Solid-state NMR (ssNMR; see Figure A2
, Appendix A
) investigations of SCD and PSC50 reveal the drastic change in material and structural properties upon pyrolysis. The comparison of 13
C CP NMR spectra of the same material before (SCD, Figure A2
a) and after pyrolysis (PSC50, Figure A2
b) shows the influence of the temperature treatment. In the spectrum of raw SCD, well-resolved signals corresponding to all organic matter present in SCD such as cellulose, hemicellulose, and lignin, as well as proteins and lipids, can be observed [36
]. Carbons of different polysaccharides such as cellulose and hemicellulose cause the intense and narrow signals between 60 and 110 ppm (Figure A2
a). In the aliphatic region between 20 and 40 ppm, signals of lipids and protein side-chains appear. The broad signals between 120 and 160 ppm are probably due to lignin aromatic carbons, and the signal at 173 ppm can be assigned to carbonyl groups of lignin, hemicellulose, and proteins.
In contrast, after pyrolysis of this material, a broad signal at 128 ppm dominates the NMR spectrum. This signal can be assigned to highly condensed aromatic carbon atoms from carbon black. Furthermore, a shoulder at ca. 150 ppm is caused by phenolic carbons [41
]. Very weak signals in the aliphatic region (20–40 ppm) are due to residual side-chains (Figure A2
b). Solid-state NMR spectroscopy, therefore, clearly shows that the vast majority of organic building blocks in the raw SCD are broken down during pyrolysis and form condensed aromatic carbons afterward.
3.1. Activated Carbons from Chemically Pretreated Spent Coffee
a shows a representative thermogravimetric analysis (TGA) curve of SCD. A small step (5.7% weight loss) is visible at around 100 °C. This step is likely due to evaporation of residual water. A second and rather broad step (70.9% weight loss) is observed between 200 and 450 °C. This step can be assigned to the slow decomposition of hemicellulose, cellulose, and lignin (in that order), the main components of spent coffee [35
]. Subsequently, the TGA data indicate a continuous weight loss until 1000 °C showing the decomposition of the sample [40
b shows representative X-ray diffraction (XRD) data of selected powders. XRD patterns of dried spent coffee (SCD) show two broad halos along with a series of reflections sitting on these halos (Figure 1
b). Such a pattern is typical for cellulosic materials [44
]. After pyrolysis, the sharp diffraction rays are not observed anymore and only two broad halos remain. These reflections are typical for graphitic carbon-type materials and can be assigned to the carbon (002) and (100) planes [35
]. Occasionally, the XRD patterns show a few remains of the sharp reflections, but this is not always the case.
shows the infrared (IR) spectra of SCD and of selected treated materials. The IR spectrum of the starting material SCD shows strong bands between 1000 and 500 cm−1
and a series of weaker bands at around 1700 and 2900 cm−1
. After the pretreatment with NaOH and before pyrolysis, the bands in the fingerprint area are weaker than before. The bands at around 2900 cm−1
are still visible, and additional new bands between 1700 and 1000 cm−1
and at ca. 3400 cm−1
appear (Figure 2
a). All bands can be assigned to the cellulose and lignin present in the materials [35
] (Table 3
After pyrolysis (Figure 2
b) the bands between 1500 and 1000 cm−1
are broader and show a significantly increased intensity, while bands between 1000 and 500 cm−1
lose some intensity. No further changes can be observed in the IR data. While the ssNMR spectra of pure pyrolyzed SC (Figure A2
) shows that the organic material was reduced to highly condensed aromatic carbon atoms, a treatment with NaOH beforehand seems to leave some functional groups on the material surface, which can be confirmed via ATR-IR.
shows scanning electron microscopy (SEM) images of SCD after different treatments. The SCD material is rather heterogeneous in terms of both the particle size and the particle shape. There are elongated and rather thin particles with lengths reaching up to 100 µm, along with rounded particles, and numerous aggregated particles of various sizes and shapes. The surfaces of these particles are smooth and without any holes (Figure 3
a). Pyrolysis of pure SCD at 500 °C has no obvious effect on the surface of the resulting carbons (Figure 3
Pretreatment with 2 M aqueous NaOH at 90 °C leads to quite drastic changes in the morphology of the pyrolyzed products (Figure 3
c). This process produces very small and round particles with diameters of about 0.6 µm with no visible pores on the surface (on the SEM length scale). Pretreatments with 0.2 M NaOH (Figure A3
) lead to different effects after pyrolysis at 500 °C. The material now has clear edges and rougher surfaces on the particles including plenty of pores with maximum diameters of ca. 2.2 µm.
The nitrogen sorption isotherm of PNa90 (2 M) is displayed in Figure 4
. The shape of the isotherm suggests that PNa90 (2 M) likely has a fairly large pore size distribution. The surface area calculated from the data is 24 m2
/g. This is much smaller than the surface areas typically observed in commercial AC (up to 1500 m2
/g). Similarly, the pore volume of 0.033 cm3
/g is also much lower than commercial AC [1
]. The pore diameter of the material is 9.9 nm.
