Doesn't work this way. With an ideal transmission line model (infinite common mode impedance), the circuit would work as a phase splitter. But infinite or reasonable high common mode impedance isn't feasible unless you place a ferrite core (transmission line transformer).
L=100um suggests you want to implement the transmission line on-chip. But at 10 MHz, it's like zero electrical length.
In terms of circuit modelling, you need to add a second transmission line instance that represents the common mode (coax sheet) impedance.