I don't think the tick parameters are customizable between the top and bottom spines; note that ax.spines['top'].axis is the same object as ax.spines['bottom'].axis.

If you want a different spine in the middle of your figure with no labels, you could try using ax.secondary_xaxis (https://matplotlib.org/stable/api/_as_gen/matplotlib.axes.Axes.secondary_xaxis.html). It should be completely configurable independent of the main x-axis. The only issue is that I don't know that you can pin secondary_xaxis to a data value (eg. 0). versus a location on the axes (eg. 0.5)

Okay thanks a lot for clarifying. I could solve the problem completely like this:

import matplotlib.pyplot as plt


def get_location(ax, value, axis='x'):
    lb, ub = ax.get_ylim() if axis == 'y' else ax.get_xlim()
    assert lb <= value <= ub, f"value must be between {lb} and {ub}, found: {value}"
    return value - lb / (ub - lb)


fig, ax = plt.subplots()
ax.plot(range(2021, 2025),
        range(-1, 3))

ax.tick_params(labelbottom=True, bottom=False, top=False)
ax2 = ax.secondary_xaxis(location=get_location(ax=ax, value=0, axis='y'))
ax2.tick_params(labelbottom=False, direction='in', length=10)
ax.set_xticks(range(2021, 2025))

plt.show()

Closing the issue.

Slight follow up in #25119 - would be nice to be able to place the secondary axes in data co-ordinates