The first question in tool selection is not "which tool is best" but how carbon fiber actually wears a tool out. The answer is abrasive wear: hard fibers continuously scrape the cutting edge and grind it away layer by layer. That explains something counterintuitive: cutting temperatures in CFRP edge trimming are only 180–350°C, far below the hot-hardness limit of HSS (still above 55 HRC at 600°C). HSS fails on carbon fiber not because it softens, but because it is ground away. A more heat-resistant tool won't help; a harder one will.
The One-Minute Summary
- Burr end mills (multi-flute, staggered teeth) are the mainstream choice for CFRP edge trimming, not PCD. The literature reports that multi-flute diamond-coated tools give a surface finish better than PCD, and that PCD is "not cost-effective."
- Reliable measured tool life: PCD reached 28,000 mm of cutting length in CFRP slotting without flank wear exceeding 0.3mm; micro-grain carbide with a 4 µm TiAlN coating can reach 40–50 m.
- Vendor claims of "10–20x tool life" deserve a discount: most do not define the baseline or the operation.
- Real list prices span a huge range: in the 6 mm class, chipbreaker/burr end mills run about USD $28–$218, CVD diamond-coated about $181–$216, and PCD from $542 to $1,873.
- Compression end mills are only suited to profiling; in slotting and plunging, the advantage of the opposing up/down spirals is completely lost.
- On CFRP, tools wear by cutting edge rounding (CER), not classic flank wear, so "keep using it until VB reaches 0.3mm" is the wrong criterion for CFRP.
Why HSS Can't Cut Carbon Fiber
Let's clear up a common misconception first. Most people assume HSS fails on composites because it "gets too hot and softens." Put three numbers side by side and you can see that is not the case:
| Item | Value | Source type | Interpretation |
|---|---|---|---|
| Cutting temperature in CFRP edge trimming | 180 – 350 °C | Peer-reviewed literature | Actual working temperature |
| Onset of carbide oxidation | > 400 °C | Peer-reviewed literature | Not reached |
| HSS hot hardness | > 55 HRC at 600 °C | Peer-reviewed literature | Far from reached |
| This conclusion is our inference from the three data points above, not a direct statement from any single source: the working temperature range in CFRP edge trimming is below the oxidation threshold of carbide and far below the hot-hardness limit of HSS. Therefore, HSS does not fail by thermal softening. | |||
The real mechanism is abrasive wear (abrasion). Multiple peer-reviewed studies agree that this is the dominant wear mechanism in CFRP machining: "wear caused by the presence of hard fibers is the dominant wear mechanism," and "tool abrasion, the rubbing of hard surfaces against the flank and rake faces, emerges as the main problem when drilling FRP composites."
We could not find any reliable source to support this common claim. The figure that circulates online, "carbon fiber has a Mohs hardness of 2–3," comes from forums and content farms and does not meet our citation standard. What reliable sources actually say is that "the high elastic modulus of CFRP makes it highly abrasive" (standard-modulus carbon fiber is 230 GPa; high-modulus fiber reaches 588 GPa), and that the material is removed as dust, more like a sanding operation than chip formation.
So the correct picture is not a hardness contest but a huge number of fine, hard particles scraping the edge continuously, like sharpening a knife on sandpaper.
The Geometry of Four Tool Types
Burr End Mills / Staggered-Tooth Burrs (Diamond Cut)
This is the most common geometry for CFRP edge trimming. The principle is to break the continuous cutting edge into teeth and overlap them strategically, so that every point along the edge falls in the path of one or more teeth. The teeth do two jobs at once: they cut the substrate and evacuate chips, and they concentrate cutting force enough to easily shear and fracture any fibers they meet.
Structurally, it differs from a standard end mill in three obvious ways: a thicker core, more flutes (up to 6–12), and shallower flutes. The shallow flutes buy structural strength, at the cost of feed rates lower than those of a standard end mill of the same diameter.
Cross-section: how a burr end mill's staggered teeth shear fibers
Left: a continuous edge pushes fibers aside. Right: staggered teeth concentrate cutting force at the tooth tips and shear the fibers.
CVD Diamond-Coated Carbide
A polycrystalline diamond coating is grown on a carbide substrate by chemical vapor deposition, with thicknesses ranging from sub-micron to tens of microns (Harvey Tool's composite tool coating is 4 µm; MAPAL claims up to 50 µm).
