Nanocellulose, a naturally occurring biopolymer consisting of β-1,4-Danhydroglucopyranose monomer units, 1 is derived from macrocellulose via chemical treatment, 2 sonication, 3 mechanical milling, 4 or enzymatic digestion. 5 Nanocellulose possesses desirable mechanical properties (e.g., Young's modulus, tensile strength) comparable to Kevlar and steel. 6,7 These mechanical properties, along with the nanoscale width, natural abundance, biodegradability, and biocompatibility of nanocellulose, elevate its use in a variety of applications, including polymer nanocomposites, 1,2,8 biomedicine, 9 sensors, 9,10 water treatment, 9,11 coatings, 12 and smart materials. [13][14][15] Many applications of nanocellulose require hydrophobic surface modification (e.g., coating, functionalization) to improve its dispersion in organic media and reduce hydrogen bond-induced homoaggregation prior to use in material applications. 10,[16][17][18] Roughly 35 million tons of nanocellulose are produced globally each year, and production is projected to further increase by 2030. 19,20 Release of nanocellulose composite materials into the environment is therefore inevitable, necessitating understanding the effect of surface modification on its microbial mineralization.
Products featuring cellulosic materials often advertise the complete biodegradability (i.e., mineralization of carbon into CO2 and/or CH4) of cellulose as an attractive feature compared to traditional carbon-based reinforcement options such as carbon nanotubes and carbon fibers due to reduced environmental persistence and impact. [21][22][23][24] The biodegradation of cellulosic materials proceeds through different mechanisms and microorganisms in anaerobic and aerobic environments. Specifically, in aerobic environments, cellulose is generally degraded by cellulase and β-glucosidase enzymes secreted by bacteria and fungi. Cellulases initiate degradation of the cellulose structure by hydrolyzing internal bonds (endoglucanases) and chain-ends (cellobiohydrolases) to yield cellobiose molecules. 25 β-Glucosidase then concludes the depolymerization process by converting cellobiose into glucose which is mineralized to CO2 and water by aerobic microorganisms. 25 In contrast, anaerobic microorganisms utilize cellulosomes or multiprotein complexes of enzymes to mineralize cellulose into water and biogas consisting of CO2 and CH4. [25][26][27] This conversion to biogas is achieved at over 80% efficiency for cellulose, demonstrating the high biodegradability of the material. 28 Furthermore, while a single aerobic microorganism species is sufficient to fully mineralize cellulose (e.g., the fungus Trichoderma reesei), multiple anaerobe species are required to work in concert to produce the enzymes necessary for conversion of cellulose to biogas. 26 Examples of bacterial phyla responsible for cellulose degradation include Acidobacteria and Firmicutes. 29,30 While native cellulose is readily fully biodegraded (mineralized), hydrophobic modifications have the potential to interfere with the enzymatic degradation of macrocellulose, 31- 33 a process which depends on the composition and activity of the microbial community involved (e.g., aerobic vs. anaerobic). 23,25,26,29,30,34 In a previous report, we demonstrated that this interference held true for cellulose nanofibrils (CNFs) modified with hydrophobic siloxane coatings, which blocked enzymatic access to nanocellulose and inhibited its anaerobic mineralization. 16 In contrast, covalent functionalization strategies utilizing ether, ester, and urethane linkages avoid the formation of surface coatings, and have been widely applied to macrocellulose and nanocellulose to improve dispersion in organic media and polymers, [35][36][37][38] yet their impact on nanocellulose biodegradation has not been previously investigated. Results from previous studies lead to the expectation that for functionalized nanocellulose, the rate limiting step in biodegradation will involve the removal of functional groups (e.g., ester groups by hydrolysis) to regenerate the functionalizing reagent and the hydroxyl groups present in native cellulose. 32, 39- 41 After this initial cleavage, the biodegradation pathway of functionalized cellulose proceeds through biodegradation of the native cellulosic component and the functionalizing reagent.
