Microfluidic steam-based synthesis of luminescent carbon quantum dots as sensing probes for nitrite detection

2.1 Materials and instruments

The watermelon juice (absolutely pulp-free) was fresh-squeezed from the watermelon, which was purchased from the local market (Wuxi, China). Chemical agents (NaCl, Co(NO₃)₂·6H₂O, Ni(NO₃)₂·6H₂O, HgCl₂, CuCl₂·2H₂O, ZnCl₂, FeCl₃, Cr(NO₃)₃·9H₂O, NH₄F, KBr, KI, NaNO₃, NaNO₂, Na₂CO₃, PbSO₄, and Na₃PO₄) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Deionized water was employed throughout all experiments.

The size and shape of CQDs were characterized using a JEOL JEM-2100 plus transmission electron microscope (TEM) operated at 200 kV. The photoluminescence properties and UV/Vis absorption spectra were recorded with a CARY Eclipse spectrophotometer (Varian, USA) and a TU-1950 spectrophotometer (Persee, China), respectively. The Fourier transform infrared spectra (FT-IR) of CQDs were recorded with a Nicolet 6700 spectrometer (Thermo Fisher, USA) ranging from 600 to 4,000 cm⁻¹. X-ray diffraction (XRD) was performed using a D8 (AXS Bruker, Germany) diffractometer with a mono X-ray source Cu Kα excitation (40 kV) and a Lynxeye array detector. Dynamic light scattering (DLS) was measured using a ZetaPALS instrument (Brookhaven, USA). Fluorescence decay curves were measured by a Lifespec ll ultrafast time-resolved fluorescent lifetime spectrometer (Edinburgh Instruments, UK).

2.2 Synthesis of CQDs

A total of 30 mL of fresh-squeezed watermelon juice was injected into a capillary microreactor (length = 3 m, inner diameter = 1 mm) by a syringe pump, of which the flow rates were set at 0.234, 0.078, 0.026, and 0.0196 mL·min⁻¹, respectively. The temperature of the oil bath was maintained constant at 130°C. After a certain reaction time, the CQDs were pushed out by steam and collected by condensation. The obtained CQD solution was then dialyzed against water (MW cutoff: 500 Da) for 24 h to achieve purification.

2.3 CFD analysis

CFD analysis is an effective way to calculate the process parameters, such as pressure, temperature, and flow rate. It has been applied to solve numerous research and engineering problems in many fields, including aerodynamics and aerospace analysis, weather simulation, fluid flows, and heat transfer processes. In this study, CFD analysis was carried out with COMSOL software. To choose the right flow conditions, the Reynolds number (Re) should be determined first. Re is used to predict flow patterns in fluids, where the Re ranges divide transporting into three regimes: laminar, transitional, and turbulent. Low Re values correspond to laminar flow, whereas high Re values describe turbulent flow [15,16]. This dimensionless unit is a ratio of the inertial force to the viscous force:

Low density (ρ, kg·m⁻³), flow rates (V, m·s⁻¹), viscous liquids (high μ, Pa·s), and small channels diameter (D, m) generally produce laminar flow (Re < 2,040), with the range Re < 1,000 being called truly laminar and 1,000 < Re < 2,040 being transitional. Typically, micro-flow liquids in microreactors are laminar. In addition, we assumed that the temperature of the inner wall of the microchannel is constant and the entire system does not involve heat exchange.

2.4 Detection of NO 2 − using CQDs

The as-prepared CQD solution (1.50 mg·mL⁻¹) was diluted with deionized water. Various concentrations of NaNO₂ were separately added into the above diluted CQD solution. The resulting solutions were recorded for their emission spectra at 470 nm and absorbance values at 295 and 540 nm. The selectivity of CQDs to NO 2 − was investigated by adding different anions or metal ion stock solutions instead of NO 2 − in the same way.

3.1 CFD analysis of the pressure and velocity profiles in the microchannel

The defined setting of the phase state of the reaction medium is crucial for the outcome of the reaction. Suppose that the liquid phase is followed by the gas phase in the microchannel, as shown in Scheme 1.

Scheme 1

Illustration of the synthetic procedure of CQDs and their sensing mechanism for the detection of NO 2 − .

