hydrogenation experiments the sample was reduced with 10% hydrogen in argon (99.999%) by heating the sample to 550 C (furnace temperature) for 30 minutes. Aer this reduction step, 5.01% carbon monoxide in helium was pulsed (pulse loop volume ¼ 0.920 mL) to a helium carrier gas stream (99.999%) to titrate the amount of active sites. To conrm the obtained results of the adsorbed amount of carbon monoxide, the titration was repeated with a smaller pulse loop (0.326 mL). Aer thermal desorption of the carbon monoxide, 5.01% carbon monoxide in helium was pulsed to a mixture of 10% hydrogen in argon (pulse loop size ¼ 0.920 mL). The furnace temperature was increased from room temperature to 400 C (ramp rate 10 K min À1 ) and the reaction products were analyzed by mass spectrometry.
Another experiment was performed with Ru/p-BNC_3.8% and hydrogen (10% in argon) as carrier gas. Nitrogen was applied as the pulse gas (pulse loop size ¼ 0.920 mL), while the furnace temperature was increased from room temperature to 800 C with a ramp rate of 10 K min À1 . The reaction products were analyzed by means of mass spectrometry.
In addition, an experiment with a pure carbon supporting material (p-BNC (1250 C under Ar), 4.0 mg) was performed to investigate the reduction of the support material under hydrogen (10%, in argon). The reaction temperature for this experiment was increased from room temperature to 950 C with a ramp rate of 10 K min À1 .