Influence of Mineral Composition of Chars Derived by Hydrothermal Carbonization on Sorption Behavior of CO2, CH4, and O2.

The doping of SiO2 and Fe2O3 into hydrochars that were produced by the hydrothermal carbonization of cellulose was studied with respect to its impact on the resulting surface characteristics and sorption behavior of CO2, CH4, and O2. During pyrolysis, the structural order of the Fe-doped char changed, as the fraction of highly ordered domains increased, which was not observed for the undoped and Si-doped chars. The Si doping had no apparent influence on the oxidation temperature of the hydrochar in contrast to the Fe-doped char where the oxidation temperature was reduced because of the catalytic effect of Fe. Both dopants reduced the micro-, meso- and macroporous surface areas of the chars, although the Fe-doped chars had larger meso- and macroporosity than the Si-doped char. However, the increased degree in the structural order of the carbon matrix of the Fe-doped char reduced its microporosity relative to the Si-doped char. The adsorption of CO2 and CH4 on the chars at temperatures between 273.15 and 423.15 K and at pressures up to 115 kPa was slightly inhibited by the Si doping but strongly suppressed by the Fe doping. For O2, however, the Si doping promoted the observed adsorption capacity, while Fe doping also showed an inhibiting effect.

Introduction

Hydrothermal carbonization (HTC) chars, which are pyrolyzed hydrochars derived via hydrothermal carbonization under high pressure from lignocellulosic biomass and water, have aroused interest in many fields of science and application over recent years.[1] One promising research focus is the use of hydrochars as a fuel in biomass-fired power generation or as a possible substitute in coal-fired power plants[2,3] because such an application could help meet increasing global energy demand while reducing associated CO2 emissions. After pyrolysis, the resulting HTC chars possess large surface areas, which make them also a promising adsorbent material for subsequent use in the removal of environmental pollutants from flue gases or waste water.[4]

The elemental composition of solid biofuels can differ a lot. Depending on different raw plants, woods, or agricultural wastes, the mineral content of, e.g., iron, silica, or potassium can vary significantly, which can have a catalytic effect on morphology, pyrolysis, gasification, and combustion.[5−8] Trubetskaya et al.[6] have shown for two biomasses with different silica and potassium content that the high silica content of rice husk leads to a preserved shape of the char particles during pyrolysis, whereby the low silica and high potassium content of wheat straw lead to a broader deviation of particle shapes. The higher potassium content of the wheat straw also resulted in a higher reactivity of the char, which was also confirmed by various studies for different biomasses.[5,9,10] Khelfa et al.[7] have shown the catalytic effect of Fe2O3 on pyrolysis and gasification of miscanthus chars. For these studies, the catalytic effect was shown using different natural biomasses with different mineral compositions[6] by the addition of further mineral particles to the biomass particles[7] or by acid leaching of most of the original mineral content with additional doping of the investigated mineral.[5,8−10] Therefore, interactions between the different mineral components or influences of the acid leaching cannot be distinguished clearly.

Due to the rudimental composition of the hydrochar derived via HTC from pure cellulose, interactions between different minerals can be excluded. By doping minerals into the hydrochar matrix, the nature and the arrangement of active sites on the surface of the char can be modified, which leads to different combustion and adsorption properties. The doping of iron oxide into the hydrochar results as well in a faster oxidation of the char due to the catalytic effects of iron,[11−13] whereas the doping of silica does not show a significant effect on the oxidation rate.[12] As diffusion and adsorption precede oxidation, this observation suggests that such doping might similarly influence these mass transport mechanisms. Any differences in the diffusion and adsorption of gas molecules in the porous structure of the char would also lead to changed reaction rates during combustion and gasification given the importance of the these mass transport mechanisms to such reactions.[14] Several studies investigated the adsorption behavior of HTC chars, originating from different biomass materials such as different wood and agricultural residues.[15−19] Since the biomasses consist of various amounts of cellulose, minerals, and functional groups, a closer look on those components is herein expedient as well. Very few studies focusing on the influence of the mineral content or the functional groups of the hydrochars on the adsorption properties have been carried out. Yu et al. investigated the influence of the iron content of hydrochars on the removal of estrogen, and Lei et al. studied the influence of the nitrogen-containing functional groups on the removal of chromium, both from waste water.[20,21] To the best of our knowledge, no study exists on the influence of the iron or silica content on the adsorption of CO2, CH4, or O2.

