Introduction

Metal nanoclusters (NCs), which are tiny colloidal nanoparticles composed of a few to some hundreds of metal atoms, show great potential in the biomedical field, optoelectronics, and sensor development as supplements to molecular level phosphors and luminescent semiconductor nanocrystals, due to their small size, biocompatibility, large Stokes shift, and reasonably strong photoluminescence (PL) [1,2,3,4,5]. Compared with conventional noble Au and Ag NCs, copper NCs display the advantages of earth-abundant precursors, relatively low-cost, and rich surface chemistry [6,7,8].

Inspired by their unique molecule-like optical properties and subsequent application prospect, the synthesis of Cu NCs has gained growing research interest in the past decade [2, 9,10,11,12,13,14,15]. As they suffer from easy oxidation in air and aggregation during the preparation process, appropriate stabilizing agents are indispensable for achieving stable Cu NCs. Up to now, various surface ligands including DNA, proteins, peptides, polymers, and small molecules have been employed to synthesize Cu NCs [16,17,18,19,20,21,22]. Recently, small molecules especially thiol-containing molecules have been widely used, because they can serve as both capping ligand and reductant and are convenient to use under mild conditions and much lower cost compared with the others [10, 23, 24]. For example, Li et al. developed a special strategy for the fast fabrication of red-emitting Cu NCs with the assistance of green-emitting Cu NCs in the presence of dihydrolipoic acid (DHLA) as a protecting ligand and bis[tetrakis(hydroxymethyl)phosphonium] sulfate (THPS) as a reducing agent, maintaining the water solubility and tiny size of the NCs [25]. Wang and coworkers prepared self-assembled Cu NCs by mixing tetrafluorothiophenol (TFTP) with a copper source in ethanol, which exhibited intense saffron yellow (590 nm) PL via self-assembly-induced emission [1]. Zhao et al. prepared hydrophobic 4-methylthiophenol-capped Cu NCs and stable aggregation-induced emission particles with intense red luminescence via self-assembly mediated by the hydrophobic interaction [26]. Wang et al. reported two Cu NCs stabilized by l-cysteine and glutathione which exhibited green and red emission respectively [10]. Zhang group prepared Cu NC self-assembled ribbons with tunable emission in a range of 548 to 698 nm by employing aromatic thiols with different substituent groups and substituent positions as the capping ligands [27]. Das et al. used glutathione as the surface ligand to prepare glutathione-capped Cu NCs in alkaline solution under 40 °C and 14 h of vigorous stirring [28]. Despite the above achievements, there remains a lack of a systematic study about the surface ligand effect on the optical properties of Cu NCs, which has been intensively investigated for Au and Ag NCs [5, 29, 30]. On the other hand, for achieving color-tunable Cu NCs and assembly, most of the preparation routes are based on the non-aqueous phase synthesis, and the resulting hydrophobic particles are not compatible with biological systems before further surface modification. There already exist some reports of small molecule-capped Cu NCs via aqueous synthesis [10, 23, 24, 31], but many of them suffered from tedious and time-consuming processes. Therefore, it is necessary to develop cost-effective synthetic methods and explore the surface ligand effect for achieving water-soluble Cu NCs with tunable PL.

Following our previous studies on preparation and sensing applications of Cu-based nanocrystals and noble metal NCs and ligand effects on iron oxide nanocrystals [32,33,34,35,36], herein, we report an aqueous synthesis of Cu NCs with tunable emission by employing small molecules with different functional groups as surface ligands. The ligand effects on the PL properties were investigated by spectroscopic characterization in combination with structural and compositional analysis on the resulting NCs. Exploring mild and rapid strategy for Hg2+ determination is highly demanded due to its toxicity, and the commonly used elemental analysis methods such as atomic fluorescence spectrometry (AFS) suffer from rigorous and time-consuming. The prepared Cu NCs selectively exhibited decreased PL towards Hg(II), even in the presence of various interferences.

