Hello, 

If you get a linear plot of your data when you plot log(intensity (counts)) versus time, then fit the data to a single exponential decay y(t)=A*Exp(-(t-t0)/tau)

You may need to exclude the first portion of your data, because this function is only a decay, and it does not account for the initial rise when your laser pulse arrives. The time of your first point is your t0. Your least square fit procedure (use Igor, Origin or even excel) with Tau and A adjustable will give you Tau the emission lifetime.

If the plot of log(int) versus time is not linear, (this appears as a decay with several nearly linear segments each of which have a different slope. Then you need to fit the data to  a sum of two or more exponentials.

This is the easiest way but it  may not be accurate. 

There are better ways to do this using a convolution of the instrument response with a sum of exponentials, This is called re-convolution fit procedure. This topic already appeared in this question-answer page. You will find more discusions here if you search among the questions answered here.

I hope this helps.

Hello, 

If you get a linear plot of your data when you plot log(intensity (counts)) versus time, then fit the data to a single exponential decay y(t)=A*Exp(-(t-t0)/tau)

You may need to exclude the first portion of your data, because this function is only a decay, and it does not account for the initial rise when your laser pulse arrives. The time of your first point is your t0. Your least square fit procedure (use Igor, Origin or even excel) with Tau and A adjustable will give you Tau the emission lifetime.

If the plot of log(int) versus time is not linear, (this appears as a decay with several nearly linear segments each of which have a different slope. Then you need to fit the data to  a sum of two or more exponentials.

This is the easiest way but it  may not be accurate. 

There are better ways to do this using a convolution of the instrument response with a sum of exponentials, This is called re-convolution fit procedure. This topic already appeared in this question-answer page. You will find more discusions here if you search among the questions answered here.

I hope this helps.

Thank you very much. It's a very big help for me.

Hard to say without any information.... What is your sample? What are your experimental conditions? In general a basic model (one exponential) is selected and made more complicated (two-exponential, three-exponential) until one obtains random residuals.

If a priori knowledge is known about your sample then a more complicated model can be selected that's more appropriate. If its your first time using the instrument, measure with mono-exponential standards to build up experience with the system before trying a more difficult sample.

Compare the residuals of popop to a mono-exponential fit to the likes of BSA which requires a three-exponential fit (or more complicated model).

My sample is CdSe quantum dots (520 nm), it is a fluorescent material （Results like the attached file）. The pulsed laser excitation is 405 nm, and all of measure data is from TCSPC system.

But that seems like you will just choose the more complicated modal, then you can get the better fitting results?（because the more random residuals you can get）

It's seems like it will lose the physical meaning just the math...

Or there is some misunderstanding?

The more exponentials you fit, the more parameters so the better the residuals... it is possible to fit pretty much anything with a 5 exponental decay. The results as guessed are often rather meaningless...

As a rule of thumb fit the minimum number possible until you get an acceptable fit. Quantum dots however may exhibit a more complicated decay than a single exponential however especially if multiple species are formed and if there is energy transfer. etc etc.

Since you are using a 405 laser excitation source make sure you use polarizers... a vertically orienated polarizer on the excitation and a magic angle orientated polarizer on the emission otherwise you will measure polarization artifacts. You should test your system using Coumarin 6 in ethanol which should fit a mono-exponential decay under your experimental conditions.

If you are unsure of the model. Sometimes more data is required to justify the use of a multi-exponential model. Measurement wise this can be in the form of a time-resolved emission spectra (decays with respect to emission wavelength to form a spectrum). e.g. if you have a two exponential fit at the peak wavelength one component may be  very strong and the other very weak so a 1 exponential decay may appear to fit well. A two-exponential fit may do slightly better but you may be unsure of it... You would be however more sure of it if the second component is significanty stronger at a higher or lower emission wavelength. You might also want to look at the steady-state excitation emission matrix to see if steady-state data reports multiple peaks. Also kinetic measurements. See the link below for some more details on these more advancd techniques.

You may also look at alteration of the samples local environment, e.g. temperature ramps etc. etc.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4600163/

Thank you very much, it's a big help for me.

There is another question I want to ask (like attached picture), what is the exponential modal *tailfit* meaning in this picture? （This picture is from the Global Fluorescence Decay Data Analysis Software）

I have known the reconvolution process will minus the instrument signal, it can get the original signal of your sample ideally from Jose Hodak.

So what is the *tailfit* meaning?

The use of reconvolution analysis or tail fitting depends on the IRF and the decay times being investigated...

There is an instrumental response IRF which depends on the source, detector, timing electronics and optics. To measure the IRF we use Rayleigh scattering. Rayleigh scattering takes place within the fs-ps regime and is significantly faster than the response of the instrument. When measuring the IRF the emission monochromator is set to the same wavelength as the excitation source and both polarizers are set to vertical. A dilute Ludox colloid is typically used (again to maximise Rayleigh scattering and to reduce fluorescence impurities from displaying in the IRF). ND filters are used to attenuate the stop/start ratio to

Hi, can anyone help me to plot TRPL graph. I have my data. i don't know how to plot the graph. Please anyone help me to do it?. I can share my data also. For one system calculation part if anyone teach me I will do for rest of the sample.

Dear sangeetha,

Initially, select the columns from you data sheet (Ch, prompt and decay values from data sheet) put these values into excel sheet / origin graph before graphing convert channels in to nano seconds in your case 1.20456E-10 per channel.

Hi, can anyone tell me the difference between ExpDec 2 and ExpDecay 2? The lifetimes fitting by these two model are totally different? But the Adj. R-Square values are the same. Which model should I chose?

Min Zeng

Hi, the difference is that ExpDec 2 (I assume you're talking about the Origin program fitting functions here) is a regular 2-phase exponential decay fit meanwhile ExpDecay 2 have an additional time offset for more detailed explanations, the help tools of origin available here ( https://www.originlab.com/doc/Origin-Help/ExpDec2-FitFunc and https://www.originlab.com/doc/Origin-Help/ExpDecay2-FitFunc ) will probably give you all the required explanation.

What I usually do is cut the data before and use ExpDec 1 or 2 depending on the needs.

Hello,

Suppose, the curve is fitted with bi-exponential function which gives two decay constants (in semiconductors, one corresponds to trap-assisted recombination and other one corresponds to radiative recombination), but if the curve fits using single exponential function then what will be the recombination process for that decay constant?