Direct Observation of Carbonate Chemisorption on Barium Titanate Surfaces by Tip‐Enhanced Raman Spectroscopy

Surface chemistry significantly influences the physicochemical and functional properties of nanoparticles, thus necessitating the investigation of surfaces, especially when subjected to chemisorption. Seeking a method that delivers both high‐resolution imaging and precise chemical information, tip‐enhanced Raman spectroscopy (TERS) emerges as a critical technique. By the combination of scanning probe microscopy with Raman spectroscopy, TERS effectively addresses the limitations of traditional optical microscopes in the study of nanoparticles. The results provide detailed nanoscale structural insights, shed light on surface chemistry, and enable the detection of Raman‐forbidden modes caused by the substantial electric field at the tip apex (≈3 × 108 V m−1). This electric field plays a crucial role in altering selection rules, thereby broadening the scope of TERS beyond merely detecting Raman modes. By combining topographical with TERS mapping, the presence of a carbonate monolayer on the barium titanate (BaTiO3) nanoparticle surface is unveiled – an observation elusive to common characterization techniques.


DOI: 10.1002/admi.202300993
various nanoscale properties of BaTiO 3 are still subject to investigation, particularly when it comes to the impact of surface adsorption on, e.g.catalytic properties. [3]9][10][11] Understanding the nature of these changes is crucial for the optimization of the performance of BaTiO 3 in its applications.
Common techniques for the investigation of surfaces at nanometer scale, such as scanning probe microscopy (SPM) and transmission electron microscopy (TEM) yield highresolution images but lack the ability to reveal the chemical structure. [12]Spectroscopies based on X-ray give information about the structural properties of samples, but their resolution is on the micrometer scale owing to the difficulties of focusing X-rays. [13]Nevertheless, X-ray photoelectron spectroscopy (XPS) pictures with a 100 nm resolution or better may be generated by zone-plate technology to concentrate the X-rays, but relatively complex equipment (e.g., synchrotron sources) and specialized conditions are necessary. [14,15]aman spectroscopy is a powerful, laboratory-scale tool for investigating chemical and structural properties of materials due to the wealth of data it provides through the vibrational spectrum.Modern Raman spectrometers exhibit exceptional spectral resolution, enabling the detection and precise tracking of even the smallest band shifts caused by molecule interactions and orientation changes. [16]Despite its versatility and non-destructive capabilities, Raman spectroscopy faces significant challenges when examining interfaces and nanostructures due to its intrinsically low scattering cross-section as well as the diffraction limit of light. [17,18]This bottleneck has been successfully addressed through the use of surface plasmons to confine visible light to nanoscale regions, overcoming the aforementioned limitations and creating novel avenues for research. [19]ip-enhanced Raman spectroscopy, an aperture-less technique, is able to collect Raman spectra of surfaces with high spatial resolutions well below the diffraction limit of light. [20]TERS combines scanning probe microscopy and Raman spectroscopy to concentrate visible photons into nanometer-scale regions, inducing an electromagnetic near-field intensity and yielding a substantially larger scattering cross-section. [21]The plasmonic enhancement mechanisms in TERS include a localized surface plasmon resonance (LSPR), gap enhancement, and chemical enhancement. [22]SPR enhancement occurs when the surface electrons of a metal nanoparticle, such as the gold tips used in our study, are resonantly excited by incident light of a specific wavelength.This interaction leads to a collective oscillation of the conduction electrons in the plasmonic tip, generating a confined electromagnetic field known as a plasmon resonance.Consequently, this amplifies both the incident laser as well as the Raman scattered signal emitted by the sample positioned nearby, effectively enhancing the intensity of Raman signal at the nanoscale.Given the large spectral width of the plasmon resonance phenomenon in comparison to the Stokes shift of the Raman effect, it is often appropriate to consider the same amplification factor for the incident and backscatter light.
Gap enhancement occurs when the metal tip is within proximity (typically <5 nm) to a metal substrate.This proximity generates a stronger local electric field at the tip apex via a virtual image charge in the metal substrate, thus amplifying the Raman scattering signal from the sample through a virtual plasmonic dimer.LSPR generally prevails over gap enhancement in TERS, [23] but gap TERS has so far been considered a crucial contribution to push TERS to the best possible enhancement.
