Carbon nanoparticles preparation and characterization
A suspension of beer wastes particles in aqueous citric acid was used as starting solution for the hydrothermal carbonization process. After reaction, the solid charcoal was separated from a colloidal solution by centrifugation. For analysis purposes, the carbon-based nanoparticles were precipitated upon aggregation by addition of ammonia solution (1 M) up to pH of approximately 9.
Morphological characterization of the nanoparticles
The carbon-based solid and nanoparticles were first observed by scanning electron microscopy and/or transmission electron microscopy in order to determine their morphology. Figure shows the SEM images of the hydrochar produced by the HTC process. It can be seen that the particles are micrometric to millimetric in sizes, highly heterogeneous, and partially nanostructured in surface. This structure is presumably mimicking the one of the biomass before carbonization.
SEM images of the biochar obtained by HTC conversion of beer waste.
In contrast, the solid collected by destabilization of the colloid solutions is composed of agglomerated nanoparticles (Figure ). Figure a,b shows field emission gun-SEM images of the as-obtained solid. The lowest quality of the image Figure b collected at higher magnification is due to the sample preparation procedure that did not contain any metallization step. However, this magnification allows the observation of the particle diameter with an improved accuracy. The nanoparticles exhibit a homogeneous size distribution, between 5 and 9 nm. Figure c,d shows typical TEM images of the nanoparticles. It is interesting to notice that the TEM grids were prepared from ethanol suspension of nanoparticles. The TEM analysis clearly underlines therefore that the agglomeration process obtained by ammonia addition is completely reversible. The morphology of these nanoparticles is very similar to the one reported for the particles obtained by HTC conversion of glucose [10
SEM (a, b) and TEM (c, d) images of carbon-based nanoparticles generated by the HTC process.
The biochar and nanoparticles were analyzed by FTIR spectroscopy. Figure shows typical infrared spectrum of dried biochar. By comparison with references from the literature, different stretching and vibration bands were attributed (see Figure ) [11
]. As a result, the crude biochar is obviously not fully mineralized and contains a large amount of lipid groups and some carbohydrates.
FTIR spectrum of beer-waste-derived biochar obtained by the HTC process.
In contrast, Figure shows a typical FTIR spectrum of nanoparticles. Important differences with the infrared spectrum of the biochar can be noticed. Similar bands have been detected, underlining the common origin of these two products. However, the signals corresponding to the carbohydrates (OH, C-O, and C-O-C vibrations) are significantly more intense in this spectrum. The nanoparticles contain therefore a more important proportion of carbohydrates to lipids than the corresponding biochar. We assume therefore that the fraction of carbohydrates, in water suspension during the HTC process, plays a key role in the formation of the nanoparticles. Further experiments will be conducted in order to collect experimental evidences for confirming or refuting this hypothesis.
FTIR spectrum of beer-waste-derived nanoparticles obtained by the HTC process.
Biochar and nanoparticles were analyzed by Raman spectroscopy. Spectra for polycrystalline graphite usually show a narrow G peak (approximately 1,580 cm-1
) attributed to in-plane vibrations of crystalline graphite, and a smaller D peak (approximately 1,360 cm-1
) attributed to disordered amorphous carbon [11
]. As shown in Figure , the two peaks featuring amorphous carbon (D, 1,360 cm-1
) and crystalline graphite (G, 1,587 cm-1
) are present, but their relative intensity is different than in polycrystalline graphite. This result is in good agreement with works conducted on other nanoshaped carbons like nanopearls [27
] and nanospheres [20
Raman spectrum of biochar produced by the HTC process.
The Raman spectrum recorded for the nanoparticles did not show any peaks. This result was also obtained by other groups on nanoshaped carbons [19
]. It was attributed to the fraction of graphitized carbon inside the nanoparticles which is too low to gain any significant signal. These authors used silver nanoparticles and surface-enhanced Raman scattering effect to overcome this drawback. We had a different approach by carbonizing the nanoparticles under nitrogen up to 1,400°C. The expected effect was to increase the ratio between the graphitized part of the nanoparticles and the non-mineral surface region. The different Raman spectra are presented in Figure . It is important to notice that the same amount of matter was analyzed during these different experiments. It is obvious that an increase of the heating temperature of the nanoparticles induces an improvement in the collected Raman signal. On the spectrum recorded for nanoparticles fired at 1,400°C, the D, G, and D’ bands were clearly identified. The relative ratio between these three peaks clearly shows the large amount of defects in the nanoparticles.
Raman spectra of the nanoparticles, crude sample, and after carbonization under nitrogen up to 1,400°C.
Carbon membrane preparation and characterization
A carbon membrane was elaborated from the as-obtained colloidal solution, using the slip-casting technique. The agglomerated nanoparticle layer formed after deposition on the inner surface of commercial tubular alumina support was heated under argon for 2 h at 1,000°C for consolidation purposes. The formation of the carbon-based membrane was easily and visually detected by the formation of a glossy black inner surface. Figure shows the SEM image of the membrane deposited on the asymmetric alumina support (cross-sectional view). The gray coloration of the alumina below the carbon layer clearly indicates the partial infiltration of colloids inside the support during the slip-casting process. The membrane exhibits a homogeneous thickness of about 50 nm. The surface appears to be rough, remembering its colloidal origin (see also Figure ). Some particles are also observable on the surface of the layer, which were presumably generated upon breaking the membrane and support system.
