Near-field infrared microscopy of nanometer-sized nickel clusters inside single-walled carbon nanotubes

Nickel nanoclusters grown inside single-walled carbon nanotubes (SWCNT) were studied by infrared scattering-type scanning near-field optical microscopy (s-SNOM). The metal clusters give high local contrast enhancement in near-field phase maps caused by the excitation of free charge carriers. The experimental results are supported by calculations using the finite dipole model, approximating the clusters with elliptical nanoparticles. Compared to magnetic force microscopy, s-SNOM appears much more sensitive to detect metal clusters inside carbon nanotubes. We estimate that these clusters contain fewer than ≈700 Ni atoms.

One of the unique applications of carbon nanotubes is their use as nanocontainers for various encapsulated species. Nanoscale metal clusters present a special perspective in this regard as the tubes give both a natural constraint and an effective protection from the environment. Following the early description of the lling procedure, 1 a considerable variety of both single-and multiwalled nanotubes combined with several metals were produced (for a review, see ref. 2), and their structural, 3 magnetic 4 and superconducting 5,6 properties investigated. Metallocenes in single-walled carbon nanotubes (SWCNTs) 7 represent a special class among these hybrid systems, as they constitute nanoreactors for both metal cluster formation 8 and inner nanotube growth. 9 In a recent study, 10 upon annealing nickelocene encapsulated in SWCNTs, superparamagnetic nickel clusters were formed that are considered as high performance single domain magnets with high coercivity. Here we measure such clusters by scattering-type near-eld optical microscopy (s-SNOM) to both probe the metallicity of such small nanostructures and to establish the sensitivity of the method.
Details of encapsulation of Ni(II) acetylacetonate in e-Dips single walled carbon nanotubes with tube diameters of 1.7 AE 0.1 nm are given in ref. 10. The encapsulated molecules were transformed to nickel clusters by annealing. The size of these clusters can be controlled by the annealing temperature: in order to get fewer but well separated, long clusters we heated the sample in vacuum at 700 C for 2 hours. Previous results using similar conditions 10 showed the formation of nickel clusters with aspect ratio ranging approximately from 1 to 15. Near-eld microscopy was performed on samples deposited on a silicon substrate by vacuum ltration. 11 In order to follow and control the annealing process we veried the disappearance of Ni(II) acetylacetonate molecules using attenuated total reection (ATR) spectroscopy in the midinfrared (MIR) region (Fig. S1 †). The disappearance of the acetylacetonate-related peaks indicates the successful decomposition of the molecules and the possible formation of Ni clusters.
We also observed nickel clusters being created inside the nanotubes via transmission electron microscopy (TEM). The nanotubes were dispersed in toluene, sonicated for 1 hour and collected onto a TEM grid with ultrathin carbon lm, then images were taken using both a JEOL 3010 and an FEI THEMIS microscope. A typical image is shown in Fig. S2. † The clusters look spatially separated enough to enable the measurement of the optical response via s-SNOM where the possible resolution limit is around 20 nm.
We applied scattering-type near-eld optical microscopy (s-SNOM) to image nano-sized nickel atom clusters inside single walled carbon nanotubes based on their infrared optical response with spatial resolution well beyond the diffraction limit. The s-SNOM setup (Neaspec GmbH) is based on an atomic force microscope with a metal-coated tip illuminated from the side by a focused laser beam (in our case, the source was an infrared (980 cm À1 ) quantum cascade laser (QCL)). The illuminated metal tip acts as an optical antenna 12 and enhances the electric eld under the tip as depicted in Fig. 1. The extension of this well-localized, high amplitude electric eld depends on the tip apex radius. 13 As this nano-sized light probe is scanned in the proximity to the surface, an optical interaction occurs between the probe and the sample.
This interaction results in propagating waves via scattering from the volume below the probing tip. The amplitude and the phase of the scattered wave are determined by the local optical properties of the sample. 14, 15 The very weak near-eld scattered light is then demodulated at the higher harmonics of the tip oscillation frequency and further analyzed using pseudoheterodyne detection 16 based on a Michelson-type interferometer, shown in Fig. S3. † This complex setup enables the simultaneous measurement of the sample topography and both the amplitude and the phase of the near-eld scattered light. The s-SNOM microscopy yields very high wavelength-independent spatial resolution (z20 nm) and high optical response that gives the opportunity to study nanostructures consisting of only a few hundred atoms.
