318
I. I. Smolyaninov and V. N. Smolyaninova
the droplet boundary was used as an efficient 2D parabolic mirror for propagating
surface plasmons excited inside the droplet by external laser illumination. Since the
plasmon wavelength is much smaller than the droplet sizes, the image formation in
such a mirror can be analyzed by simple geometrical optics in two dimensions.
The resolution test of the microscope has been performed using a 30 × 30 µm
2
array of triplet nanoholes (100 nm hole diameter with 40 nm distance between the
hole edges) shown in Fig. 13.3c. This array was imaged using a glycerine droplet
shown in Fig. 13.3a. Periodic nanohole arrays first studied by Ebbesen et al. [19]
appear to be ideal test samples for the plasmon microscope. Illuminated by laser light,
such arrays produce propagating surface waves, which explains the anomalous transmission of such arrays at optical frequencies. The image of the triplet array obtained
at 515 nm using a 100x microscope objective is shown in Fig. 13.3b (compare it
with an image in Fig. 13.3d calculated using 2D geometrical optics). Even though
some discrepancy between the experimental and theoretical images can be seen (the
image pattern observed in Fig. 13.3b looks convex looking from the left compared
to the concave pattern observed in the calculation in Fig. 13.3d), the overall match
between these images is impressive. The most probable reason for the observed convex/concave discrepancy is the fact that the droplet shape is not exactly parabolic,
which produces some image aberrations. Although the expected resolution of the
microscope at 515 nm is somewhat lower than at 502 nm, the 515 nm laser line is
brighter, which allowed us to obtain more contrast in the 2D image. The least distorted part of the image, Fig. 13.3b (far from the droplet edge, yet close enough to
the nanohole array, so surface plasmon decay does not affect resolution), is shown at
higher digital zooms of the charge-coupled device (CCD) camera mounted onto our
conventional optical microscope in Fig. 13.3e, f. These images clearly visualize the
triplet nanohole structure of the sample. Moreover, using the experimentally measured point spread function (PSF) of the SPP microscope, resolution of 2D plasmon
microscopy may be further improved to the ~30 nm scale (as shown in Fig. 13.6) by
implementing digital resolution enhancement techniques [14].
The spatial resolution of the optical images (the PSF of the microscope) may
be measured directly by calculating the cross-correlation P * E between the optical
image P and the scanning electron microscopy (SEM) image E of the same nanohole:
P
∗ E (r) =
P(r 1 )E(r 1 + r)dr 1
(13.3)
The results of these calculations in the cases of triplet nanoholes from Figs. 13.3
and 13.4 demonstrate that a resolution of the order of PSF ≈ 70 nm or ~λ/8 is
achieved in these particular imaging experiments. Such an improved resolution in
an SPP microscopy experiment is due to the fact that the SPP wavelength is shorter
than the wavelength of guided modes at the same laser frequency. Photonic crystal
effects and the effects of negative refraction also play some role in achieving better
resolution.
Even though quite an improvement compared to a regular optical microscope, the
~70 nm resolution is not sufficient to achieve clear visibility of many nanoholes in
I. I. Smolyaninov and V. N. Smolyaninova
the droplet boundary was used as an efficient 2D parabolic mirror for propagating
surface plasmons excited inside the droplet by external laser illumination. Since the
plasmon wavelength is much smaller than the droplet sizes, the image formation in
such a mirror can be analyzed by simple geometrical optics in two dimensions.
The resolution test of the microscope has been performed using a 30 × 30 µm
2
array of triplet nanoholes (100 nm hole diameter with 40 nm distance between the
hole edges) shown in Fig. 13.3c. This array was imaged using a glycerine droplet
shown in Fig. 13.3a. Periodic nanohole arrays first studied by Ebbesen et al. [19]
appear to be ideal test samples for the plasmon microscope. Illuminated by laser light,
such arrays produce propagating surface waves, which explains the anomalous transmission of such arrays at optical frequencies. The image of the triplet array obtained
at 515 nm using a 100x microscope objective is shown in Fig. 13.3b (compare it
with an image in Fig. 13.3d calculated using 2D geometrical optics). Even though
some discrepancy between the experimental and theoretical images can be seen (the
image pattern observed in Fig. 13.3b looks convex looking from the left compared
to the concave pattern observed in the calculation in Fig. 13.3d), the overall match
between these images is impressive. The most probable reason for the observed convex/concave discrepancy is the fact that the droplet shape is not exactly parabolic,
which produces some image aberrations. Although the expected resolution of the
microscope at 515 nm is somewhat lower than at 502 nm, the 515 nm laser line is
brighter, which allowed us to obtain more contrast in the 2D image. The least distorted part of the image, Fig. 13.3b (far from the droplet edge, yet close enough to
the nanohole array, so surface plasmon decay does not affect resolution), is shown at
higher digital zooms of the charge-coupled device (CCD) camera mounted onto our
conventional optical microscope in Fig. 13.3e, f. These images clearly visualize the
triplet nanohole structure of the sample. Moreover, using the experimentally measured point spread function (PSF) of the SPP microscope, resolution of 2D plasmon
microscopy may be further improved to the ~30 nm scale (as shown in Fig. 13.6) by
implementing digital resolution enhancement techniques [14].
The spatial resolution of the optical images (the PSF of the microscope) may
be measured directly by calculating the cross-correlation P * E between the optical
image P and the scanning electron microscopy (SEM) image E of the same nanohole:
P
∗ E (r) =
P(r 1 )E(r 1 + r)dr 1
(13.3)
The results of these calculations in the cases of triplet nanoholes from Figs. 13.3
and 13.4 demonstrate that a resolution of the order of PSF ≈ 70 nm or ~λ/8 is
achieved in these particular imaging experiments. Such an improved resolution in
an SPP microscopy experiment is due to the fact that the SPP wavelength is shorter
than the wavelength of guided modes at the same laser frequency. Photonic crystal
effects and the effects of negative refraction also play some role in achieving better
resolution.
Even though quite an improvement compared to a regular optical microscope, the
~70 nm resolution is not sufficient to achieve clear visibility of many nanoholes in