2 Interferometric Scattering (iSCAT) Microscopy and Related Techniques
37
If one manages to deal with the instrumental laser intensity noise, one is still
confronted with the fundamental limit of shot noise which is the noise associated
with the fact that the number N of the photons in a laser beam varies according
to a Poisson distribution, i.e., as
√
N for large N [124]. As a rule of thumb, the
shot-noise-limited SNR improves by
√
N as N is increased.
Detector background noise: Any electronic device has an intrinsic noise, stemming from, for example, thermally generated electrons in the detector (referred to
as dark noise) or errors introduced in the voltage reading circuitry. Modern cameras
and photodiodes can be extremely quiet, and since iSCAT is typically performed on
a high background level due to the reference field E r , detector dark noise is not a
major concern. Nevertheless, depending on the experimental arrangement, it might
become a limiting factor.
Dynamic range and analogue-to-digital conversion noise: Realistic detectors
have a limited working range on both sides of small and large signals. The limit
for the largest signals is given by detector nonlinearities and saturation effects, while
the lower limit often has to do with the fundamental sensitivity of the particular
detector technology. The ratio between the largest and smallest signal values defines
the dynamic range, and for imaging cameras the read noise level is often taken as
the smallest signal quantity which is detectable. From the sensor dynamic range, an
appropriate bit depth is selected for analogue-to-digital conversion. For example, an
image rendered into 12-bit imposes a read-out resolution of 1 in 4096, i.e., 2 × 10
−4 .
Mechanical stability: Although iSCAT is an interferometric method, it can be
extremely robust against mechanical instabilities if the reference and the scattering
beams share identical paths (see Fig. 2.3). Nevertheless, lateral vibrations at the
sample cause problems since even a few nanometers of motion could translate into
fluctuations of the contrast when the background is subtracted (see section below)
[134].
2.3.3 Background Removal
Fluorescence detection exploits highly efficient spectral filtering to eliminate spurious backgrounds caused by the illumination or unwanted fluorescence. Similarly,
dark-field microscopy, including variants using total internal reflection, employ spatial filtering to reject background illumination in order to detect Rayleigh scattering.
In iSCAT, however, one does not exclude the background but instead records it in an
intense reference beam, just as in resonant extinction measurements of a quantum
emitter [135].
As mentioned earlier, this would not pose any problem if one could subtract
a constant background level from the measured signal, even if the signal were to
be arbitrarily small. Figure 2.4a, which displays an iSCAT image of a coverslip
supporting 10 nm GNPs in water, shows that in practice, one is confronted with
37
If one manages to deal with the instrumental laser intensity noise, one is still
confronted with the fundamental limit of shot noise which is the noise associated
with the fact that the number N of the photons in a laser beam varies according
to a Poisson distribution, i.e., as
√
N for large N [124]. As a rule of thumb, the
shot-noise-limited SNR improves by
√
N as N is increased.
Detector background noise: Any electronic device has an intrinsic noise, stemming from, for example, thermally generated electrons in the detector (referred to
as dark noise) or errors introduced in the voltage reading circuitry. Modern cameras
and photodiodes can be extremely quiet, and since iSCAT is typically performed on
a high background level due to the reference field E r , detector dark noise is not a
major concern. Nevertheless, depending on the experimental arrangement, it might
become a limiting factor.
Dynamic range and analogue-to-digital conversion noise: Realistic detectors
have a limited working range on both sides of small and large signals. The limit
for the largest signals is given by detector nonlinearities and saturation effects, while
the lower limit often has to do with the fundamental sensitivity of the particular
detector technology. The ratio between the largest and smallest signal values defines
the dynamic range, and for imaging cameras the read noise level is often taken as
the smallest signal quantity which is detectable. From the sensor dynamic range, an
appropriate bit depth is selected for analogue-to-digital conversion. For example, an
image rendered into 12-bit imposes a read-out resolution of 1 in 4096, i.e., 2 × 10
−4 .
Mechanical stability: Although iSCAT is an interferometric method, it can be
extremely robust against mechanical instabilities if the reference and the scattering
beams share identical paths (see Fig. 2.3). Nevertheless, lateral vibrations at the
sample cause problems since even a few nanometers of motion could translate into
fluctuations of the contrast when the background is subtracted (see section below)
[134].
2.3.3 Background Removal
Fluorescence detection exploits highly efficient spectral filtering to eliminate spurious backgrounds caused by the illumination or unwanted fluorescence. Similarly,
dark-field microscopy, including variants using total internal reflection, employ spatial filtering to reject background illumination in order to detect Rayleigh scattering.
In iSCAT, however, one does not exclude the background but instead records it in an
intense reference beam, just as in resonant extinction measurements of a quantum
emitter [135].
As mentioned earlier, this would not pose any problem if one could subtract
a constant background level from the measured signal, even if the signal were to
be arbitrarily small. Figure 2.4a, which displays an iSCAT image of a coverslip
supporting 10 nm GNPs in water, shows that in practice, one is confronted with