Therefore, a novel approach towards the influence of iron and silica doping on the adsorption behavior of CO2, CH4, and O2 is presented in this study, followed by a comprehensive estimation of the volumetric adsorption measurement uncertainty. Changes in morphology and the amount of active surface sites due to mineral doping and pyrolysis of the hydrochars lead to different adsorption behavior of the chars. Since the adsorbed phase is a precursor for the following solid–gas interactions during pyrolysis, gasification, and combustion, the investigation of the influence of the mineral composition leads to a better understanding of the process.

Results and Discussion

Characterization of the HTC Chars

Elemental Mapping

Electron microscopy imaging and energy-dispersive elemental mapping of the doped HTC chars were performed to obtain information about the mode of incorporation of the dopants and are presented in Figures and 2. According to the micrographs, the particles consist of agglomerates of spheres differing in size. Elemental mapping of the Si-doped (see Figure b–d) and Fe-doped (see Figure b–d) HTC char reveals that the dopant particles are in immediate vicinity of the carbonaceous spheres. Thus, we assume that the dopant particles are incorporated into the hydrochar during the hydrothermal carbonization process and remain in contact with the char matrix after pyrolysis. For the Fe-doped char, elemental mapping shows a higher dispersion of the dopant in the char, which is presumably due to the lower particle size of the added Fe2O3 particles compared to those of SiO2.

Figure 1

Scanning transmission electron microscopy (STEM) micrograph (a) and elemental mapping with respect to carbon (b) and silicon (c) for two different particles of the Si-doped char. The overlay (d) is based on the individual distributions of carbon and silicon.

Figure 2

STEM micrograph (a) and elemental mapping with respect to carbon (b) and iron (c) for two different particles of the Fe-doped char. The overlay (d) is based on the individual distributions of carbon and iron.

X-ray Diffraction (XRD)

To identify the phases of organic and inorganic matter in the HTC chars, XRD patterns were recorded and are shown in Figure . Also shown are the peak locations corresponding to the reference patterns of graphite, quartz, iron, and cohenite.

Figure 3

XRD patterns of the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC char (reference patterns: graphite: 01-075-1621; SiO2: 01-070-7344; Fe0: 03-065-4899; Fe3C: 01-089-2005).

In the XRD pattern of the undoped char, no sharp reflections can be observed in the range 2θ = 10–80°, suggesting that this carbon material is X-ray amorphous. TEM imaging of the char verifies a low degree of structural order (see Figure S1 in the Supporting Information). Thus, pyrolysis up to T = 1073 K does not generate a high fraction of highly ordered domains in the undoped HTC char. The XRD pattern of the Si-doped HTC char contains several sharp reflections, which can be assigned to quartz (SiO2). This indicates that the dopant maintains its phase composition during the hydrothermal synthesis and the subsequent pyrolysis. In contrast, a phase transition occurs for the Fe-doped char: carbothermal reduction of the Fe2O3 dopant during the thermal treatment results in metallic iron (Fe0) and iron carbide (Fe3C), as revealed by the detail investigations shown previously.[22] This is possibly accompanied by an increase in the structural order of the carbonaceous matter, as indicated by the reflection at 2θ = 26° referring to graphitic carbon. Given that any structural order is much less pronounced for the undoped and the Si-doped char, the diffraction pattern and the TEM image (see Figure S1 in the Supporting Information) reveal that the in situ addition of the iron dopant introduces structural order to the char by enabling far more highly ordered domains than observed in the other chars.

Surface Area Characteristics and Pore Size Distribution (PSD)

For all three HTC chars, the meso- to macroporous surface area was investigated by conducting a BET analysis of sorption isotherms measured for N2 at T = 77.36 K (Figure ). For all HTC chars, a pronounced increase in the amount of N2 adsorbed is observed for very low relative pressures, which gives a first hint of the more microporous carbon structures. For p/p0, in the range 0.1–0.8, the doped chars exhibit reduced adsorption capacity, suggesting a low fraction of small mesopores relative to the undoped char. Stronger adsorption occurs as the saturation pressure is approached, with capillary condensation evident in large mesopores and macropores.

