Valorization of oat hulls and spent coffee grounds into pyridinium-modified granular adsorbents for water purification

To obtain granular adsorbents with 50% biomass by dry weight, 10 g of biomass (50%; Oh or SCG) were mixed with 8 g chitosan (40%) and 2 g kaolinite (10%). Approximately 40 ml of 0.2 M acetic acid was added and mixed until a uniform paste was obtained. The paste was spread across small pistol primer trays (S&B; Czech Republic) until all spots were filled and the excess was removed. The pellets were removed after drying for 24 h at 22 °C.

SCG, kaolinite, and chitosan were used as received without further grinding. Particle size ranges (based on dry weight; wt.%) for Chi were 88.6 wt.% between 425 µm and 150 µm, 7.5 wt.% between 150 µm and 125 µm and 3.9 wt.% between 125 µm and 75 µm. For SCG, 20.2 wt.% was > 425 µm, 77.0 wt.% ranged between 425 µm and 150 µm, 2.8 wt.% ranged between 150 µm and 125 µm. K had 70.3 wt.% that ranged between 425 µm and 150 µm, 16.5 wt.% between 150 µm and 125 µm, and 13.1 wt.% between 125 µm and 75 µm. Oh had 65 wt.% that range between 425 µm and 150 µm and 35 wt.% was < 150 µm.

The prepared pellets (10 ± 0.1 g) of the prepared, unmodified pellets (same for Oh and SCG pellets), were added to 250 ml 0.5 M NaOH with 1% (2.5 ml) ECH and mildly agitated for 24 h at 22 °C. The pellets were washed multiple times and soaked in Millipore water until a neutral pH was obtained. The pellets were tamped dry via a paper towel to remove excess moisture and employed for the next step without further drying.

To obtain pyridinium-modified pellets, 10 ± 0.1 g of the CL pellets (SCG50-CL and Oh50-CL; tamped dry) were added to a 500 ml round bottom flask with condenser and 300 ml 50% ethanol solution was added. 1 ml formic acid and 25 ml furfural were added to the mixture and refluxed for 3 d. Then, the pellets were neutralized first in sodium bicarbonate solution and further washed until a neutral pH was reached by washing with copious amounts of water. Next, the pellets (Oh50-Pyridinium; SCG50-Pyridinium) were dried for 48 h at 22 °C.

To enable comparison for NMR and thermogravimetry analysis (TGA) studies only, only, 6-8 pellets (Oh and SCG) were combined with the same concentration ratio together with furfural and HCOOH, which sat for 3 d without heat and similarly washed (Oh50-Fu; SCG-Fu).

¹³C solids NMR spectra were obtained with a 4 mm DOTY cross polarization with magic angle spinning (CP-MAS) probe with a Bruker AVANCE III HD spectrometer operating at 125.77 MHz (¹H frequency at 500.13 MHz). The ¹³C spectra under CP with total suppression of spinning sidebands (CP/TOSS) were obtained at a sample spinning speed of 7.5 kHz, a ¹H 90° pulse (5 µs), and contact time (2.0 ms) with a ramp pulse on the ¹H channel. Spectral acquisition required 2500 scans with a 1 s recycle delay, along with a 50 kHz SPINAL-64 decoupling sequence. ¹³C NMR chemical shifts were referenced to adamantane at 38.48 ppm (low field signal).

The weight loss profiles were obtained with a Q50 TA Instruments thermogravimetry analyzer (TA Instruments, USA). The samples were placed in an open aluminum pan under N₂ atmosphere, equilibrated at 30 °C for 1 min and then heated to 500 °C at a fixed rate (10 °C min^-1).

¹³C solids NMR spectral results provide structural insight on the short-range order and local environment of the ¹³C nuclei of carbonaceous materials. The effect of pyridinium-modification is shown in figure 1.

