1: \begin{abstract}

2: We study phase-ordering dynamics 

3: of a ferromagnetic system with a scalar order-parameter on fractal 

4: graphs. We propose a scaling approach, inspired by

5: renormalization group ideas, where a crossover between distinct

6: dynamical behaviors is induced by the presence of a length

7: $\lambda $ associated to the topological properties of the

8: graph. The transition between the early and the asymptotic 

9: stage is observed when the typical size $L(t)$ of the growing 

10: ordered domains reaches the crossover length $\lambda $. We consider

11: two classes of inhomogeneous 

12: substrates, with different  activated processes, where the effects of the free energy barriers can

13: be analytically controlled during the evolution. On finitely ramified graphs the free energy

14: barriers encountered by domains walls grow logarithmically 

15: with $L(t)$ while they

16: increase as a power-law on all the other structures.

17: This produces different asymptotic growth laws (power-laws

18: vs logarithmic) and different dependence of the crossover length

19: $\lambda $ on the model parameters. Our theoretical picture agrees very well with

20: extensive numerical simulations.

21: 

22: 

23: 

24: \end{abstract}