Detection of Few Hydrogen Peroxide Molecules Using Self-Reporting Fluorescent Nanodiamond Quantum Sensors

Hydrogen peroxide (H2O2) plays an important role in various signal transduction pathways and regulates important cellular processes. However, monitoring and quantitatively assessing the distribution of H2O2 molecules inside living cells requires a nanoscale sensor with molecular-level sensitivity. Herein, we show the first demonstration of sub-10 nm-sized fluorescent nanodiamonds (NDs) as catalysts for the decomposition of H2O2 and the production of radical intermediates at the nanoscale. Furthermore, the nitrogen-vacancy quantum sensors inside the NDs are employed to quantify the aforementioned radicals. We believe that our method of combining the peroxidase-mimicking activities of the NDs with their intrinsic quantum sensor showcases their application as self-reporting H2O2 sensors with molecular-level sensitivity and nanoscale spatial resolution. Given the robustness and the specificity of the sensor, our results promise a new platform for elucidating the role of H2O2 at the cellular level.

10 μL of a 1 mg/mL ND-NV-10 or ND-NV-40 aqueous solutions were deposited on a silicon substrate and dried at room temperature. XPS was conducted using a Kratos Axis UltraDLD spectrometer 3 (Kratos, Manchester, England) using an Al Kα excitation source with a photon energy of 1487 eV. The data was acquired in the hybrid mode using a 0° take-off angle, defined as the angle between the surface normal and the axis of the analyzer lens. Detailed region XP spectra were collected with setting analyzer pass energy at 80 eV, and a linear background was subtracted for all peak quantifications. The peak areas were normalized by the manufacturer supplied sensitivity factors and surface concentrations were calculated using CasaXPS software.
C 1s high-resolution spectra were collected with analyzer pass energy of 20 eV. Neutralizer was always used during spectra collection.

Density functional theory (DFT) calculations
Density functional theory (DFT) calculations were performed using the Gaussian 09 software package 1 and structural representations were generated with CYLview20. 2 All the geometry optimizations were carried out using the Minnesota hybrid meta-GGA functional M06-2X and valence double-zeta 6-31G(d) basis set. All of the optimized geometries were verified by frequency computations as minima (zero imaginary frequencies) or transition states (a single imaginary frequency corresponding to the desired reaction coordinate). The free energy values presented in Figure 3 in main text and Table S2 were derived from the electronic energy corrected by using the thermal and entropic corrections based on structural and vibration frequency data.

NV center spin relaxation time measurement
The spin relaxation time (T1) of NV centers in nanodiamonds are performed in a customized confocal microscope. We use a 532nm laser to excite the NV center. The laser power (measured before the objective) is fixed at 50 μW for all the experiments. The laser is focussed onto the sample using an oil-immersion objective (Olympus UPlanSApo 60x oil, N.A. = 1.35). The fluorescence from the NV center is collected through the same objective and detected using an avalanche photodiode (APD). We use a 650nm longpass filter in front of the APD to eliminate the excitation laser and minimize the detection of NV 0 .

S4
The pulse sequence for the T1 measurement is shown in the main text (Fig 4). It consists of a laser pulse (10μs long) to polarize the NVs in the ms = 0 state. After a variable waiting time τ, the subsequent laser pulse is used to read the spin state of the NV center. We use fluorescence photons detected in the first 500 ns of the laser pulse to constitute the signal. This all-optical T1 measurement is prone to charge state fluctuation of the NV, which would manifest as a decrease in NV fluorescence as a function of time. To measure the actual T1 time of the NV we use an additional control sequence with a linear chirp microwave pulse to invert the population from the ms = 0 to ms = ±1 state. The frequency range for the linear chirp microwave pulse is from 2850 MHz to 2950 MHz and is generated using an arbitrary waveform generator. The duration of the chirp and the Rabi frequency is adjusted to achieve the adiabatic passage condition. The T1 measurements without inversion pulse is subtracted from the T1 measurements with inversion pulse to obtain the spin relaxation time due to magnetic noise. The experiment is repeated several times with a total acquisition time of 10 minutes. The resulting T1 data is fitted with a monoexponential decay function.
To measure the T1 time statistics of the nanodiamonds, we used cleaned glass slide with a lithographically patterned MW structure. We placed a silicone gasket (cell well volume ~30μL) on top of the glass slide to confine the nanodiamonds and the analyte solution in the subsequent measurements. Next, we added ~10 μL of 10 μg/mL sample into the cell well and the sample was dried for several hours under ambient conditions. During the T1 measurement, we added ~5 μL of buffer (acetate buffer for pH 4 and DPBS for pH 7) and covered the silicone well with a glass slide to avoid evaporation. The T1 time is measured as described above on selected isolated fluorescence spots (with fluorescence counts between 400-600 kcts/s). Following the T1 measurements in the buffer, we added ~5 μL of 100 mM H2O2 solution and repeated the T1 measurement on the same fluorescent spots as before.  Figure S8 A and D). We observed only a small change in T1 time with the addition of the H2O2 solution. As discussed earlier, the small responsivity of ND-NV-40 to H2O2 molecules could be attributed to both the size of the NDs (relatively bigger than ND-NV-10, therefore the NVs are less sensitive to the surface noise) and the presence of fewer surface groups producing the radicals.

Simulation of spin relaxation times
Without the OH or O2H radicals, the = 0 spin relaxation is determined by the intrinsic NV relaxation and the electron spin noise at the surface of the nanodiamond, with the relaxation rate given by Here we take the intrinsic NV relaxation time 1 i = 1 ms, a value similar to the ones of NVs in bulk diamond. The relaxation rate 1/ 1 noise due to the noisy electrons at diamond surface is calculated in a way similar to that of the OH or O2H radicals.
The correlation times for the noisy electrons at diamond surface and for the OH or O2H radical spins are obtained using the results of Tetienne et al. 4  where vib = 50 GHz takes into account the intrinsic vibrational spin relaxation. 3 The parameter related to the minimum allowed distance among the electron spins is 0.15 nm for the noisy electrons at diamond surface and is 0.2 nm for the OH or O2H radicals. We assume a small thickness (0.1 nm, which is much smaller than the diamond sizes) for the surface electrons layer on top of the diamond surface. To obtain ~56% reduction of NV spin relaxation times observed in the experiments, the noisy electrons at the diamond has a density of 0.18/nm 3 while the OH or O2H radicals has a density of 0.05/nm 3 for the simulation.

Sensitivity estimation
The smallest number of radicals that can be detected is given as 4 ,       Figure S12. Optimized structures (M06-2X/6-31G(d)) for the reaction of H2O2 decomposition. standard deviation of 0.3D. This distribution has a probability density function (PDF) similar to a normal distribution but the PDF is zero when the length L is smaller than 0.