Monitoring the charge-transfer process in a Nd-doped semiconductor based on photoluminescence and SERS technology

Surface-enhanced Raman scattering (SERS) and photoluminescence (PL) are important photoexcitation spectroscopy techniques; however, understanding how to analyze and modulate the relationship between SERS and PL is rather important for enhancing SERS, having a great effect on practical applications. In this work, a charge-transfer (CT) mechanism is proposed to investigate the change and relationships between SERS and PL. Analyzing the change in PL and SERS before and after the adsorption of the probe molecules on Nd-doped ZnO indicates that the unique optical characteristics of Nd3+ ions increase the SERS signal. On the other hand, the observed SERS can be used to explain the cause of PL background reduction. This study demonstrates that modulating the interaction between the probe molecules and the substrate can not only enhance Raman scattering but also reduce the SERS background. Our work also provides a guideline for the investigation of CT as well as a new method for exploring fluorescence quenching. Sensors that use laser light scattering to detect molecular-scale changes to samples can be improved with a new substrate that naturally reduces background interference. Surface-enhanced Raman spectroscopy (SERS) setups typically use a combination of metal nanoparticles and fluorescent probes to capture vibrational signals from biomolecules including DNA. Ming Gao from Jilin Normal University in Siping, China, and colleagues now report a technique to keep probe emissions from interfering with SERS signals. To accomplish this, the team synthesized a neodymium-doped zinc oxide spiky nanostructures substrate. Spectroscopic measurements revealed that neodymium’s electronic properties enabled it to transfer photoexcited electrons to the probes. This charge transfer reduced the probe’s fluorescent intensity and simultaneously enhanced SERS emissions beyond that available on pure zinc oxide surfaces. This study provides a new way to naturally eliminate the SERS fluorescence background by doping with Nd ions. These fundamental discussions here provide a path to enhance Raman scattering and reduce the SERS background and a guideline for the investigation of CT and a new method for exploring fluorescence quenching.


Introduction
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 response [1][2][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 substrates 5 . 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 tag 6,7 . 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 background 8 . 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 highenergy-level instability, these electrons usually transition to a low energy level and emit photons 9,10 . 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 junctions 11,12 . 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 semiconductor 10,13 . The intra-4f emission spectra of Nd 3+ 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 electrons 14 . Thus, incorporating Nd 3+ ions into a semiconductor might be an effective way to enhance the SERS signal. On the other hand, the characteristic PL peaks of Nd 3+ 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 nanomaterials 15,16 .
In this work, we successfully synthesized Nd-doped ZnO (Zn 1 − x Nd x O) 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 Nd 3+ 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. Zn 1 − x Nd x O (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  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 Zn 1 − x Nd x O (x = 0.00, 0.01, 0.02) are shown in Fig. 1a. It is noteworthy that there are some characteristic peaks of Nd 3+ ions at 879.4, 805.2, 742.5, 685.1, 578.2, and 521.0 nm corresponding to intra-4f shell electron transitions of Nd 3+ ions 18 , 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. 1b 19 . The band gaps of Zn 1 − x Nd x O (x = 0.00, 0.01, 0.02) were derived to be 3.23, 3.14, and 3.11 eV, respectively. Zn 1 − x Nd x O has a smaller band gap value, indicating that it has a preferable optical absorption property, making the interband charge transition easier.
To explore the relationship between SERS and PL, we obtained SERS and PL spectra measured with laser lines of 514.5 nm. 4-Mpy (C 5 H 5 NS) was selected as a probe, and the SERS spectra of 4-MPy adsorbed on Zn 1 − x Nd x O are shown in Fig. 2a (several representative concentrations are shown in Fig. 2a, and other concentrations are given in the Supporting Information, Fig. S4)  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 Zn 0.98 Nd 0.02 O. Figure 2c displays the PL spectra of Zn 1 − x Nd x O. The Nddoped 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 ( 4 F 3/2 , 4 F 5/2 + 2 H 9/2 , 4 F 9/2 , and 4 G 7/2 + 2 G 5/2 ), corresponding to the UV-vis absorption spectra 20 . Figure 2d shows the Nd-ion doping concentration-dependent PL intensity of Zn 1 − x Nd x O, 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 further investigate the relationship between the PL and SERS of Zn 1 − x Nd x O, we examined the change in the PL spectrum of Zn 1 − x Nd x O (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 + Zn 0.98 Nd 0.02 O is reduced in comparison with that of 4-MPy molecules. The intensities of the luminescence peaks at 897.6 and 815.1 nm ( 4 F 3/2 and 4 F 5/2 + 2 H 9/2 ) significantly decrease, while that of the luminescence peak at 604.0 nm ( 4 G 7/2 + 2 G 5/2 ) shows a relatively small decrease. In addition, the SERS signal of 4-MPy + Zn 0.98 Nd 0.02 O is significantly increased (Fig. 2a), indicating that there is obvious CT between Zn 0.98 Nd 0.02 O substrates and 4-MPy molecules.
