Small variations in reaction conditions tune carbon dot fluorescence

The development of robust and reproducible synthetic strategies for the production of carbon dots (CDs) with improved fluorescence quantum yields and distinct emission profiles is of great relevance given the vast range of applications of CDs. The fundamental understanding at a molecular level of their formation mechanism, chemical structure and how these parameters are correlated to their photoluminescence (PL) properties is thus essential. In this study, we describe the synthesis and structural characterization of a range of CDs with distinct physico-chemical properties. The materials were prepared under three minutes of microwave irradiation using the same common starting materials (d-glucosamine hydrochloride 1 and ethylenediamine 2) but modifying the stoichiometry of the reagents. We show that small variation in reaction conditions leads to changes in the fluorescent behaviour of the CDs, especially in the selective enhancement of overlapped fluorescence bands. Structural analysis of the different CD samples suggested different reaction pathways during the CD formation and surface passivation, with the latter step being key to the observed differences. Moreover, we demonstrate that these materials have distinct reversible response to pH changes, which we can be attribute to different behaviour towards protonation/deprotonation events of distinct emission domains present within each nanomaterial. Our results highlight the importance of understanding the reaction pathways that lead to the formation of this carbon-based nanomaterials and how this can be exploited to develop tailored materials towards specific applications.

H and 13 C (HSQC) NMR spectra were measured in D2O at 500 MHz on a Varian spectroscope and their chemical shifts are quoted in parts per million (ppm). Fourier-transform infrared resonance was conducted on a Bruker ATR. Atomic force microscopy (AFM) was carried using a multi-mode VIII microscope with Nanoscope V control utilising a Fast-scan head unit whilst operating under PeakForce feedback control, using a SCANASYST-AIR-HR cantilever (Bruker, CA, USA) with a nominal spring constant of 0.4 N/m. Tip deconvolution using a Ti roughness sample (RS-12M; Bruker, CA, USA) revealed a tip radius of 4.98 nm 4 nm from the tip apex. High-resolution transmission electron microscopy (HRTEM) was conducted on a Jeol 2100F microscope with an accelerating voltage of 200 kV; for this characterization, a drop of the CDs in methanolic suspension (5mg/mL) was carefully applied to a 200 nm mesh carbon-coated copper grid and dried at ambient temperature. X-ray diffraction characterization was done using a Kappa Apex II diffractometer. Absorbance measurements were conducted on Cary UV-Vis 50 spectrophotometer in 3500 µL quartz cuvettes (ThorLabs). 2D Fluorescence measurements were obtained with a Perkin-Elmer LS45 in 3500 µL quartz cuvettes (ThorLabs) and samples were diluted prior analysis in order to avoid effects of self-absorption on the emission spectra (concentrations adjusted to an absorption lower than 0.1 at 390 nm). The conversion of the 2D fluorescent spectrum to x and y coordinates in the chromaticity space was achieved concerning the CIE colour matching functions. The approximation to sRGB from the chromaticity coordinates was done with PAL standard primaries using floating-point operations and a nonlinear correction using γ = 2.2.

Time-Correlated Single-Photon Counting (TCSPC)
An ultrafast oscillator (Coherent, Chameleon Ultra II) producing pulses at 80MHz was used to drive TCSPC experiments. The oscillator laser output was frequency-doubled in a b-barium borate crystal. The repetition rate of the laser was reduced to 3MHz using a Bragg cell driven by a high power radio frequency pulse generator (APE). These pulses were lightly focused into the sample (1cm pathlength cuvette) and the fluorescence detected at 90° relative to the laser excitation. The collected fluorescence was passed through a broadband polarizer set to magic angle and a band-pass filter to remove laser scatter and selectively acquire the sample fluorescence. The resulting light was focused onto an avalanche photodiode (IDQ, ID100-20-REG). The photon counts from the detector were acquired by a time-to-digital converter (Swabian Instruments, Time Tagger 20) and binned into histograms. TCSPC data were acquired in LabView. All data-analysis were performed in Matlab, where fluorescence decays were fit to a convolution of the fitted IRF function and two exponential decays.

X-ray Photoelectron Spectroscopy (XPS).
A Kratos Axis Ultra DLD system was used to collect XPS spectra using monochromatic Al K α X-ray source operating at 140 W (10 mA x 14 kV). Data was collected with pass energies of 160 eV for survey spectra, and 20 eV for the high-resolution scans with step sizes of 1 eV and 0.1 eV respectively. Samples were either pressed on to doubled sided Scotch tape (type 665), or for viscous samples, spread on to a UV cleaned Si wafer. The system was operated in the Hybrid mode, using a combination of magnetic immersion and electrostatic lenses, and acquired over an area approximately 300 700 µm 2 . A magnetically confined charge compensation system was used to minimize charging of the sample surface, and all spectra were taken with a 90° take of angle. A base pressure of 1 x 10 -9 Torr was maintained during the collection of the spectra. Data were analysed using CasaXPS (v2.3.23) after subtraction of a Shirley background and using modified Wagner sensitivity factors as supplied by the manufacturer.

