Dielectric Contrast Imaging Using Apertureless Scanning Near-Field Optical Microscopy in the Reflection Mode

Transcription

1 Journal of the Korean Physical Society, Vol. 47, August 2005, pp. S140 S146 Dielectric Contrast Imaging Using Apertureless Scanning Near-Field Optical Microscopy in the Reflection Mode Debdulal Roy, S. H. Leong and M. E. Welland Nanoscience Centre, University of Cambridge, 11 J J Thomson Avenue, West Cambridge, Cambridge CB30FF, UK (Received 6 September 2004) A reflection-mode, apertureless (scattering type) scanning near-field optical microscope (S- SNOM) in reflection geometry with a tip focus monitoring facility has been developed and used for imaging dielectric contrast at the nanoscale in both direct-detection and heterodyne interferometric mode. The approach curves (plot of near-field signal with tip-sample separation) were obtained and these were used to predict extent of near-field signal collection. The quality of the images obtained by adopting the heterodyne interferometric technique has been compared with those obtained by direct detection. For the former, independent phase imaging was also possible and performed. PACS numbers: C, C Keywords: Near field, Scattering, Atomic force microscopy, Interferometry, Dielectric I. INTRODUCTION The concept of near-field microscopy is known since 1928 when Synge [1] proposed to extend microscopic resolution into the ultramicroscopic regime. However, it was not possible at that time to realize the technique for material characterization. It was only in 1984, with the advent of laser and other sophisticated instrumentation that Pohl et al. [2] first demonstrated the technique using visible light. The use of a fibre-probe aperture for delivering/collecting the light [3] limited the spatial resolution to about 50 nm. Recently, after Zenhausern et al. [4] achieved an optical imaging resolution of 1 nm by using apertureless scanning near-field optical microscopy(s-snom), much interest has again garnered into developing the technique and exploiting it for material characterization. It may be noted that apertureless SNOM is often termed as scattering-type SNOM or S-SNOM. In the newer apertureless technique the field enhancement at the end of a sharp dielectric tip is used as the source of light and the scattering point. The scattered light is collected by a far-field objective lens. S-SNOM techniques have seen realisation in various other applications including fluorescence [5], infrared [6] and Raman spectroscopy [7 9]. However, most of these applications involve transmission mode configurations, while only a few utilize reflection-mode geometry [10,11], with the same wavelength used for illumination and detection. The latter is plagued by the difficulties in providing a good and effective illumination and col- d.roy.98@cantab.net; Currently at National Physical Laboratory, Hampton Road, Teddington, Middlesex, UK TW110LW -S140- lection from the tip due to space constraints, as well as inherent difficulties in separation of the useful near field signal from the huge background light of the same optical frequency. In addition, interference in the form of slowly varying standing waves also degrades the quality of the near-field image [2]. II. THEORY Near-field scattering occurs when an atomic force microscope (AFM) tip is situated very close to the sample surface ( nm). The tip end is usually coated with gold, silver, platinum or some related alloy. When a laser is shone onto the tip end keeping the polarization of the light along the tip-axis, the electric field strength is increased many folds at the tip. This is similar to the antenna effect: the field enhancement at the tip end depends upon the sharpness of the tip and the dielectric characteristic of the tip end/coating. The tip end acts as a dipole in the oscillating electric field of the laser beam. When the tip end is brought very close to the sample surface it creates an image dipole at the sample. The interaction of the two dipoles with each other gives the near-field scattering signal which can be expressed in terms of an effective polarizability, α eff as has been modelled by Knoll et al. [12] in the following form: α = 4πa 3 (ε t 1) (1 + β) ε s = (ε t + 2) (1 β) E nf α(1 + β) α eff E i = 1 αβ 16π(a+z) (1)

