A talk that explores the physics connection to crystalline cohomology in the context of string theory is Motives and Strings by Jan Stienstra, with a challenge for each community:
Why look for a MOTIVE-STRING relation?
Computations in Type IIb string theory proceed by manipulating solutions of certain differential equations. During the computations there are many denominators. In the end these drop out and true integers remain.
Many differential equations in Type IIb string can be recognized as
Picard-Fuchs equations in De Rham cohomology of families of varieties.
The integrality statements can be recognized as consequences of theorems about the crystalline cohomology of families of ordinary varieties.
Challenge for Motive people: Crystalline cohomology deals with only
one prime $p$ at a time and puts out statements about $p$-adic
integrality. What mechanism synchronizes the primes and leads to true
integers?
Challenge for String people: Crystalline cohomology implies extra
symmetries in the differential equations. Where are these extra
symmetries in Nature?