Material and methods

Chemicals

Copper sulfate pentahydrate (CuSO4·5H2O, ≥ 99.0%, Tianjin Bodi Chemical Co. Ltd., https://bdhg.company.lookchem.cn/), reduced l-glutathione (GSH, ≥ 98.0%, Aldrich, https://www.sigmaaldrich.com/), l-cysteine (Cys, 99%, Aladdin, http://www.aladdin-e.com/), ethanediamine (EDA, ≥ 99.0%, Aldrich, https://www.sigmaaldrich.com/), and ascorbic acid (AA, AR, Sinopharm Chemical Reagent Co. Ltd., http://www.sinoreagent.com/) were used as received. Other reagents including HCl solution (12 mol/L) and acetone of analytical reagent from Yantai Far East Fine Chemical Co. Ltd. (http://191131.biochemsafebuy.com/), ethanol of analytical reagent from Sinopharm Chemical Reagent Co. Ltd. (http://www.sinoreagent.com/), and NaOH of analytical reagent from Aladdin (http://www.aladdin-e.com/) were used as received without further purification. Mercury perchlorate trihydrate (Hg(ClO4)2·3H2O), NaCl, NaNO3, Na2SO4, AgNO3, PbCl2, CdCl2, BaCl2, NiCl2, CoCl2, MgCl2, MnCl2, and CuCl2 of analytical reagent from Aladdin (http://www.aladdin-e.com/) is used as received. Millipore ultra-pure water (Milli-Q water) (resistivity is 18.2 MΩ·cm) was utilized throughout the experiments.

Synthesis of water-soluble Cu NCs

Typically, for synthesizing Cu NCs with blue PL, CuSO4·5H2O (0.0800 mmol), EDA (1.36 mmol), and AA (0.800 mmol) were dissolved into 24-mL ultrapure water in a 50-mL flask under magnetic stirring to form a homogeneous solution with the pH value kept at about 4.5 using 1 M NaOH. The resultant solution was then heated to 37 °C and kept for a certain reaction time. Afterwards, the solution was cooled to room temperature. The resultant nanoclusters were precipitated by 2-propanol (volume ratio of 2-propanol and sample solution: 3:1), collected by centrifugation at 4000 rpm for 10 min, washed with 2-propanol (volume ratio of 2-propanol and added water: 3:1) for three cycles, and finally redispersed in water for further experiments. For synthesizing Cu NCs with green and red PL, EDA was replaced by the same amount of Cys and GSH respectively, while the other reaction conditions were maintained the same.

Spectroscopic characterization

Steady-state UV−Vis absorption and PL spectra were recorded at room temperature on a Shimadzu UV-2600 UV−Vis spectrophotometer and a Hitachi F-4600 fluorescence spectrophotometer, respectively. The optimal excitation wavelengths for steady-state PL measurements were as follows: 350 nm for the Cu NCs capped with EDA, and 420 nm and 370 nm for NCs capped with Cys and GSH ligand respectively. Time-resolved PL decay measurement was carried out on an Edinburgh Instruments FLS980 spectrometer equipped with a picosecond pulsed diode laser (EPL-360) as a single wavelength excitation source. The absolute PL QY of the NCs was determined by a Binsong c1515947s spectrometer with the excitation wavelength of 350 nm. The fluorescent stability of the as-prepared Cu NCs along with placing time and pH was monitored by a Hitachi F-4600 fluorescence spectrophotometer.

Structural and compositional characterization

Transmission electron microscope (TEM) and high-resolution TEM (HRTEM) images of the QDs were taken on a JEM-2100F electron microscope at an acceleration voltage of 200 kV. The particle size was determined by averaging at least 300 particles per sample by ImageJ. The Fourier transform infrared (FTIR) spectra were recorded on a Nicolet iS 10 spectrophotometer (Germany) in the range of 4000–900 cm−1. X-ray photoelectron spectroscopy (XPS) measurements were carried out with a Thermo-VG Scientific ESCALAB 250Xi spectrometer with a monochromatic Al Kα X-ray source.