Magnetic force microscopy (MFM) was also performed with a separate AFM instrument (Bruker Dimension Icon) using a standard Bruker MESP magnetic AFM tip.
In the applied mid-infrared spectral region the near-eld signal originates from the excitation of free charge carriers. In common metals, such excitations result in a frequencyindependent high phase contrast compared to the silicon substrate. Although carbon nanotubes show near-eld contrast themselves, 17,18 their conductivity is negligible compared to real metals like nickel, therefore we do not expect observable contribution from the nanotube walls.
In order to predict the near-eld contrast we performed calculations based on the extended nite dipole model (EFDM). 19,20 In the infrared region where the wavelength is much longer than the characteristic size of the tip-particlesubstrate system, the scattering problem can be approximated as an electrostatic problem at each time step (Rayleigh scattering). The system is modeled as depicted in Fig. 2. The tip is approximated as a prolate spheroid, the nanotube as an innite long cylinder. The nickel clusters are also cylinderlike objects as previous studies 10 suggested. In order to t them to the analytical model we replaced them with prolate spheroids because their polarizability is very close to that of a cylinder.
Neglecting the effect of the carbon nanotube (see above), the main part of the near-eld interaction occurs between the probing tip, the nickel cluster and the substrate. The electric eld is considered to be perpendicular to the surface of the substrate as the tip enhances the electric eld component that is parallel to its axis of revolution. In such a model we can take into account the electric eld response of the nickel cluster with several dipoles generated inside the nickel nanoparticle by the tip and the mirror charges of the substrate. Those dipoles are described by the local electric eld and the polarizability of the object (p ¼ aE loc ). The polarizability of prolate spheroids, perpendicular to their semi-major axis is given by the formula: 21 Here 3 is the dielectric permittivity of the nanoparticle at the wavelength of interest, R is the radius, and L is the length of the elliptical nanoparticle, and N t is the depolarization factor. The latter describes how much the internal eld within the spheroid is attenuated by the polarization and it depends on the geometry of the object. This can be expressed as Fig. 1 Illustration of the near-field scattering process and the illuminated probing tip.
where e ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 À R L=2 s is the eccentricity of the spheroid. The diameter is chosen to be D ¼ 1.3 nm to t inside a nanotube with diameter 1.7 nm. The dielectric permittivity of nickel was determined from the Drude model with a size-dependent term for the damping constant: 22 where v F is the Fermi velocity and C is a factor that depends on the electron scattering process inside the particle. The Drude model parameters for nickel were taken from ref. 22. These parameters were used in the EFDM model to calculate the 3rd harmonic demodulated near-eld phase contrast of nickel nanoparticles compared to the silicon substrate versus the aspect ratio of the nanoparticle (L/D). The result is presented in Fig. 3. Fig. 3 illustrates that the near-eld phase contrast does not vary noticeably with the aspect ratio: the near-eld contrast changes only by D4 O3 ¼ 0.001 rad until it starts to saturate when the aspect ratio is around four. This amount of change cannot be detected under the present experimental conditions.
Results on as-prepared Ni(II) acetylacetonate-lled nanotubes transferred onto silicon are shown in Fig. 4. We were looking for nanotube bundles with as small diameter as possible. We expect no phase signal from such bundles as Ni(II) acetylacetonate molecular vibrations are too weak to provide an observable near-eld signal. We found that nanotube bundles smaller than 10 nm have no contrast on the near-eld optical maps. The gure presents the AFM topography and the O3 near-eld phase map of the sample with multiple carbon nanotube bundles. The lack of a near-eld signal veries our predictions.
Next, we applied the annealing process (700 C, 2 h) to this sample to create nickel clusters 10 and repeated the optical characterization of the nanotubes. During the process the surface morphology of the sample changed substantially and it was not possible to nd the same nanotube. Therefore, we searched for nanotubes bundles with identical diameter in the two images. We chose the bundle in the middle of Fig. 4 with a height of 3 nm as standard. Since the diameter of an individual e-Dips nanotube is 1.7 nm, such a bundle would consist of three nanotubes. In Fig. 5, we show the AFM topography and the third harmonic phase signal of a bundle of similar size.