Its advantage over PCD is that it can be applied evenly to complex geometries. Shapes such as drills and multi-flute burrs cannot be made with brazed PCD inserts, but they can be CVD diamond coated. As one tool maker puts it: "When you need an extremely sharp edge, nothing beats PCD. But in many other cases, CVD diamond coating is a real alternative."
PCD (Polycrystalline Diamond)
A PCD tool consists of a PCD insert brazed onto a carbide body. Its strengths are excellent wear resistance and the lowest cutting temperature. Its weaknesses are just as clear:
- Lowest bending strength: the literature describes it as having "good surface roughness, but the lowest bending strength"
- High cost: one study states directly that PCD is "not cost-effective due to its high cost"
- Few flutes: usually 1–3 flutes, less efficient than a multi-flute burr
Compression (Down-Up Spiral)
The upper section has a left-hand spiral (pushing chips down) and the lower section a right-hand spiral (pulling chips up). The two opposing spirals compress the material inward in the middle, which prevents burrs, tearing, and delamination on both the top and bottom faces.
Tool makers state explicitly that compression end mills are suited only to profiling, because "the benefits of up-cut/down-cut geometry are completely lost in slotting or plunging operations." If your toolpath is slotting or plunging straight down, the extra money for a compression tool is wasted.
The Maker's Chart: Onsrud's GOOD / BETTER / BEST
Rather than guessing, look at what the tool maker itself recommends. Onsrud's catalog gives a clear three-tier tool recommendation for CFRP:
| CFRP operation | GOOD | BETTER | BEST | Depth of cut | Chip-load adjustment |
|---|---|---|---|---|---|
| Finishing | — | 66-700 (DFC low-helix multi-flute) | 68-000 (PCD 2-flute) | 1 × D | Use recommended value |
| Semi-finishing | — | 66-775 (DFC rougher/finisher) | 68-200 (PCD SERF) | 2 × D | Reduce 25% |
| Roughing | 66-900 (coated composite end mill) | 66-500 (DFC multi-flute burr) | 68-300 (PCD SERFIN) | 3 × D | Reduce 50% |
| DFC = Diamond Film Coating. Flute counts: 66-500 has 6 flutes at 1/8″, 8 at 3/16″, and 10 at 1/4″; 66-700 has 6 flutes at 1/4″, 8 at 3/8″, and 10 at 1/2″. Notably, Onsrud itself says the 66-700 multi-flute design "cuts quieter and faster in carbon fiber than typical two- or three-flute PCD tools." The maker's BEST is PCD, but it also admits that multi-flute coated tools are better in some respects. | |||||
Tool Life: Measured Data vs. Vendor Claims
This section has to keep two kinds of information apart. Peer-reviewed data with clearly defined test conditions and vendor marketing claims carry completely different levels of credibility.
| Comparison | Quantified result | Conditions | Source |
|---|---|---|---|
| PCD vs. DLC-coated carbide | PCD reached 28,000 mm of cutting length with VB under 0.3mm | CFRP slotting | El-Hofy et al. (2011) |
| DLC-coated carbide | All wore rapidly | Same as above | Same as above |
| Micro-grain carbide + 4 µm TiAlN | Poor cutting appeared only after 40–50 m of cutting length | CFRP edge trimming, multi-flute tool | López de Lacalle et al. (2009) |
| CVD-coated vs. uncoated (drilling) | CVD flank wear was less than half that of uncoated | CFRP drilling | Guo et al. (2024) |
| CVD-coated vs. uncoated (turning) | 3–5x tool life | Turning | Same as above |
| Uncoated carbide burr | After 7.2 m of trimming, coupon flexural strength dropped 10.5% (p=0.029) | CFRP edge trimming, 6 mm 8-tooth | Ashworth et al. (2022) |
| CVD diamond-coated burr | No drop in coupon strength within 16.2 m | Same as above | Same as above |
| Note that DLC (diamond-like carbon) and CVD polycrystalline diamond are different materials. El-Hofy's negative result applies to DLC and must not be applied to CVD diamond coatings. | |||
| Claimed by | Claim | Credibility note |
|---|---|---|
| Sandvik Coromant | PCD gives 10–20x the life of uncoated carbide | Operation and material undefined |
| LMT Onsrud | Coatings can extend tool life 10–20x; new geometries can cut machining time by up to 90% | Baseline undefined |
| Telcon Diamond | PCD countersinks average 1,500 holes (CFRP) vs. about 60 holes for carbide | Specific operation, but self-reported by the vendor |
| RobbJack | 25x the life of carbide in carbon fiber | Baseline undefined |
| Amana Tool | PCD tools last up to 50x longer than HSS or carbide tools | Range too wide and baseline vague |
| These numbers are not necessarily false, but without a baseline (compared to which tool? in which operation? judged by which criterion?) they cannot be verified. In practice it is more reasonable to read them as "right direction, questionable multiplier." | ||
Real List Prices: A Range of Nearly 300x
CFRP tools: real list prices on a log scale
List prices from actual product pages, 3–6mm class (USD, captured September 3, 2026)
How to Tell a Tool Is Dull: VB Is a Poor Indicator on CFRP
Metal machining judges tool life by flank wear (VB). That criterion breaks down on CFRP, because CFRP wears tools in a different way:
"Wear of the staggered PCD cutter is mainly concentrated in the cutting edge area. As milling length increases, the cutting edge radius gradually increases under the abrasive wear mechanism, while flank wear is not obvious."
