Carbohydrate-Derived Hydrothermal Carbons: A Thorough

Aug 1, 2012 - Charge referencing was done by setting the lower binding energy C 1(s) ..... The surface charge of the materials is in all cases negativ...
1 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/Langmuir

Carbohydrate-Derived Hydrothermal Carbons: A Thorough Characterization Study Linghui Yu,† Camillo Falco,†,§ Jens Weber,† Robin J. White,†,∥ Jane Y. Howe,‡ and Maria-Magdalena Titirici*,† †

Max Planck Institute for Colloids and Interfaces, Research Campus Golm, D-14424 Potsdam, Germany Oak Ridge National Laboratory, 1 Bethel Valley Road, Oak Ridge, Tennessee 37831, United States

‡

S Supporting Information *

ABSTRACT: Hydrothermal carbonization (HTC) is an aqueous-phase route to produce carbon materials using biomass or biomass-derived precursors. In this paper, a comprehensive physicochemical and textural characterization of HTC materials obtained using four different precursors, namely, xylose, glucose, sucrose, and starch, is presented. The development of porosity in the prepared HTC materials as a function of thermal treatment (under an inert atmosphere) was specifically monitored using N2 and CO2 sorption analysis. The events taking place during the thermal treatment process were studied by a combined thermogravimetric/infrared (TGA-IR) measurement. Interestingly, these inexpensive biomass-derived carbon materials show good selectivity for CO2 adsorption over N2 (CO2/N2 selectivity of 20 at 273 K, 1 bar and 1:1 gas composition). Furthermore, the elemental composition, morphologies, degree of structural order, surface charge, and functional groups are also investigated.

■

INTRODUCTION Hydrothermal carbonization (HTC) has been intensively researched over the past few years as a sustainable synthetic route to carbon materials with numerous applications.1−4 This methodology was first described by Bergius5 in 1913 as a synthetic way to mimic coal formation in only few hours under self-generated pressure in closed systems and mild temperatures (i.e., 130−250 °C). The HTC process has recently been revisited by Huang et al.,6 Li and Sun,7 Sevilla and coworkers,8,9 and Titirici and co-workers,10,11 and is currently considered of the most sustainable approaches toward functional carbon based materials.4,11 As compared to classical carbonization methods, HTC takes place in water at subcritical temperatures and pressures. A prerequisite for this process is the use of carbohydrates or carbohydrate-rich biomass as precursors. Such compounds are renewable and highly abundant within the global biosphere (i.e., 1880 billion tons of biomass). The HTC process relies on the initial dehydration of carbohydrates to hydroxymethylfurfural (HMF) followed by a reaction cascade comprising ring-opening reactions, substitutions, cycloadditions and polycondensations to form the final HTC structure.12,13 HTC occurs in an efficient energy-saving manner, capable of producing carbon and carbon hybrid materials with controlled morphologies and structures.14−17 Additionally, due to the mild carbonization conditions the resulting particles are hydrophilic with useful surface functionalities.7,17−19 This feature facilitates chemical surface mod© 2012 American Chemical Society

ifications via relatively simple routes; a task which is comparatively more challenging for the classical pyrolyzed carbons.20−24 Furthermore, using a postsynthesis thermal treatment step, it is possible to fine-tune the surface/bulk properties of HTC material, while concurrently enhancing the porous character.25,26 Thus, HTC is a flexible synthesis platform which allows modification of the final properties by simply applying different thermal treatments under an inert atmosphere (e.g., N2 or Ar). With regard to thermal treatment of HTC material, there is as yet, to the best of our knowledge, no systematic study on the properties of HTC materials prepared at increasing temperature with particular respect to chemical composition, porosity, structure, and functionality. In this study, four different carbohydrate precursors, namely, two monosaccharides, one pentose (xylose) and one hexose (glucose), one disaccharide (sucrose), and one polysaccharide (starch) were used as HTC material precursors in this study. The polysaccharide composite starch was chosen as in the majority (for both amylose and amylopectin) glucose monomers are connected via α(1−4) glycosidic linkages which make essentially starch easily hydrolyzable as compared to cellulose, where the production of HTC is not as straightforward due to the comparatively Received: June 14, 2012 Revised: July 31, 2012 Published: August 1, 2012 12373

