Biomass-Derived Nitrogen Functionalized Carbon Nanodots and Their Anti-Biofouling Properties

: The prevalence of the antibiotic resistant bacteria remains a global issue. Cheap, sustainable and multifunctional antibacterial membranes are at the forefront of ﬁltrating materials capable of treating multiple ﬂow streams, such as water cleansing treatments. Carbon nanomaterials are particularly interesting objects shown to enhance antibacterial properties of composite materials. In this article, amino-functionalized, photoluminescent carbon nanodots (CNDs) were synthesized from chitosan by bottom-up approach via simple and green hydrothermal carbonization. A chemical model for the CNDs formation during hydrothermal treatment of chitosan is proposed. The use of urea as an additional nitrogen source leads to the consumption of hydroxyl groups of chitosan and higher nitrogen doping level as pyridinic and pyrrolic N-bonding conﬁgurations in the ﬁnal carbonaceous composition. These functionalized carbon nanodots that consist of carbon core and various surface functional groups were used to modify the commercially available membranes in order to enhance their anti-biofouling properties and add possible functionalities, including ﬂuorescent labelling. Incorporation of CNDs to membranes increased their hydrophilicity, surface charge without compromising membranes integrity, thereby increasing the factors affecting bacterial wall disruption. Membranes modiﬁed with CNDs effectively stopped the growth of two Gram-negative bacterial colonies: Klebsiella oxytoca (K. oxytoca) and Pseudomonas aeruginosa (P. aeruginosa).


Introduction
Various carbon-based starting materials, such as soot [1], carbohydrates [2][3][4][5], activated carbons [6,7], graphene oxide and graphite [8,9] have been reported in recent years as precursors to luminescent [10] carbon-based nanomaterials using straightforward oxidizing conditions at elevated temperatures. Chitosan, an abundant material in nature, has also been reported as the precursor for the synthesis of carbon dots [11]. It is a biocompatible polymer [12] with amino groups present on its surface that can be a source of nitrogendoped carbon dots. Some of the synthesis methods are followed by a specific chemical treatment like oxidation by mineral acids to tune the photoluminescent [13] properties and improve the water solubility of the carbon dots. The advantage of using chitosan is the ability to obtain nitrogen-doped carbon dots in a one-step [14], high yield synthesis [15] without complex and time-consuming treatment by other compounds.
These nanomaterials have been proposed as alternatives to their conventionally used counterparts (e.g., semiconductor quantum dots) and carry a variety of benefits such as enhanced biocompatibility and smaller sizes [16,17]. These so-called carbon dots (CDs) are generally described as nanocrystalline (graphitic) or amorphous in nature, pseudospherical, and generally are smaller than 10 nm in diameter [18]. mixture was sonicated for 0.5 h at room temperature and then the solution was placed in a glass inlet sealed in a Teflon autoclave and maintained at 180 • C for 18 h (p~10 bar). This sample was named CS180. The second sample was prepared by adding 1 g of chitosan to a mixture of 6 mL of 99.7% CH 3 COOH, 10 mL H 2 O, 200 mg of urea. The mixture was sonicated for 0.5 h at room temperature and then the solution was placed in a glass inlet sealed in a Teflon autoclave and maintained at 180 • C for 18 h (p~10 bar). The second sample was named CSU180. In order to remove large particles/aggregates, the synthesized dark brown solution was centrifuged at 10.000 rpm for 10 min, and the supernatant was filtered through a 0.22 µm filter membrane.

Incorporation of CNDs into Membranes
In this research, the commercial cation exchange membranes CMI-7000 (Membranes International INC, Ringwood, NJ, USA) were used. CNDs were introduced into activated membranes surface by a dip-coating method and maintained in the CNDs solution for 24 h, at ambient temperature. Afterward, membranes were dried in an air atmosphere. In order to ensure the incorporation of CNDs, each membrane was immersed in distilled water which supernatant was further tested for the released concentration of CNDs using UV-Vis spectroscopy (UV-2600, Shimadzu, Kyoto, Japan).

Antimicrobial Activity Tests
For antimicrobial activity tests, all bacteria were cultured overnight at 37 • C. The overnight bacterial cell culture was diluted 1:100 in 1/5 LB without and with the same concentrations of CNDs for all tested bacterial cultures and then inoculated into the Petri dish plates (Costar, Corning, CN, USA) with 100 µL/well. The inoculated Petri dish plates were incubated for 1-5 days at 28 • C without shaking.