Chemical enhancement is another mechanism to enhance the Raman scattering signal in TERS and surface-enhanced Raman spectroscopy (SERS).It relies on direct contact between the tip and the sample through, e.g., contact mode feedback of the scanning probe microscope. [24]The plasmon resonance of metal nanoparticles depends on their size, shape, and composition.27] The near-field effect in TERS, confined to a few nanometers near the plasmonic tip by the electromagnetic field's limited range, makes TERS exceptionally sensitive to surface phenomena. [28]Consequently, TERS is becoming a versatile tool for the analysis of surface processes across multiple disciplines, including materials science, catalysis, and biotechnology.TERS has also been reported to possibly reduce the symmetry in samples, leading to a change of Raman selection rules through intense electric field and their gradients.Activation of additional modes for Raman scattering offers qualitative information on nanoscale surface features by capturing otherwise Ramaninactive modes. [29]Recent research revealed that in addition to the quantitative improvement of resolution and sensitivity, tipenhanced Raman spectroscopy thus has the potential to qualitatively widen the scope of investigation through the activation of conventionally Raman-forbidden modes and surface modes.These modes provide additional and possibly crucial insights into material structural distortions, symmetry breaking and reconstruction.TERS therefore continues its contributions to the advancement of surface science. [30,31]rium titanate at room temperature has a tetragonal perovskite structure with an a lattice parameter of 0.3994 nm and a c/a ratio close to 1.01 in a group P4mm where the {100} surfaces may be terminated with TiO 2 and BaO terminations, see Figure 1a. [32]At ≈130 °C, it undergoes a phase transition into a cubic, centrosymmetric phase, which is Raman-silent.
Cleaving a BaTiO 3 crystal along a certain direction exposes surface atoms with unsaturated bonds, leading to possible surface reconstruction, chemi-, and physisorption.Under ambient conditions, the chemisorption of CO 2 onto BaO-terminated terraces frequently occurs to minimize surface energy.Despite numerous theoretical studies [33,34] that corroborated this effect, direct experimental evidence detailing the formation and structure of the carbonate monolayer on the surface of BaTiO 3 is still in its infancy.The shortage of experimental findings, especially on individual particles and surfaces is mainly due to the limited sensitivity and resolution of conventional surface probing techniques.
In this work, we identify a carbonate monolayer on the surface of an individual barium titanate nanoparticle using highresolution TERS at a noticeable distance from the gold substrate, suggesting a small, possibly negligible contribution of the image charge.This study gives insight into the strength and nature of the chemical bond between the CO 2 molecule and the BaTiO 3 surface, paving the way to further investigate changes in the electronic structure and chemical reactivity of the BaTiO 3 surface due to the adsorption of CO 2 molecules.Furthermore, we report on the detection of previously non-reported Raman modes that we associate with the activation of infrared modes.Our calculations demonstrate that the electric field at the apex of the tip is indeed sufficiently strong to cause bond distortion and consequently, change the symmetry elements of the sample under investigation.This electric field therefore stimulates the appearance of novel modes in the Raman spectra, which exhibit a clear correlation in terms of both Stokes shift and intensity with the infrared modes.

Microstructural Analysis
Figure 2a illustrates a typical XRD pattern of the BaTiO 3 nanoparticles.The pattern matches the reference (JCPDS card #: 01-089-1428), which has a characteristic intense peak at 2 = 31.65°corresponding to the (101) diffraction plane, indicating a high crystallinity of the tetragonal phase. [35]However, a detectable peak associated with carbonate contamination is also observed at 2 = 24.1°.The emergence of this thermodynamically stable barium carbonate product arises from the remarkable surface reactivity of barium titanate nanoparticles with atmospheric CO 2 on BaO-terminated surfaces. [36]Barium titanate nanoparticles in the tetragonal phase are distinguished from the cubic counterpart following the splitting of peak ≈2 = 45°into (002) and (200) diffraction planes, wherein the crystal distortion leads to a larger c-axis in comparison to the a-axis. [37]In nanosized tetragonal particles, however, such peak splitting is usually obscured due to the Debye-Scherrer-induced peak broadening.
A representative TEM image of the powder is shown in Figure 2b.The nanoparticles are slightly agglomerated and composed of grains with low shape anisotropy.The average particle The visualization of the crystal structures in (a,b) was conducted using VESTA 3. [70] size is estimated from a TEM image of ≈100 particles shown as a histogram in the inset of Figure 2b, where the average particle size is 25 ± 5 nm.

