Insight into the differences in carbon dots prepared from fish scales using conventional hydrothermal and microwave methods

The preparation of carbon dots (CDs) from waste fish scales is an attractive and high-value transformation. In this study, fish scales were used as a precursor to prepare CDs, and the effects of hydrothermal and microwave methods on their fluorescence properties and structures were evaluated. The microwave method was more conducive to the self-doping of nitrogen due to rapid and uniform heating. However, the low temperature associated with the microwave method resulted in insufficient dissolution of the organic matter in the fish scales, resulting in incomplete dehydration and condensation and the formation of nanosheet-like CDs, whose emission behavior had no significant correlation with excitation. Although the CDs prepared using the conventional hydrothermal method showed lower nitrogen doping, the relative pyrrolic nitrogen content was higher, which was beneficial in improving their quantum yield. Additionally, the controllable high temperature and sealed environment used in the conventional hydrothermal method promoted dehydration and condensation of the organic matter in the fish scales to form CDs with a higher degree of carbonization, uniform size, and higher C = O/COOH content. CDs prepared using the conventional hydrothermal method exhibited higher quantum yields and excitation wavelength-dependent emission behavior.


Introduction
There has been a significant increase in the production of aquatic products worldwide, resulting not only in increased consumption but also in increased amounts of biological waste from fisheries (Golden et al. 2021). The internal organs, heads, and scales of fish in biowaste have attracted increasing attention due to their low commercial value, the lack of effective high-value utilization methods, and the threats posed to human health and the environment. The rational use of fish scales, skin, and bones, as well as waste products from aquatic production, can not only promote the sustainable and healthy development of the aquaticprocessing industry but also reduce environmental pollution (Krishnani et al. 2022;Lee et al. 2022;Rodrigues et al. 2021;Sierra et al. 2021).
The main chemical components of fish scale waste include collagen and hydroxyapatite, which are rich in carbon, hydrogen, oxygen, and nitrogen. They constitute inexpensive and readily available nitrogenous biomass waste (Deb et al. 2019;Luo et al. 2020;Niu et al. 2019). Among the many high-value conversion pathways, the use of fish scale waste as a starting material for the preparation of nitrogen-doped CDs has the advantages of the ready availability of precursors, simple reaction conditions, and high added value of the products, as well as providing environmental protection and economic benefits (Zhang et al. 2018).
The production of CDs from waste such as fish scales is relatively complicated and has not yet been thoroughly detailed. However, it is generally believed that it mainly includes the Responsible Editor: George Z. Kyzas pyrolysis of organic matter in the precursor, the transfer of sp 3 -to sp 2 -hybridized carbon, the aggregation and recombination of species to form carbon clusters, and the carbonization of intermediates to form carbon core, among other steps. The hydrothermal method is the primary technique used to prepare CDs from fish scales, and calcium phosphate by-products are usually formed. Ashraf et al. prepared four high-value nanoproducts including CDs using Labeo rohita scales as raw material and using the hydrothermal reaction (Ashraf et al. 2021). Athinarayanan et al. treated Lethrinus lentjan scales at 280 °C. The hydrothermal treatment converts the organic and inorganic substances in the fish scales into CDs and hydroxyapatite nanoparticles. An evaluation of cytotoxicity showed that the fish scale-derived materials were biocompatible and nontoxic and that they could be suitable for bioimaging and bone tissue engineering applications (Athinarayanan et al. 2020). Campalani et al. synthesized CDs from Dicentrarchus labrax scales using the hydrothermal method and applied them to photocatalytic reactions (photoreduction reactions of methyl viologen) (Campalani et al. 2021). Zhang et al. used Crucian carp scales as a precursor to prepare highly fluorescent CDs with nitrogen and oxygen functional groups by the hydrothermal reaction. These CDs exhibited strong fluorescence emission at 430 nm with a relative quantum yield of 6.9%, low cytotoxicity, strong fluorescence stability, and good ionic strength for photobleaching. Moreover, the fluorescence of the CDs was efficiently and selectively quenched by Fe 3+ ions, rendering the CDs suitable for use as fluorescent Fe 3+ nanoprobes ). Dhandapani et al. synthesized green-fluorescent hydrophilic CDs using the hydrothermal method with fish scales as the raw material (Dhandapani et al. 2020). In addition to the hydrothermal method, Yao et al. prepared CDs with blue fluorescence using ultrasonic physical peeling using waste Ctenopharyngodon idella scales as raw material, and applied them for the analysis and detection of Fe 3+ ). Chen et al. pretreated whitefish scales for use as starting material; they carbonized the scales at 260 °C for 2 h in an atmosphere of purified nitrogen to obtain CDs with multicolor fluorescence (Chen et al. 2016). Table 1 shows a summary of the preparation of other CDs from waste materials, including the preparation method, quantum yield, fluorescence color, and application.
Optimization of the preparation process to further improve the quantum yield of biomass-derived CDs is an important research direction in CD field. In addition to the commonly used  Atchudan et al. (2022) hydrothermal method described above, microwave heating has also been investigated due to its simultaneous, uniform, efficient, and rapid response, and short reaction time (Ghanem et al. 2020;Sharma et al. 2017;Xin et al. 2022;Zhao et al. 2019). However, there are no reported comparisons of the structure and performance of CDs prepared using the conventional hydrothermal and microwave methods. It is, thus, important to compare the differences in the fluorescence performance and composition of CDs prepared using the conventional hydrothermal and microwave methods. In this study, CDs were prepared from silver carp scales. Accordingly, the key process parameters, such as hydrothermal time, hydrothermal temperature, microwave time, and microwave power, were optimized with quantum yield as the optimization goal. The optical properties of the quantum dots prepared using the two methods were analyzed and characterized by fluorescence spectroscopy and ultraviolet-visible (UV-Vis) absorption spectroscopy. The microstructure and crystallization of CDs were studied using X-ray diffraction (XRD) and transmission electron microscopy (TEM). The structural differences in the quantum dots prepared by these two methods were evaluated using X-ray photoelectron spectroscopy (XPS) and Fourier-transform infrared spectroscopy (FTIR). The effects of the different preparation methods on the structural properties of the CDs were analyzed. The findings provide a reference for the further optimization and enhancement of the fluorescence performance of CDs prepared from biomass.