To probe the CT mechanism between Zn 1 − x Nd x O (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  Figure 3b shows the PL spectra of pure ZnO and Zn 0.98 Nd 0.02 O 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 nm 21 . The DLE is attributed to intrinsic defects, such as oxygen vacancies or various surface states 22 . 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 eV 13 , located above the top of the VB (Fig. 3b). In addition, theoretical calculations indicate that the electronic ground state ( 4 I 9/2 ) of the Nd 3+ ions is located~1 eV below the top of the VB 23 . Thus, the excited-state energy levels of the Nd 3+ 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 Zn 1 − x Nd x O (x = 0.00, 0.01, 0.02, taking Zn 0.98 Nd 0.02 O 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 photon 24 . 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.
As shown in Fig. 4b, for Zn 0.98 Nd 0.02 O, the 514.5-nm laser is able to excite the electrons of the 4f shell of the Nd 3+ ions from the ground state ( 4 I 9/2 ) to the excited states ( 4 F 3/2 , 4 F 5/2 + 2 H 9/2 , 4 F 9/2 , and 4 G 7/2 + 2 G 5/2 ). Then, the excited-state electrons transfer to the LUMO level and finally return to the ground state of the Nd 3+ ions, releasing Raman photons. CT from the excited-state electrons of the Nd 3+ ions to the 4-MPy molecule reduces the number of electrons returning to the ground state, resulting in fluorescence quenching of Nd 3+ ions. The energy required for transition from the excited-state 4 F 3/2 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 4 F 5/2 + 2 H 9/2 transitions to a higher unoccupied molecular orbital, CT resonance can also occur. Thus, the luminescence peaks ( 4 F 3/2 and 4 F 5/2 + 2 H 9/2 ) significantly decrease. However, the energy required to transfer electrons at 4 G 7/2 + 2 G 5/2 to the 4-MPy molecule is much lower in comparison with that for 4 F 3/2 . The CT resonance cannot occur, resulting in a lower probability of CT. As a consequence, the intensity of the 4 G 7/2 + 2 G 5/2 luminescence peak has only a relatively small decrease. To better prove this electron transfer between Zn 0.98 Nd 0.02 O and 4-MPy, we measured the fluorescence lifetime of Zn 0.98 Nd 0.02 O before and after absorbing the 4-MPy molecule and collected their emission decays at 4 G 7/2 + 2 G 5/2 (Fig. S8a) and 4 F 3/2 (Fig. S8b). It was found that when the 4-MPy molecule is adsorbed, the lifetime of the 4 F 3/2 energy level (897.6 nm) decreases rapidly, whereas the lifetime of the 4 G 5/2 + 2 G 7/2 energy level (604.0 nm) remains almost unchanged. Therefore, the fluorescence quenching of Nd 3+ ions significantly increases the SERS intensity of Zn 1 − x Nd x O, realizing that the utilization of unique optical characteristics of Nd 3+ 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 Nd 3+ ions can be excited into the higher empty 4f-5d levels, and then these excited 4f-5d electrons jump To further verify these CT processes, the SERS spectra of Zn 1 − x Nd x O (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 4 I 9/2 to 4 F 9/2 (1.84 eV) and then transfer to the LUMO level (1.93 eV). However, the 785nm laser (1.58 eV) preferentially excites electrons from 4 I 9/2 to 4 F 5/2 + 2 H 9/2 (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.
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 Nd 3+ 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.
We evaluated the structural quality of the samples with X-ray diffraction (XRD, Rigaku D/Max 3C). X-ray photoelectron spectroscopy (XPS, VG ESCALAB 250X) was used to analyze the element content of the samples. The morphology was characterized by field emission SEM