Ideas Previas

stage). In the following measurements we haven’t fixed this issue yet.

For recording an approach curve, necessary for proving true near-field character, the mov- ing mirror is placed in the white-light position (all wavelengths interfere constructively) of the interferometer. This position can be found easily by monitoring the interferogram and de- creasing the moving distance of the mirror until the interferogram shows only the white-light position (max. signal strength).

5.3

Results and discussions

Recording a continuous near-field spectrum in one single measurement represents the next log- ical step forward concerning s-SNOM. The possibility of recording an interferogram presents an additional new source of analyzing the near-field tip-sample interaction. This is a unique feature currently only provided by our technique. In the following, we present first contin- uous near-field spectra in amplitude and phase taken on silicon carbide and gold samples. Approach curves confirm their near-field character. Finally, we discuss the interferograms and show their potential for further research.

5.3.1 Silicon carbide and gold near-field spectra in amplitude and phase Near-field spectra in amplitude and phase of silicon carbide (SiC) and gold (Au) are presented in Fig.5.5 for the demodulation order of n1 and n2. The n1 demodulated spectra, called s1,

Figure 5.5: Continuous near-field spectra in amplitude and phase for SiC (a,b) and Au (c,d) are dominated by the background as expected. Therefore the spectrum is similar to the so- called ”illumination spectrum” [100] which is the backscattered light of the tip and sample without demodulation. It is interesting that both s1 spectra are similar to each other, which

shouldn’t be as for the Au spectra no SiC Fresnel coefficient has to be multiplied to achieve the correct ”illumination spectrum”. A reason for the similarity could be the illumination onto the tip, if directly or over reflection on the sample2 _{[100]. At n}

2 the background is

2

64 5. Nanoscale mid-infrared near-field spectroscopy

much better suppressed, and we record near-field spectra in amplitude and phase similar as it has been measured with monochromatic infrared s-SNOM. In the case of Au (d) no spectral dependence of the near-field interaction is expected. For SiC (b) a sharp peak with a width of 20cm−1 _{at 930cm}−1 _{should arise [101], but we observe only the Restrahlenbande. There}

are two reasons why we haven’t detected the near-field resonance of SiC. First, the spectral resolution was probably a bit too bad and second, it is well-known that using a damaged tip will not show anymore the near-field resonance due contamination of the tip. A tip can be damaged within a few hours or even during some minutes, depending on the tapping conditions, cantilever and sample properties3_{. Nevertheless we have observed first evidence} of the SiC near-field resonance as shown in Fig.5.6where a peak at the expected wavelength appears, though the peak is to small probably due missing spectral resolution. In Fig.5.7we

Figure 5.6: First evidence of the SiC near-field resonance

compare in (a) directly the obtained s2 signals, and in (b) we have divided the s2 SiC signal

through the s2 Au one as well as the Au phase is subtracted from the SiC one. The obtained

curves follows the same shape compared to sequential s-SNOM measurements, supporting that we are measuring near-field signals [100].

The recorded signals are very weak but nevertheless we have found many indications that they represent true near-field spectra in amplitude and phase. The final proof for true near-field spectra is done through so-called approach curves which we present in the next subsection.

5.3.2 Approach curves

True near-field signals are currently approved through the analysis of so-called approach curves. Concerning broadband illumination this has not been shown yet. Thereby the sample is retracted from the tip, continuously recording the signal whereas the moving mirror is kept stable in the white-light position of the interferometer. As the near-field is confined to the surface of the probe, one should see an exponential decay of the intensity if the background is sufficient suppressed. If not, the measured signal (from the background) will periodically rise and fall again over a larger distance because of an interference effect [83]. Due the weakness of the broadband signal, approach curves are very difficult to record. The retraction had to last several seconds for a strong enough signal. Fig.5.8 shows the s1 and s2 near-field

approach curves for SiC. As expected for s1 the background is not suppressed leading to a rise

in the signal strength at a greater distance [83,102]. The s2 should contain already much less

3

5.3 Results and discussions 65

Figure 5.7: SiC and Au comparison, see text for details

Figure 5.8: Approach curves on SiC for s1 and s2. Background signal is still contained in s1.

background signal which is clearly observed as the signal does not rise anymore like s1. Hence,

the s2 spectrum is dominated by the near-field but not the background signal. It has been

shown with CO2 lasers that for mid-infrared near-field spectra one should record at least the

second demodulated order signal. In Fig.5.9we compare the s2approach curves taken on gold

by the broadband illumination (red) and an attenuated CO2 laser (black). The same decay

characteristic is observed which is an additional confirmation that the measured broadband approach curve results from the near-field. With these approach curves we have proven for the first time that continuous near-field spectra in amplitude and phase can be recorded. 5.3.3 ”Near-field” interferogram

With our technique we can obtain the sample’s properties not only by the near-field spectra in amplitude and phase but also through the recorded interferograms. This is a unique

66 5. Nanoscale mid-infrared near-field spectroscopy

Figure 5.9: Broadband approach curve on Au compared with an approach curve by a CO2

laser.

feature of our Fourier-transform s-SNOM. The potential of this analysis method has to be still explored. Following we will show some first aspects gained through the interferogram analysis. Fig.5.10 presents the interferogram taken at the surface of SiC (a) and 3µm above (b) at the demodulation order n1. In the approached case (a) a long interference part (A)

is visible which disappears in the retracted case (b) which consists only of background. We assume that this is the contribution rising from the near-field interaction. Furthermore, as the approach curves have already shown, a larger background dominates the interferogram.

Figure 5.10: Interferograms (n1) taken on SiC in the (a) approached case and (b) 3µm above

the sample’s surface.

5.3 Results and discussions 67

Figure 5.11: Corresponding spectra of Fig.5.10.

Fig.5.12 shows the subtraction of the approached interferogram minus the retracted one. Part A is again the near-field contribution and part B very probably still presents a larger background signal. This can be explained by the way of illumination onto the tip apex-sample region [100]. Another way of illumination can completely avoid part B [78] and only show Part A as in [78]. The near-field contribution can be used as well for confirming true near-field signals, alternatively to approach curves.

Figure 5.12: Resulting interferogram by subtraction of Fig.5.10 (a)-(b). A: near-field contri- bution, B: residual background

The prolonged response due to the phonon resonance in SiC of part A marks the high quality factor due the substantial energy stored in the cavity formed by the tip apex and the sample. This prolonged response is probably the first evidence of free induction decay (FID) in a classical near-field coupled system, namely a scatter close to a material surface [78]. Differently materials will have a different long FID as shown in Fig.5.13 Therefore we compared the interferograms (n1) of SiC and Au. The prolonged response of Au is more ”covered” by the

background which indicates that it is shorter as the one of SiC. This is better emphasized in the measurements in [78]. But going into the field of FID is beyond the scope of this chapter [78]. In Fig.5.13 we observe another interesting behavior. As we can see, the SiC interferogram (including the background) precedes the gold one, although these measurements were done in the same measurement series without changing the illuminating or recording conditions. The reason is not yet clear but as the background is shifted too one could think about a slightly different attraction strength between the gold-to-tip-apex and SiC-to-tip-apex system which

68 5. Nanoscale mid-infrared near-field spectroscopy

Figure 5.13: Comparison of the SiC and Au interferogram (n1).

would result in a slightly different path length.

On the basis of the interferograms, an additional source of information has been added to explore material properties and interactions on the nanoscale.