The expected phase setting requires knowledge and control of the pressure and velocity profiles in the microchannel. Specifically, these determine the quality of the CQD nanoproducts, which is required to influence, in a gross, overarching manner, the manifold chemical and particle size transformations along the microchannel axis. This motivation guides the CFD investigations as follows.

Scheme 1 shows that, at position 0, the inlet flow changes by evaporation from liquid to gas at a pressure P ₀. If the above assumption holds, P ₀ is equal to the saturated vapor pressure of water at that temperature. Position L is the outlet, where the pressure (PL) is easy to measure. The static pressure energy of the compressible fluid flowing in the straight pipe decreases, since one part of it is used to increase the kinetic energy of fluid expansion, and another part is used to overcome the frictional resistance loss. The kinetic energy conservation for steam transmission can be estimated by the following formula:

which is based on the principle of conservation of mass and momentum, derived from the basic differential equation of the compressible fluid in a tube. In Eq. 2, P (Pa) is the pressure, ρ (kg·m⁻³) is the density, and D (m) is the pipe diameter. Because the gas density is small, its potential energy change (g·dz) can be ignored. u can be replaced by the mass flow rate (G, g·s⁻¹) and axial velocity (v, m·s⁻¹) and then the above formula can be rewritten as follows:

This simplifies further to

where M is the molecular weight of steam, R (=8.314 J·mol⁻¹·K⁻¹) is the universal gas constant, and T (K) is the temperature. When G = 4.3 × 10⁻⁴ g·s⁻¹ and T = 403.15 K, the P ₀ of the solution is almost equal to P _L. Therefore, the variation in the fluid expansion kinetic energy is minimal, and the first term in the bracket on the right-hand side of Eq. 4 can be ignored, which is consistent with the Bernoulli equation [17] for incompressible fluid flowing in a straight horizontal pipe.

When P ₀ is much smaller than the saturated vapor pressure (2.66 × 10⁵ Pa) of water vapor at 403.15 K, the above assumption does not hold and the quantum dot precursor would only be exposed to the gaseous reaction medium. The precursor residence times in the microreactor are 0.33, 1, 3, and 4 s and correspond to the volume flow rates (u) of 0.234, 0.078, 0.026, and 0.0196 mL·min⁻¹, respectively.

Based on the above calculations, the whole flow process was simulated as shown in Scheme 2. Considering a reaction temperature of 130°C and a residence time of 4 s as an example, the gas in the microchannel has the following characteristics: isobaric at any position; the flow velocity only depends on the distance from the center of the channel cross section; and the greater the distance, the smaller the flow velocity and v _Max is 1.71 m·s⁻¹.

Scheme 2

The simulated results of velocity distribution in the microchannel.

3.2 Characterization

In order to study the mechanism of the microfluidic method for the CQD synthesis, the process temperature and process time have to be considered. As shown in Scheme 1, the fluorescence intensity of the as-prepared CQD initially increased at a temperature of 110°C to 130°C and showed the strongest PL intensity at 130°C. After that, the fluorescence intensity decreased with a further increase of temperature (Figure A1 in the Appendix). This is ascribed to the generation of increasing carbon nuclei, where the fluorescence intensity of CQDs increased with the temperature accordingly. As the process temperature further increased, the size of CQDs gradually increased and became uneven (Figure A2), leading to a decrease in the fluorescence intensity [18].

FT-IR, UV/Vis absorption, and PL spectra as well as the XRD pattern reveal clearly the transformation of the functional groups in the molecules of the juice, the generation of carbon nucleus crystal planes, and the fluorescence effect along the duration of the heating process. FT-IR spectra of CQDs are displayed in Figure 1a, revealing the presence of peaks at 3,350, 2,950, 1,680, 1,410, and 1,100 cm⁻¹. They are assigned to the stretching vibrations of ν(OH), ν(C–H), ν(C═C), δ(C–N), and ν(C–O), respectively. This demonstrates that CQDs have plentiful functional groups, such as hydroxyl, amide, and/or amino. Two peaks at 1,720 and 1,220 cm⁻¹ only appear on the FT-IR spectrum of CQDs synthesized at a process time of 0.33 s, corresponding to ν(C═O) and ν _as(C–O–C). Therefore, at a residence time of more than 1 s, the carboxyl groups are mostly carbonized into carbon bonds.

Figure 1

The FT-IR (a), UV/Vis absorption (b), fluorescence spectra (c), and XRD pattern (d) of CQDs prepared from watermelon juice by a microfluidic method at different process times and 130°C.

