Surface-enhanced Raman scattering (SERS), as a powerful spectral technology, has been widely used in the fields of chemistry, pharmaceuticals, biosensors, food detection, and environmental monitoring owing to its high sensitivity and fast response1-3. In general, the enhanced magnitude of SERS is associated with two mechanisms. One is the electromagnetic mechanism, which is related to the localized surface plasmon resonance of the metal nanoparticles (NPs)4. The other is the chemical mechanism, which mainly originates from the charge transfer (CT) between adsorbed molecules and SERS-active substrates5. On the other hand, the photoluminescence (PL) of the substrates and adsorbed molecules, acting as a broad-continuum background of SERS spectroscopy, has a great effect on SERS spectroscopy and evenly reduces the distinctiveness of the Raman tag6, 7. Thus, controlling and utilizing PL, contributed by substrates and adsorbed molecules, to enhance the SERS signal is a key problem that urgently needs to be solved. Recently, Ren et al. successfully resolved these limitations by proposing a method for recovering native chemical information from SERS using plasmonic PL and quantitatively investigated the relationship between the PL and the SERS background8. However, the development of simpler and more effective methods to remove the negative effects of PL on SERS and to directly analyze the relationship between SERS and PL is of significance for both fundamental research and practical application.
In the PL generation mechanism, photoexcited electrons transition to a high energy level. Because of high-energy-level instability, these electrons usually transition to a low energy level and emit photons9, 10. This inspires us to use electrons as a link to explore the relationship between PL and SERS in the CT mechanism and investigate how PL affects the SERS signal. SERS technology has been widely used to monitor the charge transport of substrate–molecular junctions11, 12. Generally, when a semiconductor is used as a SERS substrate, only the CT enhancement mechanism contributes to SERS signals. However, exploring the relationship between SERS and PL of pure semiconductors is not obvious and thus is not convenient to analyze. Previously, we introduced impurity ions to optimize the matrix semiconductor10, 13. The intra-4f emission spectra of Nd3+ are characterized by narrow lines with high color purity because the 4f electrons of rare-earth ions are shielded from external forces by the outer 5s and 5p electrons14. Thus, incorporating Nd3+ ions into a semiconductor might be an effective way to enhance the SERS signal. On the other hand, the characteristic PL peaks of Nd3+ ions can be used to analyze the relationship between PL and SERS. In this study, the nanomaterial ZnO was selected as a SERS substrate owing to its good optical stability and relatively high SERS activity among the reported semiconductor nanomaterials15, 16.
In this work, we successfully synthesized Nd-doped ZnO (Zn1 − xNdxO) as a SERS substrate in which Nd doping was performed using a simple chemical method. Here, we creatively used the CT mechanism to establish the relationship between SERS and PL and examined the change in SERS and PL caused by the CT mechanism in detail. We found that both Nd3+ ions and probe molecule fluorescence quenching promote CT, enhance the SERS effect, and reduce the SERS background. This is thus the first example of using PL to enhance SERS and provides a new method for exploring fluorescence quenching.