The most commonly used metric to express the extent of covalent functionalization of cellulose is the degree of substitution (DS), representing the average number of cellulosic hydroxyl groups functionalized per anhydroglucose monomer unit (DS = 0-3). The conventionally determined DS value represents the extent of functionalization of both the surface and bulk regions of the material and can therefore be regarded as a measure of the overall DS (DSoverall). For cellulosic materials, the DSoverall is generally determined using elemental analysis 42 or nuclear magnetic resonance (NMR) spectroscopy. 43,44 For covalently modified macrocellulose, biodegradability depends on both DSoverall and the nature of the chemical linkage (i.e., ether, [45][46][47] ester 31,32,48 ). For example, degradation of macrocellulose fibers functionalized with carboxymethyl groups (ether linkage) by a cellulolytic enzyme complex decreased as the DSoverall increased from 0.41 to 1.30. 46 Furthermore, nanocellulose esterified with acetyl groups to DSoverall >1.25 exhibited significant inhibition of anaerobic biodegradation as compared to un-modified macrocellulose. 48 Past studies typically quantified the extent of modified macrocellulose biodegradation in terms of the production of low molecular mass byproducts (e.g., cellobiose, glycolic acid) rather than evolution of CO2 or CH4. Furthermore, many studies on modified macrocellulose employed model enzymes (e.g., cellulase, esterase) or a single microbial species to effect biodegradation. 46,47,49 While the information from these studies is useful in identifying trends in biodegradation as a function of material properties, such approaches do not measure complete biodegradation of the cellulosic material and do not represent the microbial communities encountered in natural environments. 23,24,34,46,47,49 Failure to discern the complete mineralization of functionalized cellulose has led to disagreement with respect to the degree of inhibition resulting from chemical functionalization. 31,33,45 Additionally, as mineralization of cellulosic materials generally proceeds via the cooperation of a microbial community, 27 more complex systems utilizing environmentally relevant microorganisms are best suited for assessing the environmental persistence of functionalized nanocellulose, rather than model enzymes or pure microbial cultures. 23,34 One established method of measuring the anaerobic biodegradation of cellulose involves the quantification of biogas produced during the mineralization of carbon into CO2 and CH4. 28,31,48,50 The aerobic biodegradation of cellulose is typically quantified using mass loss measurements to compare the amount of carbon converted from the solid (i.e., cellulosic) phase into the gas phase (i.e., CO2). 31,51,52 Another potentially important factor to consider is that the extent to which the biodegradation of a functionalized nanomaterial is inhibited may be more closely linked to the degree of functionalization of the surface (DSsurface) than to DSoverall. This distinction is important as the preliminary step in the biodegradation of a solid-phase material involves the adsorption and colonization of microorganisms at the surface. [53][54][55][56] In the case of cellulosic materials, this initial biodegradation step requires biofilm formation or the interaction of highly specific microbesecreted cellulosome complexes with its surface. 26,[57][58][59] As nanocellulose fibers are composed of numerous cellulose chains woven together into a nano-scale cord, the chains at the fiber surface are distinct from those within the bulk of the material. 2,60,61 During chemical functionalization with liquid reagents, cellulose chains in both the bulk and surface of nanocellulose are targeted, 2, 62, 63 while gas phase reagents selectively functionalize the nanocellulose surface due to their inability to penetrate into the bulk of the material. [64][65][66] Despite the potential for achieving different levels of surface vs. bulk functionalization, studies of cellulosic materials typically use only bulk-sensitive analytical techniques (e.g., NMR), and thus quantify only DSoverall. 42,44,[67][68][69][70] The effect of surface substitution is likely to be particularly important for the biodegradation of CNFs compared to macrocellulose due to the large surface area-to-volume ratio of nanocellulose as well as the decreased swelling capacity of CNFs which limits access to bulk cellulose chains. 2 In this study, we compare the influence of surface vs. bulk functionalization as well as the influence of different covalent linkages on CNF mineralization by aerobic and anaerobic microbial communities. This study is the first to investigate the biodegradability of a range of functionalized nanocellulose in both aerobic and anaerobic environments. Selective functionalization of the surface and bulk regions was accomplished using liquid-phase and gas-phase (i.e., surfacespecific) 64,65 techniques to esterify nanocellulose with long-chain hydrocarbons that are often used to improve CNF dispersion in polymer nanocomposites. 63,71 Attenuated total internal reflectance Fourier-transform infrared spectroscopy (ATR-FTIR), solid-state 13 C-nuclear magnetic resonance spectroscopy ( 13 C-NMR), and CHN elemental analysis were used to confirm functionalization.
Elemental analysis was used to determine DSoverall, while X-ray photoelectron spectroscopy (XPS) was utilized to measure DSsurface. 72 To assess the effect of different covalent linkages, CNFs were functionalized with different esters (phenyl, hexyl, dodecyl) and ethers (hexyl, dodecyl), which were also compared to common TEMPO oxidized nanocellulose carboxylates with H + or Na + counterions. Biodegradation of these samples by anaerobic and aerobic microorganisms was assessed via biomethane potential (BMP) tests and mass loss, respectively. Results from our study reveal the importance of materials characterization, particularly the surface coverage of added functional groups, in understanding the biodegradation behavior of functionalized CNFs.
Functionalization of CNFs. Freeze-dried cellulose nanofibrils (CNFs) were purchased from the University of Maine Process Development Center and either used as-received or milled into a powder with a Flack-tek mill (DAC 150, 2800 rpm, 4 min). Ethyl cellulose (48.0/48.5 % w/w ethoxyl basis) was purchased from Sigma-Aldrich. Carboxylated CNFs were purchased from the University of Maine Process Development Center as a slurry of TEMPO-oxidized CNFs (1 wt% CNF, 1.5 mmol COOH/g cellulose) and dried either as-received (Na + counterion) or after washing with dilute HCl (H + counterion).
Esterified CNFs were prepared by liquid-phase reactions with carboxylic acid reagents, 73 or with acyl chloride reagents. 63 CNF functionalization with carboxylic acids was performed by dispersing 200 mg of CNFs in 200 mL deionized water followed by a 2 h sonication before adjusting to approximately pH 4 with 4 M HCl. The mixture was then heated to evaporate water followed by addition of excess phenyl acetic acid (phenyl ester CNF), hexanoic acid (hexyl ester CNF), or dodecanoic acid (dodecyl ester CNF, DA-CNF) before melting at ~140 °C to form the reaction medium. Sample solutions were stirred with a magnetic stirrer for 14 h and subsequently quenched with ethanol (Pharmco, 200 proof ACS/USP grade). Functionalized CNF powders were recovered via vacuum filtration, washed with ethanol, and dried in a vacuum oven at 60 °C for 12 h.
Liquid-phase esterification reactions using acyl chlorides were carried out using a modified method derived from literature by dispersing 200 mg of CNFs in 12 mL of diethyl ether and 0.5 mL of triethylamine in a vented round bottom flask equipped with a magnetic stirrer. 63 After dropwise addition of 1 mL of lauroyl chloride, samples were gently mixed at room temperature for 6 h. At the end of the reaction time, samples were quenched with 30 mL of deionized water and recovered by vacuum filtration followed by a dilute HCl (100 mL, pH 5.5) and a deionized water (800 mL) wash. Samples were then dried in a vacuum oven at 50 °C for 72 h to yield lauroyl chloride esterified CNFs (LC-CNF).