Materials
Silver carp scales were purchased from Wuhan Liangzihu Aquatic Processing Co., Ltd. The structure and composition of the silver carp scales are shown in Fig. S1. The main components of the fish scales were collagen and hydroxyapatite, with an elemental composition of 48.2% carbon, 11.2% nitrogen, 30.2% oxygen, 6.8% calcium, and 3.6% phosphorous. Deionized water was used for all experiments. All reagents used in this study were purchased from Sinopharm Chemical Reagent Company.

Preparation of CDs
Conventional hydrothermal method A specific quantity of fish scales was weighed, and 25 mL NaOH solution (0.01-0.15 mol/L) was added to completely wet the scales. The mixture was sealed within the hydrothermal reactor and placed in a constant temperature oven (DHG-9140A, Shanghai Yiheng Scientific Instrument, China) and allowed to react at 170-210 °C for 1-5 h. The mixture was then centrifuged at 12,000 rpm for 10 min at room temperature (about 25 °C) and the supernatant was transferred to a dialysis bag (MW cutoff: 1000) and dialyzed against deionized water. The resultant carbon dot sample was identified as CD-HT.

Microwave method
The process was similar to that used for the hydrothermal method, except that the fish scales and NaOH solution were placed in a microwave digestion tank and allowed to react in a microwave reactor (MCR-3, Gongyi Yuhua Instrument, China) at 80-720 W for a specific time (1-3 min). The purification method was similar to that described above. The resultant carbon dot sample was identified as CD-MW.
Specific amounts of the above sample solution were analyzed using UV-absorption and fluorescence spectroscopy, and the freeze-dried samples were used for structural analysis.

Characterization of CDs
The UV-Vis absorption spectra of the aqueous CD solutions were recorded using a UV-2800 UV-Vis spectrophotometer (Unico Instruments, China). Photoluminescence emission measurements were recorded using fluorescence spectroscopy (F98; Lengguang Technology, China). The CD quantum yields were calculated using quinine sulfate (quantum yield of 0.54) in 0.1 mol/L H 2 SO 4 as the standard (Guo et al. 2020).
TEM images were acquired using Tecnai G2 F20 instrument (FEI, Netherlands) at an accelerating voltage of 100 kV. The CD crystal structures were analyzed using an XRD-7000 instrument (Shimadzu, Japan) under CuKα X radiation. The elemental composition and valence states were determined by XPS using an EscaLab 250Xi instrument (Thermo Scientific, USA) and the surface groups on the CDs were evaluated using FTIR (NEXUS, Thermo Nicolet, USA) using KBr pelleting.