3.2. Activated Carbons from Spent Coffee/Calcium Carbonate Mixtures
An alternative approach toward carbonaceous adsorbents using solid CaCO3
as activator was also explored. In this series, the SC powder was thoroughly mixed mechanically with dry calcium carbonate powder in a 1:1 (w
) proportion. Pyrolysis was then done between 500 and 850 °C, i.e., below and above the point of decomposition of calcium carbonate at 825 °C. Table 2
shows the results from elemental analysis (EA) of these materials after washing. The washing process removes the CaCO3
and leaves the pure carbonaceous material behind, as evidenced from the EA data.
The IR spectra of these materials are shown in Figure 2
b. Treatment temperatures up to 600 °C produce materials that exhibit IR spectra very similar to the spectra shown in Figure 2
a. However, higher treatment temperatures of the SCD/calcium carbonate mixtures produce materials where the typical IR bands disappear and the corresponding IR spectra only show a straight line. This indicates that the vast majority of the surface groups vanishes during the heat treatment. Similar to the broadening effects observed in the IR spectra of PNa90 (2 M), a band broadening between 1500 and 1000 cm−1
is noticeable for SCC50 and SCC60 (Figure 2
b). The spectrum of SCC65 shows an additional broad band around 3300 cm−1
which can be assigned to adsorbed water. However, as the pyrolysis temperature is increased, the resulting materials exhibit IR spectra where the bands around 1000 cm−1
are drastically reduced in intensity.
shows representative SEM images of the SCC materials after washing out the calcium carbonate with aqueous hydrochloric acid. SEM clearly shows that the treatment temperature does not significantly affect the sample morphology. However, SEM also demonstrates that the calcium carbonate present in the reaction mixture leads to drastically different morphologies when compared, e.g., to the materials pretreated with NaOH (Figure 3
The presence of the calcium carbonate leads to the formation of heavily interconnected flake-like carbons with a sponge-like appearance. The individual flakes are rather thin but are densely aggregated and tightly connected to neighboring flakes. This flake-based assembly produces a wide range of macropores that are clearly visible in the SEM (Figure 5
). SEM also shows that the large, aggregated particles have a wide range of sizes and shapes, yet their architecture on the micrometer level is rather uniform and always constructed from the flaky building blocks just described.
The SEM images show no significant differences in the feature sizes or shapes vs. increasing carbonization temperatures. All powders contain particles with a broad size distribution and a variety of particle shapes (Figure 5
). The mixture of SCD and calcium carbonate produces materials with a much finer substructure than the powders PNa90 (2 M) described above. The particles appear more “porous” (or appear to have a much rougher surface) and seem to have also broken down into smaller particles. Some particles show holes and larger open areas.
shows the corresponding nitrogen sorption data obtained from the SCC materials. All data indicate the presence of mesopores with a rather broad size distribution. Higher pyrolysis temperatures favor the formation of materials with higher surface areas with the highest surface area of ca. 166 m2
/g obtained for the materials that were pyrolyzed at 850 °C. At the same time, the pore sizes decreases. Table 4
summarizes these results. Again, even the materials with the highest porosities are far from commercial AC in terms of both specific surface area and pore volume.
In summary, both groups of materials, PNa90 and SCC, show drastic property changes in comparison to the regular SC and its pyrolyzed form. A pyrolysis temperature of 500 °C is sufficient to break down the organic (carbohydrate, lipid, and protein) components to aromatic carbon (ssNMR, XRD), but does not lead to a complete removal of surface functional groups (ATR-IR). With increasing pyrolysis temperatures, more surface functional groups are removed from the materials, leaving behind an essentially carbon-only material. Both PNa90 and SCC materials exhibit an increased surface area compared to the original spent coffee (SEM, nitrogen sorption data). However, the surfaces obtained from BET analysis along with the corresponding total pore volume are still significantly smaller than those of commercial AC [1
Clearly, the process of material synthesis is simple and effective; therefore, the materials were also investigated for the adsorption capabilities for dye removal from aqueous solution. This is a model case, and the removal of other substances can of course also be considered.
3.3. Dye Removal
Dye adsorption experiments were done with methylene blue (MB) and methyl orange (MO). MO and MB are common model compounds for dyestuff-polluted aqueous media because they are (1) water-soluble and (2) easy to detect and quantify via UV/Vis spectroscopy. As some dyes are health hazards [50
], MO and MB are suitable model cases for dye-contaminated water.
a shows the UV/Vis spectra of MB and MO solutions with 23 mg/L dye concentration along with the spectrum of a 1:1 MO/MB mixture. According to these spectra, the adsorption maxima of 664 nm for MB and 461 nm for MO were chosen for the comparison of dye uptake by the materials.
