The advent of near-field optical microscopy [1,2] allows to overcome the lateral resolution limit of the classical optical microscopes imposed by far-field diffraction phenomena. This limit is given by the well-known Rayleigh-Abbe criterion. The principle of SNOM consists in the detection of the evanescent components of the light scattered by a nanostructure. These evanescent components of the electromagnetic field contain information on the high spatial frequencies of the sample surface. The SNOM is a complementary optical tool to other Scanning Probe Microscopes such as AFM and Scanning Tunneling Microscope (STM) which can only provide topographic and mechanical and electronic information respectively. A particular advantage of SNOM is to allow a various and powerful means of local analysis such as fluorescence near-field imaging and spectroscopy which are highly sensitive and selective techniques to determine the physical and chemical properties of surfaces. SNOM microscopy in fluorescence mode has been demonstrated for the first time by Harootunian [3]. This result was confirmed a few years later in fluorescence imaging and Single Molecule Detection (SMD) experiments by several groups [4,5]. Until now, only aperture probes, generally metal coated fibre tips, were used in these experiments.

Among the large variety of SNOM configurations, the apertureless scheme [6–10] has proved high lateral resolution capabilities. This system can use both commercial probes [6,9,10] and easy-fabricated metallic tips [7,8]. These AFM probes act as local scattering centres of the sample’s optical near-field. Aperture probes present less advantages than apertureless tips because they have a limited lateral resolution imposed by the finite skin depth of the metal. They also provide a weak throughput intensity owing to the existence of a cut-off frequency for the fundamental mode of a metallic waveguide.

Earlier spectroscopic experiments have been reported [11] with an apertureless SNOM where a new form of optical spectroscopy, called scattering spectroscopy, was demonstrated. It consists in measuring the local scattering from the interaction between a silicon tip and a sample as a function of the excitation wavelength. However, the technique using interferometry prevents from selecting light at shifted wavelengths, as needed for fluorescence spectroscopy and imaging. Fluorescence imaging with an apertureless SNOM has been recently performed using one photon [12] or two and three photon excitations [13]. The results displayed showed that the tip field enhancement effect can lead to a good rejection of the background fluorescence stemming from the global illumination [12], and that together with non linear photon absorption true confinement of the excitation can be achieved [13,14]. However, in all these samples studied, the fluorescence detected is associated to topographic features. Other apertureless experiments have been performed [15], showing the fluorescence near-field detection through an approach curve, or through the simple comparison of near-field and far-field spectra [16].

In this letter, we report near-field fluorescence imaging of dots of colour centres with an apertureless SNOM, which has been developed for spectroscopy applications [17]. Illumination in p polarized light together with a vertical modulation of the tip are used to extract the near-field fluorescence light from the far-field background. Near-field fluorescence microscopy is the sole technique enabling a characterization of the quality of the dots fabrication process. To our knowledge, the fluorescence of submicrometric colour centres dots has never been reported in near-field experiments. Colour centres are particularly interesting test samples for SNOM detection because no topography (no AFM signal) can be correlated with. Moreover, we have recently investigated colour centres in LiF layer embedded in one dimensional optical microcavities [18,19]. As a first step of a two-dimensional (2D) confinement, it is proposed to investigate luminescence properties of periodical arrays of submicrometric coloured areas.

The investigated sample consists in a LiF film containing colour centres, mostly F₂ and ${{\mathrm{F}}_3}^{+}$ (two electrons bound to two and three anion vacancies respectively) produced by low energy electron irradiation on its surface. F₂ centres in LiF are well known laser-active defects which exhibit high quantum efficiency, high optical gain and a large tunability over their wide emission band in the visible. Their main absorption occurs at around 450nm as a broad band due to phonon interactions with the lattice [20]. Resonant excitation of F₂ centres with the argon laser line at 458nm produces an intense red photoluminescence peaked at 665nm, as a result of strongly allowed vibronic transition.

