Breakage is a system failure, not bad luck
When an end mill fails, the temptation is to grade the tool - good box, bad box. Sometimes that verdict is fair; more often the tool was asked to do something its geometry, its holder or its program could not support. Solid carbide is stiff and hard but not tough: it forgives a great deal of heat and very little overload. That asymmetry shapes the whole subject. Almost every cause below is a way of asking carbide to carry a load or a heat it was never specified for, and almost every fix is a way of asking less. Reading breakage as a system failure - tool, holder, program, workpiece - turns a mystery into a checklist, and the checklist is what this note is.
Chip packing: the quiet breaker
In slots, deep pockets and any cut where the flutes run close to the walls, chips have nowhere to go but up the flute. When they pack - compacted into the gullet, recut between the edge and the wall - the tool has effectively no room to cut, and the load climbs until something fails. Aluminum is the classic offender because its chips are stringy and its tendency to weld to the flute makes packing worse; the aluminum selection notes build geometry around exactly this problem. The fixes are mechanical and plain: fewer flutes for more chip room in deep slots, air blast or flood to evict chips, ramping or pecking the depth rather than one straight plunge, and stepping over rather than pocketing full width when the depth is long. A tool that keeps breaking in slots is usually telling you about chip room, not about quality.
Engagement beyond the tool's stiffness
A carbide end mill is a slender column. Load it sideways past what its diameter and reach can carry and it deflects; deflect it far enough and the edge, meant to shear, takes a load in bending - and carbide snaps in bending. The recipe for this failure is always the same: a long stickout, a wide radial or deep axial engagement, and a feed that assumes the rigidity of a setup the shop does not actually have. The fixes follow directly. Shorten the overhang to the minimum the part allows - the single most effective change in most breakage post-mortems. Take the engagement in steps: lighter radial width, more passes. Move to a stub or a larger diameter where the drawing permits. And recalibrate the chip load to the setup you have, not the one the chart assumed - the adjustment logic is in the chip load guide.
Runout: one edge doing two jobs
A tool spinning with runout does not share the work. One flute takes an oversized chip every revolution while its neighbors skim, so that single edge sees load far beyond the calculated chip load and fails first - often quietly, as accelerated wear, and then suddenly, as a fracture that seems to come from nowhere. Runout comes from the holder, the seating, the collet condition or the tool itself, and the check costs one indicator reading at the cutting edge. The desk treats runout acceptance as a spec-sheet line checked at inspection - see the quality control page - because a batch with inconsistent runout produces exactly this pattern of mysterious field breakage. The mechanism and its costs get a full note in runout and tool life.
Entries that load the corners
How a tool enters material decides what its edges meet first. A straight plunge into solid stock demands a center-cutting tool - edges that meet at the middle of the face - and even then the corners take the initial load. A non-center-cutting tool plunged into solid material does not cut at all; it hammers, and the corners chip or the tool snaps. The distinction is the subject of the center cutting note, and the tools built for plunging live in the slot drill family. The gentler entries avoid the question: ramping into the cut at an angle, or helically interpolating down, lets the periphery do the work it is built for. Entries through scale, skin or a hard inclusion deserve special care in any material - a full-thickness bite through grit on the first rotation is an edge-shortening move no matter how good the tool.
Holder slip and pull-out
A tool that pulls out of the holder mid-cut has usually been run with too little clamp force, a contaminated seat, or a holder worn past its useful life. The failure is dramatic - the tool stops spinning with the spindle and something has to give - and it tends to recur until the holder is addressed. Clean seats, correct torque, tools seated fully to depth, and holders retired when their gripping surfaces are worn: none of this is sophisticated, and all of it shows up in the breakage statistics. The holders page covers selection and condition per holder family.
Thermal and material shocks
Two quieter mechanisms round out the list. Thermal cycling - coolant arriving in bursts on a hot carbide edge - cracks the edge through repeated expansion and contraction; the coolant note on this blog treats the steady-or-dry rule in full. And hard inclusions in the workpiece - scale, weld spatter, a hard spot in a casting - deliver a momentary load spike that a healthy setup shrugs off and a marginal one does not. Neither mechanism argues against the tool; both argue for knowing the conditions the tool will meet and writing them into the program.
Reading the broken edge
| Cause | What the edge shows afterward | Prevention |
|---|---|---|
| Chip packing | Compacted material in the flutes; edge failed mid-flute | Fewer flutes, air or flood, ramp or peck the depth |
| Overload beyond stiffness | Clean fracture; tool bent or deflected before failure | Shorter overhang, lighter engagement, stepped passes |
| Runout overload | One edge worn or fractured, neighbors intact | Indicator check, holder and seating discipline |
| Non-center-cutting plunge | Corners chipped; face undamaged | Center-cutting tool, or ramp or helical entry |
| Holder slip or pull-out | Shank scored; tool stopped with the spindle | Clean seats, correct torque, retire worn holders |
| Thermal cracking | Fine cracks across the edge; coolant present | Steady coolant or none - never intermittent |
The buyer's version of the question
When a customer reports breakage, the useful first questions are not about the box but about the cut: what material, what engagement, how long the reach, what the chips looked like, whether the entry was a plunge. The answers usually place the failure on the list above within minutes. The residual share that points back to the tool - inconsistent runout, edge prep that drifted from the approved sample, substrate inconsistency - is exactly what a disciplined spec sheet and inspection regime catches before shipment, which is the desk's side of the arrangement. If your customers machine aggressive profiles and you want the geometry chosen for toughness where toughness matters, say so through the quote form; corner radii, flute counts and substrate choices are program decisions, confirmed in writing like everything else.