2 Dielectric Contrast Imaging Using Apertureless Scanning Near-Field Debdulal Roy et al. -S141- Table 1. Table containing dielectric constants (ε = ε 1 + iε 2) of some materials used in the calculation. [21] Material ε 1 ε 2 Au Ag Pd Si C Cu Al W Fe Cr Ni SiO where E nf and E i denotes the near-field response and incident excitation respectively, ε t and ε s are the dielectric functions of the tip and the sample respectively, a is the radius of the tip end, z is the tip-sample separation. To extract the near field signal from the background, the tip is usually dithered slightly and a lock-in technique is applied [13,14]. Since the spurious background light intensity is also modulated at the same frequency of the cantilever oscillation it passes through the lock-in amplifier and results in artefacts or slowly varying standing wave interference patterns at low harmonics (n = 1) of the oscillation frequency. The use of higher harmonics (n 2) serve to suppress this [15,16], but at the expense of working with lower intensity of the signal. The weaker higher harmonic near field signal however can be further amplified when detected with heterodyne interferometry [17]. Evidence of near-field signal is usually in the form of repeatable near-field approach curves that show a sharp increase in signal as the tip-sample gap is reduced during approach. Without approach curves as opposed to signal subtraction techniques, it is difficult to determine whether a true near-field signal is actually obtained. As far as subtraction techniques are concerned, near-field and background signals add coherently. The optical near-field response has been previously modelled using scattering in dipolar approximation, described by the near-field interaction between the tip s dipole and a mirror dipole induced in the sample [12,14]. Using Equation 1, the near-field signal intensities for several elements used in this system can be calculated in order to predict the level of contrast expected in the SNOM images. The dielectric constants used in these calculations for a 633 nm wavelength are shown in Table1. The intensities at higher harmonics are obtained by differentiating α eff with z [10]. The third harmonic near-field contrast is shown in Figure 1. Metals such as gold, silver provide a higher near-field signal over the semiconductors such as Fig. 1. Third harmonic near field intensities (normalized against that of gold) obtained from Equation 1. The dielectric functions are given in Table 1. The radius of curvature of the tip end was 50 nm, tip sample separation was 40 nm and laser wavelength of 633 nm. Fig. 2. Theoretical approach curves (third harmonic) with gold tip on various materials obtained from the intensities at different tip sample separation calculated using Equation 1. The tip radius was 50 nm and laser wavelength was 633 nm. silicon and insulators such as glass. These predictions are wavelength dependent as the dielectric functions are also wavelength dependent. The near-field signal intensities at different tip-sample separation are plotted in Figure 2 to obtain the SNOM approach curves for a gold tip on various materials. All the curves in the near-field region show sharp increase of detected signal in the near field. It may be noted that the degree of enhancement is significantly dependent on the dielectric properties of the materials. For example, there is almost no enhancement from glass compared to that from gold. This difference brings in the contrast in a SNOM image and differentiates one material from another.

3 -S142- Journal of the Korean Physical Society, Vol. 47, August 2005 III. S-SNOM BY DIRECT DETECTION 1. The S-SNOM Set-up The apertureless SNOM (Figure 3) consists of a home built, non-contact mode Atomic Force Microscope (AFM) coupled with the illumination and collection optics for focusing light from a HeNe laser onto the tip end. Z-distance (tip-sample separation) regulation is based on optical cantilever deflection method. The SNOM illumination light path includes a Faraday isolator (I), a broadband beam splitter (BS), and an infinity corrected Olympus (50, 0.5 NA) long working distance objective lens (OL). Back-scattered light from the tip-sample junction is then diverted by the beam splitter onto a glass slide (GS) that transmits more than 90 % of the signal to the PMT. The remaining weakly deflected light is fed via a Tube Lens (TL) and projection lens (PL) to a CCD to allow for observation of the tip focus. Both TL and PL with OL combine to act as an infinity-corrected microscope with a magnification of about 500. The cantilever used is a gold-coated AFM tip from Micro-masch, with a resonant frequency of about 300 khz. S and PZT refer to the sample and quadrant piezoelectric tube scanner (PZT-5H, Staveley NDT) respectively. An additional far field optical microscope (not shown) views the sample and tip from the top and provides white light illumination as necessary. The entire AFM setup is mounted on a precision X-Y -Z stage (Ultra-align, Newport), to allow precise movements of the whole AFM and in particular, the tip, to the point of best focus under the objective lens, which is fixed. The optical axis of the objective lens is at 30 degrees to the sample plane. To use the S-SNOM, the AFM tip is first oscillated at near its resonant frequency of 300 khz. The ampli- tude of oscillation is about 40 nm. Before the sample is approached to the tip, it is necessary to perform an initial alignment of the AFM by means of the X-Y -Z stage to position the AFM tip directly in the focus of the SNOM objective. This is done through observing the diffraction patterns of