Gas-phase (GP) esterification was performed by adding ~10 mg of CNF powder to a custom-designed Schlenk line vessel 74 suspended above 1 mL of either lauroyl chloride (GP-LC-CNF) or hexanoyl chloride (GP-HC-CNF). The bottom of the vessel was submerged in liquid nitrogen to freeze the reagent, followed by headspace evacuation. After sealing the vessel, the reagent was allowed to thaw and vaporize into the headspace of the vessel to react with the CNF powder.
Etherification was performed by swelling 200 mg of dried CNF in 200 mL of dimethyl sulfoxide (DMSO, Fisher, 99.9%) via sonication for 3 h. After swelling, 200 mg K2CO3 (Aldrich, 99.99%) was then added, and the sample sonicated for an additional 3 h. A 30 mL aliquot of 1bromohexane (hexyl ether CNF; Aldrich, 98%) or 1-bromododecane (dodecyl ether CNF; Aldrich, 97%) was added to the sample before heating to 90 °C and magnetically stirring for 45 min under reflux. 75 The reaction was then quenched with ethanol and the functionalized CNF powder recovered via vacuum filtration followed by thorough washing with ~1 L of ethanol (Pharmco, 200 proof ACS/USP grade) before being dried in a vacuum oven at 60 °C for 12 h. CNF Characterization. Cellulose nanofibril characterization techniques are briefly described, with complete details in the SI. Functional groups in the unmodified and functionalized CNFs were identified using ATR-FTIR; the bonding and concentrations of C and O at the surface of the unmodified and functionalized CNFs were assessed using XPS; the carbon structure of the nanocellulose before and after functionalization was evaluated via solid-state 13 C-NMR; the wt% C, N, O, and H of unmodified and functionalized CNFs was determined by elemental analysis.
functionalized with esters and ethers DS values were calculated from the wt % carbon (Table S1) relative to unmodified CNF (C6H10O5) 44,73,76 with an uncertainty of approximately 0.3 %. 77 For example, an increase in carbon content to 53.0 wt % after esterification with dodecanoic acid (C12H24O2) reflects a DS of 0.45, which corresponds to an average addition of approximately one dodecyl ester group per two glucose monomer units. Because elemental analysis measures the degree of CNF functionalization from the entirety of the sample, DS values determined from elemental analysis are hereafter referred to as DSoverall.
Development Center was listed as having 1.5 mmol COOH/g cellulose. Each gram of cellulose features roughly 6.2 mmol of glucose monomer units, which corresponds to 0.243 COOH groups per cellulose unit (1.5 mmol COOH/6.2 mmol cellulose), representing a DSoverall of 0.24.
were based on the fitted contribution from the C-C component (285.0 eV) to the C(1s) XPS envelope. As the C-C content in unmodified CNF was 14.5 ± 3% (due to adventitious carbon), any increase was assumed to be due to functionalization of the nanocellulose surface by ethers or esters. For example, upon esterification with dodecyl ester groups, an increase in the C-C component to 48% would require an average addition of approximately 1 dodecyl ester group per 7 glucose monomer units, corresponding to a DS of 0.43. Since DS values determined by XPS are surface specific and represent the degree of CNF functionalization within only the topmost 2 nm to 3 nm of the sample, they are hereafter referred to as DSsurface (Table S2).
DS for Gas-Phase CNF Samples. We estimated the degree of substitution for gas-phase CNF samples (GP-HC-CNF and GP-LC-CNF) from a combination of CHN analysis data and ATR-FTIR spectra. Due to the low sample mass attainable from gas phase functionalization, we were not able to measure CHN on the full set of samples. Instead, the elemental composition of one sample from each set (GP-HC-CNF-4 and GP-LC-CNF-4) was determined, and the DSoverall calculated as described above. The DSoverall values of GP-HC-CNF-4 and GP-LC-CNF-4 were then ratioed to the C=O (ester) : C-O (cellulose) peak intensities obtained from ATR-FTIR analysis.
As detailed in the results and discussion section, this provided a means to convert the C=O : C-O peak intensities measured on the remaining samples in the two series (GP-HC-CNF and GP-LC-CNF) into their respective DSoverall values.