dx.doi.org/10.1021/la3024277 | Langmuir 2012, 28, 12373−12383

Langmuir

Article

Figure 1. (a) Thermogravimetric analysis of glucose-derived HTC carbon at 180 °C before extraction and (b) after three consecutive extractions with ethanol, hexane and THF. (c) Gas chromatograph of the extract obtained after Soxhlet extraction with ethanol of glucose derived HTC carbon at 180 °C. (d) Mass spectrum of the peak indicated in (a) (red, literature data;27 black, experimental data). (e) Extracted traces from TG-IR analysis at selected wavenumbers (cm−1) for glucose derived HTC carbon at 180 °C before extraction. The legend shows either the name of the gas, the trace can be attributed to, or the wavenumber, the trace is detected at.

more stable β(1−4) linked monomer units and the resulting inter/intra molecular hydrogen bonds.13,28 These precursors were hydrothermally carbonized at 180 °C, followed by drying and thermal treatment at four representative temperatures: 350, 550, 750, and 950 °C. The materials produced in this study were then characterized in terms of porosity, structure, composition and functionality. Materials were denoted as HTC-X-Y where X represents the initial of the precursor and Y represents the thermal treatment temperature. For example and clarity, HTC-G is the hydrothermal carbon from glucose (180 °C) while HTC-G-350 is the same material after thermal treatment under N2 at 350 °C.

■

X-ray photoelectron spectroscopy (XPS) was performed using a Thermo Scientific K-Alpha ESCA instrument equipped with Al Kα monochromatized radiation with a 1486.6 eV X-ray source. An electron flood gun was used to minimize surface charging. Neutralization of the surface charge was performed by using both a low energy flood gun (electrons in the range 0−14 eV) and a low energy argon ion gun. Photoelectrons were collected from a takeoff angle of 90° relative to the sample surface. The measurement was done in “Constant Analyser Energy” mode (CAE) with a pass energy of 100 eV for the survey scane and 20 eV for high resolution spectra. Charge referencing was done by setting the lower binding energy C 1(s) photo peak at 285.0 eV. Surface elemental composition was determined using the standard Scofield photoemission cross sections. Zeta potential measurements were carried out on a Malvern Nano ZS instrument. The carbon powders were dispersed in 0.0167 M NaCl solution. 0.03 M HCl and NaOH were used to adjust the pH values. The solutions were stirred for 48 h at room temperature. All samples were filtered using PTFE 5 μm disposable filters. Disposable clear zeta cells (DTS1060c) were used for the measurements. N 2 and CO 2 sorption analysis were performed using a QUADRASORB SI/MP and an Autosorb-1 MP (Quantachrome Instruments) at 77.3 K and at 273 K, respectively. Prior to measurement, the samples were degassed at 150 °C for 20 h. Brunauer−Emmett−Teller (BET) and nonlinear density functional theory (NLDFT) methods were used for the surface area and pore size distribution (PSD) determination using N2 adsorption data, and grand canonical Monte Carlo (GCMC) method was used for CO 2 adsorption. The data evaluation was done using the QuadraWin 5.05 and Origin 8.0 software. High purity gases were used for all measurements. High resolution tunneling electron microscopy (HRTEM) was carried out using a Hitachi HF3300 cold-field emission TEM/STEM instrument at 300 kV. The STEM images were taken using the TEM, ZC, and SE modes. The carbon sample was dispersed onto a holey carbon film supported on a 200-mesh copper grid. The heating experiments were carried out using a heating system (Aduro In Situ System, Protochips, Inc. Raleigh, NC) capable of reaching 1200 °C. The temperature on the chip was controlled by varying the current