Performance of Hg2+ determination

For Hg2+ determination, a series of Hg2+ aqueous solutions were separately mixed with the aqueous solution of EDA-capped Cu NCs at pH 6, and the Hg2+ concentration in the final solution was 0.1, 0.2, 0.3, 0. 4, 0.5, 0.6, 0.8, 1, 2, 3, 4, and 5 mM respectively. The mixture was incubated at room temperature for 10 min and then the fluorescence emission spectra were recorded with the excitation wavelength of 350 nm. When measuring the fluorescence of Cu NCs in the presence or absence of Hg2+ in different pH environments, phosphate-buffered saline (PBS) with a pH value from 2 to 8 was used respectively. Apart from Hg2+, various metal ions including Na+, Ag+, Pb2+, Cd2+, Ba2+, Ni2+, Co2+, Mg2+, Mn2+, and Cu2+ and anions including ClO4−, Cl−, OH−, NO3−, and SO42− were adopted in the selectivity experiments, and the individual concentration of all the ions was set to 1.0 mM. The treatment processes were the same as those for Hg2+ determination as stated above. For the real sample detection, the tap water was used directly for Hg2+ assay by recovery experiments. The added Hg2+ concentrations were 0.5 mM, 1.0 mM, and 2.5 mM respectively. For assessing the accuracy of the current method, the water samples were also tested by the atomic fluorescence spectrometry (AFS) method (instrument model: XGY 1011A).

Results and discussion

Influence of surface ligand on optical properties of water-soluble Cu NCs

The water-soluble Cu NCs were synthesized in water at 37 °C employing CuSO4 as the metal precursor and ascorbic acid as the reduction agent in the presence of different surface ligands, as described in the “Material and methods” section. It is important to note that the reaction conditions including metal precursor, reduction agent, pH value, and reaction temperature for all samples were kept constant to ensure comparability. In order to study the impact of surface ligand on the optical properties of Cu NCs, ethanediamine (EDA), l-cysteine (Cys), and l-glutathione (GSH) were selected as the surface ligands for each preparation (Scheme 1). As shown in Fig. 1a, blue-, green-, and red-emitting Cu NCs were obtained respectively. To achieve NCs with the brightest PL, the reaction process was monitored and a series of aliquots were extracted at different reaction time. As shown in Fig. S1b, 10 min of reaction led to a remarkable enhancement of the PL of blue-emitting NCs, and the PL reached the strongest at 40 min. Meanwhile, the absorption spectra in Fig. S1a showed the same variation tendency. For the NCs capped with Cys and GSH, the reaction time of 30 min and 20 min was needed for achieving the highest PL intensity (Fig. S1c-f). In the UV−Vis spectra shown in Fig. 1b, one well-resolved absorption peak appears at 340 nm for EDA-capped Cu NCs, while for Cys- and GSH-capped NCs, the absorption peaks are rather structureless. In contrast, well-resolved peaks at 350 nm, 420 nm, and 370 nm are observed in the photoluminescence excitation (PLE) spectra (Fig. S2) for the blue-, green-, and red-emitting NCs. Figure 1c presents the normalized PL spectra of the three Cu NCs, with the emission peaks locating at 420 nm, 505 nm, and 630 nm respectively. The full width at half-maximum (FWHM) are 57 nm, 89 nm, and 110 nm for the three NCs. Among the three Cu NCs, the EDA-capped Cu NCs possess the highest PL intensity, as presented in Fig. S3.