the laser light from the AFM tip end. Subsequently, using a precision stepper motor driven translation stage, the sample is approached until the AFM feedback mechanism kicks into a place, which is about 90 95% of the initial free oscillation amplitude. When the sample is approached, our 500 optical microscope allows us to monitor the focus condition at the tip end and make fine adjustment to the focus as necessary. For our reflection-mode, oblique (30 degrees to the sample) tip illumination situation, without the novel 500 optical viewing facility, it would be difficult to realign the focus at the tip after it is approached and the stress on overall system stability becomes very critical. The tip focus condition when approached may be slightly changed from its initial unapproached condition. Added to that, drift with time would have adverse effects on the illumination and collection of the near-field signal. Figure 4 shows an optical image of the AFM tip and reflection of the cantilever underside as viewed from the 500 optical facility. 2. Near-field Signal Collection by Direct Detection As an initial experiment, it was necessary to show that a near-field signal can indeed be collected in such a direct detection scheme. Figure 5 shows the near-field approach curves for lock-in detection at the first, second and third harmonics of the tapping frequency. The sample used is a topographically flat gold layer deposited on a Si substrate. Such a sample ensures that the influence of topography induced z motion crosstalk in the directly detected optical signal is negligible and that the signal enhancement observed is of near-field origin. The obtained approach curves show typical near-field enhancement. For higher harmonics the region of near-field en- Fig. 3. Schematic of the set up for directly detected SNOM measurements. (LD: laser diode, PSD: position sensitive detector, HeNe: HeNe laser, I: Optical isolator, BE: beam expander, BS: beam splitter, OL: objective lens, S: sample, C: cantilever, PL: projection lens, TL: tube lens, GS: glass slide). Fig. 4. Image of the optical scattering from the tip end as viewed from 500 optical viewing facility.

4 Dielectric Contrast Imaging Using Apertureless Scanning Near-Field Debdulal Roy et al. -S143- Fig µm scan size. (a) AFM topography and directly detected SNOM images of gold, aluminium and silicon multilayer structures at (b) first harmonic (c) second harmonic lock in detection. Metallic regions (gold and aluminium) appears brighter than silicon in SNOM as predicted by theory. Bright features in topography (encircled) that do not appear in the SNOM images. The height of the structures were between 0 and 50 nm. Fig. 5. Near field approach curve with the signal obtained from first, second and third harmonic of the oscillation frequency using lock in detection. hancement reduces as evidence of increased near-field sensitivity. In the third harmonic case, the region of near field enhancement is in the 20 nm region, and the far field effects such as interference fringes are completely suppressed. The use of higher harmonics thus results in the improvement of spatial resolution [10]. Near-field imaging of a gold structure on silicon was next performed. Figure 6 shows the AFM topography (Figure 6(a)) and direct detection SNOM images (Figure 6(b)-(c)) of the said structures. The bright small topographical features, probably contamination particles, in the AFM images do not appear at all in the SNOM images while the contrast between gold and silicon surfaces are evident. In the second harmonic case, the edges of the gold structures are sharper and the contrast between metallic and silicon surfaces is more distinct. For large structures comparable to the wavelength of illumination, simulations using the generalized field propagator method [18] indicate that the optical field is no longer confined in the objects and strong field gradients follow the outline of the objects. This may be the reason why in the first harmonic image, bright contrast is seen beyond the metallic edges. In any case, lock-in at second harmonic helped to suppress these and sharper edge resolution was obtained. The faint interference lines present in the images as well as the lack of contrast between gold and silicon suggest that the detected signal contains a fair amount of far field components. As opposed to the 1.5 times theoretical predicted contrast of gold over aluminium, the lack of contrast could be the result of the similarity of far field (bulk) reflectance values (at 633 nm) of 0.96 and 0.92 between gold and aluminium respectively, and the far-field contribution is significant Fig µm scan size. (b) AFM (c) second harmonic SNOM image of chromium islands on gold surface. An approach curves is shown in (a). The height of the structures were 20 nm. in these images. Direct detection S-SNOM imaging was also performed for chromium islands on gold film. This combination of materials was chosen to