Visible to near-IR fluorescence from single-digit detonation nanodiamonds: excitation wavelength and pH dependence

Detonation nanodiamonds are of vital significance to many areas of science and technology. However, their fluorescence properties have rarely been explored for applications and remain poorly understood. We demonstrate significant fluorescence from the visible to near-infrared spectral regions from deaggregated, single-digit detonation nanodiamonds dispersed in water produced via post-synthesis oxidation. The excitation wavelength dependence of this fluorescence is analyzed in the spectral region from 400 nm to 700 nm as well as the particles’ absorption characteristics. We report a strong pH dependence of the fluorescence and compare our results to the pH dependent fluorescence of aromatic hydrocarbons. Our results significantly contribute to the current understanding of the fluorescence of carbon-based nanomaterials in general and detonation nanodiamonds in particular.

. Schematic drawing of the custom built in-solution fluorescence spectroscopy setup.

Experimental parameters used for data acquisition using the above setup
Pulse repetition rate: 80 MHz (fluorescence spectra) / 10 MHz (fluorescence decay) Spectrometer/ CCD camera: 5 pixel binning, 10 seconds integration time   Table S1. Nanoparticle (NP) solution, HCl and NaOH solutions used to prepare the samples investigated in this study. Sample 4 (no HCl or NaOH) was used to investigate the excitation wavelength dependence shown in Figure 2 and 3 in the main text.  Sample / pH 4 / 6.1 5 / 7.7 6 / 9.7 7 / 10.5 8 / 11.8 9 / 12.7 Figure S3. Energy-dispersive X-ray spectroscopy (EDS) results for DND particles show carbon (Kα at 0.277 keV) to be the predominant element in our sample. It also contains significant amounts of oxygen (Kα at 0.525 keV) and we find trace amounts of Cu, Si, Zr and Ca. Particles were deposited on a holey carbon TEM grid. Figure S4. Electron energy loss spectroscopy (EELS) results for the DND particles compared to glassy carbon (GC) used to determine sp 2 and sp 3 carbon content in our samples. A: The low loss EELS spectra for DND and GC. The dominant feature is the plasmon peak which is be a measure of the effective density of the material, assuming a free elctron gas model. For example, GC has a plasmon peak at 22.5eV which equates to a density of 1.54 g cm -3 . The plasmon peak of DND however consists of two distinct components with peaks at 22.3 eV (1.51 g cm -3 ) and 34 eV (3.51 g cm -3 ). This suggests the presence of diamond as well as graphitic material in our sample. B: Ionization K-edge spectra of the same samples. The presence of both sp 2 and sp 3 bonded carbon is also evident in the ionisiation K-edge spectra where the π* transition is minimal for DND compared to GC and shows that about 18% of the carbon bonds in our sample are sp 2 hybridised and 82 % sp 3 bonded as calculated below.

