fix(x, value) sets the lower and upper variable bounds of x to value.
@constraint(model, x == value) adds a new linear constraint where the left-hand side is the function 1.0 * x + 0 and the right-hand side is value.
In general, it’s better to use fix, since it’s one less constraint to deal with and solvers can efficiently use variable bounds.
It’s unusual that Ipopt struggles, but it’s hard to say why without a minimum working example.
What value is tatm₀?