EXPERIMENTAL SECTION

Hydrothermal Carbons and Post-Carbonized Samples. Hydrothermal carbons were obtained from glucose (C6H12O6·H2O), sucrose (C12H22O11), starch (C6H10O5)n, and xylose (C5H10O5). In a typical synthesis, 4.5 g of precursor was dissolved in 15 mL of deionized water and sealed into an autoclave and kept at 180 °C for 24 h. After reaction, the HTC products were washed several times with water and dried at 80 °C under vacuum. The dried HTC powders were thermally treated at four different temperatures (i.e., 350, 550, 750, and 950 °C) under a constant flow of N2 gas at a heating rate 5 K min−1 below 350 °C and 2 K min−1 above 350 °C followed by an isothermal period at the desired temperature of 2 h. Characterization. Scanning electron microscopy (SEM) images were collected on a Gemini Leo-1550 instrument. Elemental composition was determined using a Vario El elemental analyzer. Thermogravimetric analysis coupled with infrared (TG-IR) was performed using a Netzsch STA 409 apparatus at scan rates of 20 K min−1 under N2 (flow rate: 10 mL min−1), coupled to a Brüker EQUINOX-55 instrument equipped with a liquid N2 cooled mercury cadmium telluride (MCT) detector. Fourier transform infrared (FTIR) spectra were acquired using a Varian 1000 FT-IR spectrometer. Powder X-ray diffraction (XRD) data for the prepared materials were collected with a Bruker D8 Advance diffractometer using Cu Kα radiation. 12374

dx.doi.org/10.1021/la3024277 | Langmuir 2012, 28, 12373−12383

Langmuir

Article

400 °C, the strongest band intensities correspond to the loss of CO2 and CO. Subsequently at ca. 450 °C, methane evolution is detected and the intensity of its corresponding band increases until ca. 550 °C. Above this temperature threshold, all the detected signals fade away at different rates. Furan ring-opening is expected to take place over in the temperature range where the second event of the TGA curves is observed.13,33−35 The evolution of CO and CO2 can be attributed to the successive loss of oxygenated functional groups, while CH4 trace is likely related to the removal of the methylene bridges which act as cross-linkers within the HTC carbon framework.13,33−35 Although CO, CO2, and CH4 are more pronounced here, there should be also other eliminations (e.g., H2O) during this broad charring step.33−35 As a consequence of the observed thermal decomposition processes, the structure of the HTC carbon undergoes a rearrangement process due to ringopenings and aromatization.13 While the furan-based structure is the main structure for the HTC carbon, 12 after decomposition the material becomes a turbostratic type carbon as determined by XRD (see Figure 8 and associated discussion). Elemental analysis (Table 1) of the HTC carbons treated at increasing temperatures shows major compositional changes in

applied to the heating stage according to the calibration curve provided by the vendor.

