Snapshots into carbon dots formation through a combined spectroscopic approach

The design of novel carbon dots with ad hoc properties requires a comprehensive understanding of their formation mechanism, which is a complex task considering the number of variables involved, such as reaction time, structure of precursors or synthetic protocol employed. Herein, we systematically investigated the formation of carbon nanodots by tracking structural, chemical and photophysical features during the hydrothermal synthesis. We demonstrate that the formation of carbon nanodots consists of 4 consecutive steps: (i) aggregation of small organic molecules, (ii) formation of a dense core with an extended shell, (iii) collapse of the shell and (iv) aromatization of the core. In addition, we provide examples of routes towards tuning the core-shell design, synthesizing five novel carbon dots that all consist of an electron-dense core covered by an amine rich ligand shell.

SP were obtained via microwave irradiation of an aqueous solution of L-arginine and ethylenediamine (0.5 mmol). Typically, L-arginine (87.0 mg), EDA (33.0 µL) and Milli-Q water (100.0 μL) were heated at 200 W and 26 bar with a Tmax of 250 °C for 16 cycles of 15 s heating and 5 s cooling. The reaction mixtures were filtered (0.1 μm microporous membrane), washed with Milli-Q water, and dialyzed against Milli-Q water. Regarding the yield of these fractions, we have attempted to quantify them, but this procedure posed some issues. We noticed a progressive change of color of the material (from light yellow to dark brown) during their handling (dialysate recovery, rotary evaporation of the aqueous solutions and freeze-drying) which would lead to incorrect characterizations (or conclusions) regarding the progress of the reaction. For this reason, we have collected only the dialysates from the first two water changes (first 8 hours), concentrated them, and used those for characterization purposes. Therefore, for the photophysical studies these fractions were used to have a qualitative comparison with the CNDs purified over time.
AFM. Atomic force microscopy (AFM) images were obtained with a Nasoscope IIIa, VEECO Instruments. As a general procedure to perform AFM analyses, tapping mode with a HQ:NSC19/ALBS probe (80kHz; 0.6 N/m) (MikroMasch) from drop-cast samples in an aqueous solution (concentration of few mg/mL) on a mica substrate was performed. The obtained AFM images were analyzed in Gwydion 2.46.
Kaiser Test. Kaiser test kit was purchased by Sigma Aldrich and to determine the amount of free terminal amino groups on the carbon dots we followed the standard literature protocol. 1 FLQY. The quantum yield measurements were performed with quinine sulphate in 0.10 M H2SO4 (literature quantum yield 0.54) as the standard. 2 Samples were excited at 320 nm. The fluorescence quantum yields were calculated according to Supplementary Equation 1: I is the measured integrated fluorescence emission intensity, f is the absorption factor, h is the refractive index of the solvent and F is the quantum yield. The index x denotes the sample, and the index st denotes the standard.