Figure 4

N2 physisorption isotherms at T = 77.36 K of the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC chars. Hollow symbols correspond to adsorption data, while filled symbols correspond to desorption data.

Although the shape of the sorption isotherms of the undoped and Si-doped char are similar, the amount of N2 adsorbed on the latter is significantly lower. This is reflected by the Brunauer–Emmett–Teller (BET) surface areas of the chars, which amounted to 416, 350, and 310 m2/g for the undoped, Fe-doped, and Si-doped HTC char, respectively (see Table ). In contrast to the undoped and the Si-doped chars, the adsorption and desorption isotherms of the Fe-doped char exhibit a more pronounced hysteresis. On the basis of previous studies,[22] this feature is presumably caused by the desorption of the adsorbate from cavities in the carbonaceous material produced by the structural changes occurring during pyrolysis. The increase for all three samples at very small pore sizes below 2 nm indicates a pronounced micropore structure.

Table 1

Comparison of Surface Areas of the Undoped (HTC-800), the Si-Doped (Si-HTC-800) and the Fe-Doped (Fe-HTC-800) HTC Chars Determined by N2 Adsorption at T = 77.36 K (BET) and by CO2 Adsorption at T = 273.15 K (DA and BET)

	HTC-800	Si-HTC-800	Fe-HTC-800
SN2,BET [m2/g]	416	310	350
SCO2,DA [m2/g]	784	666	502
SCO2,BET [m2/g]	357	341	236

To investigate the micropore structure (<2 nm), CO2 adsorption isotherm data at T = 273.15 K (see Figure ) were used to determine a micropore surface area of each char (see Table ) via the Dubinin–Astakhov equation, as extended by the equation of Medek.[23,24] In general, the micropore surface area of all three chars is significantly larger than the BET surface area. We note that comparisons of the absolute values obtained from the models should be treated with caution, although differences in both SN and SCO are likely indicative of the HTC chars having different pore structures. The micropore surface area of the Si-doped char (666 m2/g) is much larger than that of the Fe-doped char (502 m2/g), which possibly reflects the reduction in microporosity caused by the generation of highly ordered domains induced by iron doping. In a few literature studies,[25,26] the BET theory has also been applied to the CO2 isotherm data to calculate another estimate of the microporous surface area, SCO. This is done here, and the results presented in Table also indicate that Fe-HTC-800 has the smallest surface area.

Figure 7

Adsorption loadings of CO2 on the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC chars at different temperatures. (a) T = 273.15 K; (b) T = 298.15 K; (c) T = 323.15 K; (d) T = 373.15 K; and (e) T = 423.15 K. The isotherms for the undoped HTC char at T = 273.15 and 298.15 K were reported previously.[27]

By conducting NLDFT calculations using the data from the initial slope of the N2 adsorption isotherm, the pore size distribution (PSD) of the three chars for the micropore regime was calculated (see Figure ). For all three chars, the regularization parameter of λ = 1.625 was applied based on the optimized result of the L-curve for the Si-doped HTC char. For the undoped and Si-doped char, a bimodal distribution was obtained, with a sharp peak at around 0.4 nm for both and a broader peak at around 0.9 nm for the undoped and around 1.2 nm for the Si-doped char. Having in mind that N2 is kinetically hindered at cryogenic temperatures to diffuse into the narrow micropores, the absolute values of the pore size distribution have to be considered carefully but give at least a hint of the distribution of the micropores. The larger peak areas of the undoped char show a more pronounced microporous structure than those of the Si-doped chars. A unimodal distribution with a broad peak around 0.8 nm was obtained for the Fe-doped char, resulting in a less pronounced microporosity for the Fe-doped char. These observations confirm the results of the microporous surface area SCO. The unimodality also indicates a different structural order of the Fe-doped char, as described in Section .

Figure 5

Pore size distributions obtained by nonlinear density functional theory (NLDFT) analysis of the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC char.