Characterization
Zeta-potential was determined using Malvern Zetasizer (Software v7.02 PSS0012-33 EN-JP, Malvern Instruments Ltd., Malvern, UK). The optical properties were defined using UV-Vis and fluorescence spectroscopies. UV-Vis absorption spectra were recorded on a Shimadzu UV-Vis Spectrophotometer (UV 2600). The photoluminescence (PL) measurements were carried out at room temperature using a photoluminescence spectrometer LS-50B (PerkinElmer, Waltham, MS, USA) equipped with a xenon discharge lamp as an excitation source and a R928 photomultiplier as a detector. Fluorescence spectra were recorded at different excitation wavelengths in the range of 250-350 nm. The morphology and structure of CNDs were measured using an atomic force microscope NanoScope V (Veeco, Plainview, NY, USA) and a high-resolution transmission electron microscope (HRTEM). HRTEM images were recorded on a field emission electron microscope (JEOL JEM-2100F, 200 kV FE (field emission), Tokyo, Japan). Functional groups of CNDs were determined by X-ray photoelectron spectroscopy (XPS, PREVAC Sp. z o. o. Rogow, Poland) and Fourier-transform infrared spectroscopy (FTIR, Thermo Scientific Instrument Co., Boston, MA, USA). XPS images were obtained with an MX-650 electron analyzer from the Gammadata Scienta company. Electrons were excited using monochromatic X-rays in which the anode was Mg/Al. Fourier transform infrared spectra of CNDs were recorded using by Thermo Nicolet 8700 spectrometer (Thermo Scientific Instrument Co., Boston, MA, USA). The wettability of the CNDs modified membranes were evaluated using a goniometric technique in the pendant drop mode, with a video based optical contact angle meter OCA 15, Dataphysics, Germany.