Anaerobic Biodegradation of CNFs. Biomethane Potential (BMP) Tests. Mineralization was assessed using biomethane potential tests, adapted with minor modifications from Owen et al., 50,78 to monitor biogas (CO2 and CH4) production after unmodified and functionalized CNFs were incubated with an anaerobic microbial community. Microbial BMP media (Table S3) was prepared as previously described 16 and heated at 100 °C for 30 min while sparging with N2 to achieve anoxic conditions before adding anaerobic digestor sludge (10% v/v) obtained from Back River Wastewater Treatment Plant (Baltimore, MD). The BMP media was adjusted to pH ~7 using 20% CO2 gas and kept anoxic via continuous N2 sparging. Duplicate 100 mg or 150 mg (DA-CNF only, due to increased sample availability) functionalized CNF samples were mixed with 100 mL of anaerobic media in 150 mL serum bottles and capped with rubber septa. We also assessed 150 mg of each functionalizing reagent (e.g., dodecanoic acid, hexanoic acid) independently to determine their biogas production potential. Samples were incubated at mesophilic temperature (35 °C) for up to 424 d, and biogas production was volumetrically assessed via intermittent headspace measurements with a glass syringe. In each set of samples, blank controls (i.e., anaerobic media including the same concentration of sludge in the absence of a CNF sample) were incubated in triplicate to account for biogas produced by the residual organic matter in the media (< 10% total solids, ~55% volatile solids before dilution to 10% of media volume). [79][80][81] Separate control studies were performed with the native (i.e., unfunctionalized) CNF to determine the extent of biogas production in the absence of functionalization and to confirm that the overwhelming majority of CNFs biodegrade to liberate biogas. Importantly, the carbon contributed by cellulose in each sample (42 mg C per 100 mg unfunctionalized CNF, more for functionalized CNF) vastly outweighed the contributed carbon from the BMP media (< 5 mg C in nutrients, most of which is not mineralized). Given the well-known propensity of cellulose to form biogas during biodegradation, the biogas produced by CNF samples was dominated by CNF mineralization. To account for biogas produced from the BMP media, the biogas production from CNF samples at each timepoint was reported as the difference between the volume produced by the CNF sample (typically yielded > 5 mL at each time point) and the average volume of biogas produced from the blank media (< 3 mL per timepoint). In this way, any biogas contribution from the media is removed and the reported biogas data arises solely from the mineralization of the CNF sample.
All biogas values were normalized to account for differences in sample mass (comparison between biogas production from 100 mg and 150 mg CNF; Figure S1). Biodegradation of CNFs in our BMP tests lead predominantly to biogas formation over the course of a few weeks, producing between 680-700 mL/g of biogas, representing over 80% CNF mass loss. This is consistent with the rapid and extensive mineralization of cellulosic materials typically observed. 28,82 Despite the efficiency of biogas production during CNF biodegradation, some of the carbon is converted into biomass, as is typical of biodegradation processes. This situation also holds true for functionalized CNFs.
The biodegradation of functionalized CNFs proceeds via the hydrolytic cleavage of the linkage formed between the functional group and the cellulosic monomer, and the subsequent biodegradation of the species generated in this step. For example, in the case of the CNFs functionalized with acyl chlorides, the process will be initiated by the generation of the native CNFs and carboxylic acids, followed by their subsequent biodegradation. For each functionalized CNF, we therefore performed independent biodegradation studies to determine the partitioning between biogas and biomass production for each component, after subtraction of the small amount of biogas produced due to residual carbon and biomass present in the media itself (determined in separate control studies). This information enabled us to determine the biogas each component would generate in the case of complete biodegradation under our experimental conditions.
Combined with knowledge of the chemical composition of each functionalized CNF, we could then determine the biogas we would predict in the event of complete biodegradation. For example, a CNF functionalized with a dodecyl ester with a DSoverall of 0.45 would be composed of roughly 66 wt % CNF and 34 wt % dodecyl ester. The total biogas produced from this functionalized CNF in the event of complete (100% biodegradation) is expected to be 0.66x + 0.34y, where x and y are the per gram biogas production potentials of cellulose (680 mL g -1 ) and dodecanoic acid (1280 mL g -1 ), respectively (Figure S2). This equates production of 883 mL/g biogas, considerably more than produced from cellulose alone (i.e., 680 mL/g). This predicted value was nearly met for DA-CNF-2 (94% of calculated biogas production was achieved), providing evidence for almost complete biodegradation of this sample, thereby also providing support for the validity of this normalization strategy in computing biogas potentials. By reporting data in these normalized terms (i.e., experimental data/calculated maximum data), we were able to compare samples in terms of their ability to achieve "maximum" biodegradation (a normalized Vmax value of 1). Explicitly, this value represents the extent to which each functionalized CNF reaches its maximum biogas production based on the overall partitioning predicted for biogas and biomass production from the components. The biogas values that would be produced in the event of 100% mineralization along with the experimentally measured (empirical) biogas produced for CNFs and all of the functionalized CNFs used in this study are reported in Table S4.
Gompertz Modeling. Anaerobic biogas production rates were quantified using the modified Gompertz kinetic model (Eq. 1) [83][84][85] 𝑉 𝑉 𝑒 * (Eq. 1)
where Vmax is the experimental ultimate biogas yield (mL g -1 ), K is the maximum specific rate constant (mL g -1 d -1 ), λ is the lag phase time constant (d), and ti is the total incubation time (d). The Solver optimization tool in Microsoft Excel was used to estimate the model parameters for each sample by minimizing the root mean square deviations (RMSE, Table S5), and the agreement between predicted and experimental values was evaluated by comparing the RMSE and R 2 values.