SEM images of the section (cross-sectional view) of the carbon membrane derived from beer wastes.
SEM images of the membrane surface. These were taken before (a) and after (b) heating up at 200°C during gas permeance measurements.
The N2 adsorption/desorption isotherm was recorded for the membrane and support system (Figure ). For that purpose, the alumina support was sanded in order to reveal the contribution of the carbon layer. This curve clearly shows a hysteresis loop featuring the mesoporosity of the layer. This analysis, in the BET approximation, yields a pore diameter of approximately 3.6 nm (low mesoporosity). However, it is not possible to determine if this measured porosity is only due to the presence of the porous carbon membrane or partially due to the residual alumina support not totally discarded by sanding. We decided therefore to conduct dynamic water and gas separation measurements.
N2adsorption/desorption isotherm of the HTC-processed carbon membrane.
For a further dynamic characterization of the carbon membrane, water permeability has been measured by recording the water flux through the membrane as a function of the applied nitrogen pressure on the feed solution at room temperature. Figure a shows the water flux through the commercial alumina support as a function of the applied pressure, in the range of 3–15 bars. As expected, we obtained an almost linear evolution in which values are in good agreement with the ones reported by the manufacturer. In Figure b, the water flux through the carbon membrane deposited on alumina nanofiltration support is evidenced.
Water flux as a function of the applied pressure for the different membranes. (a) The starting alumina nanofiltration membrane and (b) the carbon membranes.
As illustrated in Figure b, no water flux was measured with carbon membranes below 6 bar of applied nitrogen pressure. The measured permeability is 0.005 L h-1·m-2·bar-1, a value which is 1,000 lower than the commercial alumina system. This result can be interpreted as a very low porous volume accessible inside the membrane. Therefore, this membrane cannot be used for water filtration applications. The quasi-dense behavior of the carbon membrane for low applied external pressure emboldens us to evaluate this material for gas separation.
From the 1970s, carbon membranes have been extensively used for gas separation [6
]. Different studies were conducted on membranes originating from different sources such as polymers and carbohydrates (glucose) and have demonstrated promising permeance values in the range of 10-8
, associated with high selectivity. For instance, a carbon membrane elaborated by pyrolysis of commercial polymers and having a pore diameter between 3 and 5 Å has demonstrated a He/CO2
selectivity of 4, and a He/N2
selectivity between 20 and 40 [9
]. In our case, the gas separation test was driven using three types of gases, namely helium (He, kinetic diameter = 2.6 Å), carbon dioxide (CO2
, kinetic diameter = 3.3 Å), and nitrogen (N2
, kinetic diameter = 3.64 Å). The permeances of these gases were recorded as a function of the pressure at different temperatures of 25°C (T01), 100°C, and after cooling down again to 25°C (T02) (Figure ). At 25°C, the membrane gave a 10-9
permeance value for He, CO2
, and N2
, which is in good agreement with the values reported in the literature [6
]. At 100°C, a stable flux was obtained exhibiting a permeance in the range of 10-7
. We also observed an increase in the permeance while increasing the temperature up to 100°C whatever the used gas and the applied pressure were [26
]. We assume that this result may reflect a Knudsen diffusion mechanism for the gas separation. For CO2
, these systems enter into an apparent stationary regime by varying the pressure, and their permeances appear to become almost constant whatever the applied pressure was. In contrast, the permeance of helium increases with the applied pressure (lower kinetic diameter). As a consequence, the selectivity of helium versus the other gases increases with the pressure up to approximately 2 (Figure ). This value is lower than the one reported in the literature [6
Permeances of (a) helium, (b) nitrogen, (c) carbon dioxide as a function of the differential pressure. These were taken at different temperatures: 25°C (T01), 100°C, and 25°C after an exposure of up to 100°C (T02).
He/CO2selectivity (a) and He/N2selectivity (b) as a function of the applied pressure at 100°C.
After measurement at 100°C, the membrane was cooled down to 25°C, and its permeance was measured again for each gas (Figure ). By comparing T01 and T02, we have observed a significant increase of the permeances (by a 102 factor) whatever the studied gas was. By considering this result, we underwent measurements at 200°C. At this temperature, it was not possible to obtain a stabilized flux for any studied gas and therefore to measure any permeance value. After cooling down to 25°C, we have measured again the permeance using helium (Figure ). As illustrated by this figure, the permeance of the carbon membrane towards helium is increased after the membrane was exposed to higher operating-temperature conditions. Our assumption is that the membrane underwent a microstructural evolution during the high-temperature measurement. In order to confirm the latter, the membrane surface was analyzed by SEM after the experiment, done at 200°C (Figure ). We can clearly conclude from the images of Figure that the surface of the membrane underwent a microstructural evolution upon heating which yielded to an increase of its surface roughness. Fracture surface view analysis did not reveal any significant evolution of the membrane thickness.
Permeances of helium at different temperatures using the same membrane. Permeances at 25°C (T01), at 25°C but after an exposure at 100°C (T02), and the same membrane after an exposure at 200°C (T03).