— Chen, Wang, Gao & Liu (2017), Applied Sciences 7:934
| Indicator | How to judge | Reliability |
|---|---|---|
| Cutting length | Track the cumulative length of profiles cut; change the tool at a set value | Most practical; allows preventive tool changes |
| Sudden drop in edge quality | Burrs and protruding fibers suddenly increase | Intuitive, but by then it is too late |
| Rising cutting force | Spindle current or sound changes noticeably | Needs a baseline for comparison |
| Cutting edge rounding (CER) | Inspect the edge under a microscope for dulling | Most accurate, but requires equipment |
| Flank wear VB | Measure the width of the flank wear land | Not obvious on CFRP; easy to misjudge the tool as still usable |
| Tool makers' advice is blunt: once a tool starts to dull, cut quality drops quickly. Use preventive tool changes on CFRP; don't wait until you see signs of wear. | ||
Practical Tool Selection
| Scenario | Recommended tool | Reference price | Reason |
|---|---|---|---|
| Entry level / occasional cutting | Uncoated burr end mill, 4 or 6 mm | $6 – $30 | Low unit price, just change often; but note it affects finished-part strength |
| Everyday R&D prototyping | CVD/DFC diamond-coated multi-flute burr | $66 – $218 | Best balance of tool life and quality; the mainstream choice |
| Structural parts, strength-critical | CVD diamond-coated (required) | $181 – $216 | An uncoated tool already cut coupon strength 10.5% after 7.2m |
| Low-volume production | CVD-coated multi-flute, or PCD SERFIN | $200 – $1,900 | Amortize over batch size; PCD only pays off at high repeat volume |
| Narrow slots, inside corners R2 or smaller | 2 mm reduced-neck multi-flute tool | $6 – $60 | Tool diameter is dictated by the smallest inside radius; not a matter of preference |
| Outside-profile side milling only | Compression | Depends on spec | Clean top and bottom faces at once; no benefit in slotting |
Comparing tools on unit price alone leads to bad decisions. A better measure is cost per part. One vendor has published a countersink example: carbide at $35 initial tool price and $0.67 per hole; PCD at $80 initial tool price and $0.033 per hole. The initial price is 2.3 times higher, but the cost per hole is only 1/20.
However, these are the vendor's own numbers, and they are for drilling, not milling. What you should really do is log your own cutting lengths and tool-change points and calculate with real numbers, rather than copying someone else's multiplier.
FAQ
Is a higher flute count always better on a multi-flute tool?
No. More flutes mean more cuts per unit time and a lower load per tooth, which helps on CFRP. But more flutes also mean shallower flutes and weaker chip evacuation. Tool makers state plainly that composite end mills run "feed rates lower than expected for the same diameter and depth of cut, because the flutes are much shallower." In practice, 4–6 mm composite tools typically have 6–10 flutes. That is the balance point between geometry and chip evacuation; there is no need to chase more.
Why do tool makers give feed per revolution instead of feed per tooth?
Multi-flute burr tools have irregular cutting edges and a high flute count, so calculating with feed per tooth gives misleading results. Onsrud even notes that some series should be "treated as a single flute" for feed calculations. In practice, multiply feed per revolution by spindle speed to get mm/min, and do not multiply by the flute count again, or your feed will be set several times too high.
PCD is so wear-resistant. Why does the literature say it is not cost-effective?
Three reasons. First, cost: PCD runs 6–10 times the price of a carbide tool. Second, low flute count: PCD tools usually have 1–3 flutes, which is less efficient than a 6–10 flute burr tool, and tests show that multi-flute coated tools can even produce a better surface finish than PCD. Third, PCD has the lowest bending strength, which is a real risk with the small tool diameters used on desktop machines. PCD's sweet spot is high-volume, long continuous runs, not R&D prototyping.