Oxidation Behavior

The stability of the HTC chars can be determined by temperature-programmed oxidation (TPO), as the conversion of carbon materials by oxidation usually occurs at lower temperatures than required for gasification. The differential thermogravimetric (DTG) curves shown in Figure were calculated by evaluating the first derivative of the sample mass changes measured during the TPO experiments.

Figure 6

DTG curves during TPO in synthetic air in a thermobalance of the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC char.

The DTG profiles of the three HTC chars exhibit one broad signal in the temperature range between 673.15 and 923.15 K, corresponding to a single-step mass loss. The oxidation temperature corresponding to the maximum of the DTG peak was T = 865.15 K for the undoped char and was not affected by the addition of SiO2. The decreased height of the DTG signal obtained for the Si-doped char is caused by the lower relative fraction of oxidizable carbonaceous material present relative to that in the same initial amount of undoped char. Doping with iron produced a slight decrease of the char oxidation temperature, which can be attributed to the catalytic effect of the dopant. Nevertheless, the TPO and DTG data reveal that all three HTC chars are stable at the temperatures at which the sorption of CO2, CH4, and O2 were measured.

Sorption Equilibria of CO2, CH4, and O2

The adsorption loadings of CO2 on the three different chars were measured at five temperatures between 273.15 and 423.15 K and are shown in Figure . For each temperature, similar trends can be observed. The largest CO2 loadings were measured at p = 115 kPa for the undoped char (3.46 mmol/g at T = 273.15 K to 0.38 mmol/g at T = 423.15 K), followed by the Si-doped char (3.06 mmol/g at T = 273.15 K to 0.34 mmol/g at T = 423.15 K) and the Fe-doped char (2.24 mmol/g at T = 273.15 K to 0.24 mmol/g at T = 423.15 K).

The large adsorption loading of the undoped char can be explained by the fact that it had the largest BET surface area (416 m2/g) and/or as a result of the reduction in attractive forces between the CO2 and carbonaceous material caused by the doped minerals. However, the lower CO2 loading of the Fe-doped char relative to the Si-doped char cannot be related to their BET surface areas of 350 and 310 m2/g, respectively. However, the different microporous pore structure of the Fe-doped char (see Section ) resulting from the enhanced structural order likely impedes the CO2 adsorption process. A similar trend for the different chars was obtained for the adsorption measurements with CH4 at the four temperatures between 273.15 and 373.15 K, although the adsorption loadings were considerably lower than for CO2 (see Figure ). At 1.80, 1.62, and 1.15 mmol/g for the undoped, Si-doped, and Fe-doped chars, respectively, the resulting adsorption loadings at p = 115 kPa and T = 273.15 are all about 48% lower than that for CO2. This trend simply reflects the generally lower adsorption affinity of CH4 on any material (see here for another example[28]). The difference in adsorption loadings for the three chars can also be explained in terms of the changes in microporosity resulting from the doping process, since the adsorption of CH4 is related to the microporosity as well.[29] However, a directly inhibiting effect of the doped minerals is also possible.

Figure 8

Adsorption loadings of CH4 on the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC chars at different temperatures. (a) T = 273.15 K; (b) T = 298.15 K; (c) T = 323.15 K; (d) T = 373.15 K. The isotherms for the undoped HTC char at 273.15 and 298.15 K were reported previously.[27]

Figure shows the adsorption of O2 on the chars at T = 273.15 and 298.15 K, which exhibit a different trend than that observed for CO2 and CH4. At T = 273.15 K, the largest O2 loadings were measured for the Si-doped char (0.81 mmol/g), followed by the undoped char (0.78 mmol/g) and the Fe-doped char (0.47 mmol/g): this sequence cannot be explained by either the microporosity or the meso- to macroporosity of the samples. Furthermore, for the Si-doped char, a repeatable desorption hysteresis was observed for the lowest pressures at both temperatures, which indicates a behavior other than regular physisorption. To explain this behavior, it is helpful to consider the XRD patterns of the Si-doped char, which show the presence of quartz. At the temperature of the doping process of T = 473.15 K, the crystalline structure of the doped SiO2 is α-quartz.[30] Above a temperature of T = 846 K, a displacive transformation between α-quartz and β-quartz occurs, whereby one of the tetrahedra rotates slightly with a resulting decrease in density of SiO2.[30] During the pyrolysis of the Si-doped char at a temperature of 1073 K, the quartz structure should thus change to β-quartz but change back to α-quartz during the subsequent cooling. Some parts of the crystalline structure may be preserved in a so-called paramorph state of the β-quartz. The chemisorption of O2 on β-quartz has been reported,[31] so we hypothesize that hysteresis observed in this work may reflect the presence of this β-quartz paramorph in the Si-doped char. However, we have no proof for this hypothesis and further verification is necessary but during several adsorption cycles with the same sample, this desorption hysteresis was found to be reproducible.

