22.07.2013 Views

Principles of Fluorescence Spectroscopy

Principles of Fluorescence Spectroscopy

Principles of Fluorescence Spectroscopy

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

PRINCIPLES OF FLUORESCENCE SPECTROSCOPY 143<br />

acceptor, which is given by eq. 4.31. When using the results<br />

<strong>of</strong> a multi-exponential analysis, the transfer efficiency<br />

should be calculated using values, since these are proportional<br />

to the steady-state intensity.<br />

4.11.2. Lifetime Distributions<br />

There are many situations where one does not expect a limited<br />

number <strong>of</strong> discrete decay times, but rather a distribution<br />

<strong>of</strong> decay times. Such behavior may be expected for a<br />

fluorophore in a mixture <strong>of</strong> solvents, so that a range <strong>of</strong> environments<br />

exists. One can imagine a fluorophore being surrounded<br />

by one, two, three, or more polar molecules, each<br />

resulting in a different intensity decay. Another possibility<br />

is a protein with many tryptophan residues, so that it is not<br />

practical to consider individual decay times.<br />

In such cases the intensity decays are typically analyzed<br />

in terms <strong>of</strong> a lifetime distribution. In this case the α i<br />

values are replaced by distribution functions α(τ). The component<br />

with each individual τ value is given by<br />

(4.33)<br />

However, one cannot observe these individual components<br />

with lifetime τ, but only the entire decay. The total decay<br />

law is the sum <strong>of</strong> the individual decays weighted by the<br />

amplitudes:<br />

(4.34)<br />

where Iα(τ)dτ = 1.0.<br />

Lifetime distributions are usually used without a theoretical<br />

basis for the α(τ) distribution. One typically uses<br />

arbitrarily selected Gaussian (G) and Lorentzian (L) lifetime<br />

distributions. For these functions the α(τ) values are<br />

α G(τ) <br />

α L(τ) 1<br />

π<br />

I(τ,t) α(τ)e t/τ<br />

∞<br />

I(t) α (τ)e<br />

τ0<br />

t/τdτ 1<br />

σ√2π exp{1<br />

τ τ ( ) 2 σ 2}<br />

Γ/2<br />

(τ τ ) 2 (Γ/2) 2<br />

(4.35)<br />

(4.36)<br />

where (τ) is the central value <strong>of</strong> the distribution, σ the standard<br />

deviation <strong>of</strong> the Gaussian, and Γ the full width at half<br />

maximum (FWHM) for the Lorentzian. For a Gaussian the<br />

full width at half maximum is given by 2.345σ. For ease <strong>of</strong><br />

interpretation we prefer to describe both distributions by the<br />

full width at half maxima. An alternative approach would<br />

be to use α(τ) distributions that are not described by any<br />

particular function. This approach may be superior in that it<br />

makes no assumptions about the shape <strong>of</strong> the distribution.<br />

However, the use <strong>of</strong> functional forms for α(τ) minimizes<br />

the number <strong>of</strong> floating parameters in the fitting algorithms.<br />

Without an assumed function form it may be necessary to<br />

place restraints on the adjacent values <strong>of</strong> α(τ).<br />

By analogy with the multi-exponential model, it is possible<br />

that α(τ) is multimodal. Then<br />

α(τ) ∑ i<br />

(4.37)<br />

where i refers to the ith component <strong>of</strong> the distribution centered<br />

at α i , and g i represents the amplitude <strong>of</strong> this component.<br />

The g i values are amplitude factors and α i 0(τ) the<br />

shape factors describing the distribution. If part <strong>of</strong> the distribution<br />

exists below τ = 0, then the α i (τ) values need additional<br />

normalization. For any distribution, including those<br />

cut <strong>of</strong>f at the origin, the amplitude associated with the ith<br />

mode <strong>of</strong> the distribution is given by<br />

α i <br />

(4.38)<br />

The fractional contribution <strong>of</strong> the ith component to the total<br />

emission is given by<br />

f i <br />

g i α 0 i (τ) ∑ i<br />

∞<br />

0 α i(τ) dτ<br />

∞<br />

0 ∑ i<br />

∞<br />

0 αi(τ)τdτ ∞<br />

0 ∑ i<br />

α i(τ)dτ<br />

α i(τ)τdτ<br />

α i(τ)<br />

(4.39)<br />

In the use <strong>of</strong> lifetime distributions each decay time component<br />

is associated with three variables, α i , f i and the half<br />

width (σ or Γ). Consequently, one can fit a complex decay<br />

with fewer exponential components. For instance, data that<br />

can be fit to three discrete decay times can typically be fit<br />

to a bimodal distribution model. In general, it is not possible<br />

to distinguish between the discrete multi-exponential<br />

model (eq. 4.27) or the lifetime distribution model (eq.<br />

4.34), so the model selection must be based on one's knowledge<br />

<strong>of</strong> the system. 197–199

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

Saved successfully!

Ooh no, something went wrong!