Fixed a bug where the SEI thickness decreased at some intervals when …

…using the 'electron-migration limited' model. (#3622)

* The 'electron-migration limited' model has been corrected by considering only negative contributions to eta_sei in the j_sei

* Update pybamm/models/submodels/interface/sei/sei_growth.py

Co-authored-by: Eric G. Kratz <kratman@users.noreply.github.com>

* Logical condition (eta_inner < 0) used instead pybamm.EqualHeaviside for more clarity

---------

Co-authored-by: Eric G. Kratz <kratman@users.noreply.github.com>

pybamm/models/submodels/interface/sei/sei_growth.py

@@ -134,8 +134,7 @@ def get_coupled_variables(self, variables):
         elif SEI_option == "electron-migration limited":
             # Scott Marquis thesis (eq. 5.94)
             eta_inner = delta_phi - phase_param.U_inner
-            j_sei = phase_param.kappa_inner * eta_inner / L_sei_inner
-
+            j_sei = (eta_inner < 0) * phase_param.kappa_inner * eta_inner / L_sei_inner
         elif SEI_option == "interstitial-diffusion limited":
             # Scott Marquis thesis (eq. 5.96)
             j_sei = -(