measuring mass loss after exposure to an aerobic microbial community obtained from the primary effluent of the Back River Wastewater Treatment Plant (Baltimore, MD). Mass loss was used as the metric for aerobic biodegradation as an open system was required to maintain an oxygenated environment. Triplicate 50 mg samples of CNF powders were sedimented via centrifugation in conical vials (Figure S3) containing an aqueous medium composed of 200 mg/L sodium acetate trihydrate and 10% v/v salt stock (7.18 mM K2HPO4, 2.79 mM KH2PO4, 0.757 mM (NH4)2SO4, 0.0406 mM MgSO4•7H2O), and trace elements necessary for bacterial growth. Microbial media was made by adding 10% v/v primary effluent supernatant to the vials and shaking at 125 rpm at 28 °C for 60 d. These samples were then incubated for 60 days before the powders were recovered from the media, washed with ultrapure water (18.2 MΩ•cm, Millipore, USA), washed three times with ethanol to remove any adhered biomass/biofilm and then dried in a vacuum oven at 60 °C for 12 h to evaporate any adsorbed water, before being weighed. This approach is analogous to the one used in other aerobic biodegradation studies of cellulosic materials. 28,51,52 To account for any native material which was dispersed or otherwise lost in the media, an identical set of samples was incubated for 60 days in uninoculated media and the mass loss observed in these control studies was subtracted from the biotic mass loss values. Consistent with our expectations, the mass loss experienced by the native CNF samples was reproducible and close (80%) to the mass of the CNF added, supporting the idea that comparisons between the extent of mass loss produced by different samples could serve as the basis to compare the extent of biodegradation amongst our functionalized CNF samples. Furthermore, the products of incomplete CNF biodegradation (i.e., cellulose hydrolysis without complete conversion to CO2) such as cellobiose and glucose monomer units are water soluble and would therefore contribute to the observed mass loss. 86 Thus, the final mass measured in our studies should be composed predominately of undegraded CNF or functionalized CNF samples, as intended. The mass loss for each sample was determined by the difference between the average mass lost in bacterial culture minus the average mass lost in the abiotic media. This difference was then ratioed to the initial mass (50mg) to determine the adjusted % mass loss reported in Figure 1.
In the present investigation, mass loss data from aerobic biodegradation studies was used as a semi-quantitative measure of biodegradation amongst functionalized CNFs. Unlike the BMP tests, mass loss was determined at a single time point, precluding application of the modified Gompertz model. Moreover, in the aerobic studies we did not directly assess the mass loss that would accompany the biodegradation of the added functional groups, although the influence of this unknown is not expected to impact the qualitative conclusions drawn from these studies.
CNF naming convention: CNFs were functionalized with different ester, carboxylic acid, and ether groups (Figure S4,S5). CNF ethers functionalized using 1-bromohexane and 1bromododecane are referred to as hexyl ether CNF and dodecyl ether CNF, respectively. CNFs functionalized using phenyl acetic acid, hexanoic acid, and dodecanoic acid (DA) are referred to as phenyl, hexyl, and dodecyl ester CNF, respectively. Samples functionalized with liquid-phase DA are referred to as DA-CNF-X, where X represents the relative DSsurface rank in that sample series. For example, DA-CNF-4 represents a CNF functionalized with liquid phase DA at the highest DSsurface value of the four samples within the series. 87 Similarly, CNFs functionalized with liquid phase lauroyl chloride (LC) are referred to as LC-CNF-X, and follow the same naming convention. Functionalizations achieved using gas-phase (GP) acyl chlorides are denoted by a GP naming scheme. For example, the CNF with the surface most extensively functionalized by gas phase hexanoyl chloride (HC) in a series of four is labeled GP-HC-CNF-4.
Figure 1a shows that unmodified CNFs were completely mineralized after 60 d (Figures 1a andS6) by an anaerobic microbial community. In subsequent discussion, the extent of biodegradation will be expressed relative to (calculated) full mineralization of the sample, unless otherwise noted. Among the three esterified CNFs, the hexyl esterified CNF (DSoverall: 0.09) exhibited a biodegradation rate comparable to unmodified CNF, while dodecyl (DSoverall: 0.45) and phenyl (DSoverall: 0.14) esterified CNFs displayed considerably slower biodegradation rates (Figure 1b), although all three esters were almost completely biodegraded (> 90%) over 424 d (Eq. 1, Table S5). TEMPO-oxidized CNFs containing carboxylate groups with Na + and H + (Figure S6) counterions (DSoverall: 0.24) also biodegraded at markedly slower rates, but ultimately underwent almost complete biodegradation analogous to the behavior of unmodified CNFs. In contrast, etherified nanocellulose was dramatically less susceptible to mineralization even at extremely low DSoverall values, with hexyl (DSoverall: 0.05) and dodecyl (DSoverall: 0.11) etherified CNFs only biodegrading to 15% and 10% of the levels exhibited by unmodified CNF, respectively, after 424 d of incubation. Biodegradation of functionalized CNFs by an aerobic microbial community, as found in aerobic wastewater, for 60 d revealed similar trends of functional group-induced inhibition towards biodegradation (Figure 1c, 1d), with unmodified CNFs exhibiting 80 % mass loss, hexyl and phenyl CNF esters and carboxylated CNFs all exhibiting mass loss in excess of 9%, and etherified CNFs exhibiting no measurable mass loss.
In the present study, we have measured the effect of different functionalization strategies on the biodegradation of nanocellulose exposed to the same microbial community, revealing that CNF ethers become non-biodegradable at low DS values (0.1), in contrast to the behavior of CNF esters at similar DS values. Qualitatively, the trends observed in the relative inhibition of biodegradation induced by introducing specific functional groups mirror those observed for macrocellulose. For example, previous studies on cellulose functionalized with carboxyl groups have also shown over 50% sample biodegradation in soil burial tests. 88 This behavior that has been attributed to the increased swelling of the cellulosic fiber that occurs upon addition of the hydrophilic carboxylic acid functional groups which facilitates enzymatic ingress into the interior of the cellulosic material. 88,89 The biodegradability of esterified cellulose observed in this study has also been observed for macrocellulose and has been previously attributed to the susceptibility of ester linkages to enzymatic hydrolysis which regenerates the glucose monomer unit of cellulose. 32,39,40,90 The recalcitrance of ether linkages to biodegradation observed in Figure 1 has been attributed to their resistance to enzymatic attack/hydrolysis. 32,41,91 Indeed, we observed a complete lack of biogas production from ethyl cellulose, a commercial, non-biodegradable, macrocellulose ether that produced no biogas over 424 d (Figure 1a). This recalcitrance to mineralization was found for etherified CNFs with very small DSoverall values ( 0.1).