In a first study, PNa90 (2 M), which was washed neutral, and all as-prepared SCC materials were used for MB and MO removal in batches exposed to the solutions for 3 h. The SCC50 material (i.e., the material pyrolyzed at the lowest temperature of 500 °C) only removes below 20% of dye and was, therefore, not considered any further. It is noteworthy, however, that the non-pyrolyzed SCD raw material shows a remarkable adsorption capability for MB while it adsorbs nearly no MO. Similarly, PNa90 (2 M) shows a preferred adsorption of MB, but also takes up ca. 60% of the MO present in the solution. All other materials show a near-quantitative removal of both MO and MB (Figure 7
shows the adsorption isotherms (dye removal vs. treatment time at room temperature) for washed PNa90 (2 M) and SCC materials. Figure 8
a shows the data for dye adsorption on PNa90 (2 M). Clearly, MB removal is faster and more effective. Already after 1 min, MB is removed almost completely from the aqueous phase. Quantitative (i.e., >99%) removal is achieved after 5 min and no re-release of MB into the solution is observed. In contrast, MO is taken up somewhat less effectively and the uptake is not as rapid. After an initial, rather fast uptake reaching ca. 80% MO removal, there is a second, much slower uptake regime starting at ca. 60 min. This slower uptake continues until the end of the experiment, but only contributes to a further uptake of ca. 6% after 48 h, resulting in an overall removal of around 86% of MO at the end of the process.
At this point, it is worthwhile to note that a significant hypsochromic shift is observed in the UV/Vis spectra of MO for longer adsorption times (Figure 8
b). Already after 24 h, the spectra begin to show a peak broadening with a shoulder toward shorter wavelengths. This is further pronounced at longer treatment times of 48 h. Longer adsorption experiments running for 1 and 2 weeks show an even stronger change in the absorption bands, making a true quantification of the data obtained at long adsorption times difficult.
The reason for this hypsochromic shift is an increase in the pH value of the MO solution over time. As MO is a pH indicator, it reacts to this change in proton concentration via a color change. Indeed, measurements using a pH electrode show that the pH of the PNa90/MO mixed system starts out at pH 5 and increases to well above 7 after 24 h. This also indicates that, despite washing the material with 2 M HCl after pyrolysis, some NaOH remains in the material. The longer the material remains in the aqueous solution, the more NaOH seems to be released from the sample. This observation suggests that the current, prototype materials may need to be subjected to an improved cleaning process before application. On the other hand, such long batch exposure times are unrealistic for a larger-scale application, and we, therefore, performed further experiments at much shorter times.
c,d show the adsorption isotherms obtained from the analogous experiments with the SCC materials obtained after removing the calcium carbonate. All SCC materials are effective adsorbents for MO and MB but there are differences. In the case of MB (Figure 8
c), two different types of behavior can be distinguished: (1) SCC60 and SCC75 show a somewhat slower adsorption kinetics, and the final fraction of adsorbed dye is lower (around 98%) than in the case of SCC80 and SCC85. In contrast, (2) adsorption on SCC80 and SCC85 is very rapid and essentially reaches a quantitative removal of the dyes within 20 min.
A similar observation can be made for MO (Figure 8
d). While the dye removal process was rather rapid and efficient, the SCC60 and SCC70 materials are again somewhat less effective than SCC80 and SCC85. In contrast to the data shown in Figure 7
, these data show a slightly lower adsorption at around 60 min before returning to the very high fractions of removed dye at the end of the treatment. The reason for this slight “dip” is not clear at the moment and will be a matter of future research with materials improved from the current, first-generation adsorbents.
Overall, the SCC60 and SCC75 materials show a roughly quadratic increase in adsorption up to nearly 100%, resulting in a plateau. SCC80 and SCC85 essentially take up 100% of both dyes immediately after exposition, i.e., full dye removal is realized in less than 5 min. No significant adsorption/desorption fluctuation is noticeable in these materials.
These findings are especially interesting when compared to the analytical data described above. While PNa90 (2 M), SCC60, and SCC65 still show a significant presence of surface groups in the IR spectra (Figure 2
), SCC75 through SCC85 show no more signals for any kind of surface groups. Hence, the observed dye adsorption can not be correlated to ionic interactions between material surfaces and type of dye. Rather, we currently speculate that the main interaction could be a van der Waals type interaction between the dye molecules and the adsorbent surface.
In addition to kinetic experiments, an evaluation of the effect of adsorbent dose was carried out with one of the best-performing materials, SCC85 (Figure 9
). Overall, the data show that SCC85 is a highly effective adsorbent for both dyes with ≥90% of MO and MB taken up even at very low adsorbent doses of only 5 mg. Moreover, exposure to mixed solutions containing both dyes simultaneously has no impact on the adsorption capacity of the material. Again, all solutions are completely clear at the end of the treatment. In spite of this, MO adsorption is slightly faster at very low adsorbent doses of 1 and 2 mg. These two experiments also show a slight tint of the glassware, indicating that, at very low dosages, the sample environment may also have an influence, similar to a recent study on MO and MB uptake using a hydrogel/3D printed hybrid setup [50
]. Adsorbent doses above 2 mg do not exhibit this behavior, indicating that the adsorbent dominates the removal process as intended.