Hybrid nanodiamond quantum sensors enabled by volume phase transitions of hydrogels

Diamond nitrogen-vacancy (NV) center-based magnetometry provides a unique opportunity for quantum bio-sensing. However, NV centers are not sensitive to parameters such as temperature and pressure, and immune to many biochemical parameters such as pH and non-magnetic biomolecules. Here, we propose a scheme that can potentially enable the measurement of various biochemical parameters using diamond quantum sensing, by employing stimulus-responsive hydrogels as a spacing transducer in-between a nanodiamond (ND, with NV centers) and magnetic nanoparticles (MNPs). The volume phase transition of hydrogel upon stimulation leads to sharp variation in the separation distance between the MNPs and the ND. This in turn changes the magnetic field that the NV centers can detect sensitively. We construct a temperature sensor under this hybrid scheme and show the proof-of-the-principle demonstration of reversible temperature sensing. Applications in the detection of other bio-relevant parameters are envisioned if appropriate types of hydrogels can be engineered.

solution (50 µl) that contains hybrid sensors is firstly transferred to the silicon substrate by pipette.
And the silicon substrate is fixed on the bottom of the sample chamber. After about 20 minutes, some of the hybrid sensors settle down to the surface of the silicon substrate, then the chamber is filled with deionized water (or FBS) and covered with cover glass. The hybrid sensors are randomly distributed on the silicon substrate. Under excitation power of about 100 W, the fluorescence counts of a single ND are about 5 M per second. The temperature of the chamber is controlled by the TEC heater and monitored with a resistance thermometer. Supplementary Fig. 11: ODMR sample loading and temperature control. The home-built sample chamber is filled with deionized water (blue color in side view). The temperature of the chamber is controlled by the TEC heater and monitored by the nearby resistance thermometer (purple).
Hybrid sensors (solid red triangles) are randomly distributed on the silicon substrate. Laser excitation and fluorescence collection are carried out with the same objective. MW pulses are delivered to the NDs by the copper wire (gold line).

Supplementary Note 4. Lorentzian double-peak fitting
The measured ODMR spectra ( ) of both sensor NDs and bare NDs show a two-peak feature in all profiles recorded. They are normalized to the counts of off-resonance MW driving (other experiment conditions are the same). The normalized ODMR spectra are then fitted by nonlinear least-square method with the Lorentz double-peak function as, where is the zero-field splitting of the NV centers, is the peak shifts from , 1,2 are the contrast, and 1,2 are the FWHM (full width at half maxinum) of the correponding Lorentz peaks.
The zero-field splitting D, peak shifts, and FWHM of the corresponding Lorentz peaks in all plots are obtained from the fitted experimental data (for both bare NDs and ND@pNIPAM-Ni hybrid sensors).

Supplementary Note 5. Temperature calibration
Due to heat dissipation, the sample temperature is slightly higher than that measured by the resistance thermometer. We use NDs (and hybrid sensors) themselves to calibrate the local temperatures. Three assumptions are made in the temperature calibration: (1) At room temperature, when the TEC heater is turned off, the sample (NDs or hybrid sensors) and the thermometer measure the same temperature, that is, the room temperature; (2) The zero-field splitting D of NV centers in the ND is linearly dependent on its local temperature, with a slope of ~74 cooling processes, as shown in Supplementary Fig. 12b, c. At each voltage, a double-peak Lorentz fitting is used to extract the resonant frequencies of the two peaks. Then the zero-field splitting D is obtained by averaging the two frequencies. Supplementary Figure 12d plots the zero-field splitting as a function of the TEC voltage, and a linear dependence can be observed. The bare ND shows a good reversibility of temperature response in the heating and cooling processes. The measurements on reference ND suggest the linear dependence of temperature on TEC voltage, as well as the consistent temperature dependence for heating and cooling processes. , which is ~0.9 K Hz −1/2 (olive dash line in Fig. 5d in the main text). In this estimation, the intrinsic parameters of the ND (such as 1 and 1 ) are extracted from the ODMR spectrum of the hybrid sensor at lowest temperature (28 °C). At such a low temperature, the hydrogel shell is in the fully swollen state and the Ni MNPs are far away from the ND and hence have negligible effects on the ODMR spectra of the ND.

Supplementary Note 8. Temperature sensing in fetal bovine serum (FBS)
Supplementary Figure 15a shows the ODMR spectra of one ND@pNIPAM-Ni hybrid sensor measured in FBS at a series of temperatures (using the same methods as measurements in water).
The temperature-dependent peak shift fs and FWHM 1,2 are plotted in Supplementary Fig. 15b. temperature increase. An LCST of ~35 °C is estimated from the temperature-dependent peak shift/width plots ( Supplementary Fig. 15b). The LCST in FBS is slightly lower than that in water (~37 °C), due to the high concentration of salt ions (>100 mM) of FBS, which is consistent with literature 4,5 . The shot-noise-limited sensitivity of the hybrid thermometer in FBS is estimated ( Supplementary Fig. 15c) by adopting the same method as in water. Optimal sensitivity of the hybrid sensor in FBS is estimated to be 152 ± 2 mK Hz −1/2 (shot noise error from the ODMR spectrum) around the LCST, an improvement by ~6 times from that of a bare ND (~0.93 K Hz −1/2 , grey dashed line in Supplementary Fig. 15c). The temperature sensitivity of the hybrid sensor in FBS is similar to that measured in water.