Figure 1b and c shows the absorption and excitation/emission profiles recorded for CQDs at four different residence times. The UV/Vis absorption profile of the CQDs exhibits a narrow band at 285 nm, which is attributed to the presence of aromatic π–π* transition [8]. However, in the PLE/PL spectrum, we found that, when excited with 370 nm incident light, CQDs show a strong peak around 470 nm, which is ascribed to band gap transition corresponding to conjugated π-domain and surface defects present in the quantum dots [19]. In addition, the prolongation of the reaction time leads to an increase in the yield of CQDs and the degree of carbonization, resulting in an increase in the UV/Vis absorption intensity and fluorescence property of CQDs. A decrease in the fluorescence intensity is observed after a process time of more than 3 s, as the carbonization degree of CQDs increases, the defect states decrease, the non-radiative recombination process is inhibited, and the fluorescence is mainly derived from the eigenstate [20].

The XRD patterns of CQDs shown in Figure 1d revealed distinct Bragg peaks located at diffraction angles (2θ) of ∼20°, 28.48°, 40.62°, 50.34°, and 58.72°, corresponding to the {002}, {220}, {100}, {102}, and {103} planes, respectively. The first four facets originate from graphitic (sp²) carbon and the last one is from diamond-like (sp³) carbon [21,22]. The peak of {002} facets centered at 19.2° in the CQDs synthesized in 0.33 s has been shifted to 20.9° in the case of CQDs with residence time of 3 s, corresponding to the lattice spacing from 0.46 to 0.42 nm. The decrease in d values indicates a decrease in the amorphous nature attributed to the less oxygen-containing groups [23].

The CQDs prepared from watermelon juice at 3 s and 130°C were also used for the following characterization. TEM images of CQDs are shown in Figure 2a, the as-prepared CQDs are well dispersed, and the selected area electron diffraction (SAED) pattern is shown in the inset. The rings correspond to the {100} {111} {220} {311} planes, which present both graphite- and diamond-like structures [18]. A high-resolution TEM (HRTEM) image in Figure 2b shows that most of the CQDs contain defects, and the lattice spacing observed varies from 0.20 to 0.23 nm. The fast Fourier transform (FFT) patterns of representative CQDs are displayed in Figure 2b. The observed hexagonal lattices in the FFT image reveal that the CQDs are crystalline hexagonal structures [24]. The size distribution of the CQDs is in a narrow range, 3–7 nm, and is shown in Figure 2c along with the maximum values of the fitted Gaussian peak. The photographs illustrate that the as-prepared CQD suspension shows a blue color under irradiation with a 365 nm ultraviolet (UV) lamp (inset in Figure 2c), suggesting the luminescence nature. Figure 2d presents the PL spectrum, which shows, with the excitation wavelength ranging from 330 to 490 nm, that the emission wavelength changes from 440 to 550 nm. The excitation-dependent emission is mainly caused by surface defects [25]. The blue PL (excited at 370 nm) intensity is much stronger than the green PL (excited at 410 nm) value. Using quinine sulfate as the reference, the quantum yield (QY) of the obtained CQDs was calculated as 42.97% (Figure A3a).

Figure 2

Representative TEM images and fluorescence spectra of CQDs synthesized for 3 s at 130°C, along with direct photographs of the suspension under a 365 nm UV lamp. (a) TEM image of CQDs; the inset is the SAED pattern of CQDs; (b) HRTEM image of CQDs’ the insets are the FFT pattern of highlighted areas; (c) the size distribution of CQDs and direct photograph of sample A under a 365 nm UV lamp; and (d) emission profiles at different excitation wavelengths (λ _ex = 330–490 nm) recorded for CQDs.

The analytical characterization reveals even more mechanistic insight. The long duration allows a smooth stepwise seed formation and growth of the seed nanodots (Figure A2). As shown in FT-IR spectra, the sucrose molecules in the watermelon juice quickly decompose into small molecules with carboxyl groups in the steam atmosphere. Subsequently, soluble polymers and aromatic rings are generated under the synergistic activation of high temperatures and by hydronium ions. When the concentration of the aromatic clusters reaches a critical supersaturation point, a burst nucleation takes place, and then π-domain-conjugated and surface-defected CQDs are formed. The narrow internal space of the microchannel not only enhances the collision opportunities for the precursors (transport intensification) but also works in unusual process windows and superheated processing (chemical intensification). Meanwhile, the CFD simulation paved the ground for a robust and efficient process. The key to a high-quality CQD nanoproduct is knowledge and control of the pressure and velocity profiles in the microchannel.