test for the darker contrast of chromium on gold (brighter). Figure 7 shows the nearfield (2nd harmonic) approach curve, topography and 2nd harmonic SNOM image obtained for the sample. The differences between morphology and the SNOM image suggest that optical contrast is present although it is possible that there is z-motion influence. Dark contrast is seen in the SNOM images and attributed to chromium islands as predicted from theoretical dielectric contrast calculations (Figure 1). It is possible that the dark edges around some of the features are caused by the edge darkening effect [19] typical of broad tips. Although the approach curve shows signal increase in the near-field, the large decay length of the near-field curve suggests the detection of far field components. Detection at 2nd harmonic in this case was insufficient to extract a pure near field signal. An attempt at the 3rd harmonic resulted in no useable signal. The results from using a direct detection S-SNOM showed that it was difficult to obtain a largely near field signal. In comparison, the latter could be more easily achieved using a heterodyne interferometric approach first described in Ref. [17]. A similar setup was also implemented and results from its use reported here.

5 -S144- Journal of the Korean Physical Society, Vol. 47, August 2005 IV. S-SNOM BY HETERODYNE DETECTION 1. The Heterodyne Detection Set up The schematic for the heterodyne detection set-up is shown in Figure 8. The collection optics in this set-up is similar to that used in direct detection geometry. On the beam delivery side, it is split into two parts using an acousto-optic modulator (AOM) where one part goes straight through the beam expander and beam splitter, and the other part is given a frequency shift of 60 MHz (this is termed as the reference beam, Iref). The angle of deflection of the reference beam is < 5. The reference beam goes through a variable zoom beam expander and is made upon reflection, co-linear with the collected beam. The difference in path travelled by the direct beam and the reference beam to the detector is kept as small as possible (within the coherent length of the laser). At the detector the scattered beam (I sig ) interferes with the reference beam and thus the signal intensity is amplified. This signal is detected by a Si-PIN diode. The lock-in technique is used to extract the near-field signal. 2. Signal Enhancement by Heterodyne Technique With the optics aligned properly and using a Si-PIN detector, a heterodyne signal gain of at least three orders of magnitude could be achieved. Even though the sensitivity of the Si-PIN diode is lesser than that of PMT in this wavelength-range, around 50 times more signal was detected over direct detection in the first harmonic due to the heterodyne gain. This signal disparity greatly increased with higher harmonics of detection, probably due to the rapid decrease in sensitivity of the PMT with frequency as opposed to the Si-PIN diode that had a bandwidth of 120 MHz. For example, with a cantilever oscillation of 40 nm, a third harmonic signal of 50 to 100 µv produced by the interferometer would correspond to less than 1 µv from direction detection. 3. Near-field Signal Collection by Heterodyne Detection Similar to the case of direct detection (section 3.2), a first step in testing the system was to observe any signal enhancement in the near-field. Figure 9(a), 9(d), 9(g) show the heterodyne interferometric SNOM approach curves obtained using a gold-coated tip on Cr. The approach curves indicate that the signal to noise ratio decreased with higher harmonics. At third harmonic it is about 3 : 1 compared to 10 : 1 at the second harmonic. This indicates why the less sensitive direct detection case failed to perform well at higher harmonics. Also in direct detection, large oscillation amplitudes were required for obtaining a good lock-in signal. However, the large-tip sample gap also meant that the third harmonic signal that is onset at very close distances to the sample could not be detected. A set of S-SNOM images of Ni clusters on Cr surface is shown in Figure 9. Figure 9(a), 9(d) and 9(g) show the approach curves obtained before measurements. In the first and second harmonic images (Figure 9(c), 9(f)), presence of interference fringes and weaker contrast between the Cr and Ni is evident. The influence of topography and far-field scattering phenomena is also significant. The effect of topography can be clearly seen in the first harmonic image (Figure 9(c)) where the Ni clusters appear darker than the Cr. This is possibly due to the movement of the probe along the approach curve where an overall increase in tip-sample distance results in a drop in the overall SNOM signal. For the first two harmonics, the approach curves did not fall off to zero even far beyond 100 nm from the sample surface. Clearly this explains the presence of far-field effects such as interference fringes in the corresponding SNOM images. In sharp comparison, the SNOM image obtained at third harmonic (Figure 9(i)) reveals significant material specific contrast between the Ni islands and the Cr substrate, and the approach curve also shows effective suppression of the far field signal beyond 80 nm of tip-sample separation. Fig. 8. Schematic of the heterodyne interferometric apertureless backscattering SNOM. HeNe: HeNe laser, I: optical isolator, AOM: acousto-optical modulator, M: mirror, BE: beam expander, BS: beam splitter, OL: objective lens, GS: glass slide, Si-Pin: Silicon pin diode, I sig and I ref refer to the signal beam and frequency shifted reference beam, respectively. 