MXenes induce epitaxial growth of size-controlled noble nanometals: A case study for surface enhanced Raman scattering (SERS)

1. Introduction

Noble metal nanoparticles with interesting physical and chemical properties have been extensively studied because of fundamental scientific interest and technological applications in diverse research areas. These areas include but not limited to catalysis [1,2], electronics [3], sensing [4], photonics [5], imaging [6], and biomedicine [7,8]. Numerous scientific fields have taken advantage of the localized surface plasmon resonance (LSPR) of noble metal nanoparticles. The LSPR is dependent on the size, shape, and composition of the nanoparticles as well as external factors [9]. Therefore, extensive efforts have been devoted to the synthetic studies on noble metal nanoparticles.

The sythetic methods for noble metal nanoparticels can be categorized into two major classes: physical fabrication and wet-chemical synthesis. In physical fabrication where lithographic method is becoming more and more dominated, noble metal nanoparticles are typically deposited using various patterning methods, under vaccum, and immobilzed on supporting substrates to form nanoparticles arrays [10,11]. The stability of the physically fabricated nanoparticles is outstanding during long-term storage and use [12]. In wet-chemical synthesis, chemical reduction dominates, nanoparticle size and morphology can be tunned by various reaction parameters, such as chemical precursor choice, temperature, pH, or reaction time [10,13]. Stabilizing agents must be present during and after nucleation and growth to imbue the nanoparticles with colloidal stability. Different from the physical fabrication that always requires expensive and complex equipements, chemical reduction synthesis is cost-efficient, affordable and flexible to implement. However, chemical reduction synthesis needs to remove two obstacles. First, the presence of surfactants as stabilizing agents that will introduce interference especially in the detection of analytes with weak adsorption. Second, the absence of supporting substrates that significantly improve the stability during long-term storage and use. How to take advantage of chemical reduction method while avoiding its inherent shortcomings has always been a crucial issue in variety of areas, such as fuel cells [14,15], electrochemical sensors [16], and surface enhanced Raman scattering (SERS) [17]. The key to this issue is to find solid reducing agents with two-dimensional (2D) characteristic, which match energetically with noble metal cations as oxidants.

MXenes [[18], [19], [20], [21]], a new class of 2D compounds, show promise as such agents. As pioneered by research groups led by Gogotsi and colleagues [[22], [23], [24]], MXenes, are demonstrated to be of particular interest for both fundamental and applied research, because its composite is a diverse set of atomic structure, with a chemical formula of M_n+1X_nT_x, where M is a transition metal [25], X is C and/or N [26], T_x stands for surface termination groups of O, F, or OH. Importantly, in stark contrast to other 2D materials [27], MXenes have unambiguous characteristics, namely, transition metal M has evident electron-donating ability as the oxidation number of M in MXene is much less than that of the corresponding oxide [28]. The intrinsically conductive nature along with the electron-donating capability of MXenes offers a novel and universal platform for the construction of functional noble metal nanoparticles immobilized on supporting substrates.

Herein, we report a general approach to nanostructure-controlled synthesis of noble metal using MXene as solid reducing agent. We show that MXene does have the ability of donating electrons to noble metal cations, directly reducing the cations into metallic nanostructures via in situ redox without invoking any external reductants. Moreover, MXene transfers electrons directly to the noble metal cations, rendering the formed metallic nanostructures to firmly anchor on MXene. The strategy proposed here for in situ redox reaction between noble metal solutions and MXene offers a platform for constructing powerful, reproducible and surface-active substrates with noble metal nanostructures, as evidenced by primary SERS evaluations. This work addresses a long-standing issue of external reductant interference and points the way to new classes of solid reductant for directly anchoring noble metal nanoparticles on solids, which definitely promotes basic and applied research in the fields of fuel cells and heterogeneous catalysis in which noble metals are critically important.

2. Experimental methods

2.1. Materials

Ti powders (99%, -300 mesh) were provided by Haixin (China), Nb powders (99.9%, -325 mesh) were purchased from Beijing Goodwill Metallic Co., Ltd. Al powders (99%, D₅₀ = 10 μm) were obtained from AIFA (China), and graphite powders (99%, D₉₀ = 6.5 μm) were purchased from TIMCAL (Switzerland). Concentrated hydrochloric acid, LiF, HAuCl₄⋅4H₂O, PdCl₂, H₂PtCl₆⋅6H₂O, AgNO₃ and methylene blue trihydrate were of analytical grade and all were provided by Sinopharm Chemical Reagent Co., Ltd., Shanghai, China. All the chemicals were used as received without further purification.