3.3 Fluorescence and spectrophotometric detection of NO 2 −

The selectivity and sensitivity of CQDs for NO 2 − detection were evaluated to explore the feasibility of CQDs in environmental applications. Fluorescence screening experiments were performed to identify potential ions that interfere with NO 2 − in the CQD aqueous solution, including Na⁺, Co²⁺, Ni²⁺, Hg²⁺, Cu²⁺, Zn²⁺, Fe³⁺, Cr³⁺, F⁻, Cl⁻, Br⁻, I⁻, NO 3 − , CO 3 2 − , SO 4 2 − , and PO 4 3 − (note that the same mechanism is followed for both fluorescence and spectrophotometric readout and representative interferences study for only fluorescent readout was recorded). All results complied with the same conditions (10 μM ions’ concentration, excited at 370 nm, and equilibrated for 10 min). As shown in Figure A3b, only NO 2 − ions significantly reduced the fluorescence intensity at 470 nm of CQDs (quenching 55% of the blank case at 370 nm); the signal was slightly changed for the remaining ions. This result indicates that CQDs have excellent selectivity for NO 2 − detection.

For sensitivity studies, fluorescence detection and spectrophotometric detection were both performed to assess the response of CQDs to different NO 2 − concentrations. As shown in Figure 3a, a gradual decrease of PL intensity at ∼470 nm is observed with the increase of NO 2 − concentrations from 0 to 100 μM. Figure 3b shows a good linear response between F ₀/F (where F ₀ and F are the PL intensity of CQDs at 470 nm in the absence and presence of NO 2 − , respectively) and the NO 2 − concentration in the range of 0.2–100 μM (R ² = 0.9917). The limit of detection (LOD) for NO 2 − was calculated approximately to be 0.11 μM (S/N = 3).

Figure 3

(a) Fluorescence emission spectra of the CQD solution in the presence of different concentrations of NO 2 − (0–100 μM); (b) the linear relationship between F ₀/F and NO 2 − concentration; (c) UV-Vis spectra of the CQD solution in the presence of different concentrations of NO 2 − (0–10 μM); and (d) the linear relationship between Abs_{540 nm}/Abs_{295 nm} and NO 2 − concentration.

With increasing NO 2 − concentrations from 0 μM to 10 mM, the peak intensity at ∼290 nm decreased, while the peak at 540 nm became stronger (Figure 3c). A good linear correlation between the absorbance (ratio values of two absorbance bands (Abs_{540 nm}/Abs_{295 nm})) and the concentrations of NO 2 − can be obtained in the concentration range of 2–80 μM (Figure 3d). The LOD was estimated to be 80 nM. In addition, the detection performance of the obtained CQDs was compared with the previously reported CDs in monitoring NO 2 − (Table 2).

3.4 Possible mechanism of NO 2 − quenching the fluorescence of CQDs

As demonstrated in Figures 1b and 3c, a narrow band appears on the CQD curve at 285 nm, which redshifts to 295 nm with the addition of NO 2 − . Due to the p–π conjugation effect of electrons provided by NO 2 − and π electrons provided by CQDs, the π → π* transition energy decreases, and the absorption peak of the chromophore shifts to the long wave, indicating that NO 2 − might have interacted with CQDs to form the new complex [8]. To confirm this, DLS measurements of CQDs with/without NO 2 − were carried out, as shown in Figure 4a. The average hydrodynamic diameter of CQDs increased from 9 ± 3 to 52 ± 34 nm after the addition of NO 2 − . Moreover, the average lifetime of CQDs (4.57 ± 0.15 ns) was almost the same as that of CQDs + NO 2 − (4.72 ± 0.12 ns), as shown in Figure 4b. Therefore, the mechanism of the optical response of CQDs to NO 2 − should be attributed to static quenching from NO 2 − -induced aggregation of CQDs.

Figure 4

(a) DLS measurements of CQDs with different concentrations of NO 2 − ; (b) Fluorescence decay traces of CQDs in the absence (red curve) and presence (black curve) of NO 2 − .