During the course of our experiments in anaerobic media, and in contrast to our expectations, we observed that an esterified CNF with relatively high DSoverall (dodecyl, DSoverall 0.45) biodegraded similarly (Figure 1a) to a CNF ester with significantly lower DSoverall (phenyl, DSoverall 0.14). Based on existing literature for macrocellulose, a threefold increase in DSoverall would be expected to decrease the biodegradability of esterified CNFs by over 90%. 48 These data suggested that conventional DSoverall values may not be predictive of the relative biodegradability of functionalized nanocellulose. One possible explanation is that the large surface area-to-volume ratio of nanocellulose causes the surface of functionalized CNFs to take on increased importance relative to macrocellulose. To explore this possibility, a series of esterified CNFs with varying degrees of surface (DSsurface) and overall (DSoverall) functionalization was synthesized, characterized, and biodegraded by anaerobic microorganisms where comparisons of biodegradation behavior were made easier by virtue of our ability to track biogas formation as a function of incubation time.
CNFs were functionalized with dodecyl ester groups using liquid-phase dodecanoic acid (DA-CNF; Figure 2a-d) and lauroyl chloride (LC-CNF; Figures S7,S8). Elemental analysis revealed that by varying reaction conditions, functionalized CNFs with a range of DSoverall values (Table S1) could be prepared for both sets of CNF esters (DA-CNF and LC-CNF). ATR-FTIR provided spectroscopic evidence of functionalization through the observation of CH2 (2920 cm -1 and 2850 cm -1 ) and C=O (1700 cm -1 ) stretching modes (LC-CNF; Figure S7) in addition to the characteristic O-H (3339 cm -1 ), C-H (2905 cm -1 ), and C-O (1031 cm -1 ) stretches of cellulose.
Moreover, a linear relationship between the DSoverall values obtained from elemental analysis and the C=O (ester):C-O (cellulose) vibrational band ratio of DA-CNFs was observed (Figure 2c), suggesting that ATR-FTIR can serve as a facile, non-destructive alternative to elemental analysis for determining DSoverall values of functionalized nanocellulose.
Solid-state 13 C-NMR qualitatively confirmed the trends in DSoverall as shown in Figure 2b (increasing for DA-CNF-4, 3, and 2).. The 13 C-NMR spectrum of unmodified CNF consists of peaks between 50 and 110 ppm arising from carbons 1-6 in cellulose (C1-C6, labeled in Figure S9 and Table S6) and includes peaks arising from amorphous and crystalline nanocellulose. In addition to these principal cellulose peaks, the spectra of esterified CNF contain ester (180 ppm) and methylene (32, 25, 15 ppm) peaks that increased in intensity with increasing reaction time (in order of DA-CNF sample 1, 4, 3, 2). The NMR spectra reveal a minor increase in crystallinity (39% to 58%) of the esterified CNF samples compared to the unmodified CNF sample (Table S7). This is not expected to significantly contribute to differences in biodegradation, however, as crystalline and non-crystalline nanocellulose exhibit similar biodegradation properties 92 (Figure S10). We calculated DSoverall from variable contact time cross polarization-magic-angle spinning experiments for two samples: phenyl ester CNF and DA-CNF-2 (Table S8, Figure S11). (0.10), but the highest DSoverall (0.45). In addition to liquid-phase esterification, gas-phase reactions using lauroyl or hexanoyl chloride were performed with the expectation that this approach would restrict functionalization to the CNF surface (Figure S12). The XPS spectra in Figure S12 confirms that measurable increases in DSsurface occurred after gas-phase CNF functionalization in the absence of significant bulk functionalization (Figure S13).
Esterified CNFs functionalized with dodecanoic acid (DA-CNFs) and lauroyl chloride (LC-CNFs) were biodegraded by an anaerobic microbial community (Figures 3 andS14) to assess the sensitivity of CNF biodegradability to changes in DSsurface, DSoverall, and/or both. The rate and extent of biodegradation of both CNF types were found to change systematically in response to changes in DSsurface, but not DSoverall (Figure 3, Table S9, Figure S15). This trend is most clearly demonstrated by DA-CNF samples where increases in DSsurface led to systematic decreases in biodegradation, while DSoverall values did not correlate with biodegradation trends (Figure 3a, Analogous behavior is observed with LC-CNFs. As the extent of surface functionalization (DSsurface) increased in LC-CNF samples, the extent and rate of biogas production decreased, with the most extensively surface functionalized CNF in this series (LC-CNF-4) exhibiting a biogas production rate and extent of biodegradation only 24% and 37% of unmodified CNF, respectively (Figure 3b). The lack of correlation with DSoverall is observed most clearly in LC-CNF-3: this sample featured the lowest DSoverall (0.56) of the LC-CNFs but was more recalcitrant to biodegradation than LC-CNF-1 and LC-CNF-2, which each featured a DSoverall of approximately 0.8 (Figure 3b).
We note that LC-CNF samples experienced more extensive overall biodegradation than DA-CNF samples despite LC-CNF samples reaching higher DSsurface values (max 2.46) than DA-CNFs (max 0.43). This behavior is a direct result of the production of HCl during CNF functionalization with acyl chlorides, which reduces cellulose chain length and particle size. 63, 94- 97 This decrease in chain length increases the overall biodegradability of cellulosic materials by offering more sites/surface area (e.g., chain ends) for the initiation of enzymatic attack. 55,98,99 In contrast, esterification using carboxylic acid reagents as used in the synthesis of DA-CNFs does not produce HCl at the site of functionalization, limiting damage to the cellulose chain, thereby producing DA-CNF samples with lower DSsurface values which undergo a smaller degree of biodegradation.