2.2. Preparation of Ti₃C₂T_x freestanding membrane

Ti₃C₂T_x MXene was derived from Ti₃AlC₂ MAX phase as precursor. The synthesis of Ti₃AlC₂ was achieved by heating a blend of elemental powders of Ti, Al and C at 1400 °C for 1.5 h, following the method described previously [29,30]. Ti₃AlC₂ powders obtained by drilling the as-prepared porous Ti₃AlC₂ monolith were etched in a mixed solution. The solution was prepared by dissolving LiF (1.0 g) in 20 mL of 9 mol L^-1 hydrochloric acid. For etching, the Ti₃AlC₂ powders (1.0 g) were slowly added to the mixed solution and etched at room temperature for 20 days. After that, the resultant was shaken and washed at least 3 times with deionized water, until the supernatant reached a pH of ~5. With filtration on cellulose filter membrane (0.22 μm pore size), a slurry-like resultant was obtained. The Ti₃C₂T_x slurry was then dispersed in 250 mL of deionized water, forming a suspension. The suspension was centrifuged at 4000 rpm for 10 min and the supernatant was collected for further use. For freestanding membranes, the as-prepared suspension was readily filtered with cellulose filter membrane. Notably, the thickness of membrane can be precisely tuned by the volume of the suspension with a determined concentration. The accurate concentration of the Ti₃C₂T_x suspension was determined by vacuum filtrating a given volume of the Ti₃C₂T_x suspension and then weighing the filtered membranes.

2.3. Fabrication of SERS substrate

For SERS substrates, the as-prepared Ti₃C₂T_x membrane with a thickness of about 10 μm was punched into small circles with a diameter of 12 mm, followed by dropping noble metal aqueous solution on the membrane. In a typical fabrication, a droplet (10 μL) of HAuCl₄ solution with a concentration of 1.0 × 10^-3 mol L^-1, was dropped onto a Ti₃C₂T_x membrane by a pipettor and allowed to dryness ambient. The membrane prepared in such a way is denoted as Au@Ti₃C₂T_x. Similarly, by using AgNO₃ and H₂PtCl₆, Ag@Ti₃C₂T_x and Pt@Ti₃C₂T_x were prepared, respectively. Pd@Ti₃C₂T_x was fabricated by using H₂PdCl₄ solution prepared by mixing PdCl₂ and hydrochloric acid. Methylene blue (MB) was used as probe molecule in the SERS evaluations. For the evaluations, 5 μL of ethanol solution of MB (1.0 × 10^-5 mol L^-1) was dropped on the Au@Ti₃C₂T_x, Ag@Ti₃C₂T_x, Pt@Ti₃C₂T_x or Pd@Ti₃C₂T_x and dried at room temperature. The unpolarized Raman spectra were randomly collected from the Au@Ti₃C₂T_x or Pd@Ti₃C₂T_x, which were labelled with MB molecules, on a LabRAM HR800 Raman spectroscope (Jobin Yvon, France) equipped with an air-cooled CCD array detector in the backscattering configuration. For the Raman measurements, a 50X long-working-distance objective with numerical aperture of 0.50 was used. A He-Ne laser (632.8 nm) was used and the laser power was kept below 4 mW on the sample surface to avoid laser-induced heating. A grating of 600 lines per mm was used (acquisition time, 30 s).

2.4. Experimental determination of work function for Ti3C2Tx freestanding membrane

Kelvin probe force microscopy (KPFM, Bruker Dimension Icon) was used to detect the work functions of samples. Two types of probe were applied to the measurements. One is SCM-PIT with a PtIr coating tip and the other is magnetic etched silicon probe (MESP) with a CoCr coating tip. Only the “potential” channel frames of the KPFM results were useful for the work function calculation, which presented the work function difference between tips and samples. The average work functions of the tips were first obtained by measuring the standard sample of HOPG, so that the work functions of the samples can be obtained by the “potential” frames. The measurements were carried out under ambient conditions at room temperature. In order to eliminate the measurement uncertainty, the KPFM scan process executed on five areas with different tips. It is critical to control the contact force between the tips and the samples to ensure that the tips were at the lowest consumption throughout the entire scan process and the consistency of the relative potential difference.