Zoom In Zoom Out Reset image size

In both cases, the polysaccharide fingerprint region covers the 60–110 ppm region, while signatures around 20–30 ppm and 175 ppm stem from the acetyl group of chitosan [32]. SCG without modification shows signals between 160–115 ppm and near 170 ppm, ascribed to components such as lignin, tannins, etc [48, 49]. To assign ¹³C signatures attributed to pyridinium modification and ligneous constituents in Oh/SCG composites, we compared imine-linked Oh50-Furfuryl/SCG50-Furfuryl and pyridinium-modified systems, along with SCG50-pyridinium and Oh50-Pyridinium samples. As noted in figure 1, aromatic and ligneous signatures occur between 120–160 ppm in Oh (A) and SCG (B). This fraction is removed upon washing in NaOH (aq) and subsequent crosslinking, whereas the signals between 145–160 ppm indicate an imine-linked furfuryl moiety. Additional signals near 135 ppm in both figure 1(A) and 1(B) reveal incomplete conversion (ca. 20%–30%) into pyridinium moieties incorporated into both structures with significant furfuryl-content, according to ¹³C NMR estimates. However, it should be noted that this is a provisional estimate due to the nature of the experimental conditions.

TGA and the resulting decomposition profiles enable the study of structural changes within composite materials at variable temperature, as evidenced in figure 2. In figure 2(A), pristine Oh observes a distinct decomposition event (maximum 280 °C) which overlaps with a second decomposition event (maximum 340 °C). The first event with an onset above 200 °C can be ascribed to hemicellulose, while the second decomposition event was assigned to cellulose [50]. Furfuryl- and pyridinium modification appears to slightly shift the decomposition events to higher temperatures, which can be attributed to introduction of arene or unsaturated moieties with double-bond functionality versus the role of ECH-crosslinking [32]. On the other hand (figure 2(B)), SCG reveals its first decomposition event near 330 °C (360 °C and 410 °C for the maximum of the second and third event). This indicated the presence of hemicellulose, cellulose, and arene/ligneous compounds with increased thermal stability [49, 51]. The absence or decrease of the event around 410 °C in the modified composites indicates removal of lignin via the NaOH-washing step during the crosslinking step. Crosslinking did not yield any notable increase to the thermal stability of the composite materials, as compared to the unmodified SCG50 or SCG50-Fu. The aromatic and imine-linked moieties (pyridinium, furfuryl) may be inferred through presence of the second event (with an event near 360 °C) for the composites, as compared to non-modified composites, which concurs with results from a previous study [32].

Zoom In Zoom Out Reset image size

The solvent uptake evidenced by a weight increase of both cyclohexane and water for both granular systems gave information about the hydrophile–lipophile balance of these adsorbents. The relative solvent uptake was evaluated and compared to the dry weight of the pellets (see figure 3).

Zoom In Zoom Out Reset image size

The cyclohexane uptake is less than 16% compared to doubling the weight of both pellet systems through sorption-based water uptake. Thus, both composites are similarly polar, whereas Oh50-Py also shows slightly greater apolar character, as evidenced by greater cyclohexane uptake. This may be due to a reduced accessibility of the arene-moieties in SCG50-Py, in conjunction with potentially lower pyridinium-content (see figure 1).

Chitosan as a binder and crucial component in all pellet systems requires acetic acid to facilitate its dissolution in water via protonation to enable formation of the pelletized adsorbents. In turn, protonated amino sites can adsorb MO via electrostatic interactions. In turn, neutralization via NaOH during the surface modification and crosslinking results in deprotonation, which favors MB adsorption. The role of the synthetic conditions on the dye adsorption properties was investigated. A comparison with controls (non-neutralized and only NaOH-washed granular adsorbents with and without waste biomass) is illustrated in figure 4.

Zoom In Zoom Out Reset image size

The Sips isotherm model (also known as Langmuir–Freundlich hybrid model) was chosen to investigate the dye adsorption properties. The Sips model combines features of the Freundlich and Langmuir models that serve to address the limitations of either respective model. The Freundlich isotherm model is valid at low concentration, which reveals a continual rise in adsorbed quantity at high concentrations and accounts for multilayer adsorption onto heterogeneous sites. In contrast, the Langmuir model assumes adsorption onto homogeneous sites of constant energy in a monolayer profile. The Sips isotherm combines features of both isotherm models whilst accounting for a breadth of adsorption properties. The case for the heterogeneity factor (n) that equals 1, the Sips isotherm converges to the Langmuir model behavior for the case of monolayer adsorption onto homogeneous sites [52, 53].