SAXS and WAXS.
Small-and wide-angle X-Ray scattering measurements were performed at the Austrian SAXS beamline of the electron storage ring ELETTRA using a photon energy of 8 keV. X-Ray scattering experiments were performed using multiple setups (sample-detector-distances), 3 hence the SAXS and WAXS patterns from different experiments might present different q-ranges. In any case, the beamline setup was adjusted to a sample to SAXS-detector [Pilatus 1M detector (Dectris, Switzerland)] distance of 760-800 mm, resulting in an accessible q-range of approx. 0.1-10 nm -1 . A secondary WAXS detector [Pilatus 100K detector (Dectris, Switzerland)] was placed at a distance of approx. 320-340 mm with a tilting of approx. 20-22° in respect to the vertical axis to result in an accessible q-range 9-19 nm -1 . Images were taken synchronized with both detectors with at least 6 exposures of 20 seconds per sample to check for radiation damage. Reference patterns to calibrate the q-scales were collected of silver-behenate (d-spacings of 5.838 nm) for the SAXS-and p-bromobenzoic acid for the WAXS-regime. All measurements were done using 1.5 mm quartz capillaries (VGM Glass, Germany). The radial averaging and the image calibration were conducted using the FIT2D software. 4 All presented data was corrected for fluctuations of the primary intensity and the corresponding background has been subtracted from each solution scattering pattern.
UV-Vis and Fluorescence. UV-Vis spectra were recorded on a PerkinElmer Lambda 35 UV-Vis spectrophotometer. Fluorescence spectra (including 2D excitation-emission spectra) were recorded on a Varian Cary Eclipse Fluorescence Spectrophotometer. All the spectra were recorded at room temperature using 10 mm path-length quartz cuvettes. Emission intensities are reported as arbitrary units (a.u.) since the FLQY was calculated for each sample (reported in Figure 3a).
XPS. X-ray photoemission spectroscopy (XPS) spectra of the samples were measured on a SPECS Sage HR 100 spectrometer using a non-monochromatised Mg-Kα radiation of 1253.6 eV and 250 W, in an ultra-high vacuum chamber at pressure below 8×10 -7 mbar. For each analysis, an aqueous solution (ca. 3 mg/mL) of material were deposited on a gold thin film. The calibration was done using the 3d5/2 line of Ag. High-resolution spectra were collected with pass energy of 15 eV and 0.15 eV/step. SpecsLab Prodigy (Specs) and XPST 1.3 software were used for data processing and fitting. Curve fittings of the C1s and N1s spectra were realized using a Gaussian-Lorentzian peak shape (GL ratio of 0.3) after performing a linear background correction, to finally obtain the relative percentage of each type of bond inside the analyzed sample.
In the following paragraphs, the model details will be explained. The refined model fits can be seen in Figure 2a and Supplementary Figure 4, 6, 10. A detailed summary of all refined and derived parameters can be found in Supplementary Table 1.

Form-Factor scattering: Schulz-distributed spheres
To describe the scattering from the electron dense CND regions (carbogenic core), we use the analytic expression for polydisperse Schulz-distributed spheres. This model is linked to the parameters *'0 and , denoting the center and width/skewness of the number-weighted size distribution. [6][7][8] The mean volume-weighted sphere-size 12(3 is calculated by the first statistical moment of the r 3 -scaled volume distribution.

Structure-Factor contribution
We use a sticky-hard-sphere potential to describe the interaction between CNDs. A graphical overview of the underlying potential ( ) between particles separated by a given distance is depicted in the scheme below (graphical illustration explaining the fitting parameters used for the calculation of the sticky-hard-sphere structure factor. The red line describes the potential ( ) as a function between two particles separated by a given distance . Adapted from literature. 5,11 ). The structure-factor *+* ( ) corresponding to this model can be calculated analytically according to the literature. 9,12 Throughout the fitting process, the potential well size and the particle volume fraction were kept constant at = 2 and = 0.003. The shell thickness "02)) is calculated by comparing the minimal interaction distance 2 • • +" with the volume-weighted mean diameter of the electron dense core 2 • 12(3 , such that "02)) = 2 • ( • +" − 12(3 ).

Porod-contribution of large-scale aggregates
We describe the scattering of large-scale aggregates with dimension beyond the resolution limit of SAXS using a Porod contribution according to Supplementary Equation 3:

Supplementary Tables
Supplementary Table 1. Fitting results corresponding to the refined model curves in Figure 2a and Supplementary Figure 4, 6, 10. Here, parameters r 6789 and t :;7<< were calculated from the refined model parameters (see description above for details).  Supplementary Figure 5. WAXS of RM. WAXS patterns of the RM after the first reaction cycles (red, blue) compared to the scattering of EDA in water (black, 33 μL EDA in 100 µL H2O -equivalent to solvent conditions for MW synthesis). Note, that the scattering patterns were corrected for scattering of H2O by background subtraction (see Supplementary Methods). Here, the strong suppression of the EDA WAXS peak after the first cycle (see black to red) indicates consumption of EDA within the given time. In the second cycle, the WAXS intensity drastically increases, concomitant with a peak center shift to lower angles (higher distances in real space).