2.5. In-situ XRD studies and surface/interface chemistry analysis

In-situ XRD patterns were recorded on a diffractometer (Bruker D8 discover, Germany) operating at 40 kV and 0.5 mA using CoKα radiation, equipped with a collimator with a diameter size of 1 mm, a Vantec-500 detector and with a 2θ step of 0.05°. Nova Nano SEM 430 and HITACHI SU-70 were used for morphological characterization. Chemical compositions and oxidation state of the samples were further analyzed using high resolution X-ray photoelectron spectroscopy (XPS, ESCALAB250, Thermo VG) with monochromated AlKα radiation (1486.6 eV). Binding energies were referenced to the C 1s of carbon contamination, which was set at 284.6 eV and the peak deconvolution was carried out using commercially available software, CasaXPS. A contact angle meter (OCA 15Pro, DataPhysics Instruments GmbH, Germany) was employed to characterize the surface modification of the samples. The specimens for transmission electron microscope (TEM) observations were prepared on an FEI-Scios Dual-Beam™ system (FEI, Brno, Czech Republic). For the TEM specimen preparation, Pt deposition was applied to protect the surface of the desired observation area with the electron beam to minimize the damage on the Au, followed by Ga⁺ ion-beam Pt deposition. A 30 kV Ga⁺ ion beam was used for focused ion beam (FIB) sectioning with the final milling performed using a 5 kV Ga⁺ ion beam to minimize ion-beam damage. The microstructural characterizations were performed using a TEM (FEI, Talos™, F200X, USA) equipped with energy dispersive spectroscopy (EDS) in the scanning transmission electron microscopy system.

2.6. DFT calculations

Theoretical calculations were performed within the framework of density functional theory (DFT) [31,32], and carried out in the Vienna Ab-initio Simulation Package (VASP) [33]. The pseudopotentials were established by the projector-augmented wave (PAW) [34] method and the Perdew-Burke-Ernzerh (PBE) [35] of exchange-correlation function. A 500 eV cutoff energy was used for the plane-wave basis set. The vacuum separation between the adjacent MXene layers was set to 15 Å to avoid any interaction due to the use of periodic boundary conditions. The geometry optimizations were performed by using the conjugated gradient method, and the convergence threshold was set to be 10^-6 eV/atom in energy and 10^-3 eV/Å in force. The Monkhorst-Pack scheme with 11 k× 11 k× 1 k points meshes was used for integration in the irreducible Brillouin zone so that the individual spacing was less than 0.05 Å^-1.

3. Results and discussion

3.1. Electronic structure of termination-functionalized MXene

We start with theoretical analysis on the electronic structure of MXenes with various terminations before moving on to experimental endeavor. For simplicity, we use mono-termination on each type of MXene. By systematically calculating and analyzing the density of states (DOS) in termination-functionalized MXene using density functional theory (DFT), we demonstrate that the identity of termination as functional group sensitively affects the electronic structure of MXenes. The surface M 3d orbitals hybridize with X 2p (X = C, O) orbitals with the upper bands showing M character while the lower bands are of more X character (Fig. 1), while the involving of the functional groups moves the Fermi level (E_F) closer to the computed X 2p band center, thereby giving a reduced energy gap between the metal 3d and X 2p band centers. No matter what the identity of termination, the electronic states near E_F are of M character in the MXenes investigated in this work. As calculated, the electronic work functions of terminated MXenes strongly depend on the identity of termination. Specifically, the work functions of Ti₃C₂(OH)₂, Ti₃C₂O(OH), Ti₃C₂F₂ and Ti₃C₂O₂ are 2.0, 3.0, 4.79 and 6.57 eV, respectively, following the order: W_OH < W_F < W_O (Fig. S1 in Supporting Information).

View original graphic| Download| PPT slide

Fig. 1.   Schematic of electronic structure for termination-functionalized Ti₃C₂T_x MXenes. All the bands are aligned by the vacuum energy level. The black dash lines indicate the Fermi level (E_F), showing the boundary of occupied states and unoccupied electronic states. The shaded areas are filled with electrons in the solid. As calculated theoretically, the work functions of Ti₃C₂(OH)₂, Ti₃C₂O(OH), Ti₃C₂F₂ and Ti₃C₂O₂ are 2.0, 3.0, 4.79 and 6.57 eV, respectively. The work function of the free-standing Ti₃C₂T_x membrane is experimentally determined to be 4.36-4.47 eV, as marked by the red asterisk in the right panel. The energy of free electron is 4.5 eV on the hydrogen scale.

It is generally accepted that the valance electrons in the vicinity of E_F of compounds relate to participate in oxidation-reduction (redox) reactions. Due to the remarkable difference in work function, the relative energy of E_F with respect to standard hydrogen electrode (SHE) differs a lot. With their ultralow work function, it is noteworthy that the E_F of OH-terminated MXene lies well above the redox energy of H⁺/H₂ couple. As a result, when the OH-terminated MXenes like Ti₃C₂(OH)₂ and Ti₃C₂O(OH) contact with an acid solution, the electrons of the MXenes will transfer to the H⁺ in the solution, thereby having the tendency to emit molecule H₂. In contrast, Ti₃C₂F₂ cannot devote electrons to the H₂ emission, as its E_F is lower than the redox energy of H⁺/H₂ couple. Similar to Ti₃C₂F₂, the O-terminated Ti₃C₂O₂ cannot contribute electrons to H₂ emission, either. As demonstrated in the schematic of electronic structure shown in Fig. 1, Ti₃C₂(OH)₂, Ti₃C₂O(OH) and Ti₃C₂F₂ except Ti₃C₂O₂ all can donate electrons to noble metal cations like Au³⁺ ions. The three former MXene species having non-bonding Ti states electrochemically offer a band for devoting extra electrons and thereby reduction capability for the MXenes, without the risk of structural destabilization. This is unlike Ti₃C₂O₂ wherein non-bonding Ti states are absent. Further increase in the oxidation state of Ti likely results in structural reconstruction. To summarize, MXenes with appropriate terminations are theoretically capable of reducing noble metal ions into metals.