TERS Measurement
The sample configuration is schematically illustrated in Figure 3a, showing BaTiO 3 nanoparticles spin-coated on Au/glass substrates.The thickness of the gold layer is ≈ 50 nm, strongly attenuating any far-field Raman signature of the substrate, which becomes negligible with the use of glass.Figure 3b displays the topography of the sample measured via tapping mode atomic force microscopy (AFM).The AFM topography image shows a relatively homogeneous distribution of the nanoparticles on the surface, with heights ranging from 10 to 50 nm, consistent with the size distribution extracted from TEM analysis.Due to the low Raman scattering cross-section and the small sample volume compared to the confocal volume, confocal Raman spectroscopy alone, i.e., spectral acquisition integrated over a diffraction-limited spot size of ≈1 μm2 for 1 s exposure time, does not provide a detectable Raman signature of either BaTiO 3 nor the carbonate contamination.
Figure 4 shows the topography scan and TERS map obtained in shear force mode utilizing Au tips.The topography scan (Figure 4a) unveils a nanoparticle of ≈12 nm height, with its profile depicted as a line scan in Figure 4b.It is noteworthy that the presence of a barium titanate nanoparticle, combined with a distance exceeding 10 nm from the gold surface, substantially reduces the efficiency of gap-mode TERS, i.e., the enhancement of the near-field Raman signal through a virtual image charge in the gold surface.Figure 4c displays a high-resolution TERS map in a false-color image, where the spectral emissions ≈540 cm −1 (corresponding to the BaTiO 3 TO 4 peak) are plotted as a function of position.This map clearly identifies the location of the barium titanate nanoparticle, exhibiting a pronounced correlation with the topography scan.
In order to compare TERS spectra between the substrate and nanoparticle, spectral averaging was performed on both regions to enhance the signal-to-noise ratio, within the specific regions marked by white broken lines in Figure 4c.Individual spectra are provided as Supporting Information (Figure S1, Supporting Information).The averaged results are presented in Figure 5, where the signal observed on the bare gold surface remains at the noise level, whereas a pronounced and well-distinguished spectrum emerges for the nanoparticle.By evaluating nanoparticle dimensions and comparing the near-field signals on both the nanoparticle and substrate, our findings indicate a substantial plasmonic enhancement at the apex of the Au tips.This enhancement results in a highly sensitive TERS signal, as previously discussed.
The amplification of the near-field spectrum can be quantified by calculating the enhancement factor (EF). [38] For TERS measurement on a sufficient thin sample, EF is defined as follows: [39] EF = where I NF corresponds to the intensity of the Raman signal captured within the near-field region, I FF represents the intensity of the Raman signal obtained under far-field conditions, A NF and A FF denote the areas associated with the near-field and far-field measurements, respectively.Comparison of areas instead of volumes takes into account that the relatively small thickness of the   barium titanate nanograin is the limiting factor for both the farand the near-field.The ratio between the near-field signal and the far-field with a value of 0.34 counts RMS yields an I NF /I FF ratio of 56.Considering the approximate diameter of the tip (≈10 nm) and the laser spot size (≈1.1 μm), the resultant enhancement factor amounts to 4.6 × 10 5 This corresponds to a local field enhancement at the apex of the tip of ≈26 under the assumption of the g 4 dependence, where g denotes the enhancement factor of the optical near field/far field.It is noteworthy to emphasize that the observed enhancement stems solely from LSPR and does not incorporate chemical enhancement as the experiments were conducted in shear-force mode at a distance of ≈1 nm.In comparison with existing literature, where enhancement factors typically range from 10 2 to 10 3 in similar experimental setups, [40][41][42][43][44] the enhancement factor we report appears remarkable and is almost certainly due to an exceptionally strong plasmonic amplification among a series of various gold tips tested.This emphasizes the impact of our approach, which now suggests the opportunity for a chemical and structural analysis with monolayer sensitivity and a high spatial resolution without requiring a specific substrate.
A comparative analysis of the acquired TERS spectra was performed in pursuit of a more comprehensive understanding.In this regard, the TERS spectra from a single nanoparticle were compared against confocal Raman spectra from a nanopowder.As depicted in Figure 6, a noticeable spectral resemblance between the confocal Raman and the near-field spectrum within the range of 100-800 cm −1 was observed.This region reflects characteristic Raman features of tetragonal BaTiO 3 .The distinctive Raman peaks of BaTiO 3 are comprehensively presented in Table 1.
Furthermore, comparing the carbonate peak located at ≈1060 cm −1 in both spectra reveals a notable enhancement of the carbonate monolayer surface.Additional peaks emerge in the TERS spectrum, located at 1260, 1449, and 1550 cm −1 .From classical theory, the interaction of an electric field with molecular bonds triggers the generation of a Raman signal, with its strength being directly related to the electric field and gradient of polarizability.Interestingly, there is a way of generating an additional Raman signal, termed the gradient field Raman (GFR) signal.With a substantial gradient in the electric field, it becomes possible to alter the selection rules.Consequently, this may lead to the activation of previously Raman-silent modes, e.g., IR-active modes.
Ayars and Hallen pioneered the investigation into electric field gradients in the early 2000s. [48]The near-field gradient effect in TERS has garnered significant attention within the physics community.[51][52] For the dipole moment, we have [48] where  describes the polarizability tensor, Q denotes the vibration coordinate, (a, b, c) refer to the spatial coordinates (x, y, z), μ p describes the permanent dipole moment and A is a tensor that determines the dipole induced by a field gradient and the quadrupole induced by a uniform field. [53,54]In proximity to a metallic surface, the jellium model proposed by Feibelman [55] suggests that the electric field experiences almost its entire amplitude change within a mere 0.2 nm distance.Consequently, a significant electric field may induce the GFR term, leading to the emergence of new vibrational modes in the Raman spectrum.To gain a deeper understanding of this phenomenon, we quantify the strength of the electric field.According to theory, the intensity of the electric field is given by: [56] E = where P stands for the laser power, A corresponds to the nearfield area, c and ϵ 0 represent the speed of light and dielectric permittivity in vacuum, respectively.With a laser power of ≈1.5 mW and considering near-field dimensions, the resulting electric field reaches ≈3 × 10 8 V m −1 .This substantial strength potentially induces distortions in atomic vibrations and impacts molecular polarizability.However, the high electric field is primarily localized at the surface of the nanoparticle, and its intensity diminishes rapidly with distance according to: where r represents the tip radius and d signifies the distance from the sample. [57,58]This rapid attenuation results in a pronounced electric field gradient at the tip apex.Our calculation provides strong evidence that the pronounced electric field gradient possesses the capacity to activate new modes from the gradient field Raman (term as defined in Equation 2. Consequently, when conventional IR-active modes are affected by changes in polarizability due to the GFR effect, they become more likely to manifest as Raman-active peaks.The preference for IR modes arises from the interplay between electric field gradients, molecular dipole moments, and polarizability, which collectively enhance the probability of observing IR-active vibrations as Raman peaks in TERS spectra.The IR modes of BaTiO 3 nanoparticles from different references are detailed in Table 2, and are consistent with our experimental results in the near-field spectrum.For instance, the peak at 1263 cm −1 corresponds to the bending vibration of oxygen atoms in the BaTiO 3 lattice, belonging to the A 1 mode symmetry group that traditionally renders the lattice "Raman inactive". [59]he 1449 and 1550 cm −1 peaks belong to the E symmetry group and culminate from the vibrations involving in-and out-of-phase stretching of oxygen atoms, respectively. [60]he interaction between the intense electric field gradients and the crystal lattice vibrations disrupts the symmetry constraints that once rendered these modes Raman inactive.This augmented interaction effectively bridges the gap between traditionally inactive IR modes and their visibility through the Raman scattering process, leading to the emergence of well-defined peaks within the TERS spectrum.The experimental achievements emphasize the extraordinary sensitivity of TERS to localized electric field gradient.Such notable advancement not only enhances our comprehension of material vibrational dynamics but also amplifies TERS' analytical capabilities, which in turn opens avenues for additional chemical and structural information from the sample, as in the case of the activation of infrared modes in near-field Raman spectroscopy.At this point, we would like to mention the possibility that these modes are a consequence of an organic contamination that gets amplified by the TERS setup.Such contaminations have commonly been reported on gold surfaces, however, in our case, we do not see these modes on the bare gold surface and it would require a high binding selectivity to a relatively passive barium titanate surface to explain the results presented in Figure 7.