4. SNOM Intensity and Phase Imaging of Regular Periodic Structures Heterodyne interferometry allows the obtainment of both SNOM intensity and phase images simultaneously. This is not possible with the direct detection technique. Figure 10 shows the AFM and SNOM images of a regular pattern of gold nanostructures on a silicon substrate.

6 Dielectric Contrast Imaging Using Apertureless Scanning Near-Field Debdulal Roy et al. Fig µm scan size. (c) First harmonic ((f) second harmonic ((i) third harmonic heterodyne interferometric SNOM images of nickel island on chromium surface with their corresponding topography images ((b), (e) and (h) respectively) and approach curves ((a), (d) and (g) respectively). The height of the structures are between 0 and 40 nm. These structures were first written by electron beam lithography. Subsequently, a 40 nm thick gold layer was deposited and a further lift-off process created the final gold nanostructures (Figure 10(a), 10(d)). For the AFM and SNOM imaging the tip-sample gap was about 40 nm. Comparing Figure 10(b) and Figure 10(e), it is obvious that the higher harmonic detection was more able to detect the near-field signal over the background. This resulted in distinct optical features and good material contrast in Figure 10(e). In addition, the SNOM image was also free from far field effects and interference patterns that were visible in the second harmonic case. Again the background suppression of higher harmonic detection leading to good near field signal detection is clearly shown. Comparing the phase images (Figure 10(c) and Figure 10(f)) with the other images, for example, topography (Figure 10(a)) and SNOM intensity (Figure 10(c), 10(e)), no direct correlation was observed. In fact, the phase image in Figure 10(c) shows optical (dark) features located in the empty region between gold structures. Clearly these are of a pure optical origin and could originate from interference effects resulting from the regular pattern of structures acting like a grating. These effects were much suppressed at the higher 3rd harmonic. However, at this stage the interpretation of the phase images remains a difficult task. This situation is made worse by the temporal instability of phase measurement itself as well as phase drift. For good phase measurements, a small path length and absolutely stable set-up is essen- -S145- Fig µm scan size. First harmonic and second harmonic heterodyne interferometric SNOM intensity and phase images of regular gold nanostructures. First harmonic SNOM intensity image (b) with corresponding phase image (c) and topography image (a). Second harmonic SNOM intensity image (e) with corresponding phase image (f) and topography image (d).the height of these structures were 40 nm. tial. Even though there is phase drift in heterodyne detection, the SNOM intensity is not affected. This makes the SNOM intensity measurement more resilient to system drifts and instabilities that directly affect the phase but not the intensity. It also indicates strongly that with a stable set-up, the heterodyne technique can be used to separately and reliably measure phase. 5. Influence of Topography It is well known that the undesired movement of the probe in the z direction can cause spurious effects or artefacts in a SNOM image [20]. Even when there is no topography on a sample surface, by merely changing the AFM set-point or forcibly changing the tip-sample distance, a movement of the operating point along the SNOM approach curve occurs. Such movements directly affect the SNOM signal. It follows therefore that in general, any irregularities or instabilities in the probe sample distance regulation results in SNOM artefacts. This is particularly expected while probing large structures, and is less prominent with smaller structures. However, it can be improved by improving the regulation system of the AFM. In these measurements (e.g. Figure 9 and Figure 10), the artefacts were not severe. In the first harmonic and second harmonic measurements there were some far-field contribution to the signal that resulted in the decrease in quality of the images. For direct measurements this problem was more intense, and was significantly improved upon by suppressing the far-field contribution through heterodyne detection at higher harmonics.