Supplementary Note 9. Numerical Simulation of ND@pNIPAM-Ni hybrid sensor
The hybrid nano-sensor is numerically simulated with a simplified model, as illustrated in Supplementary Fig. 16a. 500 NV centers with four different orientations are uniformly distributed inside the ND, which is assumed to be a cuboid. The Ni cluster is assumed to contain four spherical where is the volume of the Ni nanoparticles, is the magnetization of the th nanoparticle and is the distance between the th NV center and the th Ni particle, which depends on the pNIPAM thickness ℎ( ) and the temperature. Transition frequencies ± of the NV centers between |±1⟩ and |0⟩ spin states then are obtained by diagonalization of the Hamiltonian .
Therefore, the ODMR spectra is written in summation of Lorentzian functions as where is the ODMR contrast and is the linewidth (FWHM) of each NV center. Typical values of = 9 ± 4 MHz and = 3.6 ± 0.5 MHz are obtained by fitting the ODMR spectra of six bare NDs (for example, Supplementary Fig. 12b, c).
For the hybrid nano-sensor, the parameters of the NV centers in a bare ND are assumed to be σ E = 3.6 MHz, = 7.5%, Δf = 10 MHz and the total count rate = 8 × 10 6 s −1 . The relative angle between the ND and Ni cluster is set as = 130° (which is not optimized but the dependence of the sensitivity on the angle is weak since the ODMR spectra is averaged over NV centers along four different crystallographic directions). The dependence of pNIPAM thickness on the temperature ℎ( ) is obtained from the DLS results in Fig. 3a in the main text. Supplementary   Figure 16b shows two typical simulated ODMR spectra at high (45 °C) and low (30 °C) temperature, respectively. The simulated ODMR spectra are fitted with the two-peak function described in Supplementary Note 4 (grey lines in Supplementary Fig. 16b), and the fitting parameters s and 1,2 are shown by grey lines in Fig. 5c in main text and Supplementary Fig. 14 cluster results in a net magnetic field similar to that from a single MNP. In the case of chain-shape clusters, the net field is large along the chain axis but relatively weak (comparable to that from a single MNP) at the location of the ND, which is usually on the perpendicular plane of the chain.
The cluster's magnetic field at the location of the ND is thus approximated as that of a magnetic dipole on the pNIPAM surface with an effective moment (in units of M0, the moment of a single MNP). Given the thickness of the hydrogel in the fully swollen state ℎ s , the thickness at the LCST ℎ c is determined by the volume change ratio measured by DLS, with the assumption that the volume change ratio is independent of the thickness. The thickness change per Kelvin at the LCST is (for the sake of simplicity, the ND and pNIPAM are assumed to be spherical, and the ND diameter = 50 nm) where the volume change ratio per Kelvin at the LCST v = ( + )/ ( − ) ≈ 0.84 is measured by DLS (see Fig. 3a in the main text). The optimal sensitivity versus the effective moment of the MNP cluster and the thickness of pNIPAM (ℎ s ) is shown in Supplementary Fig.16c. In most cases, the optimal sensitivity is 60-200 mK Hz −1/2 for hydrogel thickness from 80 to 300 nm and effective moment from 0.1 to 2 M0. To improve the sensitivity, a hybrid sensor is constructed with pNIPAM hydrogel in-between a smaller sized ND (~50 nm in cubic shape) and a Co MNP with different diameter (Co MNPs have large magnetization at room temperature (1,422 kA m −1 ) and also large single-domain size

Supplementary
The estimated sensitivity versus the diameter of the Co MNP and the thickness of pNIPAM (ℎ s ) is mapped as shown in Supplementary Fig. 17a. The parameters of the ND are assumed to be the same as realized in the experiment (as shown in Supplementary Equation 6). An optimal sensitivity can be archived as ~ 31 mK Hz −1/2 at a polymer thickness (in swollen state) of ~ 380 nm, and an MNP magnetic moment about 17 times that of a Ni MNP of 50 nm diameter, as illustrated in the main text.
Sensitivity can be further improved by using NDs that contain single NV centers. Broadening of the ODMR spectra induced by the magnetic field gradient is asbent in such NDs. Simulations based on single-NV NDs are run with contrast = 20%, FWHM Δf = 5 MHz and count rate = 2 × 10 5 s −1 . The estimated sensitivity at the LCST versus the diameter of the Co MNP and the thickness of pNIPAM (ℎ s ) is shown in Supplementary Fig. 17b. An optimal sensitivity ~ 0.3 mK Hz −1/2 is reached when the thickness of pNIPAM in the swollen state is ~ 50 nm.
Supplementary Fig. 17 The sensitivity at the LCST as a function of the diameter of the Co MNP and the fully swollen thickness of pNIPAM. In (a) the ND contains 100 NV centers, and in (b) a single NV center. The ND has diameter of 50 nm.