High Spatial Resolution TERS Maps of Carbonate and Infrared Modes
Figure 7 displays the TERS maps over the Raman mode at 1060 cm −1 as well as new modes emerging from GFR effects on BaTiO 3 nanoparticles.The maps reveal the presence of carbonate at the location of the BaTiO 3 nanoparticle, which indicates chemisorption through a surface reaction with atmospheric CO 2 .While theory predicts the predominant binding of CO 2 to the BaO-terminated surface, the interaction between the Ti 3+ ions, oxygen vacancies, and CO 3 2− ions has also been suggested to facilitate the segregation of BaCO 3 as these defects can potentially act as nucleation sites. [62]The carbonate ion, CO 3 2− bonded to Ba 2+ has the D3h symmetry [63] with four (4) main asymmetric peaks between 900 and 1700 cm −1 (see Figure 7e).Although, the characteristic Raman peak of CO 3 corresponding to the symmetric stretching vibrations reported here is located at ≈1060 cm −1 , this band position differs across the literature. [60,64,65]While the variation is typically ±6 cm −1 , such a difference in Raman shift is presumed to be caused by the interaction of moisture, i.e., H 2 O with BaTiO 3 surface prior to the formation of the carbonate.The bending in-and out-of-plane vibrations in CO 3 2− are mostly IR active and appear at Raman frequencies lower than 1000 cm −1 . [63,66]The peaks labeled b, c and d in Figure 7e are both IR and Raman active and denote the asymmetric stretching vibrations of the carbonate.The carbonyl group designated by a carbon-oxygen double bond, C═O has stretching modes in the carboxylate salt, COO─Ba at 1263 cm −1 . [59]Generally, the carboxylate group is often negatively charged due to the donation of an electron pair from the oxygen to the carbon atom, which results in a resonance structure that stabilizes the charge distribution.Another characteristic Raman band of BaCO 3 structure is reflected by the asymmetric peak at 1449 cm −1 , which is deconvoluted into two peaks namely 1415 ± 9 and 1449 ± 3 cm −1 .Although both are related to asymmetric vibrations of C═O, the former occurs in the bulk BaCO 3, whereas the latter occurs on the surface of the carbonate ion. [64]Their intensities are typically weak and depend on temperature. [59]The spectrum shows another prominent band at 1560 ± 1 cm −1 with a shoulder at the higher frequency side (≈1610 ± 2 cm −1 ), both of which are characteristics of asymmetric stretching vibrations of CO 3 2− .The presence of these IR-active modes is an indication that the CO 3 2− ion is antisymmetric.The undistorted carbonate ion is Raman active and has a fundamental signature at 1060 ± 5 cm −1 due to the symmetrical trigonal planar structure of the three oxygen atoms with a carbon atom.When distorted (i.e., either by charge distribution, defects etc.) they become less symmetric and this triggers the infrared activity. [64]The maps not only reveal the active Raman peaks but also extend to active infrared peaks-a new dimension of analysis enabling the detection of molecular species that elude the traditional Raman spectroscopy techniques.