Scheme 1
scheme 1

Schematic illustration of aqueous synthesis of Cu NCs capped with ethanediamine, cysteine, and glutathione respectively

Full size image
Fig. 1
figure 1

Photographs of Cu NCs capped with ethanediamine, cysteine, and glutathione from left to right under room light (upper row) and 365 nm UV irradiation (lower row) (a), UV−Vis absorption (b), and normalized PL (c) spectra of Cu NCs. The excitation wavelengths were 350 nm, 420 nm, and 370 nm for three Cu NCs respectively

Full size image

Influence of surface ligand on compositional and structural properties

The valence state of the prepared Cu NCs was analyzed by X-ray photoelectron spectroscopy (XPS). As shown in Fig. 2a, all the NCs present two peaks at 932 eV and 952 eV in the high-resolution XPS spectra irrespective of the surface ligand, which are assigned to 2p3/2 and 2p1/2 states of metallic Cu [7]. The reduction of Cu2+ by ascorbic acid is evidenced by the absence of any satellite peaks around 942 eV. It is worth mentioning that the difference in binding energy between 2p3/2 peaks of Cu0 and Cu+ species is only 0.1 eV, so the valence state of Cu in the NCs may lie between zero (in the core) and one (on the nanocrystal surface), which is consistent with previous reports [11, 12].

Fig. 2
figure 2

X-ray photoelectron spectra (a) and Fourier-transform infrared spectra (b) of Cu NCs with different surface ligands

Full size image

Furthermore, the surface bonds of Cu NCs were characterized using Fourier-transform infrared (FTIR) spectroscopy. As shown in Fig. 2b, the broad peaks between 3550 and 3300 cm−1 were attributed to stretching of the amino group (NH2) existed on the surface of Cys- and GSH-capped Cu NCs, while for EDA-coated NCs, the above peak is not evident, suggesting that EDA is connected with the Cu core through NH2. The peak between 2660 and 2550 cm−1, which corresponds to the S–H stretching vibration mode of Cys and GSH molecules, was not found in the corresponding spectra, indicating that for the Cys- and GSH-capped NCs, thiols of the ligands may form covalent bonds to metal atoms [11, 23].

Since the synthetic conditions for obtaining the blue-, green-, and red-emitting Cu NCs were maintained the same except for the ligand species, it is clear that the ligands are key factors for the nucleation and growth processes during the reaction to form different emitted NCs. To study the morphology variations when changing the surface ligand, transmission electron microscope (TEM) measurements of the QDs were performed and the images are shown in Fig. 3. It was observed that Cu NCs were mainly distributed in the range of 2–3.4 nm with an average diameter of 2.6 nm for EDA-capped particles (Fig. 3a–b). While for Cys- and GSH-coated NCs, the average particle size was increased to 3.0 ± 0.3 nm (Fig. 3c–d) and 4.4 ± 0.5 nm (Fig. 3e–f) respectively. This suggests that in the current reaction system, the GSH ligand can modulate the Cu precursor to grow into larger NCs. During the particle formation processes, ligands coordinate to the particle surface to prevent aggregation. On the other hand, ligands can also coordinate with the Cu monomer, which can change its reactivity. Since EDA has only the amine group to bind with Cu, while Cys possesses thiol group, amine group, and carboxyl group to coordinate with Cu precursor, and GSH also has the above three functional groups but with more quantities, the reactivity of the monomer was reduced furthest by GSH due to its strongest binding effect [32]. From a thermodynamics aspect, the supersaturation degree of the reaction decreased consequently, which is favorable for growing larger particles out of the reaction system, owing to the reduced number of nuclei generated during the nucleation process [37]. The PL of Cu NCs is known to be strongly dependent on the particle size; therefore, employing GSH as surface ligand led to large Cu NCs with red fluorescence, while EDA and Cys tend to form NCs with shorter wavelength emission.

Fig. 3
figure 3

Transmission electron microscope images and size histograms of Cu NCs capped with ethanediamine (a, b), cysteine (c, d), and glutathione (e, f)

Full size image

Photoluminescence lifetime and fluorescent stability of Cu NCs

Since the EDA-capped Cu NCs possess the highest PL intensity, they were chosen as representative samples to study the transient PL property and fluorescent stability. As shown in Fig. 4a, the PL decay curve of blue-emitting NCs can be fitted by a two-exponential function: \( I\left(\mathrm{t}\right)={B}_1\exp \left(-\frac{t}{\tau_1}\right)+{B}_2\exp \left(-\frac{t}{\tau_2}\right) \), where τ1 and τ2 represent the decay time constants of the PL mission and B1 and B2 represent the normalized amplitudes of each component. Two components of 2.1 ns (28.4%) and 3.7 ns (71.6%) were obtained, providing an average PL lifetime of 3.6 ns. Such short PL lifetime can be attributed to the emission from the singlet excited state of metal core, as has been previously observed for blue-emitting Cu NCs [3]. The fluorescent stability of metal NCs is a key concern for their practical application, to assure the reliability of the signal.