Acquisition and analysis of EELS spectra
Electron Energy Loss Spectra (EELS) were collected on a JEOL 2100F TEM operating at 200keV with a Gatan Imaging Filter (GIF Tridium) in imaging mode. The Carbon K-edge and low-loss plasmon spectra were both acquired. The K-edge spectra were processed by removing the inherent background and the contribution due to multiple scattering removed. To obtain the sp 2 fraction, the 1s-2π* feature was fitted using a Gaussian distribution and the intensity was compared to the intensity of (1s-2π*) + (1s-2σ). This ratio was compared to the K-edge spectra collected from a glassy carbon sample which is 100% sp 2 according to the below formula Where !" * is the integral under the 1s-2π* feature of the sample, ! (△ ) is the integral under the (1s-2π*) + (1s-2σ) features of the sample !" * is the integral under 1s-2π* feature of glassy carbon ! (△ ) is the integral under the (1s-2π*) + (1s-2σ) features of glassy carbon Figure S5. Raw fluorescence spectra for DND samples in water at neutral pH (colored lines) compared to water only (black line) for all excitation wavelengths as indicated in the graphs.     Figure 3B in the main text. B: Fluorescence decay traces for the different excitation wavelengths as indicated in the graph. The long fluorescence lifetime component Tau2 was determined by fitting a single exponential to the decay traces a shown in the graph (black lines). We find the lifetimes determined this way to vary by ±0.35 ns depending on the exact region used for fitting, which is reflected in the error bars shown in Figure 3C in the main text. The same approach was used for the analysis of the pH dependent results shown in Figure 5 on the main text.  pH 3.7 4.5 5.4 6.2 7.7 9.7 10.5 11.8 IRF Figure S8. A: Integrated fluorescence intensity as a function of time. A DND nanoparticle solution (300 µL, 1.33 mg/mL) was excited with 500 nm light and the fluorescence collected with a spectrometer at 5 frames per second. An aqueous solution of HCl (100 µL, 1 mM) was added at time t=0 seconds. The fluorescence decreases by 75% within less than 0.5 s and remains stable thereafter. This decrease is caused by a dilution of the starting solution by 33% as well as the decrease in pH, resulting in a decreases of fluorescence in agreement with Figure 5B in the main text. Nanoparticle aggregation is a diffusion-limited process. The mean displacement of a 5 nm spherical particle due to Brownian motion after 500 ms is below 1 nm, which makes a collision with another particle (as a prerequisite for aggregation to occur) within this timeframe highly improbable in our nanoparticle solutions. B: Same data as in A, but zoomed into the region where the HCl addition occurs. C: The same experiment as in A, but using NaCL instead of HCl. Here, the intensity decreases due to the dilution of the solution, but only by around 23% instead of the expected 33% dilution. This is likely caused by incomplete mixing, which can be difficult to achieve in these measurements. D: Fluorescence spectra of DND particles dispersed in water (black line) and in 250 µM NaCl (green line), which is the final HCl concentration used in panel A and NaCl concentration used in panel C. The spectra were measured ~ 30 s after the addition of either water or NaCl. In the presence of salt the spectrum shows a slight red-shift, which is typical for partially aggregated particles. Overall, the difference in fluorescence intensity is < 1%, which is within the experimental A B igure S9     Figure S11. Absorption spectrum of DND particles in water (0.04 mg/ mL) for the spectral range from 250 nm to 800 nm. The data was acquired using a Cary 700 absorption spectrometer (Agilent Technologies) and an integrating sphere.

Calculation of the Debye length
The Debye length λ D is commonly defined by the equation 1 : where ε 0 is the permittivity of free space, ε r the dielectric constant of the solvent, k BT the thermal energy, c i the ionic concentration of the i-th ion species in solution, e the elementary charge and z i the valency of the i-th ion species. The Debye length is a characteristic length for the range of the electrostatic potential into the solvent.

Calculation of mean square displacement
The mean square displacement X of a particle in 3 dimenasions was estimated using the equation = 3 where D is the Stokes-Einstein diffusion coefficient and t is time. D was calculated using the following equation and parameters: = ! 6 k B T = 4.11 E-21 J (thermal energy) η = 8.94 E-04 kg m -1 s -1 (viscosity of water) r = 2.5 nm (particle radius) For a diffusion time of 1 second and a particle size of 5 nm (most particles are larger than this so this is an upper bound for X) the mean square displacement is ~0.3 nm.

Estimation of the average particle separation in solution
We have used the Wigner-Seitz radius to estimate the average separation of DND particles dispersed in water at a concentration of 3 µM or ~ 1.8 e18 particles per liter (1 dm 3 ). The Wigner-Seitz radius in 3D is given by: where V is the volume of the solvent and N the number of particles. Using the values given above one obtains R s = 51 nm. The nearest neighbor distance would thus be 102 nm (2 R s ). This value is more than two orders of magnitude higher than the mean square displacement of 5 nm particles after diffusing for 1 second in water.

Determination of the relative fluorescence quantum yield
The quantum yield (Φ) was determined using the equation: where Φ DND and Φ F are the quantum yields of DND and fluorescein ,respectively, and the corresponding gradients b of the fits to the data shown in Figure S12 C and F using the equation = . The quantum yield of fluorescein in 10 mM NaOH of Φ F =0.93 was used as reported by Kubista et al. 2 This yields a value of Φ DND = 0.22% for the quantum yield of the DND particles.