Conclusion
Tip-enhanced Raman spectroscopy images of BaTiO 3 nanograins on a polycrystalline gold surface were obtained with a very pronounced enhancement of the tip, as we estimate contributions from gap-TERS to be relatively weak and as we may entirely rule out chemical enhancement.A few-tens-of-atoms-thick BaTiO 3 nanoparticles were synthesized by a microwave-assisted hydrothermal technique and crystalized in the tetragonal phase with a size distribution with sizes in the range of 10-50 nm as confirmed by TEM.TERS provides monolayer sensitivity and a spatial resolution of ≈6.6 nm based on the Nyquist theorem.Its ability to detect Raman-forbidden and infrared (IR)-active modes made it possible to estimate a chemisorbed layer of BaCO 3 , a common impurity originating from chemisorbed atmospheric CO 2 .Our results demonstrate that TERS may activate non-active IR modes in the Raman response, thereby aiding the investigation of chemisorbed species on nanoscale surfaces and indicating distortions of chemisorbed molecules.The synergy between plasmonic enhancement and localized electric fields at the tip's apex resulted in a nanoscale spatial resolution with an electric field enhancement factor at the apex of the tip close to 26.This implies that chemical modifications at the surface can be probed, thus opening avenues for applications in molecular surface chemistry in heterogeneous catalysis, semiconductor-based technology, and corrosion.