Zn1 − xNdxO (x = 0.00, 0.005, 0.01, 0.015, 0.0175, 0.02, 0.0225, and 0.03) was prepared via the coprecipitation method (details regarding the synthesis are presented in the Supporting Information, Fig. S1). X-ray diffraction (XRD) patterns (Fig. S2) revealed that the solubility limit of Nd3+ ions is ~0.02 in the Zn1 − xNdxO structure and that excessive doping causes the precipitation of Nd2O3 NPs17. The morphologies of ZnO and Zn0.98Nd0.02O were examined via scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The SEM images show that pure ZnO is a NP, aggregating with a diameter of ~100 nm (Fig. S3a); after Nd doping, Zn0.98Nd0.02O is a 3D urchin-like nanostructure with a diameter of 2 μm (Fig. S3c). Figure S3b, d (e.g., TEM images) shows that the surface area increases after doping, which improves the adsorption capacity. Among these eight samples with different doping amounts, we selected 0, 1, and 2% representative data for subsequent research. The UV–vis absorption spectra of Zn1 − xNdxO (x = 0.00, 0.01, 0.02) are shown in Fig. 1a. It is noteworthy that there are some characteristic peaks of Nd3+ ions at 879.4, 805.2, 742.5, 685.1, 578.2, and 521.0 nm corresponding to intra-4f shell electron transitions of Nd3+ ions18, as shown in the inset of Fig. 1a. It is well known that optical absorption properties are associated with the optical band gap (Eg), which can be obtained by Tauc's formula (αhν)2 = hν − Eg, as shown in Fig. 1b19. The band gaps of Zn1 − xNdxO (x = 0.00, 0.01, 0.02) were derived to be 3.23, 3.14, and 3.11 eV, respectively. Zn1 − xNdxO has a smaller band gap value, indicating that it has a preferable optical absorption property, making the interband charge transition easier.
The UV-vis absorption spectra and of Optical energy band gap ZnO and Zn1−xNdxO.
a The UV–vis absorption spectra of ZnO and Zn1−xNdxO (x = 0.00, 0.01, 0.02). b Optical energy band gap of ZnO and Zn1−xNdxO (x = 0.00, 0.01, 0.02)
To explore the relationship between SERS and PL, we obtained SERS and PL spectra measured with laser lines of 514.5 nm. 4-Mpy (C5H5NS) was selected as a probe, and the SERS spectra of 4-MPy adsorbed on Zn1 − xNdxO are shown in Fig. 2a (several representative concentrations are shown in Fig. 2a, and other concentrations are given in the Supporting Information, Fig. S4). The SERS spectral signal is enhanced after Nd doping, while the signal contributed by the fluorescence background is weakened. Figure 2b shows the SERS intensity of the 1593 cm−1 band of 4-MPy plotted as a function of the Nd concentration. The SERS intensity increases with the Nd concentration and reaches a maximum for Zn0.98Nd0.02O. Figure 2c displays the PL spectra of Zn1 − xNdxO. The Nd-doped ZnO exhibits dramatically sharp luminescence peaks, which is in sharp contrast with pure ZnO. Moreover, luminescent peaks are located at 897.6, 815.1, 674.2, and 604.0 nm (4F3/2, 4F5/2 + 2H9/2, 4F9/2, and 4G7/2 + 2G5/2), corresponding to the UV–vis absorption spectra20. Figure 2d shows the Nd-ion doping concentration-dependent PL intensity of Zn1 − xNdxO, which initially increases with the enhancement of the Nd concentration. Notably, the change in the PL intensity (Fig. 2d) is consistent with the change in the SERS intensity (Fig. 2b).
To explore the relationship between SERS and PL, SERS spectra of 4-MPy adsorbed on Zn1−xNdxO and PL spectra of Zn1−xNdxO measured with laser lines of 514.5 nm.
a SERS spectra of 4-MPy adsorbed on Zn1−xNdxO (x = 0.00, 0.01, 0.02) under a 514.5-nm laser. b A plot of the SERS intensity of the 1593 cm−1 band of 4-MPy versus Nd concentration. c PL spectra of Zn1−xNdxO (x = 0.00, 0.01, 0.02) under a 514.5-nm laser. d A plot of the PL intensity of 604.0 and 897.6 nm versus Nd concentration
To further investigate the relationship between the PL and SERS of Zn1 − xNdxO, we examined the change in the PL spectrum of Zn1 − xNdxO (x = 0.00, 0.01, 0.02) with and without 4-MPy molecules. Figure 3a shows that the PL signal of 4-MPy + ZnO is a simple superposition of the fluorescence of ZnO and 4-MPy, but the PL signal of 4-MPy + Zn0.98Nd0.02O is reduced in comparison with that of 4-MPy molecules. The intensities of the luminescence peaks at 897.6 and 815.1 nm (4F3/2 and 4F5/2 + 2H9/2) significantly decrease, while that of the luminescence peak at 604.0 nm (4G7/2 + 2G5/2) shows a relatively small decrease. In addition, the SERS signal of 4-MPy + Zn0.98Nd0.02O is significantly increased (Fig. 2a), indicating that there is obvious CT between Zn0.98Nd0.02O substrates and 4-MPy molecules.