Can diamond-coated tools be reground?
Not recommended. Regrinding removes the coating and exposes the carbide substrate underneath, which effectively turns it back into an uncoated tool, and it also changes the edge geometry. PCD tools, on the other hand, can be reground (a vendor claims up to 4,500 holes in total after two regrinds) because the PCD insert itself has thickness (0.010″–0.030″). Multi-flute burr tools have complex tooth forms, so regrinding them is not practical either.
How do I know how often to change tools?
The most practical method is to keep your own record of cutting length. Your CAM software shows the total toolpath length of each file; add these up. Reference points from the literature: uncoated carbide burr tools had already reduced coupon strength by 10.5% after 7.2 meters; micro-grain carbide with a 4 µm TiAlN coating can reach 40–50 meters; PCD can reach 28 meters in slotting without VB exceeding the limit. Start with conservative numbers, watch for when edge quality begins to degrade, and then calibrate your tool-change interval from there.
Sources
- Geier, Xu, Poór, Dege & Davim (2023), A review on advanced cutting tools and technologies for edge trimming of CFRP composites, Composites Part B 266:111037 (review of tool types, PCD weaknesses, cutting edge radius/CER), accessed September 3, 2026 — https://real.mtak.hu/206251/1/review%20edge%20trimming.pdf
- LMT Onsrud, Production Cutting Tools Catalog PCT-19 (CFRP GOOD/BETTER/BEST selection chart, flute counts and depth-of-cut rules by series), accessed September 3, 2026 — https://downloads.ctfassets.net/tzrd35jgpyko/6YTPDsAQyv5rfGbKTsrgx0/73607bc77179e97064561fad38085cfe/LMT_Onsrud_Production_Cutting_Tools__Catalog_PCT-19.pdf
- PreciseBits, Chipbreakers and Diamond-Cuts (staggered-tooth geometry; design trade-off of a thick core and shallow flutes), accessed September 3, 2026 — https://www.precisebits.com/tutorials/chipbreakers_and_diamond-cuts.htm
- Ashworth, Fairclough, Meredith, Takikawa & Kerrigan (2022), Wear 498-499:204340 (effect of coating and tool wear on coupon flexural strength), accessed September 3, 2026 — https://wrap.warwick.ac.uk/id/eprint/164926/1/WRAP-Effects-tool-coating-tool-wear-surface-flexural-strength-slotted-2022.pdf
- Guo et al. (2024), CVD diamond processing tools: A review, Journal of Advanced Research 74:333-358 (CVD coating thickness, 3–5x tool life in turning, half the wear in CFRP drilling) — https://pmc.ncbi.nlm.nih.gov/articles/PMC12302480/
- Chen, Wang, Gao & Liu (2017), Experimental Study on Milling CFRP with Staggered PCD Cutter, Applied Sciences 7:934 (wear concentrated at the cutting edge; flank wear not obvious) — https://www.mdpi.com/2076-3417/7/9/934
- Feito, Díaz-Álvarez, Díaz-Álvarez, Cantero & Miguélez (2014), Materials 7(6):4258 (abrasive wear as the dominant mechanism) — https://pmc.ncbi.nlm.nih.gov/articles/PMC5455907/
- Harvey Performance, Carbon Fiber Reinforced Polymers (CFRP) (high elastic modulus makes it highly abrasive; material removed as dust; heat cannot be carried away by chips), accessed September 3, 2026 — https://www.harveyperformance.com/in-the-loupe/carbon-fiber-reinforced-polymers-cfrp/
- Harvey Tool, End Mills for Composites — Compression Cutter (compression tools suited only to profiling; no benefit in slotting or plunging), accessed September 3, 2026 — https://www.harveytool.com/products/end-mills-for-composites---compression-cutter
- CompositesWorld, Hole Ambitions: Optimize, Customize (Sandvik and Onsrud tool-life claims; PCD at 6–10x the price of carbide), accessed September 3, 2026 — https://www.compositesworld.com/articles/hole-ambitions-optimize-customize
The Right Tool Needs a Spindle and Collet to Match
Carbon-CNC machines come standard with an ER11 collet and a 20,000 rpm BLDC brushless spindle, and accept composite end mills with 4 mm and 6 mm shanks. To discuss tooling and tool-change intervals for your parts, contact us.
Get a Quote