Experimental Section
Sample Preparation: A 50 nm-thin sputtered gold film on a glass cover slip was utilized as the substrate.In order to reduce the surface roughness, the Au films were annealed at 500 °C for 1 min under an ambient atmosphere.The BaTiO 3 nanoparticles were hydrothermally synthesized from a homogeneous mixture of Ti(OBu) 4 and Ba(OH) 2 •8H 2 O solutions.Briefly, 50 mm of Ti(OBu) 4 was dropped and stirred in 5 mL of ethanol.Thereafter, 75 mm of the barium salt was dissolved in preheated distilled water (>90 °C) and then gently added to the initial solution in the presence of H 2 O 2 (2 mL, 32% by volume) under constant stirring.H 2 O 2 serves as a scavenger for residual hydroxyl groups that may deteriorate the electronic properties of barium titanate.A predetermined quantity of polyethylene glycol, PEG-8000 as surfactant was further added to the homogenously stirred solution before transferring it to a 23 mL Teflon container.This Teflon container was then inserted into a polymer autoclave from Parr Instrument and then placed in a Panasonic Inverter microwave oven (2.45 GHz).
After microwave exposure (conditions: 360 W, 4 min), the reactor was allowed to cool down to room temperature and the resultant product was thoroughly washed in preheated distilled water and ethanol before drying under ambient atmosphere in an oven at 80 °C for 15 h.Subsequently, the dried BaTiO 3 nanoparticles were ultrasonically dispersed in ethanol (0.5 gL −1 ) for 2 h prior to deposition onto the Au/glass substrates by spincoating at 3000 rpm for 3 min.It is emphasized that both polyethylene glycol as well as PEG have a Raman signature that could possibly interfere with the optical near-field experiments.Therefore, after the rinsing in preheated distilled water and sonification in ethanol, far-field experiments on these nanopowders were conducted that are reported in Figure 3a of ref. [68] in which any evidence of residual contaminations were not found, while the experiments were sufficiently sensitive to detect the presence of carbonate groups due to the large surface-to-volume ratio of the nanopowders.
Plasmonic Tip Fabrication: The plasmonic tips were produced by electrochemical etching of thermally pre-treated Au wires (Goodfellow, >99.99% purity) with a diameter of 100 μm.The Au wire, acting as the anode, was submerged ≈1 mm into a 37% HCl solution (Sigma-Aldrich) and was connected to a waveform generator to apply a pulse sequence.In parallel, a gold ring with a diameter of ≈20 mm was also immersed into the solution as the cathode.Voltage pulses of +4.5 V, 30 μs with a +0.5 V offset bias are applied to the tip with a frequency of 3 kHz, resulting in a duty cycle of 10% resulting in a smooth surface with typical tip radii ≈10 nm, as depicted in scanning electron microscopy (SEM) images in Figure 3 of ref. [68].The comprehensive procedure was reported elsewhere. [69]ip-Enhanced Raman Spectroscopy Setup: TERS measurements were conducted under ambient conditions using an AIST-NT Omega Scope 1000 scanning probe, a NanoFinder 30 Raman spectrometer, and a thermoelectrically cooled Andor charge coupled device (CCD) detector.The Raman spectrometer was confocally combined with the atomic force microscope (SmartSPM, AIST-NT), and laser excitation was achieved by a linearly polarized helium-neon laser (632.8 nm, TEM00).An optical neutral density filter of ND 0.5 results in an effective power of 1.5 mW on the sample.A Mitutoyo MPlan Apo 100× objective with a numerical aperture of 0.7 (0.7 NA) was implemented to focus on the tip via side access with an incident angle of 65°.The acquisition time of the spectral map was 1 second per pixel.
Characterization: The crystallographic structure of the powder was confirmed via X-ray diffraction (XRD).The XRD measurements were performed using a Bruker D8 Advance diffractometer equipped with a Cu K  source ( = 1.5406Å) between 20°≤ 2 ≤ 70°.Complimentary bright-field TEM images of the powder were acquired using a JEOL JEM-2100F transmission electron microscope.For this purpose, 2 mg of the powder was ultrasonically dispersed in ethanol prior to dip-coating into Lacey carboncoated Cu grids (LC400-Cu).