PL spectra of Zn1−xNdxO with and without 4-MPy molecules upon 514.5 nm and 325 nm laser excitation.
a PL spectra of ZnO, Zn0.98Nd0.02O, 4-MPy, 4-MPy+ZnO, and 4-MPy+Zn0.98Nd0.02O upon 514.5-nm laser excitation; b PL spectra of ZnO, Zn0.98Nd0.02O, and ZnO upon 325-nm laser excitation
To probe the CT mechanism between Zn1 − xNdxO (x = 0.00, 0.01, 0.02) and 4-MPy in the SERS spectrum, first, UPS and UV–vis spectra were used to determine the position of each energy level. The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) levels of 4-MPy are −9.77 and −6.34 eV, respectively (calculated from Figs. S5 and S6). The maximum valence band (VB) and minimum conduction band (CB) of Zn0.98Nd0.02O are −9.11 and −6.00 eV (calculation process in SI, Fig. S7), respectively. Figure 3b shows the PL spectra of pure ZnO and Zn0.98Nd0.02O upon 325-nm laser excitation. The two samples consist of two emission bands: a near band edge at ~383 nm and a wide deep level emission (DLE) from 480 to 660 nm21. The DLE is attributed to intrinsic defects, such as oxygen vacancies or various surface states22. On the other hand, Nd doping increases the number of surface defects of ZnO. These oxygen vacancies and surface defects induce new surface state energy levels (Ess) of 1.88–2.58 eV13, located above the top of the VB (Fig. 3b). In addition, theoretical calculations indicate that the electronic ground state (4I9/2) of the Nd3+ ions is located ~1 eV below the top of the VB23. Thus, the excited-state energy levels of the Nd3+ ions are located at −8.73, −8.61, −8.27, and −8.06 eV, respectively.
Based on the above results, we analyzed the enhancement mechanism of Zn1 − xNdxO (x = 0.00, 0.01, 0.02, taking Zn0.98Nd0.02O as an example). As shown in Fig. 4a, upon excitation at 514.5 nm (2.41 eV), the VB electrons of ZnO can be excited to Ess, transition to the LUMO level of the adsorbed 4-MPy molecules, and finally return to the VB of ZnO to release a Raman photon24. Considering that pure ZnO contains few oxygen defects, ZnO only undergoes CT in this process. Therefore, its SERS intensity is very low, and the PL signal with the probe molecules is superimposed.
Schematic representations for charge-transfer mechanisms between a 4-MPy and ZnO and b 4-MPy and Zn0.98Nd0.02O
As shown in Fig. 4b, for Zn0.98Nd0.02O, the 514.5-nm laser is able to excite the electrons of the 4f shell of the Nd3+ ions from the ground state (4I9/2) to the excited states (4F3/2, 4F5/2 + 2H9/2, 4F9/2, and 4G7/2 + 2G5/2). Then, the excited-state electrons transfer to the LUMO level and finally return to the ground state of the Nd3+ ions, releasing Raman photons. CT from the excited-state electrons of the Nd3+ ions to the 4-MPy molecule reduces the number of electrons returning to the ground state, resulting in fluorescence quenching of Nd3+ ions. The energy required for transition from the excited-state 4F3/2 to the LUMO is 2.39 eV, and the laser energy of 514.5 nm is exactly 2.41 eV. The two almost identical energy levels induce the occurrence of CT resonance. When 4F5/2 + 2H9/2 transitions to a higher unoccupied molecular orbital, CT resonance can also occur. Thus, the luminescence peaks (4F3/2 and 4F5/2 + 2H9/2) significantly decrease. However, the energy required to transfer electrons at 4G7/2 + 2G5/2 to the 4-MPy molecule is much lower in comparison with that for 4F3/2. The CT resonance cannot occur, resulting in a lower probability of CT. As a consequence, the intensity of the 4G7/2 + 2G5/2 luminescence peak has only a relatively small decrease. To better prove this electron transfer between Zn0.98Nd0.02O