Smearing, pull-out and edge rounding are preparation faults that look like results. How mechanical sectioning, broad beam ion polishing and focused ion beam work compare, and which to use for a coating, a composite or a single device.
A coating fails on a component, the section goes under the microscope, and the interface looks diffuse. Everyone starts theorising about interdiffusion. The interface is sharp; the polishing smeared the softer of the two materials across the harder one, and the diffuse layer is entirely an artefact of preparation. A cross section that has been prepared badly does not look obviously wrong. It looks like a result.
The mechanical route is fast, cheap and familiar, and for a homogeneous metal or a coarse ceramic it is entirely adequate. Getting it right is a matter of discipline at each of four stages.
Section with a precision saw and a blade matched to the material, under coolant, at a feed slow enough not to heat the specimen. A hacksaw or an abrasive cut-off wheel work hardens the region you are about to look at, and polishing does not remove damage that runs deep.
Mount with the failure mode in mind. Hot compression mounting is quick but applies heat and pressure that distorts polymers, softens low melting metals and closes cracks you were trying to image. Cold curing resin under vacuum impregnates porous or fragile material so that it survives grinding. For edge retention on a coating, use a hard filled resin or plate the surface with a sacrificial metal layer, so that the polish sees two hard materials rather than a hard one against soft resin.
Grind through the grit sequence without skipping steps, and clean the sample, your gloves and the wheel between every step; most embedded abrasive arrives on the operator's fingers. Finish with diamond and then colloidal silica, which combines a fine mechanical action with a mild chemical one and removes much of the deformed layer.
A broad argon ion beam removes material atom by atom, so there is no mechanical force, no smearing and no pull-out. It handles every combination that polishing handles badly: soft metal against hard ceramic, porous bodies, multilayer coatings, filled polymers, and any sample where the interface itself is the subject.
Two arrangements are worth distinguishing. Flat milling treats an existing surface, cleaning off the deformed layer left by polishing and revealing grain structure and channelling contrast. Cross section milling clamps a mask plate over the sample and removes everything projecting past its edge, producing a face perpendicular to the surface that has never been touched mechanically. For a coating on a substrate, that face shows the true layer thicknesses and the true interface.
The costs are time and heat. Milling is slow, so the sensible workflow is mechanical preparation to bring the region of interest close, then ion milling to finish. The beam also heats the sample, which matters for polymers and low melting alloys; specify a cooled stage if the laboratory works on those.
Broad beam milling treats a region. A focused ion beam treats a point. If the question concerns one transistor, one defect found at a known coordinate, or one particle among thousands, the focused beam is the only route, because it can be positioned on that feature under electron imaging and cut a section through it.
Two practices separate a usable focused beam section from a wasted one. Deposit a protective cap of platinum or carbon over the region before milling, or the top surface rounds off and the section shows the vertical streaks known as curtaining. And step down to lower beam currents for the final cleaning cuts, particularly when thinning a lamella for transmission microscopy, because the implanted and amorphised layer left by a high current cut can be a large fraction of the lamella thickness.
For most laboratory problems the order of preference is straightforward. If the feature is somewhere in a large area, mechanically prepare and then ion polish. If it is at a known point and nothing else will do, use the focused beam and accept the time. If the sample is homogeneous and hard, mechanical polishing alone is fine and the rest is over-preparation.