■

RESULTS AND DISCUSSION Before going into the characterization of the HTC materials prior and subsequent to thermal treatment, in terms of chemical functionalities, structure, surface charge, and especially porosity, we would like to remind the reader that upon hydrothermal carbonization of carbohydrates spherical carbonaceous particles are produced. This is because upon the decomposition of the carbohydrate into hydroxymethylfurfural (HMF) and subsequent further polymerization−polycondensation, small nuclei are formed. These nuclei grow according to a LaMer model,29 until all the “monomer” has been consumed and the final particle size is attained. Without any catalysts or additives, carbohydrate precursor of concentrations of 10−30 wt % results in particle diameters of between 500 nm and 10 μm (Figure S1a, d, g, and j) by HTC. In the present case, slight differences are observed for xylose leading to the formation of rather monodispersed particles as compared to glucose, sucrose or starch. The HTC of xylose, a 5C sugar, proceeds via a “furfural” and not “HMF” route. The HMF route is specific for hexose-based sugars.30 The morphology of HTC materials is stable and does not change upon further thermal treatment at higher temperatures (Figure S1). Since this paper focuses mainly on the characterization of HTC materials after further thermal treatment at increasingly higher temperature, the thermal decomposition behavior of glucose-derived HTC was investigated using a combined thermogravimetric/infrared (TG-IR) analysis. As observed from the TG/dTG traces, the thermal decomposition of HTC-G is composed of two relative broad decomposition events (160−270 and 350−600 °C) (Figure 1a). The first event (160−270 °C) is attributed to the thermal evolution of levulinic acid (boiling point range: 245−246 °C) embedded within the highly cross-linked HTC structure. Levulinic acid is a typical decomposition product resulting from the hydrothermal treatment of glucose, formed upon the rehydration of HMF.31,32 This molecule is then physisorbed on the surface of HTC particles during HMF formation. This is supported by a comparative analysis with a standard HTC sample simply washed with water after synthesis and a second sample, which was extracted consecutively in a Soxhlet with ethanol, hexane, and tetrahydrofuran (THF). First of all, the thermal decomposition of the extracted sample is composed of only one relative broad decomposition event (350−600 °C), indicating the materials responsible for the first event (160− 270 °C) were removed by Soxhlet extraction (Figure 1b). [NB: The first peak (Figure 1b) between 100 and 180 °C is ignored, as it is attributed to the removal of residual extraction solvent and physisorbed water]. Second, GC-MS analysis of the extracted ethanol fraction demonstrates the presence of levulinic acid, confirming its removal upon Soxhlet extraction with ethanol (Figure 1c and d), while the same analysis shows almost nothing in hexane or THF fraction (not presented). Such a result explains the absence of a dTG peak in the region of T = 160−270 °C in the TGA curve of the extracted sample. Namely, the presence of levulinic acid in the standard sample produces the first event of its TGA curve. The main decomposition event expands within the range T = 350−600 °C and is indicative of a restructuring/charring of the native carbon motifs. The loss of volatile species is described by the corresponding gas phase IR analysis (Figure 1e). Starting at

Table 1. Elemental Analysis Data of HTC and Thermally Treated Carbons

a

samples

C (wt%)

H (wt%)

O (wt%)a

HTC-G HTC-G-350 HTC-G-550 HTC-G-750 HTC-G-950 HTC-Su HTC-Su-350 HTC-Su-550 HTC-Su-750 HTC-Su-950 HTC-St HTC-St-350 HTC-St-550 HTC-St-750 HTC-St-950 HTC-X HTC-X-350 HTC-X-550 HTC-X-750 HTC-X-950

66.8 68.3 82.7 94.0 96.4 66.8 69.8 87.1 93.0 95.2 67.1 68.9 83.5 95.1 96.1 68.5 69.6 83.6 95.7 97.2

4.4 4.2 3.1 1.7 0.5 4.4 4.3 3.0 1.7 0.4 4.5 4.2 3.5 1.1 0.4 4.1 4.0 3.5 1.2 0.4

28.8 27.5 14.2 4.3 3.1 28.8 25.9 9.9 5.3 4.4 28.4 26.9 13.0 3.8 3.5 27.4 26.4 12.9 3.1 2.4

Calculated value.

the temperature interval 350−750 °C, where carbon content increases from ∼70% to ∼95% while the H and O contents are correspondingly reduced. Increasing calcination temperature to 950 °C has a continuous effect on the HTC carbon elemental composition, but this time the rate of change with temperature is small. These findings suggest the critical temperature region for HTC carbon thermal decomposition is lower than 750 °C which is consistent with TGA data having indicated that the main decomposition event is over the range 350−600 °C. The FT-IR spectra of the HTC carbons as well as the ones obtained after thermal treatment indicate the type of functional groups of the HTCs and structural and functional changes occurring upon thermal treatment (Figure 2). Abundant 12375

dx.doi.org/10.1021/la3024277 | Langmuir 2012, 28, 12373−12383

Langmuir

Article

Figure 2. FT-IR spectra of the HTC and thermally treated carbons, derived from (a) glucose, (b) sucrose, (c) starch, and (d) xylose.