3.2. Experimental determination of work function for free-standing Ti₃C₂T_x membrane

Keep in mind that the above calculations definitely provide very useful information on termination-dependent electronic structure and work function, while the models used are mostly based on the mono-terminated species that are hypothetical. In the actual MXenes experimentally obtained, the termination condition is complex in forms of composition and ratio. Hence, it is necessary to experimentally determine them and work function specifically. By means of Kelvin probe force microscopy, the work function of the free-standing Ti₃C₂T_x membrane is experimentally determined to be 4.36 ± 0.027 or 4.47 ± 0.014 eV depending on the probes used PtIr coating tip or CoCr coating tip (Fig. S2).

3.3. Real-time monitoring of noble metal solution droplet on MXene

On the basis of the above DFT calculations and the experimental determination of work function, MXene is capable of devoting electron to Au³⁺, being reduced into metallic Au. To examine whether this is the case or not, a droplet of HAuCl4 solution was dropped on a Ti₃C₂T_x membrane and immediately subjected to operando X-ray diffraction (XRD) analysis. Fig. 2 presents the XRD patterns collected in an in situ manner. By doing so, the real-time interaction between the solution and MXene can be monitored vividly. From the operando XRD, it can be seen that the characteristic Au (111) reflection is definitely recognized when the contact time is over 30 min, indicating the formation of crystalline Au after that time onset (Fig. S3). Notably, the intensities of the reflections corresponding to crystalline Au have been unchanged after 30 min, demonstrating that the conversion of Au³⁺ to Au had likely completed in the first 30 min.

View original graphic| Download| PPT slide

Fig. 2.   Evolution of HAuCl₄ solution droplet on Ti₃C₂T_x membrane. (a) optical photograph of droplet of HAuCl₄ solution just dropped on Ti₃C₂T_x membrane; (b) optical image of Ti₃C₂T_x membrane and droplets after 3 min of dropwise addition of HAuCl₄ solution. (c) optical image of Ti₃C₂T_x membrane and droplets after 15 min of dropwise addition of HAuCl₄ solution, (d) optical photograph of Au@Ti₃C₂T_x membrane, (e) the cross-sectional microstructure of Au anchored on Ti₃C₂T_x membrane, (f) contour plot of 2D XRD patterns for operando XRD to reveal the interaction between the solution droplet of 10 μL HAuCl₄ and the Ti₃C₂T_x membrane at room temperature. The positions of the Au (111) and Ti₃C₂T_x (004) reflection are marked, (g) contact angle and appearance of the droplet on Ti₃C₂T_x membrane in the course of time, (h) mass loss of HAuCl₄ solution droplet on Ti₃C₂T_x membrane as a function of time. Inset shows the volume evolution of the sphere-cap-like droplet with time. Note that the complete evaporation of the droplet to dryness at room temperature approximately requires 50 min.

To more directly unveil the conversion process, the evolution of a droplet on a Ti₃C₂T_x membrane was recorded by video (Fig. S4). Very interestingly, the appearance of the contact region of the membrane with the HAuCl₄ solution turned from grayish to goldish in a few minutes. Such significant change in appearance demonstrates that Au had formed at that time point. Compared with the operando XRD analysis, the straightforward optical recognition is more accurate in determining the beginning of the transformation. The reason for the delayed detection of crystalline Au may come from the interference of X-ray and the HAuCl₄ solution droplet. To figure out whether the broad XRD reflections originate from the interference or the reaction between the MXene membrane and the solution, we dropped HAuCl₄ solution on a Si single crystal that is inactive to the solution. As presented in Fig. S5, the broad XRD reflections resulted from the interference of X-ray and the HAuCl₄ solution droplet. Contact angle measurements also support this argument. As shown in Fig. 2, the droplet on the membrane is sphere cap like, and the height of the sphere cap decreases with time. At the time point of 30 min, the droplet height is small enough for X-ray penetration.