Surface Functional Groups and Formation of CNDs
To confirm the presence of surface functional groups the FTIR and XPS spectra of CNDs were analyzed. The clusters of CS180 contain a large proportion of C=O and -OH groups, which favor agglomeration through strong hydrogen bonding. The CNDs derived from chitosan and urea, CSU180 also contain -OH groups, but the surface concentration of the -OHs is much lower due to their condensation reactions with urea leading to amides formation. The hydrogen bonding in amides is less efficient than for -OH containing molecules, which probably accounts for missing cluster formation in the case of CSU180.
The FTIR spectra of CS180 and CSU180 samples are shown in Figure 1. The broad, convoluted bands at ca. 3000-3500 cm −1 are assigned to hydroxyl groups stretching vibration. The two sharp absorbance bands at 1572 and 1418 cm −1 presumably stem from the stretching of C=C bonds in the aromatic region and from the C-OH bending mode, respectively. Besides, the stretching vibration peaks of the C-N bonds (possibly amines) at circa 1000 cm −1 are clearly shown. When comparing the spectra of CSU180 with CS180, one can notice two additional bands at 1288 and 1675 cm −1 plus a weak shoulder at ca. 1720 cm −1 . The presence of urea in the thermally treated acidic solution of chitosan could result in chemical reactions, which influence the composition and morphology of the final carbonaceous material. One can envisage either decomposition of urea or its condensation with chitosan. It was shown that the former process, probably through ammonia release, allows obtaining high doping level of nitrogen in a product, mainly as pyridinic and pyrrolic N-bonding configurations [44]. The condensation reaction, however, leads to a polymer-like material followed by carbonization and formation of CNDs. The mentioned condensation is well-known to proceed between urea and anomeric center of glucose in the presence of acidic catalyst [45,46]. The products of reaction can be both asymmetrical ureides as well as symmetrical urea-linked disaccharides. This reaction causes the consumption of hydroxyl groups of chitosan. The presented FTIR results seem to reveal remnants of this condensation in the form of the vibration bands, which we assign to C=O stretching in -NHCO-(1675 cm −1 ) and at the CNDs edges (1720 cm −1 ). The band at 1288 cm −1 was assigned to in-plane vibration of C-N and N-H groups of the bound amides. The surface functional groups generated on the CNDs surface by the hydrotherm carbonization of chitosan precursor were analyzed by high-resolution XPS ( Figure 2). a representative example, Figure 2 shows the C 1s, N 1s and O 1s surface-level spectra a the peak-fitting for the CS180 sample. The C 1s spectrum was deconvoluted to four signa at 284.7, 286.0, 287.4 eV. The first, dominant C 1s peak at 284.7 eV stems from graphi carbon. The other samples with urea ( Figure S1,) also reveal that band, which, howev is significantly stronger than that of CS180. In the case of the second signal, two bon contribute to its intensity; C-N and C-O in C-OH. The third signal could be attributed The surface functional groups generated on the CNDs surface by the hydrothermal carbonization of chitosan precursor were analyzed by high-resolution XPS ( Figure 2). As a representative example, Figure 2 shows the C 1s, N 1s and O 1s surface-level spectra and the peak-fitting for the CS180 sample. The C 1s spectrum was deconvoluted to four signals at 284.7, 286.0, 287.4 eV. The first, dominant C 1s peak at 284.7 eV stems from graphitic carbon. The other samples with urea ( Figure S1,) also reveal that band, which, however, is significantly stronger than that of CS180. In the case of the second signal, two bonds contribute to its intensity; C-N and C-O in C-OH. The third signal could be attributed to the presence of the carbonyl groups. The presence of these oxygenated groups is confirmed by the O 1s spectra, in which two signals are identified at 531.3 and 532.2 eV (Figure 2, O1s; Figure S2) [47]. The former signal corresponds to O=C groups, whereas the peak at 533.2 eV is mainly attributed to -O-C-groups. The N 1s spectra (Figure 2 N1s, Figure S3) are deconvoluted to the dominant peaks at around 398.7-399.1 eV corresponding to pyridinic-type nitrogen, 400.1-400.2 eV ascribed N-H in amine and amide, and the last peak at around 401.3-401.6 eV assigned to protonated nitrogen atoms bound to three C atoms in the bulk of a carbon [48].  These results agree well with those obtained from the FTIR measurements, which also show the presence of aromatic carbon, as well as C-N bonds and abundant oxygenated functional groups.
The surface of CS180 is negatively charged at neutral pH (Table 3). This means that The surface atomic concentrations of C, O, and N were derived from the corresponding peak areas of the XPS spectra and are summarized in Table 1. The hydrothermal treatment of chitosan precursors generates CNDs with the oxygen surface concentration of ca. 20%. Moreover, the sample obtained with urea has the largest amount of nitrogen with an atomic concentration of 10.3% on the CNDs surface. As data in Table 2 show, an increase in nitrogen concentration on the CSU180 surface is mainly due to higher content of the amine/amide groups, and to a lesser extent to the pyridinic groups. These results agree well with those obtained from the FTIR measurements, which also show the presence of aromatic carbon, as well as C-N bonds and abundant oxygenated functional groups.
The surface of CS180 is negatively charged at neutral pH (Table 3). This means that there are amine groups on the surface, but they are in a minority related to negatively charged surface groups, for example hydroxyl groups. In the case of CSU180 we observed positively charged surface which is caused by the presence of a secondary amine formed through the consumption of hydroxyl groups. Based on the results obtained by the spectroscopic analysis (FTIR, and XPS), we propose a revised chemical reaction model for the formation of CNDs in the presence of urea during hydrothermal carbonization of chitosan. The model is depicted in Figure 3 and compared to a standard route. Figure 4a,b shows the TEM images of the synthesized CNDs. The CNDs are evenly dispersed, with strong tendency for clustering into circular aggregates for unmodified CNDs, here-forth referred as to CS180 (Figure 4a). Size distribution measurements show that the average diameter of CS180 is 9.12 ± 0.82 nm, while the urea-doped sample (henceforth referred as to CSU180) are much smaller with an average diameter of 1.43 ± 0.06 nm. The narrow size distribution and nearly spherical shape with the average height of 1.351 ± 0.733 is observed for the CSU180, in the AFM measurements ( Figure 5).

UV-Vis Absorption and Fluorescence Properties of CNDs
The UV-Vis absorption spectra of both the CDs exhibit strong absorption in UV region ( Figure 6, λ < 300 nm), which is assigned to n-π* transitions in C=O bonds in nonconductive species, sp 3 hybrid regions. A very weak additional wave can be seen for the CSU180 dots at the wavelength around 340 nm. It corresponds to the moved n-π* transitions, the same as the previous one, from aromatic system with heteroatoms C=O. This shift may be due to inductive influence of the adjacent nitrogen atoms [49].
Upon 290 nm excitation, the fluorescence quantum yields of the CNDs CSU180 and CS180 were estimated to be 17% and 13%, respectively.