Figure 9

Adsorption and desorption loadings of O2 on the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC chars at (a) T = 273.15 K and (b) T = 298.15 K. Adsorption plotted by empty and desorption by filled markers.

To summarize the inhibiting effect of the iron doping on the char adsorption behavior, loadings for each of the three gases at T = 298.15 K with regard to the absolute mass of carbon in the chars are shown in Figure . For CO2 and CH4, the adsorption loadings on the undoped and the Si-doped chars are essentially the same, while the adsorption on the Fe-doped char is much lower, reflecting a clear inhibiting effect by the dopant. For the case of O2, the promoting effect of silica can also be observed.

Figure 10

Adsorption loadings of the three gases at T = 298.15 K for the undoped (HTC-800), Si-doped (Si-HTC-800), and Fe-doped (Fe-HTC-800) HTC char.

Conclusions

To understand the catalytic influence of minerals in chars, a combined approach to investigate changes of char morphology and adsorption behavior of mineral-doped HTC chars is presented. Different characteristics (XRD, BET, DA, and TPO) and the adsorption behavior (CO2, CH4, and O2 at temperatures between T = 273.15 and 423.15 K) of an undoped, a Si-doped, and a Fe-doped char were investigated. The XRD patterns and TEM images show that the Fe doping increases the structural ordering of the char during pyrolysis while the Si doping had no apparent effect on the char structure. BET analysis (N2 adsorption at T = 77.36 K) and Dubinin–Astakhov analysis (CO2 adsorption at T = 273.15 K) showed that the doped chars had smaller macro- and micropore surface areas than that of the undoped char. While the Fe-doped char had a larger macropore surface area than that of the Si-doped char, the opposite was true for the micropore surface areas as a result of the higher fraction of highly ordered domains in the Fe-doped char.

For the first time, the influence of mineral doping on the gas adsorption behavior of HTC chars was investigated. The highest adsorption loadings for CO2 and CH4 were observed for the undoped char, reflecting its higher microporosity, followed by the Si-doped and Fe-doped chars. For O2 adsorption, the highest adsorption affinity was observed for the Si-doped char, which may be explained by chemisorption of O2 on a paramorph of the doped quartz. Due to less adsorption of all three analysis gases on the Fe-doped char, it can be concluded that Fe doping and the resulting increase in structural order during subsequent pyrolysis leads to less accessible adsorption sites on the surface of the char. The catalytic effect of iron on char conversion is therefore not a consequence of a higher concentration of adsorption sites on the surface. For future research work, the effect of the mineral influence on the rate of pore diffusion has also to be taken into account to reach a complete understanding of the influence on mass transfer.

Experimental Methods

Synthesis of Hydrochars

Doped synthetic chars were prepared by hydrothermal carbonization (HTC) using cellulose as a precursor followed by pyrolysis of the obtained hydrochars.[32] In situ doping with SiO2 or Fe2O3 imitated the incorporation of minerals into coal during the natural coalification process.[12]