Statistical analysis
Data are expressed as means and standard deviations and were compared by analysis of variance, followed by Duncan's multiple-range test. Statistically significant differences between different groups of data are indicated by different letters (adjusted P < 0.05).

Optimization of the crushing method for CD preparation
Fish scales were cleaned and dried to obtain the precursor, and the effects of different methods of crushing the fish scales on the quantum yield of CDs were investigated. The quantum yield of CDs prepared using the conventional hydrothermal method of freeze-grinding the fish-scale powder (6.05 ± 0.05%) was significantly higher than that obtained using the conventional ballgrinding method (3.87 ± 0.05%) and shearing of the fish scales (3.26 ± 0.06%). The macroscopic structure of the fish scales obtained using the three different crushing methods was examined under a stereo microscope (Fig. S2). Compared with the sheared fish-scale powder, the ground (conventional ball-grinding and freeze-grinding) fish-scale powder had smaller particles, a larger specific surface area, and a larger area of contact with the dispersion system, which would make CD synthesis more facile and efficient. The fish-scale powder obtained by freezegrinding was finer, more uniform, and more dispersible than that obtained by conventional ball milling, as the fish scales frozen in liquid nitrogen were more brittle and more easily fractured during the freeze-grinding process Xia et al. 2022).

Optimization of the reaction parameters for CD preparation
CDs were prepared using the freeze-ground fish-scale powder as a precursor, and the conventional hydrothermal and microwave preparation methods were further optimized.

Optimization of reaction parameters for CD-HT
CD-HT was prepared by a carbonization reaction under hydrothermal conditions with fish scales as a precursor. The effects of hydrothermal temperature, hydrothermal time, and NaOH concentration on the quantum yield were studied (Fig. S3). Based on univariate experiments, NaOH concentration (0.03 mol/L, 0.05 mol/L, 0.1 mol/L), hydrothermal temperature (180 °C, 190 °C, 200 °C), and hydrothermal time (2 h, 3 h, 4 h) were selected and analyzed using an orthogonal table L9(3 4 ).
The orthogonal experiment was analyzed using range analysis and variance analysis (Table S1). Hydrothermal temperature was determined to be the main factor affecting the quantum yield of CDs, followed by NaOH concentration and hydrothermal time. Of the nine trials in the orthogonal test, the preferred combination was determined to be a NaOH concentration of 0.03 mol/L, reaction time of 4 h, and temperature of 200 °C. The quantum yield of CD-HT at these conditions was 6.04 ± 0.05%. The production yield of CD-HT was 7.1 ± 1.1%.

Optimization of reaction parameters for CD-MW
Similar to the evaluation of CD-HT, the effects of microwave conditions (NaOH concentration, microwave power, microwave time) on the quantum yield of CD-MW were analyzed using single-factor analysis, and the results are shown in Fig. S4. NaOH concentration (0.01 mol/L, 0.03 mol/L, 0.05 mol/L), microwave power (240 W, 400 W, 560 W), and microwave time (1 min, 1.5 min, 2 min) were selected using the orthogonal table L9(3 4 ) for orthogonal experimental analysis.
The results of the orthogonal experiment (Table S2) suggested that the NaOH concentration with the highest R-value had the greatest influence on the fluorescence quantum yield. The influence of the other factors (microwave time and microwave power) on fluorescence quantum yield was secondary. After the verification test, the optimal scheme was as follows: NaOH concentration, 0.01 mol/L; microwave time, 1.5 min; and microwave power, 400 W. The quantum yield of CD-MW using these parameters was 5.10 ± 0.13%. The corresponding yield of CD-MW was 4.5 ± 1.2%.