The neutralized (OH) pellet systems chitosan-kaolinite pellets (CPK-OH), Oh50-OH, and SCG50-OH do not show any notable MO uptake. Similarly, protonation with acetic acid attenuates the MB uptake of CP and CPK (cf. table 3 for the adsorption capacities all pellet systems with MO). Refer to figure 5 for a visual comparison of selected pellet systems after MO and MB adsorption. With reference to the influence of NaOH-washing (neutralization in 0.5 M NaOH), a previous report revealed a maximum adsorption capacity towards MB of 54 mg g⁻¹ and 83 mg g⁻¹ for SCG40 and SCG60, and 32 mg g⁻¹ and 34 mg g⁻¹ for Oh40 and Oh60 were reached [32]. We posit that electrostatic interactions govern the binding mechanism for MO uptake. By comparison, MB shows binding affinity toward ligneous and arene components in Oh and SCG result in minor uptake through H-bonding and cation-π interactions. Visual comparison of the pellet systems prior to and after exposure to dye solutions with variable concentration reveal that variable swelling effects of the granular composites occur after 24 h of immersion. As well, we observe a relative color change based on the relative dye uptake for each dye concentration. A notable feature is the cracking and loss of shape of the unmodified Oh50 systems during the adsorption process, as compared to other granules (cf. Figure 5). A previous study of the mechanical properties of granules evaluated the hardness (N) of CL and non- CL Oh and SCG adsorbents without pyridinium modification. The assumption is that pyridinium-modification does not appreciably increase the hardness or the Young’s moduli. Further, it is assumed that the values for the surface modified pellet systems are intermediate to that of the reported adsorbents in table 1 [37]. An appreciable difference relates to lower levels of NaOH (0.5 M) for the preparation of the CL adsorbents herein.

Zoom In Zoom Out Reset image size

A brief confirmation of the applicability of the hardness estimates from table 1 to this study are evidenced by SCG50-Py as the selected material, which yielded a hardness of 123 N ± 33 N. The trend is comparable to SCG60-CL (124 N ± 35 N) and validates the assumption that the hardness values in table 1 can be extrapolated to pyridinium-modified pellets. In general, crosslinking was required to prepare stable and durable Oh-derived pellets, while SCG-derived adsorbents did not benefit from crosslinking. Without crosslinking, the hardness of Oh-derived pellets was at a maximum value of 76 N (20% Oh content). The hardness decreased with greater Oh-content, whereas pellets with 80% Oh-content were unstable in water. The role of hydration and shear stress in water may affect these adsorbents in differently than their hardness properties in the dry state. Hence, the hardness was measured for pellets in the wet state due to the relevant application of the pellets in aqueous media, where such stability counts (see table 2).

As indicated, Oh-based pellets observe the lowest hardness in their hydrated state with ca. 2.5 N, whereas crosslinking increased their hardness to ca. 10.5 N (hydrated). In contrast, SCG50-Py shows a hardness of 16.7 N in contrast to 11.5 N for SCG50 in the hydrated state, which exceed the hardness of Oh-based composites. This trend supports that Oh-based adsorbents require crosslinking for increased stability in solution, further necessitating the role of surface modification to enhance dye adsorption towards MO and MB.

These trends suggest that future research requires a more comprehensive study of the mechanical properties under variable (dry vs. wet) conditions to establish the utility of these adsorbents for practical applications.

In general, while CPK showed an adsorption capacity with MO of 485 mg g⁻¹ (similar to CP), unmodified agro- and food-waste-based pellet systems (Oh50 and SCG50) with 40% chitosan content showed 324 mg g⁻¹ and 364 mg g⁻¹ uptake (table 3). By contrast, the pyridinium-modified pellet systems showed similar dye uptake (comparable within the margin of error) between 74–121 mg g⁻¹ at equilibrium. A heterogeneity factor (n ∼ 1) is noted for the unmodified pellet systems, whereas a lower value (n = 0.5) is noted for SCG50-Py and Oh50-Py is indicative of their heterogeneous adsorption sites.

By contrast, markedly lower adsorption capacities were observed for MB (see table 4) as compared with MO. While positive charges were crucial in enabling MO uptake, MB was not favorably adsorbed onto these pellet systems. MB uptake was contingent on neutralization to avoid electrostatic repulsion, indicating that electrostatic interactions govern the adsorption process for non-modified pellet systems.