According to the online time-dependent mass change of the droplet (Fig. 2), mass loss undergoes linearly and terminates at the time onset of 50 min. In combination of operando XRD analysis, online optical recognition and time-dependent weight change, it is reasonable to elucidate that the conversion of Au³⁺ in the droplet to crystalline Au occurred when the droplet was still in the form of solution. We must emphasize that the conversion in the state of solution is very important since the homogeneous nature of solution enables the conversion in a uniform feature, which is identified in the following morphological observation. Interestingly, the operando XRD analysis also demonstrates that the crystal structure of the Ti₃C₂T_x did not undergo observable change within the diffractometer’s limitation during the whole conversion process. Similarly, real-time monitoring of H₂PdCl₄ solution droplet on a Ti₃C₂T_x membrane was also investigated (Fig. S6). This experimental result is excellently in consistent with the theoretical calculations (Fig. 1). Since there are no external reductants were involved in the redox process [17,36,37], the membrane Ti₃C₂T_x must play a critical role as reductant, establishing a redox couple between chloroauric acid and MXene. The redox can be unambiguously verified by X-ray photoelectron spectroscopy (XPS).

3.4. In situ redox mechanism

As afore-discussed theoretically, Au³⁺, as electron acceptor, is able to accept electrons, being reduced to metallic Au. XPS spectrum definitely identifies the formation of crystalline Au upon dropping the HAuCl₄ solution on the Ti₃C₂T_x membrane (Fig. S7 and Fig. S8). Correspondingly, as electron donor, MXene should be oxidized, namely, the oxidation state of the transition metal should increase. To understand the change in the Ti₃C₂T_x MXene surfaces after dropping HAuCl₄ solution, the chemical states of the Ti₃C₂T_x membranes without Ar sputtering or with Ar sputtering were carefully investigated by XPS. The results are shown in Figs. 3, S8, and S9. As shown in Fig. 3(a), high-resolution XPS spectra of Ti 2p region of the pristine membrane reveal peaks that could be deconvoluted into components corresponding to Ti-C (MXene), C-Ti-OH and C-Ti-O peaks [38] (Table S1). The surface of the pristine membrane has no clear change in Ti 2p state after sputtering for 120 s, revealing the homogeneity of Ti 2p state in the thickness dimension (Table S2). Nevertheless, in the case of Au@Ti₃C₂T_x, the components of C-Ti-O increase remarkably compared with the pristine membrane. According to the previous work on chemical origin of Ti₃C₂ MXene, terminal Ti atom in Ti₃C₂ terminated by -O functional groups has higher oxidation state than that of Ti₃C₂ terminated by -OH or -F functional group, leaving the pristine Ti₃C₂T_x MXene with electronically unsaturated terminal Ti atoms. In other words, the pristine Ti₃C₂T_x MXene membrane is prone to oxidation by oxidants like HAuCl₄ solution in this work (Fig. 1). The random thermal motion of ions in the solution causes the noble metal ions to collide with the Ti₃C₂T_x membrane surface where a redox reaction takes place. Based on the above discussion, the redox between HAuCl₄ and Ti₃C₂T_x MXene occured in an in situ manner according to the following equation:

3Ti₃C₂O_x(OH)_yF_z + δHAuCl₄ → 3Ti₃C₂O_x+δ(OH)_y-δF_z + δAu + 4δHCl(1)

View original graphic| Download| PPT slide

Fig. 3.   Redox mechanism of HAuCl₄ solution with Ti₃C₂T_x membrane. (a) XPS spectra of Ti 2p collected on pristine Ti₃C₂T_x membrane, (b) XPS spectra of Ti 2p collected on Au@Ti₃C₂T_x. The sample was prepared by placing a droplet of chloroauric acid (1.0 × 10^-3 mol L^-1) on Ti₃C₂T_x MXene membrane at room temperature and allowed to dryness. Note that the C-Ti-O composition increases in Au@Ti₃C₂T_x. The XPS spectra were recorded without Ar sputtering or with Ar sputtering for 120 s; (c) schematics of the redox mechanism of HAuCl₄ solution with Ti₃C₂T_x MXene. It illustrates the redox reaction in which Au³⁺ as electron acceptor and MXene as donor, forming gold nanometal as chloroauric acid solution contacts MXene membrane.

Additionally, the redox reaction stops, when the work function of MXene is close to electrode potential of noble metal cations. Analogous to HAuCl₄, H₂PdCl₄ solution (1.0 × 10^-3 mol L^-1) is capable of forming crystalline Pd on Ti₃C₂T_x MXene (Fig. S8 and Fig. S9). The formation of crystalline Pd is also the result of redox process in which H₂PdCl₄ acts as oxidant while the MXene as reductant.