Synthesis of CDs
Chemicals were purchased and used without further purification. D-glucosamine hydrochloride (1.00 g, 4.63 mmol) was dissolved in distilled H2O (20 mL) in a 250 mL round bottom flask. ethylenediamine (EDA, quantity depending on the experiment) was then added to the solution and stirred vigorously for 30 min to ensure homogeneity. The round bottom flask was then placed in a domestic microwave 700 W (Wilko's Homebrand) and the solution was reacted for 3 min. After the microwave synthesis, a viscous brown residue was obtained which was taken up in distilled H2O (10 mL). The crude solution was purified via centrifuge filtration (MWCO of 10 kDa, 8500 rpm, 30 min). The supernatant was concentrated in vacuo (or lyophilised) to yield a viscous brown syrup. The sample was then taken through size exclusion chromatography using a Sephadex G-10 resin acquired from GE Healthcare Life Sciences. Finally, the CDs were dialysed for 24 hours replacing with pure water frequently, employing a 500 Da dialysis membranes (purchased from Sigma-Aldrich). The resulting solution was concentrated in vacuo. Figure S1: Hydrogen NMR resonances of synthesis of CDs varying the stoichiometry S4

Diffusion ordered spectroscopy (DOSY)
An indirect estimation of the hydrodynamic radius of the CDs (σ CD ) was possible using the Debye-Einstein equation and a correction for the diffusion coefficients [1]: The Stoke Einstein equation tell us that.
Where D is the diffusion coefficient Kb is the Boltzmann constant, T is the temperature, c is a parameter that approaches 6 as the hydrodynamic radius (σ) reaches 1 nm and η is the viscosity of the solvent.
The estimation of σ CD was calculated using a simplified relationship from the DOSY NMR (which probes the diffusion coefficient for each of the components of the 1H NMR spectrum), relative to the signal of 1 H2O (D w , 2.3×10 −9 m 2 /s, σ w , 0.1375 nm) and considering the resonances that correlate to the CDs (table 1). This allows us not only to simplify the relationship of the Stoke-Einstein equation but also to ignore differences of viscosity in the sample.  Table S1: Hydrodynamic radius of CDs. Estimation relative to the water signal using the equation (1).

X-Ray Electron Diffraction (XRD)
The crystallite size (D) of samples CD-3 and CD-8 was determined based on the broadening of the XRD signals using the Scherrer equation.
The Scherrer equations tell us that: Where D is the crystallites size, λ = 1.5418 Å is the wavelength of the X-ray source, k = 0.9 is the shape factor and βd is the peak broadening caused by the crystallite domain in radians. The βd has been approximated to the total broadening (βt) of the peak, which is determined by the full width at half maximum (FWHM). However, the βt of the XRD peaks is a contributed effect caused not only by the crystallites size broadening (βd) but also by the microstrain and the instrumental broadenings (βs and βi respectively). This is: Hence, βd can only smaller or equal than the FWHM, which will imply that the estimation for D is only smaller than the real length of D.  S9 Figure S11: XRD spectroscopies of samples CD-3 and CD-8. FWHM extracted from peak centred at 12.4° and 12.5° of 2Θ respectively.

Chromaticity Space
Figure S12: Chromaticity Space separation of samples CD-3 and CD-8 at different excitation wavelengths.

Quantum Yields
Quantum Yield (Φ) estimation was performed based on IUPAC protocol [2]. The spectral data measurements for CDs were conducted on both Perkin-Elmer LS45 and Cary UV-Vis 50 spectrophotometer in 3500 µL quartz cuvettes.  Table S3: Quatum yields measurements for CD-3 and CD-8. Figure S13: Standard Gradient Curves of CDs for QY determination

Cupric Ions Photo-quenching
The fluorescence of samples CD-3 and CD-8 was quenched with the addition of cupric ions at different concentrations. A 100 ml aqueous batch solution of CD-3 was prepared to a concentration of 800 µg/ml. The solution was partitioned into 50 ml in separated round bottom flasks and 150 mg of copper (II) chloride were added to one of the partitions. Finally, serial dilutions were prepared from the two partitions to vary the concentration of cupric ions and keep the concentration of CDs. The same procedure was employed for sample CD-8. Figure S14: Fluorescence quenching of CD-3 and CD-8 upon addition of copper(II) chloride.