Optical properties of CDs
The optical properties of the two CD solutions were compared and the results are shown in Fig. 1. The UV-Vis absorption spectra revealed that both CDs have a wide weak absorption band at 250-290 nm, which may be attributed to the π-π* transition of the C = C bonds during aromatic sp 2 hybridization (Pandiyan et al. 2020;Reckmeier et al. 2016;Shamsipur et al. 2018). The fluorescence properties of the two CDs were further studied based on their fluorescence spectra. The optimal excitation of CD-HT was 360 nm and the optimal emission was 430 nm, whereas the optimal excitation of CD-MW was 310 nm and optimal emission was 400 nm. The difference was related to the particle size, domain size, and surface states of the CDs (Ma et al. 2022, Pundi andChang 2022). The illustrations are images of the two CD solutions visualized using a fluorescent lamp and an ultraviolet lamp, respectively. The two CD solutions appeared transparent in visible light and were evenly dispersed without any aggregation or precipitation. Furthermore, they exhibited a bright blue fluorescence under ultraviolet light (365 nm). The fluorescence intensity of CD-HT was stronger than that of CD-MW, which was consistent with their quantum yield.
The fluorescence characteristics of both CDs were analyzed (Fig. 2). When the excitation wavelength of CD-HT increased from 320 to 360 nm, the fluorescence intensity increased first and then decreased, and an emission peak redshift was observed. The CD-HT emission spectra exhibited excitation-related properties, which may be due to the difference in the size of the CDs and the presence of different organic groups on the surface of the CDs that resulted in different surface state distributions (Khan et al. 2019;Liu et al. 2011). Fluorescence spectrometry of CD-MW revealed that when the excitation wavelength was increased from 280 to 350 nm, there was little change in the corresponding emission positions at different excitation wavelengths, together with the observation of a single fixed emission at 400 nm, indicating an almost excitation-independent emission behavior. The nonexcited-state behavior may be because the luminescence properties of CD-MW depend on the surface state rather than the quantum size effect, likely because the amino group enhances the surface functionalization of the carbon surface group, and the surface state of these CD-MW should be fairly uniform (Karami et al. 2020;Krysmann et al. 2012;Yang et al. 2020).
Stable fluorescence performance of CDs is crucial for practical applications Thulasi et al. 2020;Yun et al. 2022). Therefore, we investigated the stability of both types of CDs under light and ion interference (Fig. 3). Both CDs were continuously excited under a xenon lamp for 120 min to measure their photostability. This showed that the luminescence intensity was almost unchanged, indicating the photostability of the prepared CDs. Next, the effect of ionic intensity on the emission spectrum of CDs was determined, and the fluorescence intensities of both CD variants were monitored at different salt concentrations. When the concentration of NaCl was increased from 0.2 to 1.0 mol/L, the fluorescence intensity was almost unchanged. In addition, both CDs yielded uniformly clarified solutions without obvious precipitation caused by salt interference. This finding showed that both CD variants had stable fluorescence characteristics at higher salt concentrations. Collectively, our findings revealed that both CD variants exhibited light stability and salt tolerance, which are important characteristics for their practical application.
In summary, both CDs exhibited good light stability and salt tolerance, but their optical properties were different. The excitation was inconsistent (CD-HT excitation at 360 nm and CD-MW excitation at 310 nm); CD-HT exhibited emission behavior related to excitation, whereas CD-MW exhibited emission behavior that was almost independent of excitation. Moreover, the quantum yield of CD-HT was slightly higher than that of CD-MW. This may be related to the surface state and was also likely affected by the size. Thus, in our subsequent analysis, we focused on the structural differences between CDs.

Morphology of CD-HT and CD-MW
The crystal structure of the CDs was analyzed using XRD and the results are shown in Fig. 4. Both CDs showed wide and obvious diffraction peaks with a diffraction angle of 2θ = 21.55° which, corresponding to the layer spacing d, was 0.41 nm. Compared with bulk graphite (about 0.34 nm), the fluorescent carbon nanoparticles showed larger spacing, which may be due to the abundance of O and N groups, resulting in weak aromatic-layer interactions between the graphite layers (Rodríguez Padrón et al. 2018;Yang et al. 2018). Therefore, both CDs have the characteristics of amorphous carbon, and the degree of graphitization was low.
The size and structure of the CDs were analyzed using TEM (Fig. 5). It can be seen from the TEM image that CD-HT is mostly uniform, appearing as quasi-spherical nanoparticles with good dispersion. The histogram of size distribution indicated that the particle size-distribution range of CD-HT was 1.7-6.5 nm, average size was 4.02 ± 0.89 nm, and the average diameter was < 5 nm. On the other hand, CD-MW appeared uneven with irregular sheets, with a particle size distribution range of 2.4-7.2 nm and an average particle size of 4.23 ± 1.03 nm. These significant differences in morphology between the CD variants may account for the differences in their optical properties. Higher magnification HRTEM images (Fig. 5C, D) showed that the two CDs had similar microscopic morphologies, with both presenting a graphitized carbon core surrounded by an amorphous shell. The degree of graphitization of the two CDs was similar, with non-uniform lattice fringe spacings of approximately 0.2 nm, indicating that the structure of carbon core was not uniform (Luo et al. 2019). These observations are consistent with the results of the XRD analysis.