Figure 1 .
Figure 1.a) crystal structure of BaTiO 3 highlighting the oxygen octahedra b) layered structure of BaTiO 3 (100) planes with BaO and TiO 2 vicinal surfaces.The visualization of the crystal structures in (a,b) was conducted using VESTA 3.[70]

Figure 2 .
Figure 2. The XRD pattern a), and TEM image b) of the as-synthesized BaTiO 3 nanoparticles." " denotes peak due to carbonate impurity.Inset of (b) is the histogram showing the particle size distribution.

Figure 3 .
Figure 3. a) Schematic illustration of the sample configuration, depicting the distribution of BaTiO 3 nanoparticles onto Au-coated glass substrates.b) 1 μm × 1 μm tapping-mode AFM topography image showing the BaTiO 3 nanoparticles on the polycrystalline gold surface.

Figure 4 .
Figure 4. a) Topography scan over a 200 nm × 200 nm region with 60 × 60 pixels revealing an individual barium titanate particle.b) Profile indicates the size of the particle as measured along the dotted white line in (a).c) TERS map simultaneously acquired in the same area as in (a) revealing a strong signal in the 480-540 cm −1 spectral range at the location of the nanoparticle with a 3.3 nm step-size.The white dotted squares serve for a spatial integration of the Raman spectrum as displayed in Figure 5.

Figure 5 .
Figure 5.Comparison of the near-field spectrum on the nanoparticle (top) and near-field spectrum on the gold surface ≈50 nm away from the BaTiO 3 grain (bottom).The near-field signature on the BaTiO 3 grains is roughly one order of magnitude above the noise floor; red dots are data points, and the black line guides the eye.

Figure 6 .
Figure 6.Comparison of the near-field spectrum from an individual BaTiO 3 nanoparticle and the confocal Raman spectrum of bulk BaTiO 3 .

2 −Figure 7 .
Figure 7. TERS map of a BaTiO 3 nanoparticle on gold in a 200 nm × 200 nm region, captured by a 60 × 60 pixels array.The signal intensity was integrated across distinct spectral ranges: a) carbonate peak, b) 1260 cm −1 , c) 1449 cm −1 , and d) 1550 cm −1 .The color bars indicate intensity levels, bounded by the minimum and maximum values with 0% representing the dark-count of the CCD.e) Cumulative Raman spectrum of the 200 nm × 200 nm region.

Table 2 .
Infrared modes observed on BaTiO3 nanoparticles with carbonate surface contamination.