For the hydrothermal synthesis, a 50 mL polytetrafluoroethylene inliner of a stainless steel autoclave was applied to the reactor. For the undoped hydrochar, 6 g of α-cellulose (Sigma-Aldrich, BioReagent) was suspended within 30 mL of deionized H2O. In situ doping was performed by the addition of either 5 wt % Fe2O3 (Alfa Aesar, 20–40 nm) or 5 wt % SiO2 (Riedel-de Haën, 40–63 μm) to the cellulose dispersion. The loaded autoclave was heated in a convection oven to T = 473.15 K, which generated an autogenous pressure of about 1.55 MPa inside the autoclave. After 24 h, the autoclave was cooled to room temperature. Subsequently, the yielded suspension was filtered and the resulting solid was washed with deionized H2O until a neutral pH was reached. The obtained hydrochar was dried in a convection oven at T = 378.15 K for 24 h. Afterwards, the hydrochar was loaded into a quartz boat, which was placed in a horizontal oven and pyrolyzed in a N2 flow (Air Liquide; purity: 99.999 mol %) by heating at a linear rate of 5 K/min to T = 1073.15 K, which was then held for 2 h. The resulting undoped, Si-doped, and Fe-doped chars are labeled HTC-800, Si-HTC-800, and Fe-HTC-800, respectively.

An overview of the composition of the synthesized and pyrolyzed chars is presented in Table . Pyrolysis causes the release of volatiles, resulting in an increase of the carbon content accompanied by a decrease of the H and O content in the remaining char. For the undoped char, the degree of carbonization was higher than 90 wt %. To specify the extent to which the added dopants were incorporated, the actual loadings are reported as a weight percentage of the particular metal component in the char. The doped amounts correspond to 2.7 wt % Si content for the Si-doped char and 3.8 wt % Fe content for the Fe-doped char. A proximate analysis revealed that the undoped char contained no mineral components in contrast to the doped chars.

Table 2

Overview of Characterization of the Undoped (HTC-800), Si-Doped (Si-HTC-800), and Fe-Doped (Fe-HTC-800) HTC Char by Elemental Analysis, AAS, and Proximate Analysis

	HTC-800	Si-HTC-800	Fe-HTC-800
C [wt %]	94.1	84.7	90.6
H [wt %]	0.6	0.8	0.6
O [wt %]	5.3	11.9	5.0
Fe [wt %]			3.8
Si [wt %]		2.7
volatiles [wt %]	4	4	4
ash [wt %]	0	8	7

Characterization of Chars

The prepared chars were characterized with respect to composition, structural properties, and reactivity in an oxidative atmosphere. A vario Micro cube (Elementar Analysensysteme) was used for elemental (CHNS) analysis of the chars. The amount of Si and Fe included in the Si-doped and the Fe-doped char, respectively, was determined by atomic absorption spectroscopy (AAS). For electron microscopy imaging, the doped HTC chars were first suspended in a mixture of ethanol and water and then dispersed in an ultrasonic bath. Afterwards, the samples were placed on an Au grid with a lacey carbon film. Scanning transmission electron microscopy (STEM) and energy-dispersive X-ray (EDX) spectroscopy were performed using a JEOL JEM-2800 microscope. The X-ray diffraction patterns in the 2θ range of 5–80° for Cu Kα radiation were recorded by an Empyrean diffractometer (PANalytical) at the Department of Mineralogy at Ruhr University Bochum. Phase identification was based on reference patterns by the International Center for Diffraction Data (ICDD), a HighScore Plus software (Malvern Panalytical). The volatile and ash fractions in the chars were derived from thermogravimetric measurements using a magnetic-suspension balance (TA Instruments, previously Rubotherm) with a coupled quadrupole mass spectrometer (Thermostar, Pfeiffer Vacuum). For proximate analysis, about 30 mg of the carbon materials was heated under a He flow of 100 cm3/min (Air Liquide; purity: 99.999 mol %) to a temperature of 1173.15 K with a heating rate of 10 K/min followed by a treatment at T = 1088.15 K in a 100 cm3/min flow of 20/80 vol % O2/He (Air Liquide; purities: 99.995, 99.999 mol %). The oxygen content of the chars was calculated by subtraction: O (wt %) = 100 (wt %) – C (wt %) – H (wt %) – Si (wt %) – Fe (wt %). The influence of dopants on the stability of the chars was evaluated by temperature-programmed oxidation (TPO) experiments in the magnetic-suspension balance. During TPO, about 30 mg of the chars was heated at 5 K/min to T = 1173.15 K in a 100 cm3/min flow of 20/80 vol % O2/He.