Fig. 4
figure 4

Normalized time-resolved PL decay curve (a) and the temporal evolutions of the PL intensity and PL peak of Cu NCs with ethanediamine as a surface ligand (b)

Full size image

As shown in Fig. 4b, the NCs stabilized by EDA possess terrific fluorescent stability against long placing time until 34 days, manifesting as both nearly unchanged PL peak position and intensity. This lays a solid foundation for the following sensing application of the as-prepared Cu NCs.

Performance of Hg2+ determination

Mercury, as one of the most toxic heavy metal pollutants, widely exists in water, soil, and foodstuffs [38,39,40,41,42,43]. In view of its toxicity, exploring effective strategies for Hg2+ determination is highly demanded. Based on the above-prepared blue-emitting Cu NCs with high PL intensity and stability, a fluorescent sensor for Hg2+ determination was developed through a mild and rapid way. Before the prepared Cu NCs were applied for assaying Hg2+, repetitive ultrafiltration was performed for removing the influence of excessive ethanediamine and assuring the same background. During the assaying process, the pH of the buffer will significantly influence the sensitivity. Therefore, the pH value of the assay was optimized as shown in Fig. S4. It turned out that a nearly neutral medium facilitates the decrease of relative PL intensity, with a maximum decrease obtained at the pH 6.0. The weak acid environment can effectively reduce the protonation of the amine group on the surface of Cu NCs which would hinder the binding of Hg2+, while alkaline solution would accelerate the hydrolysis of Hg2+.

Figure 5a presents the PL changes of Cu NCs (with an emission peak at 420 nm excited by the wavelength of 350 nm) by introducing different concentrations of Hg2+ (0.1–5 mM) within 10 min at room temperature and optimal pH value of 6.0. It is evident that the PL intensity continuously decreased along with the increase of Hg2+ concentration, and the PL of NCs was quenched by more than 90% upon the addition of 5 mM Hg2+. Apart from the change of PL intensity, the PL peak position red-shifted from 425 to 437 nm at high Hg2+ concentration (> 1 mM). This red-shift of PL peak may be correlated with the change of nanocluster surface state which resulted from the binding between Hg2+ and the amine group. To explore the mechanism of Hg2+-induced fluorescence quenching of EDA-capped Cu NCs, UV–Vis absorption spectra of EDA and EDA-capped Cu NCs before and after the addition of Hg2+ were further investigated. As shown in Fig. S5, the Cu NCs have a well-resolved absorption peak at 340 nm (curve b), which is remarkably different from that of the surface ligand EDA (curve a, structureless absorption above 300 nm). While in the presence of Hg2+, the characteristic peak of Cu NCs at 340 nm disappeared (curve c), presenting similar absorption compared with that of EDA with Hg2+ (curve d). It means that ethanediamine has a stronger coordination affinity with Hg2+ to form stable Hg-EDA complex, which induces the fluorescence quenching of Cu NCs.