7 -S146- Journal of the Korean Physical Society, Vol. 47, August 2005 V. CONCLUSIONS The development of an apertureless near-field optical microscope, both in direct detection and heterodyne detection, and dielectric contrast on various sample measurements using these set-ups are reported. It was found that the direct detection SNOM was more susceptible to far field effects and was unable to precisely resolve the dielectric contrast expected for various tip-sample configurations. By implementing the heterodyne interferometric approach, great improvement of the quality of the SNOM images in heterodyne detection was achieved with the concurrent suppression of far field effects at higher harmonics of detection and amplification of the near field signal through interferometry. Independent phase imaging using the heterodyne technique was also performed and interference (grating) effects from a regular patterned structure were observed in phase imaging. These interference effects in phase imaging could also be suppressed using a higher harmonic of detection. REFERENCES [1] E. H. Synge, Phil. Mag. 6, 1928 (1928). [2] D. W. Pohl, W. Denk and M. Lanz, Appl. Phys. Lett. 44, 651 (1984). [3] E. Betzig, M. Isaacson and A. Lewis, Appl. Phys. Lett. 51, 2088 (1987). [4] F. Zenhausern, Y. Martin and H. K. Wickramasinghe, Science 269, 1083 (1995). [5] P. Adam, Opt. Express 9, 353 (2001). [6] R. Hillenbrand, T. Taubner and F. Keilmann, Nature 418, 159 (2002). [7] X. S. Xie, A. Hartschuh, E. J. Sanchez and L. Novotny, Phys. Rev. Lett. 90, (2003). [8] A. Hartschuh, H. N. Pedrosa, L. Novotny and T. D. Krauss, Science 301, 1354 (2003). [9] A. Hartschuh, N. Anderson and L. Novotny, J. Microscopy 210, 234 (2003). [10] R. Hillenbrand and F. Keilmann, App. Phys. Lett. 80, 25 (2002). [11] R. Laddada, P. Royer, P. M. Adam and J. L. Bijeon, Opt. Eng. 37, 2142 (1998). [12] B. Knoll and F. Keilmann, Nature 399, 132 (1999). [13] F. Zenhausern, M. P. O Boyle and H. K. Wickramasinghe, Appl. Phys. Lett. 65, 1623 (1994). [14] F. Keilmann and B. Knoll, Opt. Commun. 182, 321 (2000). [15] G. Wurtz, R. Bachelot and P. Royer, Eur. Phys. J.-Appl. Phys. 5, 269 (1999). [16] M. Labardi, S. Patanè and M. Allegrini, Appl. Phys. Lett. 77, 621 (2000). [17] R. Hillenbrand and F. Keilmann, Phys. Rev. Lett. 85, 3029 (2000). [18] O. J. F. Martin, C. Girard and A. Dereux, Phys. Rev. Lett. 74, 526 (1995). [19] T. Taubner, R. Hillenbrand and F. Keilmann, J. Microscopy 210, 311 (2003). [20] B. Hecht, H. Bielefeldt, Y. Inouye, D. W. Pohl and L. Novotny, J. Appl. Phys. 81, 2492 (1997). [21] D. R. Lide and H. P. R. Frederikse, CRC Handbook of Chemistry and Physics (CRC Press, 1994).