We have fabricated lattices of coloured zones using a Scanning Electron Microscope equipped with a lithography apparatus which allows control of the electron beam deflexion. Several lattices of coloured dots were made on the surface of a 2,4µm thick polycrystalline LiF film, prepared by thermal evaporation [21]. LiF powder was heated at ~800° C in a Tantalum crucible at a pressure of 10^-6 mbar on a glass substrate maintained at 250° C during the deposition, whose rate was ~2 nm/s. Using the e-beam lithography (EBL) system described above, 100nm diameter dots of colour centres were written on the LiF layer with 2keV electron energy, a 100pA current and a dose of 450µC/cm². The dots were periodically separated by a distance of 1µm, large enough to allow the fluorescence detection of a single dot.

Our experimental system (fig 1) consists in a combination of an AFM and an apertureless SNOM. Advantages of our apertureless set-up are its versatility and flexibility which allows to implement various illumination and collection modes [10]. Our AFM is a standard commercial CP model from Thermomicroscopes. The detection of the forces acting on the cantilever is performed by a position sensitive photodetector (PSPD) associated to a laser diode (wavelength is 670 nm). We use silicon cantilevers with a conical tip [22] whose curvature radius is 40 nm. This tip is located at the very extremity of the AFM cantilever and has an high aspect ratio (cone angle is less than 20°). Thus, it allows a high numerical aperture detection of the waves scattered by the tip apex. The displacement of the sample under the tip is ensured by x-y-z piezotranslators.

 

FIG. 1. SNOM experimental setup

Download Full Size | PDF

The AFM operates in the intermittent contact mode to control the probe-to-sample distance and to detect the signal coming from the end of the probe by the modulation of the optical near-field. The resonance frequency of the AFM cantilever is close to its resonance frequency of 220 kHz. The vertical vibration amplitude is 25 nm. The optical signal is detected by a lock in amplifier tuned at the vibration cantilever frequency. Thus, mainly the signal which is at the cantilever resonance frequency is detected. With this method, the signal to background ratio is improved since the detection of the far field is reduced. Indeed, the vertical modulation of the tip acts as a high pass filter for the spatial frequencies of the electromagnetic field [23,24]. The photoexcitation of the F₂ colour centres was provided by the Argon laser (λ=458 nm) whose beam is launched into a single mode optical fibre. This fibre is connected to a beam expander to deliver a collimated beam. It is finally focused on the sample below the extremity of the tip (diameter of the spot is 10 µm) through a microscope objective (x10, N.A.=0.28) resulting in far field illumination conditions. The beam is p polarized so as to enable a concentration of the electromagnetic field under the tip [13].

The AFM tip is positioned above the area of the sample where the colour centres were created. When the probe is close to the illuminated surface of the sample, it scatters the evanescent waves of the near field. The resulting radiated waves are reflected on a prism and detected in transmission mode by a photomultiplier tube (PMT) through a microscope objective (x10, N.A.=0.28) focused under the tip in a dark field detection direction. To improve the signal to background ratio we use a spatial filter placed in front of the PMT, so that a circular area of 10 µm of diameter is selected around the extremity of the tip. In addition, this spatial filter enables to remove most of the light coming from the AFM laser diode (whose focused spot is slightly displaced from the end of the lever wher the tip is). A rejecting and a broad band filters are placed behind the objective to cut off the wavelengths of the residual light of the AFM laser diode and the illuminating argon laser respectively. The rejecting filter has an Optical Density (O.D.) of 6, centered at λ=680 nm with a full width at half maximum of 20 nm. Below λ=665 nm transmittivity is 80%. The broadband filter is O.D. 6 at λ=458 nm and transmittivity above λ=500 nm is 80%.

Fig. 2 shows an AFM image (2.5 µm×2.5 µm) of the sample surface. Fig. 3 is the simultaneously obtained fluorescence near field image. The AFM measured a low RMS residual roughness of 7.2 nm. On the SNOM image nine coloured dots are clearly identified with a good signal to noise ratio and high contrast. We stress that sample topography does not affect the optical image [25].

