I think the concept that will help you shed the light on this is "fictive temperature". The structure of a glass forming liquid at some temperature is in equilibrium. At a different temperature, the equilibrium structure will be a little bit different. As you cool the glass forming liquid, you shift the temperature, and with it, the equilibrium, and the structure tries to keep up. The faster you change the temperature, the more out of equilibrium your structure is. This effect is significant enough that you can see different values of Tg for different cooling rates, and you can "freeze in" different glass structures. Hope this helps.

The glass transition has to be seen as any physico chemical transformation. It has a change in heat capacity (Cpx, where x is the degree of advance of one (or several) physico  chemical transformation occuring in the product), plus a thermal effect of this transformation, and a (or several) kinetics, in the direction glass to liquid and liquid to glass. We made experiments using heating rates between +/- 2K/mi, down to. +/-  0.1 K/min and Tg=f(Heating rate) gives a nice curve . See Thermochim. Acta  1997,293,1-11

Much earlier we did experiments in step-heating on warming or cooling. it was much before  Perkins Elmer apparatus. See Themochim. Acta 1988,128,251,251-260; Heating (or cooling) rate was about 3 K/hour.

Since liquid -> glass -> liquid as many times as you want, then it must be seen as an equilibrium.

I know that this simple view of Tg is not accepted, but do experiments, and  have conclusions.

Good work, be lucky. I am intersested  to discuss with you, as much as you want.

Pierre Claudy

Dear Shaunak,

I am not sure 100% I understand the issue.

Let me briefly remind here the basic formula for DSC:

Heat Flow = sample mass * specific heat capacity of the sample * heating rate + (other terms, related to structural changes, i.e. enthalpy changes * heating rate)

It means that a) you see the glass transition as a step on DSC because the specific heat capacity of the glassy and the rubbery state of the sample changes and b) the step height is proportional to the heating rate.

PS: smart of you to measure the Tg also in cooling ;-)

Hope this helps, Samuele

Ive not done any work with glycerol but is it possible that it is partially crystallising during the cool?  The slower cooling rate will allow more crystallisation and therefore your amorphous content will reduce and so will the Tg

@ Samuele Giani:

the proportionallity to the heating rate is correct for the sensor signal but this would only be observed in a plot over measurement time but if you plot it, as it is typically done, over temperature this will not have an effect as time at a defined temperature is inversely proportional to the heating rate,or?

@Matthias Jaunich:

This is the example I had in mind. http://www.mt.com/ch/en/home/supportive_content/matchar_apps/MatChar_HB804.html

Check out page 22 of the attached PDF for the Tg example. Usually one expect a slightly shift in Tg temperature, i.e. 3-5 degC per heating rate decade change.

I am referring to the Y-axis, while you are talking about the abscissa representation.

Moreover (Catherine Kelly exmple), if there is an overlap of crystallization and glass transition, applying a different cooling rate might separate the two phenomena on the thermogram. This since crystallization might be (partially)suppressed at high cooling rates or shifting differently compared to the Tg.

Crystallization is a slow process. If the cooling is faster there is defect and the packing of molecules/polymer is lower. This allows enhanced void volume and the Tg shifts downward. At lower cooling rate the packing is slow and perfection is higher and  Tg  also is higher.

I suggest - as others have alluded - that you're encountering a kinetic effect. It takes time for the glass transition to occur - the fast the heating rate, the higher the observed Tg will be due to the thermal lag of the material. Conversely, the higher the cooling rate, the lower the Tg will appear to be if observing on a cooling cycle. There are mathematical models you can apply to find the "true" Tg (i.e., for a heating rate of 0, what would the Tg be?) and to predict Tg at different heating rates - same as you can apply similar kinetic models to other physicochemical processes. 

For a discussion on the effect of the cooling rate (from the melt) on the glass transition temperature, you may want to see: C.A. Queiroz, J. Šesták, "Aspects of the non-crystalline state", Physics and Chemistry of Glasses - European Journal of Glass Science and Technology Part B, 51(3) June 2010, 165-172:

Article Aspects of the non-crystalline state