UV-Vis Absorption and Fluorescence Properties of CNDs
The UV-Vis absorption spectra of both the CDs exhibit strong absorption in UV region ( Figure 6, λ < 300 nm), which is assigned to n-π* transitions in C=O bonds in nonconductive species, sp 3 hybrid regions. A very weak additional wave can be seen for the CSU180 dots at the wavelength around 340 nm. It corresponds to the moved n-π* transitions, the same as the previous one, from aromatic system with heteroatoms C=O. This shift may be due to inductive influence of the adjacent nitrogen atoms [49].  Figure 7a,b illustrate the fluorescence spectra of the CS180 and CSU180 samples, respectively. As depicted, we observed the clear wavelength dependent fluorescence for both CNDs in terms of the peaks position, their intensity and shape. This effect could result from the presence of edge/core states derived from functional groups on CNDs combined with the dots-solvent interaction. The spectra for the CS180 dots are mainly determined by carboxylic acid groups that are electronically coupled with the surrounding graphene core sheets [50]. In the case of the CSU180, in addition to oxygen-based groups, amine-based groups also contribute to their surface state, either by allowing the electronic structure to be modified by the effective orbital resonance of the amine moieties and the graphene core [51], or through formation of an additional interband within the energy gap owing to p-orbital hybridization of C-N atoms at the edge sites [52]. Here, the fluorescence is affected by the electron transfer nature, e.g., electron withdrawing and donating properties of the functional environment. The phenomenon of fluorescence red-shift could be intensified by the presence of a slowed solvation process during fluorescence emission [53], which should stronger influence the larger CS180 dots. The mechanism behind the red-shift emission properties is often associated with surface or defect states in the amorphous carbon shell of CDs near the Fermi level [54]. Upon 290 nm excitation, the fluorescence quantum yields of the CNDs CSU180 and CS180 were estimated to be 17% and 13%, respectively. Figure 7a,b illustrate the fluorescence spectra of the CS180 and CSU180 samples, respectively. As depicted, we observed the clear wavelength dependent fluorescence for both CNDs in terms of the peaks position, their intensity and shape. This effect could result from the presence of edge/core states derived from functional groups on CNDs combined with the dots-solvent interaction. The spectra for the CS180 dots are mainly determined by carboxylic acid groups that are electronically coupled with the surrounding graphene core sheets [50]. In the case of the CSU180, in addition to oxygen-based groups, amine-based groups also contribute to their surface state, either by allowing the electronic structure to be modified by the effective orbital resonance of the amine moieties and the graphene core [51], or through formation of an additional interband within the energy gap owing to p-orbital hybridization of C-N atoms at the edge sites [52]. Here, the fluorescence is affected by the electron transfer nature, e.g., electron withdrawing and donating properties of the functional environment. The phenomenon of fluorescence red-shift could be intensified by the presence of a slowed solvation process during fluorescence emission [53], which should stronger influence the larger CS180 dots. The mechanism behind the red-shift emission properties is often associated with surface or defect states in the amorphous carbon shell of CDs near the Fermi level [54].

Incorporation of CNDs Into Membranes and Their Antimicrobial Properties
In order to investigate the antimicrobial properties of the prepared CNDs, they were used to modify the commercial CMI-7000 7000 cation exchange membranes. Firstly, the