To determine the meso- to macroporous surface area according to the theory of Brunauer, Emmett, and Teller (BET),[33] N2 (Air Liquide; purity: 99.999 mol %) adsorption measurements at T = 77.36 K were carried out in a BELSORP-mini (BEL Japan). Before conducting the adsorption measurements, the samples were degassed for 6 h at a temperature of T = 473.15 K. However, at cryogenic temperatures, the diffusion of N2 into the HTC char’s narrow micropores is highly restricted and a quantitative analysis of the sample’s microporosity cannot be often achieved by such measurements.[34,35] Instead, using CO2 as the adsorbate at T = 273.15 K, this kinetic restriction can be avoided, and given the saturation pressure for CO2 at this temperature is 3.4851 MPa,[36] the initial stage of physisorption can be studied at moderate pressures in the range 0.1–100 kPa.[34] However, as CO2 has a quadrupole moment that can interact with functional groups on the char’s surface, this could in principle influence the pore filling. Nevertheless, most carbonaceous surfaces in general do not contain as many functional groups as, e.g., zeolites or MOFs. Furthermore, Lotz et al. observed a strong loss of functional groups on the hydrochars during pyrolysis. Accordingly, CO2 was used in this work to analyze the microporous structure of the chars[22,34] using the Dubinin–Astakhov theory with the extended equation by Medek.[23,24] For the CO2 sorption experiments, the volumetric system described in Section was used.

The N2 adsorption data was also used to calculate the pore size distribution by nonlinear density functional theory (NLDFT) calculations. The 2D-NLDFT model for porous carbons with heterogeneous surfaces was used,[37,38] which is implemented in the analysis software SAIEUS by Micromeritics.

Sorption Experiments with CO2, CH4, and O2

For the adsorption investigations, a volumetric measurement system (model ASAP 2020 by Micromeritics) was used, which was slightly modified from its previous configurations, as shown in Figure .[39−43] The temperature of the solid sample in the measurement cell (marked green in Figure ) was controlled by a homemade heating jacket filled with a thermostated heat transfer oil. Temperature and pressure in the manifold volume (marked red in Figure ) were measured continuously using transducers with full scales and uncertainties shown in Figure . After degassing the solid sample at T = 473.15 K, the manifold volume was filled with gas up to a certain pressure (e.g., 5, 10 kPa). Subsequently, valve V13 was opened, allowing gas to flow into the measuring cell and the pressure change associated with adsorption to be determined; after reaching a steady state pressure, the process was repeated multiple times up the isotherm. CO2, O2, CH4, and He were supplied by Coregas with a purity of 99.995 mol %.

Figure 11

Schematic of the volumetric measurement system.

Experimental Procedure

For clarity and to promote a better understanding of the following uncertainty estimation, we briefly present the working equations of the volumetric sorption measurement system. The adsorbed amount at equilibrium is obtained by calculating the difference between the dosed amount ndosed and the amount of gas in the free space nFS inside the measuring volume.The cumulative nature of the measurement requires that the amounts dosed during previous pressure steps have to be accounted for when evaluating ndosed,. The amount previously dosed is incremented by the difference between the initial and equilibrium amounts of gas inside the manifold volume VM (marked red in Figure ). The real gas behavior is considered using the nonideality factor α (see eq ).Before starting the adsorption measurement, the free gas volume in the measuring volume (marked green in Figure ) is determined twice by helium pycnometry, once at the measurement temperature set by the bath (VF,bath, Tbath) and once at room temperature (VF,room, Troom).The standard temperature is defined at TSTD = 273.15 K and the nonideality factor is related to the compressibility factor Z, as given in eq , which depends on the actual temperature and pressure of the gas.[44] The values[45,46] for the nonideality factor still widely used in sorption science are often traceable to the suggestions of Emmett and Brunauer[47] from 1937. In this work, the far more accurate reference equations of state for CO2,[36] He,[48] CH4,[49] O2,[50] and N2[51] were used to calculate Z for different temperatures, using the software package Trend 4.0.[52] Hence, the calculated values for Z and α are listed in Tables and 4, respectively. These tables also indicate at which temperature the measurements with the different gases were conducted. CO2 was measured from T = 273.15 K up to T = 423.15 K, while CH4 and O2 were only measured up to T = 373.15 and T = 298.15 K, respectively. At higher temperatures, the experimental loadings of CH4 and O2 were so small that they could not be reproduced properly.The adsorption capacity can finally be obtained through the ratio of the adsorbed amount of molecules and the mass of the sample loaded into the measuring cell.

