22.07.2013 Views

Principles of Fluorescence Spectroscopy

Principles of Fluorescence Spectroscopy

Principles of Fluorescence Spectroscopy

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

184 FREQUENCY-DOMAIN LIFETIME MEASUREMENTS<br />

Figure 5.36. Lifetime χ R 2 surfaces for the three component mixtures<br />

<strong>of</strong> In, AA, and 2-AP; dashed, 380 nm; solid, global fit at 360, 380,<br />

400, 420, and 440 nm. From [116].<br />

<strong>of</strong> χ R 2 are 1.71 and 1.81, respectively. At first glance it<br />

seems that χ R 2 has increased for the three-exponential fit.<br />

However, the increase in χ R 2 for the three-decay-time model<br />

is a result <strong>of</strong> the larger number <strong>of</strong> variable parameters and<br />

the smaller number <strong>of</strong> degrees <strong>of</strong> freedom. The value <strong>of</strong> χ 2 ,<br />

which is the sum <strong>of</strong> the squared deviations, for these three<br />

fits is 2006, 59.7, and 59.6, for the one-, two-, and threedecay-time<br />

fits. Hence, the fit is not worse for the threedecay-time<br />

fit, but is essentially equivalent to the twodecay-time<br />

model.<br />

Samples such as this three-component mixture are difficult<br />

to analyze. In this case we know there are three decay<br />

times, and the three decay times are correctly determined<br />

by the analysis. However, obtaining the correct values<br />

required that the starting parameter values are close to the<br />

correct values. Otherwise, the program stopped at incorrect<br />

values, apparently trapped in local χ R 2 minima. Additionally,<br />

the χ R 2 surface is almost independent <strong>of</strong> lifetime, as<br />

shown in Figure 5.36 for the data measured at 380 nm<br />

(dashed). Without prior knowledge <strong>of</strong> the presence <strong>of</strong> three<br />

decay times, it would be difficult to know whether to accept<br />

the two or three decay time fit.<br />

At this point in the analysis there is little reason for<br />

proceeding further. If the information is not present in the<br />

data, no amount <strong>of</strong> analysis will create new information.<br />

One can either add new experimental data, or add information<br />

by restricting parameters based on separate knowledge<br />

about the sample. If one or more <strong>of</strong> the lifetimes are known,<br />

these can be held constant during the least-square fit. Similarly,<br />

one <strong>of</strong> the amplitudes could be fixed. However, the<br />

best approach is to add new data and perform a global<br />

analysis.<br />

Figure 5.37. Frequency-domain intensity decays <strong>of</strong> the three-component<br />

mixture observed at 360, 380, 400, 420, and 440 nm. The lines<br />

are for the best fit to three global decay times and non-global amplitudes.<br />

The values <strong>of</strong> χ R 2 for the one-, two-, and three-decay-time fit<br />

are 109.8, 2.3, and 1.1, respectively. From [116].<br />

The emission from the three-component mixture was<br />

measured at five wavelengths: 360, 380, 400, 420 and 440<br />

nm (Figure 5.37). At each wavelength each fluorophore displays<br />

the same decay time, but a different fractional amplitude<br />

based on its emission spectrum (Figure 5.38). Because<br />

<strong>of</strong> the different amplitudes at each wavelength, the frequency<br />

responses are wavelength dependent (Figure 5.37). The<br />

Figure 5.38. Emission spectra <strong>of</strong> the three-component mixture <strong>of</strong><br />

indole, anthranilic acid, and 2-aminopurine. Also shown are the fractional<br />

intensities recovered from global analysis <strong>of</strong> the frequencydomain<br />

data. From [116].

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!