Incorporation of CNDs into Membranes and Their Antimicrobial Properties
In order to investigate the antimicrobial properties of the prepared CNDs, they were used to modify the commercial CMI-7000 7000 cation exchange membranes. Firstly, the hydrophilicity of modified membranes was evaluated by the contact angle (Table 4). The minimum contact angle of 24.0 • was obtained for CSU180 incorporated membranes (M-CSU180), indicating CNDs decreased the hydrophobicity of both unmodified commercial membranes (60.8 • ), as well as unmodified CNDs incorporated into membranes (M-CS180) (70.3 • ). The contact angle values strongly depend on chemical structure and surface porosity [40]. The enhancement of hydrophilicity could affect the microbial properties of membranes, because bacterial attachment is significantly stronger to more hydrophobic surfaces [40]. The antibacterial tests against Escherichia coli (E. coli), P. aeruginosa and K. oxytoca were performed to determine the antibacterial properties of as-prepared samples. All of the bacterial cultures were representative of Gram-negative type of bacteria. As can be seen, after incubation, E. coli bacteria had attached and expanded on the surface of the unmodified sample (Figure 8). The results are similar for sample modified with CS180 what indicates that the CNDs treatment had no effect on the growth of E. coli. This behavior (in the case of sample with CSU180) can be correlated with the roughness value of the modified surface that is similar for modification with CSU180 but lower than for CS180. ified sample (Figure 8). The results are similar for sample modified with CS180 what indicates that the CNDs treatment had no effect on the growth of E. coli. This behavior (in the case of sample with CSU180) can be correlated with the roughness value of the modified surface that is similar for modification with CSU180 but lower than for CS180. Better results have been received for P. aeruginosa bacteria tests. After incubation visible differences between modified and unmodified samples can be seen (Figure 8, Table  5). The modification with CS180 and CSU180 significantly inhibited growth of the P. aeruginosa bacteria, and the lowest activity of this bacterial culture is observed on the sample modified with CS180.
Similar results were obtained for K. oxytoca tests for sample with CS180 with the highest inhibition results, while for modification with CSU180 high activity of K. oxytoca is still Better results have been received for P. aeruginosa bacteria tests. After incubation visible differences between modified and unmodified samples can be seen (Figure 8, Table 5). The modification with CS180 and CSU180 significantly inhibited growth of the P. aeruginosa bacteria, and the lowest activity of this bacterial culture is observed on the sample modified with CS180. Similar results were obtained for K. oxytoca tests for sample with CS180 with the highest inhibition results, while for modification with CSU180 high activity of K. oxytoca is still observed.
Since all tested bacteria were Gram-negative, the surface charge and different affinity of the bacteria to the surface of the modified membrane is the key factor of inhibition. The cell wall charge of Gram-negative bacteria is strongly negative due to the presence of lipopolysaccharides as an outer layer. Therefore, it can be expected that they could easily interact with the positively charged surface via electrostatic interactions. There are some studies on the interaction between carbon dots and bacteria that indicated that the surface charge on CNDs is an important factor inhibiting growth of bacteria [45].
However, it has to be considered that bacterial growth inhibition differs not only depending on carbon dots solution used for membrane modification but also depending on the type of tested bacteria colonies. It means that as-prepared carbon dots show antibacterial selectivity. The Gram-type selective antibacterial activity of carbon dots has already been reported for example for quaternized carbon nanospheres [55] and can be explained by the difference between Gram-positive and Gram-negative bacteria concerning their structure and charge under physiological pH conditions. Interestingly, the membranes modified with CDs were ineffective in inhibiting the activity of E. coli. Even though all of the tested bacteria are gram negative they belong to different groups in terms of producing the energy-the facultative anaerobes (K. oxytoca, P. aeruginosa) and microaerophilic bacteria (E. coli) [56]. The presented results suggested that the as-prepared CDs could not inhibit the growth of microaerophilic bacteria, but it showed highly efficiency against facultative anaerobes. The selective inhibition was also demonstrated in literature for carbon nanodots obtained from Metronidazole [56]. Liu et al. established that carbon nanodots can only inhibit the growth of obligate anaerobes and the growth of facultative anaerobes, obligate aerobes or microaerophilic bacteria could not be inhibited. It was explained as a result of their different structures, and the possibility of the degradation of the macromolecule only in obligate anaerobes. Moreover, CNDs can act as oxidizing agents, which produce free radicals in microbial organisms. Further, CNDs induced membrane stress by increasing oxidative stress resulting in cell content leakage and cell death [56]. A characteristic feature of CNDs compared with functionalized nanoparticles, GQDs are more effective against microorganisms than metal nanoparticles (ranging in size from 30 nm to 100 nm) [57].
In addition, further research on CNDs-modified membranes can help design membranes with broad-spectrum antimicrobial activity for fuel cell technology. Generally speaking, CNDs possess varied surface functional groups which can be further easily engineered by ligands or even drugs for suitable applications [55].

Conclusions
In summary, we demonstrated a facile method to synthesize amino-functionalized fluorescent carbon nanodots with antimicrobial properties by hydrothermal carbonization of chitosan. No further modification is needed in their preparation. Our work has shown that many of the properties of CNDs depend on their functionalization, including their optical and antibacterial properties. The as-prepared CNDs consist of an aromatic core and surface functional groups, mainly amine, that are able to inhibit the growth of some types of Gram-negative bacteria. We showed that membranes modified with CNDs stopped the growth of two Gram-negative bacterial colonies: K. oxytoca and P. aeruginosa. However, this process is complex and also depends on other parameters, such as the hydrophobicity of the membrane and its surface roughness.
Our results show that bacterial growth is inhibited upon application of CNDs, which indicates their possible use in antibacterial applications.