Fig. 5
figure 5

PL spectra of Cu NCs capped with ethanediamine in the presence of different amounts of Hg2+ ions (0.1–5 mM) with an excitation wavelength of 350 nm (a), linear calibration with the standard deviation of three repeated measurements as the error bar (b, c), and normalized PL intensity (with error bars calculated from three parallel measurements) and peak shift towards low energy region of Cu NCs in the absence or presence of different ions with the concentration of 1 mM (d)

Full size image

The performance of Hg2+ sensing was quantitatively shown in Fig. 5b and c. The relative PL intensity of Cu NCs with an emission peak at 420 nm shows a good linear relationship with Hg2+ concentration in two ranges: from 0.1–1.0 mM with R2 of 0.992 (linear equation: I/I0 = 1.102–0.782C) and from 1.0–5.0 mM with R2 of 0.991 (linear equation: I/I0 = 0.399–0.0641C). The detection limit was calculated to be 33 μM (S/N = 3). The analytical performance of the obtained Cu NCs towards Hg2+ is not superior compared with previous reports as shown in Table S1, which would hinder the application for trace detection. Nevertheless, the assay is mild, rapid, and low-cost, and most importantly, the Cu NCs can be directly used as the sensing probe without further modification, which is suitable for the fast assay of Hg2+.

To assess the selectivity of Cu NCs for Hg2+, the PL spectra of NCs were measured in the presence of other ten metal ions including Na+, Ag+, Pb2+, Cd2+, Ba2+, Ni2+, Co2+, Mg2+, Mn2+, and Cu2+, and anions including Cl−, OH−, NO3−, SO42−, and ClO4−. As shown in Fig. 5d, the relative PL intensities of the NCs generally exhibit no significant changes upon the above ions compared with Hg2+ under the same ion concentration, even in the presence of commonly existing heavy metal ions such as Cu2+, Pb2+, and Cd2+. Since the fluorescence quenching mechanism of Cu NCs towards Hg2+ is the greater coordination affinity of Hg2+ with ethanediamine, the weak response of Cu NCs to other metal ions can be attributed to their weaker combining capability with ethanediamine, which is not strong enough to grab the surface ligand of Cu NCs. For example, the formation constants (log K1) of Cu(II)-EDA and Cd(II)-EDA complexes are 10.67 and 5.47 respectively, while that of Hg(II)-EDA is 14.3 [44]. More importantly, the presence of possible interfering ions can hardly shift the PL peak position of the Cu NCs except that Pb2+ caused a blue-shift of about 4 nm, while Hg2+ resulted in a red-shift of about 5 nm. The fluorescence spectra with different emission peaks further ensure the selectivity. Therefore, the interfering ions have nearly no influence on the sensing capability of the Cu NCs for Hg2+ determination. Furthermore, the Cu NCs were applied to assay Hg2+ in tap water. As shown in Table S2, the recovery rates were 100.5%, 99.46%, and 103.4% with RSD values of less than 4.9% for three samples, revealing good potential of the current method for sensing Hg2+ in real samples. To further prove the accuracy of the assay, water samples were also tested by the atomic fluorescence spectrometry (AFS) method. The results in Table S2 show that our method and AFS have comparable accuracy, demonstrating the reliability of the method for sensing Hg2+. However, the sensitivity of the developed method based on Cu NCs needs to be improved for trace detection in real samples.

Conclusions

Water-soluble Cu NCs with blue, green, and red emissions were synthesized by choosing ethanediamine, cysteine, and glutathione as surface ligands respectively. The formation of Cu NCs supported by the above ligands was confirmed by XPS, FTIR, and TEM characterizations. The dependence of emission color on the structures of ligands was mainly attributed to their different chelating ability, which can greatly affect the supersaturation degree of monomers bound to ligands. As a consequence, NCs with different sizes were formed and exhibited different emission wavelengths due to the size effect. Without further surface modification, Cu NCs capped with ethanediamine were directly employed as the probe for mild and rapid sensing of Hg2+ with satisfying selectivity. The fluorescence quenching mechanism was mainly attributed to the strong coordination affinity of Hg2+ with ethanediamine. Because the Cu element is low-toxic and earth-abundant, the current work of tuning the emission color of Cu NCs by ligand effect has provided an effective approach towards NCs with controlled properties and potentially scaling up capacity. Nevertheless, the sensitivity of the assay needs to be further improved for trace detection of Hg(II), which can find new ways by combining with other functional nanomaterials.