Journal of Microscopy, Vol. 1 7 7 , Pt 2 , February 1 9 9 5 , pp. 1 1 5 - 1 1 8 .

Received 1 3 June 1 9 9 4 ; accepted 1 4 October 1 9 9 4

Forbidden light scanning near-field optical microscopy

H. HEINZELMANN,* B. HECHT,f L. NOVOTNYt & D. W . POHLS
llnstitutfir Physik, Universitiit Basel. Klingelbergstr. 8 2 . C H 4 0 5 6 Basel, Switzerland
fIBM Research Division, Zurich Research Laboratory, Siiumerstr. 4 , CH-8803 Riischlikon,
Switzerland
Slnstitutfiir Feldtheorie und Hijchstfrequenztechnik, ETH Zurich, CH-8092 Zurich, Switzerland

Key words. Scanning near-field optical microscopy, scanning tunnelling optical

microscopy, evanescence, multiple multipole method, shear force imaging.

Summary (approximately one half the wavelength of the radiation

used) can be broken. Many of the contrast mechanisms
Near-field optics (NFO) opens the door to light microscopy
known for conventional far-field optical microscopy can be
techniques with resolutions well beyond the diffraction
exploited in SNOM as well, such as fluorescence (Betzig et
limit. The richness of optical investigations is now
al., 1994), polarization (Betzig et al., 1994) and phase
applicable on a near-molecular level. Among the novel
contrast (Vaez-Iravani & Toledo-Crow, 1993). and new
scanning near-field optical microscopy (SNOM) schemes,
effects specific to the near-field might prove useful to gain
the most prominent representatives are aperture SNOM and
additional sample information.
scanning tunnelling optical microscopy (STOM or PSTM).
During the past few years, two main experimental
New experimental and theoretical work has to be
manifestations of the method have been established. One
performed to study the phenomena specific to NFO. One
scheme called (aperture-) SNOM employs a small aperture
such example is the angular dependence of light emission in
as optical probe which is used to illuminate the sample
aperture SNOM. The detection of radiation at angles greater
surface locally (Fig. la). In other schemes the probe is used
than the critical angle of total internal reflection
to collect light transmitted through the sample, or the light
a, = arcsin(nP1), where n is the sample refractive index,
reflected off an opaque surface is detected. SNOM owes its
can represent a microscopy scheme that combines the
high resolving power to the field confinement by the probe
respective advantages of both aperture SNOM and STOM.
aperture. The localization of the illumination to the surface
Recent experiments have demonstrated the expected
area just beneath the probe has the additional advantage of
exponential dependence of light intensity on gap width
avoiding unnecessarily high radiation exposure of photo-
(for fixed emission angle a > a,). The decay length as a
sensitive materials. A shortcoming of the method, however.
function of a is in agreement with the Fresnel description of
is the large radiation background caused by light coming
the evanescent field when total reflection occurs at an
from sources other than the aperture. This background
interface. These investigations were additionally motivated
makes the detection of the near-field signal difficult.
by calculations based on the multiple multipole method.
In the other scheme, scanning tunnelling optical
microscopy (STOM, or PSTM, 'photon scanning tunnelling
Introduction microscopy') (Reddick et al., 1989; Vigoreux et al., 1989),
the sample is placed on a glass surface that is illuminated
Scanning near-fie1d m i c r O s c o (SNOM)
~~ '''1
from the back at such an angle that total internal reflection
1984; Diirig et al., 1986) is a scanning probe technique
(TIR) occurs at the samplelair interface (Fig. lb). A
that provides optical information at resolutions better than
dielectric probe, usually a bare fibre tip, dives into the
lOOnm (for a recent review see Betzig et al., 1991; Pohl et
field of the evanescent wave created by the TIR. Scattering
d., 1991; Heinzelmann & Pohl, 1994).The optical image is
converts a small amount of the energy, proportional to the
generated by the interaction of a subwavelength optical
local excitation at the probe, into a waveguide mode of the
probe held in close proximity with the sample (i.e. in the
fibre that propagates to the photodetector. Since the
near-field), and monitoring this interaction while scanning
evanescent field decays exponentially above the surface,
the probe across the sample surface. No optical lenses are
the STOM signal has a strong gap-width dependence
employed, and Rayleigh's diffraction limit of resolution
(similar to that of an electron STM, but with a larger
e-mail: heinzelrnann@urz.unibas.ch. decay length of typically 100-300nm). No radiation can

