Principles of Fluorescence Spectroscopy

PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 173
Figure 5.17. Emission spectra of 9,10-diphenylanthracene (DPA) in
ethanol for excitation at 325 nm from a helium–cadmium laser. The
solvent contained a small amount of Ludox colloidal silica as the scatterer.
The dashed line is the transmission of the Corning 3-75 filter.
Revised from [69].
observed without an emission filter (solid) there is a large
peak due to scattered light at the excitation wavelength of
325 nm. The presence of this scattered component would
not be recognized without measurement of the emission
spectrum, and would result in an incorrect intensity decay.
Scattered light is typically rejected from the detector
by using emission filters. In this case we used a Corning 3-
75 filter, which transmits above 360 nm (dashed). As a control
measurement one should always record the emission
spectrum of the blank sample through the emission filter to
ensure scattered light is rejected. Alternatively, one can
measure the intensity of the blank through the filter to
determine that the blank contribution is negligible. In such
control measurements it is important that the blank scatters
light to the same extent as the sample. Frequently, buffer
blanks do not scatter as strongly as the sample containing
the macromolecules because of the inner filter effect present
in the sample.
It is useful to understand how scattered light can corrupt
the frequency-domain data. Frequency responses for
the DPA solution are shown in Figure 5.18. For these measurements
the excitation source was a helium-cadmium
(HeCd) laser at 325 nm. The cw output of this laser was
modulated with an electrooptic modulator, as shown in Figure
5.8. The effect of scattered light is visually evident in
the frequency-domain data. When measured without an
emission filter, the phase angles (") are considerably smaller
than expected for the single exponential decay of DPA
(!). The phase angle error becomes larger at higher frequencies.
It should be noted that the fractional intensity of
the background is only 15% (f B = 0.15), so that significant
Figure 5.18. Frequency-domain intensity decay of 9,10-diphenylanthracene
in ethanol with a scatterer. The sample was in equilibrium
with dissolved oxygen. Data were measured without an emission filter
(") and then corrected (!) for the scattered light using eqs.
5.18–5.23. Bottom panels: deviations from the best single exponential
fits for the data with ("), and corrected for (!) scattered light. Revised
from [69].
errors in phase angle are expected for even small amounts
of scattered light.
It is possible to correct for background from the sample.
The solid dots represent the data corrected according to
eqs. 5.18–5.23. The corrected data can be fit to a single
decay time with τ = 6.01 ns. An alternative approach is to
fit the data with scattered light to include a second component
with a lifetime near zero. This also results in a good fit
to the data, with a decay time near zero associated with f B =
0.15. However, this procedure is only appropriate if the
background is only due to scattered light. In general autofluorescence
will display nonzero lifetimes and nonzero
phase angles.
5.5. SIMPLE FREQUENCY-DOMAIN
INSTRUMENTS
A large fraction of the cost of a frequency-domain instrument
is the light source and/or modulation optics. In
TCSPC these expensive light sources are being replaced by
LDs and LEDs. This substitution is also occurring with frequency-domain
instruments. For most experiments the