Table 2. Experimental C 1(s)/Binding Energy (BE eV)/Chemical State Assignments for Glucose- And Xylose-Derived Carbons C1

C2

C3

C4

C4(b)

C5

samples

sp2-graphitic or C−C/C-Hx

CO

CO

OCO

carbonate

π−π* shakeup satellite

HTC-G-180 HTC-G-550 HTC-G-950 HTC-X-180 HTC-X-550 HTC-X-950

285/69.4% 285/82.5% 285/85.1% 285/74.4% 285/78.9% 285/81.4%

286.3/22.5% 286.4/11.6% 286.54/7.3% 286.6/18% 286.3/16.3 286.33/10%

287.9/6.4% 288/3.3% 288/4.2% 288/4.7% 287.9/2.1% 288/3.8%

289.3/1.7% × × 289.44/2.8% 289.4/2.8% 289.5/2.1%

× 290.1/2.6% 290.4/3.4% × × ×

× × × × × 291.1/2.7%

occurs via the formation of furfural and not HMF, leading to a more condensed structure with a slightly more “aromatic” character being obtained upon HTC. This is also confirmed by 13 C NMR spectra of the HTC carbons as described previously.30 The FT-IR spectra of the HTC material treated at 350 °C have very similar features with the HTC materials. This is supported by the TG-IR curve which shows that the main decomposition event starts above this temperature. Upon further thermal treatment to 550 °C, all the characteristic bands corresponding to oxygenated functionalities decrease while the intensity of the CC increases. In addition, bands at 876, 814, and 748 cm−1 corresponding to aromatization (δ(C− H)out of plane) appear,36 due to increasingly condensed arenelike aromatic domains.13 As the temperature advances to 750 °C, the FT-IR spectra become featureless due to the increasingly light absorbing character of the carbon materials.17 At 950 °C, all the spectra are smooth with no characteristic absorption bands.

functional groups can be observed on the HTC carbons. A broad absorption band between 3700 and 3100 cm −1 corresponds to O−H (bonded) stretching vibration. The band at 2925 cm−1 indicates the presence of methylene-type groups (e.g., ν(C−H)stretch). The bands in the region between 1700 and 1600 cm−1 are attributed to CO and CC stretching vibrations, respectively, and support the concept of increasing aromatization of biomass during hydrothermal treatment.7 While the results are similar for all four HTC carbons, independent of the precursors, some differences can be observed in the case of HTC-X, especially with respect to the bands related to “aromatic” character, namely, two weak bands in the 1520−1450 cm−1 region attributed to aromatic ring stretching vibration.18 This is different in the case of the other three HTC carbons, which present only one band each. Another indication for a more “aromatic” character for HTC-X is brought by the presence of the highly intense bands at 880 and 752 cm−1 (δ(C−H)out of plane).36 Xylose as mentioned earlier is a C5 sugar, and therefore, the formation of HTC 12376

dx.doi.org/10.1021/la3024277 | Langmuir 2012, 28, 12373−12383

Langmuir

Article

Figure 3. Zeta potential as a function of pH for the HTC and thermally treated carbons, derived from (a) glucose, (b) sucrose, (c) starch, and (d) xylose.

surface charge of the materials is in all cases negative over all the chosen pH range. This was expected for HTC materials due to the acidic oxygenated groups on the surface.37 It can be clearly observed that the materials obtained after HTC have highly negative zeta potential values similar to those determined for materials treated at 350 °C. Thermal treatment at 550 °C results in significant increase in the zeta potential value, in good agreement with the TGA profile and FT-IR analysis. This is due to the loss of functional groups as described before. This effect is even more pronounced at higher treatment temperatures, although not such great differences in the zeta potential values are noticed between 550 and 950 °C, as the majority of functional groups are lost during this temperature interval. However, the values are still negative due to the acidic character of the carbon materials possessing residual groups such as isolated phenolic hydroxyl groups at carbon plane edges.37 The porosity of carbohydrate-based HTCs has not been systematically investigated, particularly with respect to the impact of subsequent thermal treatment of the base HTC material. Here material porosity was investigated using both N2 (77.3 K) as well as CO2 (273 K) sorption. N2 sorption at 77.3 K can give information about a broad size range of micro- and mesopores.38,39 However, in the case of ultramicroporous materials, the measurement may have kinetic restrictions and could not achieve equilibrium in an acceptable measurement time.40−42 Thus, the measurement may provide erroneous determination of adsorption isotherms.40 This was exactly the case when measuring the porosity of HTC carbons and thermally treated counterparts, using N2 (77.3 K) as an adsorbate. The N2 adsorption isotherms of the base HTC