3.5. Structural analysis of noble metal nanoparticles on MXene

As already recognized by XRD analysis and XPS investigation, crystalline metallic Au was formed on the Ti₃C₂T_x membrane. We denote it as Au@Ti₃C₂T_x. It was morphologically investigated by scanning electron microscopy (SEM). Fig. 4 shows the typical SEM image of Au@Ti₃C₂T_x. It is seen that the Au is in the form of nanoparticles that are homogeneously distributed on the membrane in the scale of tens of micrometers. The Au nanoparticles are primarily nanostar with corners and complex sharp structure (Fig. S10), which may benefit the SERS effect as discussed in the following section. As definitely identified by energy dispersive spectroscopy (EDS) mapping, the nanoparticles are Au. When the concentration of HAuCl₄ is 1.0 × 10^-3 mol L^-1, the mean size of the Au nanoparticles is 124 nm. In larger scale throughout the droplet imprint, the nanoparticles are also homogeneous (Fig. S11). Very strikingly, the mean sizes of Au nanoparticles are readily tunable by a simple yet efficient route via concentration modifications. For example, the mean size decreased to 39 nm as the solution was diluted to 1.0 × 10^-4 mol L^-1 (Fig. 4). Interestingly, if we drop noble metal solution on the Ti₃C₂T_x MXene membrane several times, nucleation and lateral growth of noble metal on the already established noble metal particles fill inter-particle gap, and a noble metal film will be formed on the MXene membrane consequently (Fig. S12).

View original graphic| Download| PPT slide

Fig. 4.   Morphological characterization of Au@Ti₃C₂T_x. (a-d) Au nanoparticles is prepared by 1.0 × 10^-3 mol L^-1 HAuCl₄ solution droplet on Ti₃C₂T_x membrane; (e-f) Au@Ti₃C₂T_x prepared by 1.0 × 10^-4 mol L^-1 HAuCl₄ solution droplet on Ti₃C₂T_x membrane; (a) SEM image of Au@Ti₃C₂T_x (Inset is a high magnification of (a)), (b, c) EDS elemental mapping of (b) Au, and (c) Ti, (d) size distribution of Au nanoparticles. The statistic size of Au nanoparticles is 124 nm (e) SEM image of Au@Ti₃C₂T_x (Inset is a high magnification of (e)). EDS elemental mapping of (f) Au, and (g) Ti. (h) Size distribution of Au nanoparticles. The statistic size of Au nanoparticles is 39 nm.

Similarly, contact of H₂PdCl₄ solution droplet on Ti₃C₂T_x membrane also gives rise to Pd particles via redox, which is in consistent with the theoretical calculation on work function (Fig. 1). In the case of 1.0 × 10^-3 mol L^-1 H₂PdCl₄, the mean size of Pd nanoparticles distributes bimodally with two peaks at 102 nm and 306 nm (Fig. 5). The Pd nanoparticles are primarily spherical. A similar morphology has been previously reported [39]. When the concentration was diluted to 1.0 × 10^-4 mol L^-1, the mean size decreased to 57 nm (Fig. 5). Moreover, EDS elemental mapping clearly reveals that Pd homogeneously distributed on the Ti₃C₂T_x membrane, further demonstrating the uniform dispersion nature of Pd particles (Fig. S13).

View original graphic| Download| PPT slide

Fig. 5.   Morphological characterization of Pd@Ti₃C₂T_x. (a-d) Pd nanoparticles is prepared by 1 × 10^-3 mol L^-1 H₂PdCl₄ solution droplet on Ti₃C₂T_x membrane; (e-h) Pd@Ti₃C₂T_x is prepared by 1 × 10^-4 mol L^-1 H₂PdCl₄ solution droplet on Ti₃C₂T_x membrane. (a) SEM image of Pd@Ti₃C₂T_x. EDS elemental mapping of (b) Pd, and (c) Ti. (d) Size distribution of Pd nanoparticles. The Pd particles distribute bimodally with two peaks at 102 nm and 306 nm. (e) SEM image of Pd@Ti₃C₂T_x. EDS elemental mapping of (f) Pd, and (g) Ti. (h) Size distribution of Pd nanoparticles. The statistic size of Pd nanoparticles is 58 nm.

Since the redox occurred on the Ti₃C₂T_x membrane, an interesting question raised. The question is that the noble metal particles established only on the outmost surface of the membrane or not. To unveil this concern, the membranes were cross-sectioned by FIB, and then subjected to transmission electron microscopy (TEM) investigation. As demonstrated in Fig. S14, the particles are distributed on the top surface only, without penetration into the membrane interior. More strikingly, the particles are polycrystalline: each particle is an aggregation of many smaller particles, which are adhered strongly on the membrane’s outmost surface. The reason for establishing such interesting nanostructures is that the nanoparticles are formed via in situ redox in which the noble metal ions in the solution act as oxidant while the membrane as solid reductant. It is worth noting that the MXene membrane used in this study was obtained by vacuum filtration and then dried, and the membrane consisting of nano-thick MXene sheets was somewhat solution tight. This is because the flakes are stacked with hydrogen bonds, and the interlayer coupling in Ti₃C₂T_x MXene is not so weak [40], which prevents the solution dropped on the surface of membrane from penetration into the interior of the membrane. Since epitaxial growth of noble nanometals is initiated via a mechanism that involves an in situ redox reaction between noble metal solution and MXene, nanoparticles were only distributed on the outermost side of the membrane.

