In ion thrusters, the scoop grid is charged positively and "scoops" ions from the plasma. The accel grid is charged negatively and placed at some distance to accelerate the ions by the applied voltage.
Intuitively, I'd expect the holes in the accel-grid to be bigger, so ions that pass through the hole in the scoop-grid have a high chance of clearing the accel-grid. However, apparently the opposite is the case. The accel-grid has smaller holes than the scoop grid.
Why is that?
You can find the answer to your question in chapter 5 of the book by Dan M. Goebel and Ira Katz "Fundamentals of Electric Propulsion: Ion and Hall Thrusters".
The reasons behind different hole sizes is summarized in the paragraph quoted below, where the "screen" grid is just another term used to describe the "scoop" grid:
In actual design, the diameter of each accel grid aperture is minimized to retain unionized neutral gas in the plasma generator, and the screen grid transparency is maximized so that the grid extracts the maximum possible number of ions from the plasma. The electrode diameters and spacing are then optimized to eliminate direct interception of the beam ions on the accel grid, which would cause rapid erosion due to the high ion energy.
The screen grid is needed to prevent the erosion of the accel-grid due to the impact of the high speed ions, which miss the accel-grid holes and get intercepted by it. Aligning the holes of the screen grid with the holes of the accel-grid ensures that most of the ions, extracted by the screen grid, will hit the holes of the accel-grid.
To maximize the number of the extracted ions, the holes in the screen grid have to be relatively big. To minimize the loss of neutral atoms, which could drift out of the thruster at low speed without contributing much to the propulsion, the holes in the accel-grid have to be relatively small.
You can find much more details in section 5.1.
Though the previous answer is correct, I wanted to add a bit more detail for future reference:
First up, the grids should be thought of as cylinders in an ion optics accelerator system.
Secondly, the deformation of the plasma due to the fields of the grid seems to be important, though I can't yet tell how or why. I found figure 1 and the paragraph before it in this article enlightening. It also explains that ion-density and accelerating electric field need to be tuned to fit, otherwise the plasma bugles too much or too little. Furthermore, deliberate formation of the emissive plain[1] could increase efficiency.
[1]: doi: 10.1063/1.3115488