 

Fig. 2. AFM image of the surface of a 2,4 µm thick LiF film thermally evaporated on a glass substrate

Download Full Size | PDF

 

Fig. 3. SNOM image of the surface of a 2,4 µm thick LiF film thermally evaporated on a glass substrate

Download Full Size | PDF

As mentioned above, the vertical modulation improves the detection of high spatial frequencies of the sample. However, the body of the tip along with the end of the cantilever can provide a fair amount of far-field modulated signal. To check this we performed two tests: first, the spot of the incident beam is not centred on the tip but on the cantilever (avoiding carefully a tip illumination); secondly, the tip is removed a few µm from the surface (at a distance where only far-field is modulated). In both cases, the dots are not resolved. This ensures that the SNOM image is not stemming from the fluorescence far field modulation by the cantilever or the tip body but is due to modulation of the fluorescence evanescent waves scattered by the extremity of the tip.

The measured lattice spacing is approximately 1 µm with a good periodicity along the X and Y directions (see figure 3) while the size of the dots is greater than the 100 nm expected value. Indeed, the lateral resolution in the optical image is 240 nm. It is estimated by taking the Fourier transform of the naer field data and defining the cut-off frequency where the noise is overpassing the exponential decrease of the amplitude of the spectrum [25]. The resolution is lower than one might expect considering the tip diameter. First, the main point in our SNOM image is that the resolution is purely optical as it is not related to the path followed by the tip displayed on the topography image [26]. Secondly, resolution is both related to the tip and sample features. The fabrication process of the dots on the polycristallin LiF thin film does not lead to a step-like variation of the centres concentration but rather a gradient-like one due to diffusion of electrons after the implantation. As theoretical resolution is dependent on the tip sample interaction and on the transfer function of the tip vertical modulation, no simple assessment of its value can be done. However, the sharper the tip, the closer the dots profile will reflect the concentration gradient.

Tip sample interaction has been discussed in previous apertureless fluorescence experiments with metal [13,15] and silicon tips [12,16]. As all the authors exploit a concentration of the electric field to favour molecular excitations under the tip, this effect is in competition with non radiative energy transfers from the excited molecules to the tip. This quenching effect is strongly dependent on the tip material and the emission wavelength. Metals with which stronger enhancements of the excitation field are expected [13] also will quench fluorescence more strongly than silicon, as the imaginary part of the dielectric permittivity is higher. A closer experiment to ours is that of Hamann [12] who used silicon tips to observe fluorescence from dye doped polystyrene spheres: the near-field approach curves showed no evidence of quenching effects. In contrast to all these previous works, what we detect is the signal output of a lock-in amplifier at the vibration frequency of the tip. Consequently, what we measure is the time Fourier transform of a signal related to the tip sample interactions during a full vertical motion of 50 nm peak to peak [23,24]. The choice of this vibration amplitude will enhance the detection of evanescent waves with a penetration depth comparable to this value. It will thus further reduces the influence of short range effects such as quenching (few nanometers) in the detected signal. Quenching, if present, could be observed with a much lower amplitude or by demodulating the signal at frequency harmonics [23, 24, 27, 28].

A deformation of the spots is observed along a 45° direction relative to the X axis in the optical image. No periodical structures are present in the AFM image which could be related to the fluorescence SNOM image, neither such asymmetry is observable in the AFM image, thereby ruling out an asymmetry of the tip. The deformation probably corresponds to dynamic effects induced during the dot writing. They could be related to electromagnetic oscillations in the SEM lenses that slightly affect the beam positioning. This should be avoided by lowering the beam current.

We have reported the realisation of coloured dot as small as 100 nm, produced by EBL on LiF films. The fluorescence of the colour centres has been observed with an apertureless SNOM with a high contrast and good resolution. These results show that colour centres can be used as test samples for artefact free SNOM experiments, both aperture and apertureless. Improving in the EBL procedure which implies to lower electron beam current and doses, should bring to a wide range of structures based on LiF films which could improve the understanding of the contrast mechanisms of SNOM. In this regard, spectroscopic measurements with aperture probes are currently under investigation.