Surface properties of CD-HT and CD-MW
The chemical structures and states of the elements in the CDs were analyzed using XPS. The full XPS spectrum (Fig. S5) showed that the prepared CDs contained C, N, O, and Na. A small amount of Na was derived from residual Na ions remaining after the CD preparation. Ca and P were not found in the prepared CDs, indicating no obvious entrainment of these elements into the complex amorphous carbon structure during CD synthesis. This could be attributed to the NaOH system that was used. Under alkaline conditions, the solubility of the main inorganic component, hydroxyapatite, in fish scales was low, resulting in only trace Ca and P ions in the reaction system, which cannot easily participate in CD formation. Elemental composition analysis (Table 2) revealed significant amounts of self-doped N element in both CDs. The passivation effect of the doped elements conferred excellent optoelectronic properties upon the CDs (Manioudakis et al. 2019). Further improvement showed that there were fewer doped elements in CD-HT than in CD-MW, which was likely related to the rapid reaction using the microwave method.
The high-resolution spectra of the C 1s , N 1s , and O 1s bands were deconvoluted, as shown in Fig. 6. The C element exhibited four chemical states at different peak positions: C-C/C = C near 284.8 eV, C-N near 285.9 eV, C-O near 286.5 eV, and C = O near 287.6 eV (Fang and Zheng 2021). The contents of the different chemical states calculated by peak area are as follows: for CD-HT, the C-C/C = C content was 59.8%, C-N content was 10.9%, C-O content was 6.5%, and C = O content was 22.9%. For CD-MW, the C-C/C = C content was 39.8%, C-N content was 24.2%, C-O content was 8.5%, and C = O content was 27.3%. CD-HT contained more C-C/C = C, whereas the other three chemical states (structural defects) were less; these findings indicated that the doping of heteroatoms in CD-HT was less, which was consistent with the results of the elemental analysis of the total spectrum.
In addition to C, the N 1s spectrum comprised four main peaks (Fig. 6B), which were attributed to pyridine N (398.5 eV), amine N (399.7 eV), pyrrole N (400.2 eV), and quaternary N (401.3 eV) (Liu et al. 2020;Pillar-Little and Kim 2017). By analyzing the ratios of the different types of N, we found that CD-HT had a higher proportion of pyrrole N, leading to better charge mobility and donor-acceptor ability, which is beneficial to improved quantum yields. Moreover, a small amount of oxidized N was present in CD-HT. The O 1s spectrum (Fig. 6C) can be deconvoluted into two peaks at 531.1 eV and 532.4 eV, which are attributed to the presence of C = O and C-OH/C-O-C, respectively (Liu et al. 2020). Oxygen in CD-MW tends to exist in the form of C-OH/C-O-C, whereas oxygen in CD-HT exists in the form of C = O, which is also conducive to the improvement of quantum yield.
The functional groups of the CDs were further analyzed using FTIR, and the main signal regions that were identified are shown in Fig. 7 (Dutta 2017). The total spectrum (Fig. S6) and specific peak parameters are shown in the Supporting Material (Table S3 and Table S4).
In Fig. 7A, the peak at ~ 3560 cm −1 represents the O-H stretching vibration of − COOH in CD-HT, which is significantly stronger than that in CD-MW, indicating that CD-HT contains more − COOH, which is consistent with the results from XPS. For the signal peaks representing the N-H stretching vibration at ~ 3450 cm −1 , and N-H and O-H stretching vibrations at ~ 3300 cm −1 , the peak area of CD-HT is smaller than that of CD-MW, which is related to the less  doping of N and O elements in CD-HT. The wider peak shape of CD-MW was attributed to higher hydrogen bond association, indicating the hydrophilicity of CD-MW. The signal peaks of = C-H and Ar-H (3060 cm −1 ) and of saturated C-H asymmetric vibrations (2860 cm −1 , ~ 2960 cm −1 , and 2944 cm −1 ) in CD-HT were significantly stronger than those of CD-MW, which also indicated the high carbonization degree of CD-HT. As shown in Fig. 7B, the peak area representing the C = C stretching vibration (~ 1650 cm −1 ) in CD-HT was significantly larger than that in CD-MW, which is consistent with previous XPS results. There was no significant difference between