Gas-phase functionalization was used to specifically target the role of surface functionalization in biodegradation. To this end, CNF surfaces were modified with lauroyl chloride (GP-LC-CNF) and hexanoyl chloride (GP-HC-CNF). Hexanoyl chloride enabled a wider range of and higher DSsurface values to be achieved (1.19-2.43) compared to lauroyl chloride (0.07-0.33) due to its higher volatility (hexanoyl chloride Tb 150 C vs. lauroyl chloride Tb 260 C). Because these esterified samples exhibited levels of overall substitution (i.e., DSoverall < 0.17, Figure S13) that would not slow the biodegradation of CNF esters (see Table S9), the effect of surface functionalization on the biodegradation of nanocellulose could be isolated (Figures 3c-d andS14).
The extremely low levels of DSoverall produced by gas-phase functionalization compared to the corresponding DSsurface values can be appreciated if we consider that the CNFs have diameters ~50 nm and that the XPS measurements of the near-surface region are dominated by photoelectrons from the outermost 3 nm of the CNFs. 100 Thus, even in the event that 100% of the CNF hydroxyl groups in the near-surface region were functionalized (i.e., DSsurface of 3.0), the corresponding DSoverall would be only approximately 0.2.
The specificity of surface functionalization realized with these gas-phase modified CNF samples provided a clear indication of the role that DSsurface plays in regulating CNF biodegradability. As seen in Figure 3c, GP-HC-CNF-1, the least functionalized GP-HC-CNF samples (DSsurface 1.19), displayed a 60% reduction in biogas production rate compared to unmodified CNF. In contrast, the most surface functionalized GP-HC-CNF (GP-HC-CNF-4; DSsurface 2.43) exhibited a biogas production rate only 17% of that observed for unmodified CNF. The extent of biodegradation followed the same dependence on DSsurface, with GP-HC-CNF-1 and GP-HC-CNF-4 samples undergoing 100% and 70% biodegradation, respectively. As expected, the CNFs functionalized with gas phase lauroyl chloride were also more recalcitrant to biodegradation (Figure 3d) than unmodified CNF, but less so than GP-HC-CNFs (lower DSsurface values). GP-LC-CNF samples with DSsurface values > 0.17 exhibited measurable decreases in the biogas production rate (> 40% reduction), although the extent of biodegradation reached approximately 90% of the value expected for unmodified CNFs after 75 d due to the relatively low (< 0.35 DSsurface) levels of surface functionalization. Figure 3 reveals that for both gas-and liquid-phase CNF functionalization, the extent and rate of biodegradation decrease with increasing DSsurface values, largely independent of DSoverall values. This dependency on surface chemistry is consistent with the environmental properties of other carbon nanomaterials such as carbon nanotubes, 101,102 carbon dots, [103][104][105] and graphene, 106,107 which have previously been found to be influenced-and in some cases wholly determinedby surface properties. Of significance, the observed reduction in biodegradability for functionalized nanocellulose manifested as a decreased propensity to be mineralized into biogas (i.e., complete biodegradation). Consequently, either the parent material or products of incomplete biodegradation may persist even in conditions with high microbial activity and result in environmental accumulation.
The importance of surface functionalization is likely a reflection of the biodegradation mechanism of cellulosic materials, which is typically initiated at the surface via highly specific interactions with microbial enzymes. 26,56,59 For example, during anaerobic biodegradation, cellulose is completely mineralized by microorganisms which initiate the process using extracellular cellulosomes or multiprotein complexes of cellulolytic enzymes. [25][26][27] These enzymes are particularly sensitive to the surface of the substrate material, and microorganisms alter the structural and enzymatic composition of the cellulosome to suit the substrate in question. 57 The small length of the glucose subunits of cellulose (roughly 0.5 nm) 108 compared to that of cellulosomes (roughly 18 nm) 109 suggests that functional groups covalently attached to surface subunits of nanocellulose, even at low concentrations, must be removed before conventional enzymatic degradation can proceed. 57,58,110 Specifically, biodegradation will be delayed until these functionalized surface sites have been sufficiently removed to regenerate a cellulose substrate recognizable to microbial cellulosomes. The surface-dependent hydrolysis of these functionalized sites serves as the rate-limiting step for the biodegradation of functionalized nanocellulose and likely explains why the bulk of the material is of less importance in determining the rate and extent of biodegradation.
For esterified CNFs with relatively low degrees of surface functionalization (e.g., DA-CNF-1, phenyl ester CNF; DSoverall ≤ 0.14), the presence of ester groups at the surface causes a decrease in biodegradation rate-although the CNF still ultimately biodegrades. However, as the degree of surface functionalization increases (e.g., DA-CNF-4, LC-CNF-4), our data indicates that an increasing fraction of the CNFs are recalcitrant to biodegradation (see Figure 3) despite the enzymatic susceptibility of ester groups. We ascribe this effect to the concentrations of ester groups in certain regions of the surface being sufficiently high to interfere with enzyme regioselectivity, blocking esterases from properly orienting with a single ester group and thus preventing their hydrolysis. 90 As the DS value increases, the fraction of the CNF surface covered with sufficiently high concentrations of ester groups to prevent biodegradation increases. This argument is supported by the observation that esterases are unable to biodegrade macrocellulose esters with high DS values 31,48,90 The observed inhibitory effect is enhanced when the covalent linkages are resistant to enzymatic hydrolysis, as is the case for etherified CNFs, 91 where a DSsurface of 0.16 and 0.25 was sufficient to prevent biodegradation of dodecyl and hexyl ether CNFs, respectively (see Figure 1). Indeed, the larger DSsurface vs DSoverall values for etherified CNFs (e.g., 0.16 vs. 0.11, respectively for dodecyl ether CNF; see Figure 1b) helps to explain why these functionalized CNFs were so recalcitrant to biodegradation even at low levels of DSoverall.