3.6. SERS effect evaluation

The fabricated Au@Ti₃C₂T_x and Pd@Ti₃C₂T_x were subjected to SERS evaluation by using methylene blue (MB) as the probe molecule. In the case of bare Ti₃C₂T_x MXene membrane that was drop-casted by 1.0 × 10^-5 mol L^-1 MB solution, the Raman bands of MB molecule are visible but the intensity is very low (Fig. 6(a) and Fig. S15). In stark contrast, when the same solution drop-casted on Au@Ti₃C₂T_x with an SERS enhancement factor of 10⁴, the Raman bands of MB molecule are well resolved and the intensity increases remarkably, demonstrating a good SERS effect on Au@Ti₃C₂T_x, as shown in Fig. 6(a). Pd@Ti₃C₂T_x with an SERS enhancement factor of 10³ exhibits similar SERS effect comparable to Au@Ti₃C₂T_x. Similarly, other noble metals including bimetallic nanostructure anchored on MXene, such as Ag@Ti₃C₂T_x, Pt@Ti₃C₂T_x, Ag@Au@Ti₃C₂T_x, and Ag@Pt@Ti₃C₂T_x also exhibit good SERS effect (Fig. S16). Generally, nanostructures with proper size, complex sharp structure or more edges and corners as well as short distance between nanostructures have higher SERS effect [41]. Au@Ti₃C₂T_x with complex sharp structure and corners exhibits superior SERS effect over its counterparts Ag@Ti₃C₂T_x and Pd@Ti₃C₂T_x substrates (Figs. 4, 5 and S17). Alloy nanostructure provides an effective strategy for enhancing the functionality of metal nanomaterials [41]. Nucleation of one noble metal will fill the gap between another noble metal or grow directly on another noble metal. They have distinctive characteristics giving rise to unique electromagnetic coupling effect that allows higher SERS effect. So those alloy substrates have shown superior SERS effect which are not attainable by their monometallic counterparts.

View original graphic| Download| PPT slide

Fig. 6.   SERS effect of Au@Ti₃C₂T_x and Pd@Ti₃C₂T_x. (a) Typical Raman spectra of MB recorded from Au@Ti₃C₂T_x and Ti₃C₂T_x. Inset shows the optical photograph of Au@Ti₃C₂T_x membrane. For Au@Ti₃C₂T_x, the droplet imprint on the membrane is the area where Raman spectra are collected. (b) SERS spectra of MB collected from an area of 10 μm × 10 μm on Au@Ti₃C₂T_x. (c) Raman intensity mapping of the band centered at 1621 cm^-1. (d) Typical Raman spectrum of MB recorded from Pd@Ti₃C₂T_x substrates. Inset shows the molecular structure of MB. (e) SERS spectra of MB collected from Pd@Ti₃C₂T_x. (f) Raman intensity mapping of the band centered at 1621 cm^-1.

For an SERS substrate, good point-to-point reproducibility of Raman signals is an indispensable requirement in practical application. To examine the point-to-point reproducibility of Au@Ti₃C₂T_x, over 100 spectra of MB molecules from an area of 10 μm × 10 μm with a step size of 1 μm were collected and the relative standard deviation (RSD) of these SERS intensities was also calculated (Fig. 7(a, b)). Relative standard deviations of the Raman band intensity at 1621 cm^-1 are 23% and 14% for Au@Ti₃C₂T_x and Pd@Ti₃C₂T_x, respectively. This indicates that signal uniformity for Au@Ti₃C₂T_x and Pd@Ti₃C₂T_x is not perfectly ideal because the noble metal particles and the inter-particle distance are not extremely homogeneous. We believe that the signal uniformity can be further improved if one can control the concentration of noble metal solution and make noble metal fill the gap between noble metal or grown directly on noble metal by adding noble metal solution again. As shown in Fig. 6(b), the Raman spectra have almost identical characteristics in terms of intensity, in consistent with the fact that in situ redox enables uniform gold metal nanostructures anchored on the Ti₃C₂T_x membrane. These results definitely demonstrate that the as-prepared Au@Ti₃C₂T_x exhibited high uniformity and excellent reproducibility over the entire surface as high-performance SERS substrates.