0 1995 The Royal Microscopical Society 115

116 H . H E I N Z E L M A N N E T AL

Fig. 1. Different schemes of scanning near-field

optical microscopy. (a) SNOM in illumination
mode: a dielectric aperture-type probe illumi-
nates the sample: the light is collected in the
far-field. (b) STOM or PSTM: a dielectric probe
frustrates the evanescent field at the sample sur-
face, generating waves which propagate in the
probe. (c) 'Tunnel SNOM' or 'inverted STOM':
a dielectric aperture-type probe illuminates the
sample: the light is collected at angles larger
than the critical angle of total internal reflec-
tion at the air/glass interface.

be detected when the probe is far away from the surface; the
far-field radiation background is zero.
The imaging capabilities of STOM seem less favourable
than those of aperture SNOM. First, the interaction volume
is laterally less well defined since light can enter the
uncoated fibre over the entire decay length of the
evanescent wave. Further, both scattered and evanescent
waves contribute to the detected signal, making interpreta-
tion of the collected data difficult. Additionally, the
illuminated sample area and hence the radiation exposure
is large compared with aperture SNOM.

Results and discussion

We investigated a new scheme that combines the
favourable properties of both aperture SNOM and STOM,
and eliminates their respective drawbacks (Hecht et al.,
1994). Inspired by Helmholtz's reciprocity theorem. we
inverted the light path in the STOM set-up: the light is
emitted from the tip and is detected behind the glass sample
at an angle larger than the critical angle of total internal
reflection 8, = arcsin(npl), where n is the index of
refraction (Fig. lc). By using a metal-coated fibre tip, the
exponential distance dependence of this signal can be
combined with the high spatial resolution delivered by the
aperture probe. The resulting near-field optical microscope
conceptually is an 'inverted STOM' or a 'tunnel SNOM'. It
should be noted that the emission of light in the forbidden
region when a light source approaches a glass surface was
described more than 60 years ago (Frohlich, 1921). In the
context of near-field optical microscopy, this effect was
discussed for the first time by Fischer (1985). Here, we
report on its experimental application in aperture SNOM.
The development of this scheme was additionally
motivated by recent theoretical simulations of light
propagation in a model SNOM geometry (Novotny et al.,
1994). Optical diffraction theory does not describe the
phenomena occurring in the near-field; Maxwell's equa- Fig. 2. MMP simulations of the light propagation in a model SNOM
tions have to be solved for the boundary conditions imposed geometry. (a) An ideally conducting object of 20nm diameter is
by the tip-sample geometry. Based on the multiple placed 30nm from the centre. Grey scales depict light intensity.
multipole (MMP) method, the radiation emitted from an white arrows power flux. (b) Transmitted radiation patterns for
Al-coated fibre tip with an aperture of 30 nm diameter over the object moving from 0 nm (centre position) to infinity.
 F FC A12 dl4 L

Fig. 3. Left: combined aperture SNOM/

STOM set-up. Channels A and B: optical
LD
+-+--=
c

signals in forward direction (8 = 0) and in

tunnel direction (6' > O,), respectively:
.-
channel C: shear force. L, laser; ,412 and
X/4, retardation plates; FC, fibre coupler:
F, fibre; P, piezo; H, hemisphere; An,
-
.-
Em
8 = 52.8"

-
.-
analyser; PM, photomultiplier; LD, laser
diode; PD, photo diode. Right: light inten- !\I ,
sity (lower trace) and shear force (upper o 500 nm
trace) as a function of tip-sample distance. tip - sample distance

a glass sample surface at a separation of 3 0 n m was
calculated (Novotny et al., 1994). Figure 2(a) shows the
tip-sample arrangement with an ideally conducting object
placed at 30 nm from the centre position. The shades of grey
depict the time-average of the electric field strength squared,
i.e, the electrical energy density, on a logarithmic scale.
Note the penetration of the electromagnetic field into the A1
coating at the sides of the fibre taper, a consequence of the
finite conductivity of the metal. White arrows denote the
Poynting vector, i.e, the power flux.
Under certain conditions, the maximum light intensity is
not found in the forward direction, but at an angle of about
55" from the surface normal (Fig. 2b). The same holds true
for the most prominent change in light intensity when the
fibre tip is scanned across an object (Novotny et al., 1994).
This effect is clearly a near-field phenomenon: no radiation

