Influence of Electron Donors on the Charge Transfer Dynamics of Carbon Nanodots in Photocatalytic Systems

Carbon nanodots (CNDs) are nanosized light-harvesters emerging as next-generation photosensitizers in photocatalytic reactions. Despite their ever-increasing potential applications, the intricacies underlying their photoexcited charge carrier dynamics are yet to be elucidated. In this study, nitrogen-doped graphitic CNDs (NgCNDs) are selectively excited in the presence of methyl viologen (MV2+, redox mediator) and different electron donors (EDs), namely ascorbic acid (AA) and ethylenediaminetetraacetic acid (EDTA). The consequent formation of the methyl viologen radical cation (MV•+) is investigated, and the excited charge carrier dynamics of the photocatalytic system are understood on a 0.1 ps–1 ms time range, providing spectroscopic evidence of oxidative or reductive quenching mechanisms experienced by optically excited NgCNDs (NgCNDs*) depending on the ED implemented. In the presence of AA, NgCNDs* undergo oxidative quenching by MV2+ to form MV•+, which is short-lived due to dehydroascorbic acid, a product of photoinduced hole quenching of oxidized NgCNDs. The EDTA-mediated reductive quenching of NgCNDs* is observed to be at least 2 orders of magnitude slower due to screening by EDTA-MV2+ complexes, but the MV•+ population is stable due to the irreversibly oxidized EDTA preventing a back reaction. In general, our methodology provides a distinct solution with which to study charge transfer dynamics in photocatalytic systems on an extended time range spanning 10 orders of magnitude. This approach generates a mechanistic understanding to select and develop suitable EDs to promote photocatalytic reactions.

Supplementary Note 1: On the use of stretched exponential fitting for carbon nanodot excited state kinetics.1b and S3h employ the model of a stretched exponential (equation S1) to describe the physical decay processes of nitrogen-doped graphitic carbon nanodots (CNDs).The stretched exponential intensity decay function contains two main parameters, the characteristic timescale of the decay (τ) and heterogeneity parameter (β), and takes the form:

Data in Figures
where I0 is the initial luminescence intensity.The β parameter is typically constrained to the range 0 < β < 1, with β approaching 1 representing a homogeneous system obeying a single exponential decay, and β approaching 0 denoting a heterogeneous distribution of decays of varied timescales.
In the context of excited state relaxation in semiconductors, such a physical model is typically reserved for heterogeneous systems of species whose individual relaxation kinetics are first order but not singular (i.e., adopt a value within some range), and as a sufficiently disordered ensemble can be represented by this modified relationship. 1This is particularly appropriate for nanoparticles with varying structural characteristics, 2 samples where long-range energy transfer governs luminescence kinetics, 3 or systems where local conditions vary spatially and/or temporally for emitters (e.g. in solution or biological samples), 4 leading to a distribution of decay constants for one physical system.
Studies of CNDs have frequently used this stretched exponential relationship to describe their decay kinetics, prompted by the inability to fit with single exponentials or justifiably with discrete multiexponential fits.The exact physical mechanism behind the heterogeneous decay behavior of CNDs is uncertain but possible explanations include variation in CND structure 5,6 or diverse surface-bulk state interactions across CNDs or between neighbours. 7,8In terms of the information gathered from fitting with a stretched exponential, the characteristic timescale of decay can still be used as an indicator of the approximate quenching timescale of the dots emissive states, while the heterogeneity parameter reveals the diversity of such states that exist within a given sample.

Figure S4 :
Figure S4: Additional long-time transient absorption data (pump excitation at 400 nm) for nitrogendoped graphitic carbon nanodots in aqueous solution (0.25 g L -1 ), and with added electron donors (EDTA, AA; both 0.1 M). (a-c) Three-dimensional TA data (500 µJ cm -2 pulse -1 ) for NgCNDs alone and with added EDs. (d-f) Three-dimensional TA data (980 µJ cm -2 pulse -1 ) for NgCNDs alone and with added EDs. (g,h) Spectrally integrated TA kinetics for NgCNDs at two different excitation fluences.Higher fluence excitation appears to slightly increase the overall decay rate, particularly at higher energies.

Figure S5 :
Figure S5: Additional long-time TA data for nitrogen-doped graphitic carbon dot / ascorbic acid / methyl viologen system.(a,b) 3D TA data for NgCND (0.25 g L -1 ), AA (0.1 M) and MV 2+ added with (a) 12 µM and (b) 120 µM molarity.(c) TA spectral slices extracted at a pump-probe delay of 100 µs, for varying MV 2+ concentration.(d) Spectrally averaged (530 -700 nm) kinetics for the MV 2+ concentration series.The black dashed arrow indicates the trend of the onset of the MV •+ radical absorption towards earlier time.(e) Peak absorbance signal of the MV •+ related signal, compared by concentration.(f) Delay time at which the peak absorbance of the MV •+ related signal occurs, compared by concentration.

Figure S7 :
Figure S7: Monitoring the emergence of MV •+ of a NgCND / AA / MV system with 40 µM MV 2+ infiltrated within a Kagome-style hollow-core photonic crystal fiber (HC-PCF) -See Reference 22 in manuscript for further details regarding the experimental setup.The photoreduction process was initiated via external irradiation (λirr = 355 nm) of the HC-PCF.Unlike when EDTA is utilized as a SED, no trace of MV •+ was observed to form during irradiation when AA is the SED.

Figure S9 :
Figure S9: Irreversible chemical conversion of MV species.(a) Normalized TA spectra at several pump-probe delay times for a freshly prepared NgCND-MV 2+ hybrid photocatalytic system with EDTA as a sacrificial electron donor.The MV 2+ molarity is 200 μM, and the calculated ratio of MV 2+ to dots is 5:1.Data was acquired for 26 measurement sweeps (∼50 minutes).(b) Normalized TA spectra atseveral pump-probe delay times for the same sample after 260 measurement sweeps (∼8 hours).At pump-probe delays < 1000 ns, only the signal:noise ratio is improved, but at delays of 10,000 and 100,000 ns there is a change in spectral shape in the wavelength range 530-650 nm.(c) Comparison of spectrally averaged (530-760 nm) kinetics for the short and long exposure samples.