To gain more detailed information regarding surface functionality as a result of thermal treatment of these synthesized HTC materials, XPS was performed on representative samples prepared from xylose and glucose precursors (Table 2 and Figure S2). The high resolution C 1(s) photoelectron envelope for glucose- and xylose-derived HTC material is characterized by three main contributions; at 285.0 eV (C1, C−C and C−Hx), 286.3/286.6 eV (C2, C−O−H (hydroxyl), C−O−C (ether)), and 287.9/288 eV (C3, CO (carbonyl)) with a minor shoulder at 289.3/289.44 eV (C4, OC−O (acid or ester)). The relatively high intensity of the peaks at C2 and C3 indicates the presence of a considerable amount of oxygenated functionalities mostly related to the furan and carbonyl moieties present in the HTC carbon structure.13 Calcination at 550 °C causes a major reduction of the oxygen related contributions, indicating the loss of such functionalities and a resultant increase of surface hydrophobicity. The residual C3 peaks can be attributed to the presence of plane edge phenolic groups forming during thermal treatment as confirmed by 13C solid state NMR.13 The peak at 290.1 eV (C4(b)) for the glucose-derived carbon is attributed to adsorption of CO2. Increasing the temperature to 950 °C leads to a further loss of residual oxygenated functional groups. In addition, the simultaneous appearance of a peak at 291.1 eV (C5) for xylose-derived carbon, corresponding to π→π* shakeup satellites, suggests the presence of extended pregraphinic polyaromatic domains as the major building unit of the carbon scaffold. The peak at 290.4 eV (C4(b)) for glucose-derived carbon is also attributed to adsorption of CO2. Zeta potential measurements support well the conclusions drawn from the FT-IR and XPS experiments (Figure 3). The 12377

dx.doi.org/10.1021/la3024277 | Langmuir 2012, 28, 12373−12383

Langmuir

Article

Figure 4. N2 (77.3 K) sorption isotherms of HTC and thermally treated carbons, derived from (a) glucose and (b) xylose.

material and samples treated at 350 °C show negligible N2 uptake, indicating the absence of microporosity (Table S1 and Figure 4). On the contrary, CO2 adsorption shows the presence of some microporosity, although still small in scale (Vp of 0.06− 0.08 cm3 g−1 by GCMC (Table 3)). This supports well the fact that the N2 adsorption measurements are kinetically restricted.

diameter of CO2 (DCO2 = 0.33 nm, DN2 = 0.36 nm) and the higher measurement temperature of 273 K (higher thermal energy). Besides, the influence of the high quadrupole moment of CO2 should also be taken into account, as it makes the adsorption isotherm very sensitive to the presence of functional groups and may lead to a higher CO2 uptake than N2 as described later.41,44 All CO2 adsorption isotherms acquired at 273 K and the corresponding GCMC PSDs for the HTC and thermally treated samples show no hysteresis, indicating good adsorption kinetics as compared to N2 adsorption (Figures 5 and 6). Porosity is observed to continuously develop upon increasing the treatment temperature as shown by the continuous increase in gas uptake. This increase is limited for the 180 and 350 °C materials. Again this is in good agreement with TG-IR which shows that the elimination of volatiles initiates at temperatures >350 °C. For the 550 °C materials, a leap in gas uptake is observed, correlated with the elimination of volatile species and the formation of the condensed aromatic intermediates during thermal decomposition.13 Both processes lead to the development of microvoids between the growing aromatic pregraphinic structures. There is another significant increase in gas uptake observed for materials prepared at 750 °C, due to the elimination of all volatile compounds up to that temperature. At higher temperatures, the creation of additional porosity is negligible. PSDs show pore size in the same range (mainly 0.35−0.8 nm). While the amount of ultramicropores (