Hello.

Without seeing the data, it is not possible to determine which mechanism applies. While I know that some have identified mechanisms based on R2, this is generally not the appropriate way to assign a mechanism.

There are a few reasons for this.

1) The length of time the study occurs is critical. For instance, if the data are not collected for long enough times, you can get linear plots from more than one equation. For instance, a first order and zero order are both linear at early times, but the sloes and intercepts will have completely different meanings. (If you don't know the mechanism, for instance, you may need to go well beyond one or two half lives to identify and verify a first order mechanism.)

As another example, consider M = M0 * exp(-kt) (first order) vs. M = A * sqrt(b*t) (Higuchi). It is clear that plotting M vs. sqrt(t) in the second case is linear. However, you can also get linear plots of M0*exp(-kt) vs. sqrt(t) with high R2 values. if you set k = 1 and plot M0*exp(-kt) vs. sqrt(t) for times from 0-2 (arbitrary units), it is linear over sqrt(t) = 0.5 to 1.4 (t=0.25 to 2) with R2 > 0.99. If you set k = 0.5, it is linear (R2 > 0.99) from sqrt(t) from 0.5 to 2 (t=0.25 to 4).

Thus, the linear range of the first order vs. sqrt(t) gets longer as k gets smaller. For a reaction or process with a short half life, R2 may discriminate. But for a slower process or reaction, such as release, the k may be small enough so R2 cannot tell the difference between Higuchi and first order at all.

There are too many of these examples to list here, but I hope this makes the point. By the way, this is mathematically a consequence of the Taylor series expansions for the functions, some of which may look similar enough in the earl terms.

2) Another factor to consider is the meaning of such small differences in your R2 values. In light of experimental error, these differences are too small to be reliable indicators.

The best practice is to design experiments that are long enough or otherwise have an ability to deviate from each model. The deviation from a model by experimental data is the proper way to rule out a mechanism.

Hope this helps!

Robert A Bellantone Thank you very much. Could you please recommend any documents or papers about the last point?

Best regards.

This might help.

Srushti Tambe Thank you very much.

• R2 zero order= 0.9908
• R2 first order = 0.9794
• From the data i am able to understand that you plotted drug release profiles using cumulative percentage drug release versus time and the regression value is 0.9908-this is drawn to know if the release kinetics follows zero order
• Similarly when you plotted Log cumulative percentage drug release versus time, your regression value is 0.9794. this is drawn to know if the release kinetics follows first order
• Now comparing two values, Which ever value is nearer to 1.is the order followed.
• So 0.9908 is near to 1. So the release kinetics is said to be zero order mechanism.
• Similarly if you want to know if the release is by diffusion or dissolution mechanism you can plot Higuchi plot and Hixson-crowell plot.

Raja Rajeswari Kamisetti Thank you very much for your answer.

n result of korsmeyer is not logical, since if it confirms for zero it would be close to 1. Maybe you need to repeat it with Ddsolver.