Although the adsorption capacities are an order of magnitude lower for MB vs. MO, the pyridinium-modified Oh50-Py system show measureable improvement over the unmodified systems (K-value ∼1; high adsorption capacity at low concentrations). Also, while Oh and SCG enable MB dye adsorption, SCG50-Py present a heterogeneity factor of 0.6 in contrast to Oh50-Py (∼1). This trend indicates the role of homogeneous adsorption sites, which are contrasted with the heterogeneous adsorption sites of SCG-derived adsorbents.

For SCG50-Py, the furfuryl-pyridinium modification slightly enhances the uptake of MB, while enabling appreciable MO dye uptake after NaOH-washing and crosslinking. However, SCG50 without any treatment can drastically exceed MO uptake, while still exhibiting MB adsorption. For Oh50-Py, the modification enhances the MB uptake capacity compared to Oh50, which also affords comparable MO uptake (cf. Table 5 for a survey of the literature).

Notably, while most adsorbents (except the nano-fiber-pads) are in powdered form, SCG50-Py and Oh50-Py are in granular form. Hence, lower adsorption capacities are expected for granular systems, while also providing an option for research under dynamic (column) conditions by offsetting backpressure and other concerns. Yetgin and Amlani⁴⁵ reported the effect of granular vs. powdered adsorbents for powdered and washed (but otherwise non-modified) wheat hulls, rice hulls, and garlic stalks for the uptake of MB. The corresponding uptake capacities between 55–370 mg g⁻¹ were reported (lowest for rice hull, highest for garlic stalk) [54]. As a further point of comparison, Gaber et al investigated methylene blue removal via palm peat (particle size range 1–8 mm) and noted MB removal at 24.5 mg g⁻¹ with an initial concentration of ca. 35 mg l⁻¹ [10]. Additionally, considering the inherently low stability in the wet state of Oh50 compared with Oh50-Py, SCG50 and SCG50-Py, chemical modification (crosslinking and pyridinium modification) is necessary to develop Oh-composites for practical applications in environmental remediation. For regeneration experiments, regenerant screening with methanol and 1 M NaCl solution revealed that more MO was desorbed in 1 M NaCl compared to MeOH. This trend is in contrast to MB regeneration, which showed more desorbed dye in MeOH solution than 1 M NaCl. Subsequently, MeOH was selected for MB and 1 M NaCl selected for MO (see figure 6).

Zoom In Zoom Out Reset image size

While MB appears to have reasonable regeneration capabilities over 3 cycles for SCG-Py, the removal efficiency drops to ca. 70% compared to the initial removal for Oh50-Py prior to resuming to the original value. In contrast, both pellet systems dropped to 10% of the initial removal of MO after the third cycle. To enable improved regeneration, the solvent regenerant and time sequence of the process require additional optimization.

As noted above, the focus of this study was on the role of the pyridinium-modification of the pellet surface, the conducted isotherm studies are limited based on the overall insight into the adsorption process. Further insight about the adsorption process is envisaged through a systematic study directed toward the dye adsorption kinetics. Future dynamic adsorption studies are anticipated to support practical field applications, evaluation of the adsorbent performance, and potential for improved adsorbent regeneration.

The granular adsorbents prepared herein were crosslinked to obtain more stable pellets for greater utility in practical column-based applications, in contrast to powdered adsorbents. Based on the raw material inputs outlined in a previous study [32], a cursory cost-benefit analysis is summarized in table 6.

By contrast, CPK costs approx. US$30 per kg adsorbent for chemical inputs without further modification, [32] whereas agro- and food-biomass incorporation drastically reduce the cost for the adsorbent preparation. Looking forward, we recommend that future studies assess the role of adsorbent regeneration and re-use to evaluate the adsorbent and the operational costs. Although 20 g batches are feasible for pellet preparation via the suggested methodology, 10 g batch size are favored to enable uniform paste consistency. However, scale-up and modified methodology for extrusion seems feasible for large batch systems. In turn, the chemical surface modification herein relies on widely available industrial chemicals (NaOH, EtOH, furfural, epichlorohydrin, formic acid), where further process optimization is possible.