The gure shows a typical nanotube bundle in the middle with diameter of 3 nm. Since the diameter of an individual e-Dips nanotube is d z 1.7 nm, bundles with d ¼ 3 nm probably consist of three nanotubes.
As the optical image demonstrates, bright, high contrast spots alternate along the nanotubes, with regions where no optical signal is found. These high contrast spots correspond to the transmission electron microscopy images which showed inhomogeneous spatial distribution of nickel clusters along the nanotube bundles. Our measurements show remarkable agreement with the analytical model. Phase contrast values of the brightest spots are 4 O3 ¼ 0.139 AE 0.01 rad, to be compared with a calculated value of 4 z 0.142 rad. Other examples of phase maps can be seen in Fig. 6, where the location of the nickel clusters can be easily identied. We also found very few nanotubes with no contrast, indicating that the lling of most tubes was sufficient to obtain nickel clusters. The contrast values vary within the sample; smaller values possibly correspond to smaller diameter clusters that do not ll the nanotube perfectly, larger contrast, in turn, could indicate multiple clusters measured together at the bottom of the tip. We also managed to do sequential mapping at different wavenumbers. From these images we extracted the average phase value of the clusters in a representative nanotube at every measured wavenumber. Fig. S4 † shows a near-eld phase spectrum acquired this way, together with that calculated from the EFDM. The high, frequency-independent phase value through the measured spectral range indicates metallic behavior.  We also tried to locate the nickel clusters by magnetic force microscopy (MFM). This technique was already applied to investigate nanoparticles with magnetic behavior. [23][24][25] As the excellent magnetic properties of these nickel clusters were already demonstrated, 10 we expected to obtain the signal of nickel clusters in the MFM phase images as magnetic dipoles give bright and dark spots at their opposite poles. Fig. 7 displays the AFM topography, the MFM and s-SNOM measurements on the same nanotube bundle. We repeated the MFM measurements with different tip li height (15, 25, 35, 50, 70, 130 nm) until the topographic related phase appeared. We did not nd any sign of nickel clusters with MFM probing, indicating the higher sensitivity of the optical method.
If we assume that the aspect ratio of a nickel cluster is four (this was the saturation limit in the calculated near-eld phase) and we treat it like a cylinder shaped cluster of face centered cubic (f.c.c.) structured nickel, we can roughly estimate the number of atoms measured in one spot. We used the lattice constant of f.c.c. nickel, 0.35 nm (ref. 26) and calculated how many cubes can t in the above mentioned cylinder. This and O3 near-field map (right) of a typical carbon nanotube bundle after the annealing process. Optical images were taken with a n ¼ 980 cm À1 illuminating laser. Fig. 6 O3 optical phase maps of two different carbon nanotube bundles. Images were taken with a n ¼ 980 cm À1 illuminating laser. The presence of nickel clusters is very obvious as they cause high phase contrast.
This journal is © The Royal Society of Chemistry 2019 RSC Adv., 2019, 9, 34120-34124 | 34123 Paper number was then multiplied by four because the unit cell of an f.c.c. crystal contains 4 atoms, giving the number of atoms to be around 644. We nd that near-eld optical probing gives reliable information about the location of nickel clusters inside carbon nanotubes even in the case of such a small amount of material.
In summary, we observed nickel nanoclusters grown inside single walled carbon nanotubes via near-eld microscopy based on their infrared optical properties. We found that these measurements are very sensitive to the presence of the metallic phase. With a tip-dened spatial resolution of 25 nm, we were able to detect optical signals from objects of a few nm in size, containing less than 700 atoms. Our modied EFDM model gives phase contrasts close to the measured values and is found to reliably predict the optical signal of nanoparticles. We also detected nickel clusters in most of the nanotubes, consistent with electron microscopy results which indicate the good lling ratio of the nanotubes.

Conflicts of interest
There are no conicts to declare.