Fig. 4. Simultaneously acquired topographic (a) and optical image taken at an observation angle of 48" (b) of an optical grating with 383 nm
periodicity and 8 nm height.
118 H. HEINZELMANN ET A L .
under this angle, which is larger than the critical angle of
TIR, would be observable if the light source were placed at a
distance z >> X from the sample surface. The angular
distribution of radiation is plotted for sample off-centre
positions of Onm (sample located exactly under the probe
tip), 50 nm. 100 nm and cc (no sample present). Note that
there are almost no variations in the forward direction,
which is where aperture SNOM is operated in most cases.
Our experimental set-up is shown in Fig. 3. Light from a
red HeNe laser L (A = 633 nm) is emitted by a tapered fibre
probe and collected after transmission through the sample
in optical contact with a glass hemisphere H. Retardation
plates and analysers allow polarization control. The
intensities in the forward direction (channel A) and at an
angle larger than the critical angle of total internal
reflection 0, = 41.8" for n = 1.5 (channel B) are monitored
simultaneously for comparison. Channel C delivers an
auxiliary shear force signal used for distance control and
feedback, and to map the sample topography.
On the right-hand side of Fig. 3, approach curves for the
light intensity detected at 0 = 52.8" (lower trace) and the
auxiliary shear force signal (upper trace) are shown. The
exponential decay of the evanescent field extends over
several hundred nanometres. The decay lengths as a
function of detection angle follow the Fresnel description
for total internal reflection (Hecht et al.. 1994). A detectable
shear force signal is usually found at separations below
100nm; it probably relies heavily on surface contamina-
tions such as an adsorbed water film.
The first images taken in this new mode are shown in
Fig. 4. As a test pattern we used an optical grating with a
rectangular profile, a lateral periodicity of 383 nm and a
vertical corrugation of 8 nm. The optical image was taken
in the forbidden zone at 0 = 48" (Fig. 4a). The grating is
well resolved and of good contrast. Figure 4(b) shows the
simultaneously acquired sample topography.
The image was taken at constant tip-sample distance.
In addition to the advantages described above, this new
set-up could open new possibilities for near-field optical
microscopy. One example is simultaneous observation at
different detection angles, which potentially reveals addi-
tional sample information (Novotny et al., 1994).
Acknowledgments
We would like to thank B. Curtis from the Paul Scherrer
Institute for providing the test sample, and G. Bona. P. Unger
and H. Richard from IBM Research for kindly sharing some
of their equipment. This work was supported by a grant
from the Swiss priority program OPTIQUE.
References
Betzig, E., Chichester, R.J., Lanni, F. &Taylor,D.L. (1994) Near-field
fluorescence imaging of cytoskeletal actin. Bioimaging, 1, 129.
Betzig, E., Trautman. J.K.. Harris, T.D., Weiner, J.S. & Kostelak.
R.L. (1991) Breaking the diffraction barrier: optical microscopy
on a nanometric scale. Science, 2 5 1 , 1468.
Betzig, E., Trautman, J.K.. Weiner, J.S., Harris, T.D. & Wolfe, R.
(1992) Polarization contrast in near-field scanning optical
microscopy. Appl. Opt. 31, 4563.
Diirig, U., Pohl, D.W. & Rohner, F. (1986) Near-field optical-
scanning microscopy. 1. Appl. Phys. 5 9 , 3318.
Fischer. U.Ch. (1985) Optical characteristics of 0.1 pm circular
apertures in a metal film as light sources for scanning
ultramicroscopy. 1. Vac. Sci. Technol. B, 3 , 386.
Frohlich. P. (1921) Die Giiltigkeitsgrenze des geometrischen
Gesetzes der Lichtbrechung. Advan. Phys. 65, 39.
Hecht, B., Heinzelmann, H. & Pohl, D.W. (1994) Combined
aperture SNOMIPSTM: the best of both worlds? Ultramicroscopy
(in press).
Heinzelmann. H. & Pohl. D.W. (1994) Scanning near-field optical
microscopy. Appl. Phys. A, 59. 89-101.
Novotny, L., Pohl, D.W. & Regli, P. (1994) Light propagation
through nanometer-sized structures: the two-dimensional-
aperture scanning near-field optical microscope. ]. Opt. Soc.
Am. A l l , 1768.
Pohl. D.W. ( 199 1) Scanning near-field optical microscopy (SNOM).
Advances in Optical and Electron Microscopy, Vol. 12 (ed. by
T. Mulvey and C. J. R. Sheppard), pp. 243-312. Academic Press.
Pohl, D.W., Denk, W. & Lanz, M. (1984) Optical stethoscopy:
image recording with resolution X/20. Appl. Phys. Lett. 44, 651.
Reddick. R.C., Warmack, R.J. & Ferrell. T.L. (1989) New form of
scanning optical microscopy. Phys. Rev. B. 39, 7 6 7 .
Vaez-Iravani. M. & Toledo-Crow. R. (1993) Phase contrast and
amplitude pseudoheterodyne interference near field scanning
optical microscopy. Appl. Phys. Lett. 6 2 , 1044.
Vigoureux, J.M.. Girard. C. & Courjon, D. (1989) General principles
of scanning tunneling optical microscopy. Opt. Lett. 14, 1039.