In contrast to our findings, DSoverall is generally found to be a reasonable predictor for the biodegradation behavior of macrocellulose, with the degree of surface functionalization rarely reported in biodegradation studies. 31,[46][47][48] One potential explanation for this difference between macro-and nanocellulose is that a stronger correlation between DSoverall and DSsurface may exist for macrocellulose. In this respect, we note that macrocellulose exhibits an increased swelling capacity (~48% vs. ~26% for macrocellulose compared to CNFs in aqueous media). This will almost certainly increase the ingress of chemical reagents into the interior of macrocellulose during liquid phase functionalization, likely leading to more similar degrees of bulk vs surface functionalization in macrocellulose compared to CNFs. [111][112][113] Regardless of the detailed mechanistic underpinnings, results from this investigation highlight the need to measure both DSoverall and DSsurface for functionalized macrocellulose, and to establish the influence of these two metrics on biodegradation properties.
The increased recalcitrance of surface functionalized nanocellulose to mineralization is undesirable because complete biodegradation is necessary to ensure its removal from the environment, minimizing accumulation and any consequent impact. 23,24 The present study has revealed that the extent of surface functionalization and type of covalent linkage strongly influence the degree of CNF recalcitrance to biodegradation. We note that the use of microbial communities derived from wastewater represent optimized conditions for the biodegradation of cellulose due to the diversity, activity, and concentration of microorganisms in the culture as well as the availability of nutrients in the BMP tests. 16,[114][115][116] Therefore, under environmental conditions where less diverse microbial communities exist (e.g., soils, aquatic sediments), the effect of functionalization on nanocellulose mineralization is expected to be more pronounced relative to the effects observed in this study. We note that the primary biological transformation products of functionalized nanocellulose are expected be carbon dioxide and methane in anaerobic environments and carbon dioxide in aerobic environments; both gases contribute to the greenhouse effect. Thus, although biodegradation is typically viewed as a positive environmental outcome because it acts to remove otherwise persistent materials its effects are not without consequences.
We found that although relative biodegradation trends among different functionalized CNFs were independent of the microbial community (i.e., aerobic vs. anaerobic), the magnitude of inhibition differed (Figure 1d). Specifically, the extent of biodegradation was reduced in aerobic wastewater compared to anaerobic wastewater, likely due to differences in microbial population and numbers. The decreased aerobic biodegradation of functionalized CNF suggests that anaerobic digestion should be utilized to maximize biodegradation of functionalized CNFs and reduce landfill disposal.
While sustainability has been identified as an area of focus in the production phase of surface-modified nanocellulose, 64,73,117 the end-of-life environmental fate of such nanomaterials has been largely overlooked or assumed to be comparable to unmodified nanocellulose. 1,21,[118][119][120] Importantly, products which utilize surface-functionalized nanocellulose and are marketed as biodegradable (e.g., packaging materials) 21,121 may actually feature environmentally persistent nanocellulose. For example, the combination of surface-esterified nanocellulose with the biodegradable polymer poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) to create a strengthened material deemed appropriate for use as a fully biodegradable food packaging material. 21 However, based on our data, the surface-level esterification of nanocellulose used in the reinforcement of PHBV would significantly reduce its overall biodegradability. The same argument holds true for functionalized nanocellulose used in other applications such as in displays and coatings. 12,[122][123][124][125] For example, Granstrӧm et al., developed a stearoyl ester CNF-based aerogel with projected applications in coatings and insulators 12 that our study indicates will not biodegrade as rapidly as unmodified nanocellulose. Indeed, applications of esterified CNFs in packaging materials, coatings, and lubricants are expected to spur a growth in production to meet increased demand. 21,62,121,122,126 Our study highlights that the commercial benefit achieved through functionalization of nanocellulose must be carefully weighed against the consequent changes in the persistence of these nanomaterials. For example, decreasing the cellulose chain length and DS for CNFs functionalized with esters can be anticipated to increase biodegradability, but by the same token these changes are also likely to decrease CNF dispersibility in organic solvents with potential impacts on materials properties. Another practical consequence of the findings from this study is that even relatively low degrees of CNF surface functionalization lead to a portion of the material becoming recalcitrant to biodegradation. Moreover, due to the differential influence of DSsurface and DSoverall on the biodegradation of functionalized CNF direct comparisons of the effects that different functional groups play in mediating CNF biodegradation is difficult because both DSoverall and DSsurface need to be similar to isolate the impact of different functional groups on the biodegradation of the functionalized CNF. However, due to the difference in bulk vs. surface accessibility and reagent reactivity, exerting control over these two parameters experimentally is difficult.
Detailed information about instrumental parameters for material characterization (CHN analysis, ATR-FTIR, XPS, NMR). Specific information on solid state NMR experiments, theory, and data (e.g., crystallinity and peak positions). Tables listing S4). The DS surface and DS overall values, as well as normalized maximum biogas volume (V max ) and biogas production rate (K) for each sample are provided in the inset of each plot. Accuracy for Gompertz parameters is reported in Table S5.