Table 3

Compressibility Factor Z Calculated for T and pequi = 0.1 MPa

	Z
	77.36 K	273.15 K	298.15 K	323.15 K	373.15 K	423.15 K
CO2		0.99335	0.99502	0.99620	0.99768	0.99853
He		1.00053	1.00048	1.00044	1.00037	1.00032
CH4		0.99764	0.99827	0.99873	0.99932
O2		0.99904	0.99936
N2	0.95747

Table 4

Nonideality Factor α for T and pequi = 0.1 MPaa

	α (10–9/Pa)
	77.36 K	273.15 K	298.15 K	323.15 K	373.15 K	423.15 K
CO2		–66.522	–49.793	–38.069	–23.249	–14.699
He		5.2618	4.7785	4.3688	3.7138	3.2151
CH4		–23.561	–17.299	–12.737	–6.7668
O2		–9.5601	–6.3854
N2	–425.29

Note that the ASAP 2020 device requires a different definition of α in 1/Torr.

Estimation of the Volumetric Adsorption Measurement Uncertainty

The measurement uncertainty was estimated according to the “Guide of the Expression of Uncertainty in Measurements”, abbreviated as GUM.[53] The combined standard uncertainty uc of the adsorption capacity q can be calculated by eq for every single pressure step. Due to the cumulative nature of the measurements, the uncertainties of all previous pressure steps j have to be considered for the uncertainty of the pressure step i (j defined from 1 to i). In Table , the standard uncertainties of the different contributions are listed. The standard uncertainties of the pressure and temperature measurements as well of the sample mass determination are taken from the manufacturer specifications. For the volumes, no specifications were provided by the manufacturer of the volumetric measurement system and so we conservatively estimated these values. The sensitivity coefficients ∂q/∂x were determined by conducting a sensitivity analysis, whereby the influence of the standard uncertainties on the adsorption loading q was evaluated. Using a coverage factor of k = 2, the combined expanded uncertainty Uc(q) in determination of adsorption capacity can be calculated; values are indicated in the figures of Section , and the numerical values are provided as state point uncertainties in the tables of the Supporting Information.

Table 5

Uncertainty Budget for the Relative Combined Standard Uncertainty in Adsorption Loading uc(q)/qa

uncertainty contribution	standard uncertainty	contribution to uc(qi)/qi
equilibrium pressure pequi,i	1.5 × 10–5 p	2.34 × 10–3
initial pressure pinitial,i	1.5 × 10–5 p	2.21 × 10–3
manifold volume VM	1 × 10–3 VMb	1.45 × 10–6
free volume at room temperature VF,room	2 × 10–2 VF,roomc	9.70 × 10–8
free volume at bath temperature VF,bath	2 × 10–2 VF,bathc	9.35 × 10–3
bath temperature Tbath	0.5 K	9.53 × 10–10
room temperature Troom	0.5 K	1.49 × 10–9
manifold temperature TM	0.02 K	2.94 × 10–4
compressibility factor Z	(0.02–0.1) × 10–3Zd	3.94 × 10–9
mass of sample msample	0.0005 g	5.57 × 10–6
uc(qi)/qi		1.42 × 10–2

As an example, the contributions to uc(q)/q were calculated for the adsorption of CO2 on Si-HTC-800 at T = 373.15 K and p = 110.4 kPa.

Manifold volume was provided by the manufacturer but not the standard uncertainty. Therefore, standard uncertainty was conservatively estimated.

Free volumes were determined by helium pycnometry before the adsorption measurement started. Since the apparatus routine does not provide the values to recalculate those volumes, standard uncertainties were conservatively estimated.

The uncertainty depends on the used EOS for the different fluids.[36,48−51]