View original graphic| Download| PPT slide

Fig. 7.   Reproducibility and stability of SERS substrates. Raman intensity distribution of the band at 1621 cm^-1 for (a) Au@Ti₃C₂T_x and (b) Pd@Ti₃C₂T_x substrates. Storage time dependence of Raman intensity of MB at 1621 cm^-1 for (c) Au@Ti₃C₂T_x and (d) Pd@Ti₃C₂T_x substrates.

In practical application, apart from sensitivity and reproducibility, stability is an important criterion in high-performance substrates. It is generally acknowledged that adsorption of carboneous contamination in the environment on noble metal nanoparticles surface is detrimental to the SERS effect. To eliminate the negative influence from the carboneous contamination as much as possible, the SERS substrates prepared in this work were sealed in centrifuge tubes just after the stability measurements. To evaluate the stability of the Au@Ti₃C₂T_x substrate, the time-dependent SERS measurements were performed at room temperature. As demonstrated, even after storing for 6 months in ambient, the Raman intensities of the characteristic Raman bands do not decrease, demonstrating an excellent stability of the Au@Ti₃C₂T_x substrate (Fig. 7(c)). Such excellent long-term stability is presumably ascribed to the firmly anchored nature of the noble metal nanoparticles on the MXene membrane. Referring to the pH test paper popular in the laboratory, we call the membrane-like substrates fabricated in this work SERS test paper. In each SERS test paper, the amount of gold is as low as around 2 μg, making it inexpensive. The low cost yet high efficiency of the SERS test paper may lay the foundations emerging from laboratory to household.

In summary, we have identified that in situ redox occurred between noble metal solutions and MXene membrane at room temperature. In the redox, noble metal ions act as oxidant and MXene as reductant. Density functional theory calculations, work function determination, kinetic and spectroscopic studies demonstrate that transition metal Ti of MXene donates electrons to the noble metal cations. Even diluted solutions with concentrations as low as 1.0 × 10^-4 mol L^-1 enable the cations to reduce into metallic nanostructures, while MXene membrane as solid reductant allows firmly anchoring of the nanostructures owing to the fact that redox reaction takes place between the cations and the membrane thereon without invoking any external reductant. The unique feature of in situ redox facilitates uniform distribution of noble metal nanostructures firmly anchored on MXene membrane. Very recently, our another work demonstrated that by using the in situ redox strategy, the noble metal nanoparticles anchored on MXene particulates exhibit excellent electrocatalytic properties with long-term stability [42]. The implications and importance of this work extend far beyond the results shown herein by taking the Ti₃C₂T_x MXene as an example. To the best knowledge of the authors, there are over 22 MXenes available so far [43,44]. We currently have solid evidence for Ti₂CT_x, TiNbCT_x and Nb₂CT_x that enable the noble metal cations to reduce into metallic nanostructures. In addition to the MXene membrane fabricated by means of filtration, we have examined the feasibility of other membrane or film preparation methods. For example, dropping MXene suspension onto polyethylene terephthalate film and allowing the suspension dried at room temperature also gives rise to high-qulity MXene film that reduces noble metal cations into metallic nanostructures. The nanostructures fabricated in such a way also exhibit excellent SERS effect. In stark contrast to the conventional colloidal suspension that is very environment sensitive, the in situ redox strategy proposed in this work may lead to generating powerful, reproducible and SERS-active surfaces with excellent long-term stability, which forebodes well for low cost production scale up.

4. Conclusion

In situ redox has been demonstrated to be a novel approach to firmly anchoring noble metal nanoparticles on newly developed 2D MXene. As demonstrated by density functional theory calculations, work function determination, kinetic and spectroscopic studies, the formation of noble nanometals is initiated via a mechanism that involves an in situ redox reaction. In the redox, MXenes as solid reductants that energetically match with noble metal cations spontaneously donates electrons to noble metal cations, reducing the cations into nanoscale metallic metals and anchoring them on the outmost surface of MXenes. No external reductants are used during the whole fabrication process, which addresses a long-standing issue of external reductant interference in the conventional chemical reduction synthesis. We also have shown very specifically excellent SERS effect using the constructed nanostructures as the SERS substrate. Although many future works are required to verify the uniqueness of the nanostructures fabricated via in situ redox in other fields such as electrocatalysis and heterogeneous catalysis, this is an exciting new direction for the development of high-quality nano-sized heterostructures with long-term stability.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No.51972310), the Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences (CAS), and the Youth Innovation Promotion Association, CAS (No.2011152), and the Special Program for Applied Research on Super Computation of the NSFC-Guangdong Joint Fund (the second phase) (No. U1501501). We express our sincere thanks to Zhihui Li and Zhonghai Ji from Institute of Metal Research, Chinese Academy of Sciences for their kind help.

Appendix A. Supplementary data

Supplementary material related to this article can be found, inthe online version, at doi: https://doi.org/10.1016/j.jmst.2019.09.013.