the signals of the N-H bending vibration and C-N stretching vibration of the amide at ~ 1560 cm −1 . The signal at ~ 1520 cm −1 attributed to the asymmetric stretching vibration of − NO 2 in CD-HT was significantly higher than that in CD-MW, indicating higher − NO 2 content. The peaks in the range of 1480 ~ 1350 cm −1 were mainly attributed to C-H bending vibration in − CH 3 / − CH 2 -; the corresponding signal peaks in CD-HT were more obvious, indicating a higher number of C-H bonds, which were closely related to its higher degree of carbonization. Additionally, the signal peak (1420 cm −1 ) of C-O stretching vibration and O-H bending vibration in CD-HT attributed to − COOH was stronger than that in CD-MW, which is consistent with the analysis results of − COOH (Fig. 7A).
The above analyses revealed that due to the different heating methods used, CDs prepared from fish scales using the conventional hydrothermal and microwave methods exhibited obvious differences in optical properties, structural morphology, and surface group composition (Fig. 8).
Microwave heating can rapidly and uniformly increase the temperature of the reaction system and promote the carbonization of organic matter (mainly collagen) in fish scales to form CDs. The result of the rapid reaction is that there is greater transport of the abundant N (~ 17%) in the organic matter to the CDs, resulting in higher N doping. On the other hand, the slow heating using the conventional hydrothermal method was helpful in obtaining a more balanced product with lower local energy. Thus, the N doping in the CD-HT was less, but a relatively large proportion in the form of pyrrole N was present and a certain amount of − NO 2 was formed.
Microwave heating results in higher requirements for reaction vessels. The use of traditional stainless steel-sealed reaction vessels is limited, whereas conventional PTFE or glass containers cannot provide effective sealing strength. Thus, the temperature of the reaction system is usually only slightly > 100 °C. Therefore, only a part of the protein (mainly collagen) could dissolve to form the nanosheet-like fluorescent material. To confirm this, TGA, FTIR, and XRD were used to analyze the composition of the precipitate in the CD preparation systems as shown in Fig. S7. It was found that the preparation residue of CD-MW contained more organic matter. In contrast, the heating temperature of the hydrothermal reaction has no obvious restrictions and can be preset to temperatures > 100 °C. Therefore, CDs with higher a higher degree of carbonization, smaller size, and higher C = O/COOH content were obtained. These structural features are the main reasons for the higher quantum yield and lower excitation wavelength of CD-HT.

Conclusions
The effects of conventional hydrothermal and microwave methods on the fluorescence performance and structure of CDs were investigated using fish scales as a precursor. The CD variants prepared under optimal conditions were characterized and compared. Both CDs exhibited blue fluorescence under ultraviolet light and had good dispersion, light stability, and salt tolerance. The rapid and uniform microwave heating method was more conducive to the selfdoping of N. However, due to the sealing limitations of the reaction vessel, the low temperature of the microwave method leads to incomplete dissolution of the organic matter in fish scales; moreover, the dissolved organic matter is more difficult to completely dehydrate and condense. Thus, nanosheet-like fluorescent CDs are formed wherein the emission behavior has no significant correlation with excitation. Although the N doping of CDs prepared using the conventional hydrothermal is low, the relative pyrrolic N content is high, which is beneficial in improving quantum yield. Additionally, the controllable high temperature and sealed environment of the conventional hydrothermal method can promote the dehydration and condensation of organic matter in fish scales to form CDs with a higher degree of carbonization, uniform size, and higher C = O/ COOH content, as well as higher quantum yield and excitation wavelength-related emission behavior. These findings will provide guidance for the structural optimization and practical use of biomass-derived carbon dots.