and 4-MPy, we measured the fluorescence lifetime of Zn0.98Nd0.02O before and after absorbing the 4-MPy molecule and collected their emission decays at 4G7/2 + 2G5/2 (Fig. S8a) and 4F3/2 (Fig. S8b). It was found that when the 4-MPy molecule is adsorbed, the lifetime of the 4F3/2 energy level (897.6 nm) decreases rapidly, whereas the lifetime of the 4G5/2 + 2G7/2 energy level (604.0 nm) remains almost unchanged. Therefore, the fluorescence quenching of Nd3+ ions significantly increases the SERS intensity of Zn1 − xNdxO, realizing that the utilization of unique optical characteristics of Nd3+ ions promotes the SERS signal. As shown in the theoretical calculation (Fig. S9), the empty Nd 4f impurity levels are close to the Nd 5d levels, leading to mixing of the 4f and small amounts of the 5d orbitals, referred to as the 4f–5d orbital. Under laser irradiation, the 4f ground-state electrons of Nd3+ ions can be excited into the higher empty 4f–5d levels, and then these excited 4f–5d electrons jump into the 5d levels under continuous irradiation25. At the same time, the electrons in the 5d orbital are reductive and easily lose electrons under irradiation by light26. Overall, the CT channel, originating from charge-transfer resonance, overcomes the effect of the 4f levels shielded by external 5s and 5p electrons.
Furthermore, upon laser excitation, the HOMO-level electrons of the 4-MPy molecules transfer to the excited states and then to the CB of Zn1 − xNdxO. They finally return to the HOMO level of the 4-MPy molecules, releasing Raman photons. This process inhibits the recombination of electrons and holes in the HOMO level, resulting in fluorescence quenching of 4-MPy molecules. The above CT processes work together toward the enhancement of SERS signals and significantly reduce the fluorescence background.
To further verify these CT processes, the SERS spectra of Zn1 − xNdxO (x = 0.00, 0.01, 0.02) were analyzed under 633 and 785 nm laser irradiation (Fig. 4). The SERS spectra at 633 nm irradiation were essentially the same as the spectrum at 532 nm irradiation (Fig. 5a). However, the SERS spectra at 785 nm irradiation were significantly different (Fig. 5b): the SERS background increased significantly, while the SERS intensity was only slightly enhanced. The 633-nm laser (1.96 eV) can maximally excite electrons from 4I9/2 to 4F9/2 (1.84 eV) and then transfer to the LUMO level (1.93 eV). However, the 785-nm laser (1.58 eV) preferentially excites electrons from 4I9/2 to 4F5/2 + 2H9/2 (1.52 eV), but the laser energy is not sufficient to again excite the electrons to the LUMO level (2.27 eV). In the case of an increase in the PL of Nd-doped ZnO, the inability of the substrate to reduce the PL signal via CT is responsible for the SERS background enhancement. Although the SERS spectra obtained under different excitation lines are different, they all support our proposed mechanism.
SERS spectra of 4-MPy adsorbed on Zn1 − xNdxO (x = 0.00, 0.01, 0.02) under different laser excitation: a 633 nm and b 785 nm
In summary, we analyzed the change between the PL and SERS relative intensities of Nd-doped ZnO before and after the adsorption of probe molecules to explore the CT mechanisms in Nd-doped ZnO systems. The results indicated that the unique CT between Nd3+ ions and probe molecules improves the SERS performance and naturally eliminates the SERS fluorescent background. Moreover, the mechanism is further confirmed by examining the SERS spectra under various excitation wavelengths. This work paves the way for developing novel molecular-sensing techniques.