2 Interferometric Scattering (iSCAT) Microscopy and Related Techniques
35
tion of small nanoparticles since conventional methods such as dark-field microscopy
were not able to reach this limit. The only other technique with appreciable sensitivity was reported just two years earlier based on photothermal detection [71]. In
this technique, one heats the GNP through its enhanced absorption at the plasmon
resonance and detects the heat-induced change of refractive index in its vicinity.
Interestingly, however, this latter decisive step is also achieved via interferometry
using a second laser beam.
In the following years, iSCAT was extended in our laboratory to different illumination and detection conditions [113, 114] and used to detect single unlabeled
viruses [115, 116], semiconductor quantum dots [114], lipid vesicles [117, 118]
and unlabeled proteins [119]. In more recent years, several other groups have also
successfully applied different illumination/detection variants of iSCAT to detect single proteins [108, 120, 121], single viruses [83, 106], lipids [122, 123] and other
nanoparticles [82], and even charge carriers [107].
Neglecting the scattering intensity (I s ) which is vanishingly weak for very small
particles, the iSCAT signal of interest, namely the interferometric cross-term, can
be obtained by subtracting the reference intensity that acts as a background from
the detected intensity, I det − I r ≈ 2E s E r cos φ. Thus, the contrast obtained when
comparing images with and without a nanoparticle becomes:
c =
2E s E r cos φ
I r
= 2
E s
E r
cos φ .
(2.3)
The final expression in (2.3) might prompt one to conclude that one can reach a
better sensitivity through minimization of E r in the denominator. Indeed, one can
devise an optical scheme, where E r is freely adjustable, e.g., by implementing a
separate reference arm (see, e.g., scheme of Fig. 2.3a). However, the resulting increase
in contrast is at the cost of a lower overall signal which leads to a smaller signalto-noise ratio (SNR) in shot-noise-limited detection. Furthermore, a separate arm
compromises sensitivity owing to the introduction of mechanical instabilities. We
discuss the issue of SNR and other instrumental considerations in more detail below.
Before we begin the discussion of experimental issues, we take a moment to
remark on the fundamental efficiency of interferometric detection of small objects,
down to single atoms and molecules. When an ideal two-level atom is illuminated
with monochromatic light at its resonance frequency, it undergoes a transition from
the ground to the excited state. This process can lead to absorption and successive
emission of light. The resulting fluorescence is incoherent because it follows a spontaneous process, thus not respecting the phase of the illumination field. However, in
the weak excitation regime, the interaction of the incoming light with an atom can
also be described by Rayleigh scattering [124]. It follows then that the interaction
of the incoming light and the atom can equally be described in the same way as in
(2.1), that is, via interference. Considering that the extinction cross section of an
unperturbed atom can be as large as 3λ
2
/2π , where λ is the transition wavelength,
and that light can be focused down to the diffraction limit in the order of (λ/2)
2 , one
ought to expect a single atom to be capable of casting a dark shadow on a laser beam.
35
tion of small nanoparticles since conventional methods such as dark-field microscopy
were not able to reach this limit. The only other technique with appreciable sensitivity was reported just two years earlier based on photothermal detection [71]. In
this technique, one heats the GNP through its enhanced absorption at the plasmon
resonance and detects the heat-induced change of refractive index in its vicinity.
Interestingly, however, this latter decisive step is also achieved via interferometry
using a second laser beam.
In the following years, iSCAT was extended in our laboratory to different illumination and detection conditions [113, 114] and used to detect single unlabeled
viruses [115, 116], semiconductor quantum dots [114], lipid vesicles [117, 118]
and unlabeled proteins [119]. In more recent years, several other groups have also
successfully applied different illumination/detection variants of iSCAT to detect single proteins [108, 120, 121], single viruses [83, 106], lipids [122, 123] and other
nanoparticles [82], and even charge carriers [107].
Neglecting the scattering intensity (I s ) which is vanishingly weak for very small
particles, the iSCAT signal of interest, namely the interferometric cross-term, can
be obtained by subtracting the reference intensity that acts as a background from
the detected intensity, I det − I r ≈ 2E s E r cos φ. Thus, the contrast obtained when
comparing images with and without a nanoparticle becomes:
c =
2E s E r cos φ
I r
= 2
E s
E r
cos φ .
(2.3)
The final expression in (2.3) might prompt one to conclude that one can reach a
better sensitivity through minimization of E r in the denominator. Indeed, one can
devise an optical scheme, where E r is freely adjustable, e.g., by implementing a
separate reference arm (see, e.g., scheme of Fig. 2.3a). However, the resulting increase
in contrast is at the cost of a lower overall signal which leads to a smaller signalto-noise ratio (SNR) in shot-noise-limited detection. Furthermore, a separate arm
compromises sensitivity owing to the introduction of mechanical instabilities. We
discuss the issue of SNR and other instrumental considerations in more detail below.
Before we begin the discussion of experimental issues, we take a moment to
remark on the fundamental efficiency of interferometric detection of small objects,
down to single atoms and molecules. When an ideal two-level atom is illuminated
with monochromatic light at its resonance frequency, it undergoes a transition from
the ground to the excited state. This process can lead to absorption and successive
emission of light. The resulting fluorescence is incoherent because it follows a spontaneous process, thus not respecting the phase of the illumination field. However, in
the weak excitation regime, the interaction of the incoming light with an atom can
also be described by Rayleigh scattering [124]. It follows then that the interaction
of the incoming light and the atom can equally be described in the same way as in
(2.1), that is, via interference. Considering that the extinction cross section of an
unperturbed atom can be as large as 3λ
2
/2π , where λ is the transition wavelength,
and that light can be focused down to the diffraction limit in the order of (λ/2)
2 , one
ought to expect a single atom to be capable of casting a dark shadow on a laser beam.