A titanium part does not need to exceed its yield strength to crack. In a peer-reviewed surface-finish study on Ti-6Al-4V, the same alloy gave a fatigue limit of 596 MPa when polished and 90 MPa after abusive grinding — a 6.6× spread created by the last operation on the part, not by the melt, the grade or the certificate (Salamina et al., Table 4, rotating bend, R = -1, 107 cycles). In commercially pure Grade 2, a published rotating-bend limit of 232 MPa drops to 154 MPa with a Kt = 3 notch (TIMET TIMETAL 50A datasheet), so a single shoulder radius can remove a third of the usable life. The highest-strength material on the shelf is therefore not automatically the safest choice: a badly finished Grade 5 part can sit below a well-finished Grade 2 part. Boundary: every number in this article is a laboratory result at a stated stress ratio, cycle count and surface condition. None is a design allowable for your geometry or medium. Ask your supplier for coupon data in your own section, finish and environment. Start with our grade equivalence guide if you need the chemistry behind the grade names.
How to read the numbers. We are a stockholder and light fabricator in Baoji, not a fatigue laboratory or an airworthiness authority. Each value below is quoted from a named source with its test basis attached; where a source gives a range, the range is shown rather than an average. Published fatigue limits depend strongly on stress ratio, specimen geometry, surface roughness and residual stress, so values from different tests should be compared only where the basis is stated. No number here is a substitute for a design curve, a damage-tolerance assessment or a fatigue test on your part. Where public sources do not give a single value for a medium, we say so and tell you what to request instead of inventing one.
| Factor | Published value or boundary | Mechanism | What to freeze on the RFQ |
|---|---|---|---|
| Surface finish, Gr5 | 596 MPa polished / 90 MPa abusive grind | Roughness + tensile residual stress | Ra / Rz and finishing route on the drawing |
| Notch, Gr2 | 232 MPa smooth / 154 MPa at Kt=3 | Local stress concentration | Fillet radius and thread profile |
| Grade swap | Not a fatigue upgrade by itself | Strength and notch sensitivity move together | Finish first, then grade |
| Alpha case | ~70 µm layer promoted crack initiation | Brittle oxygen-rich surface | Removal method + depth acceptance |
| In-service oxygen layer | Distinct from furnace alpha case | Diffusion without beta denudation | State temperature and exposure hours |
| Hard alpha / LDI | Bulk chemistry can stay compliant | Local hard inclusion, sub-surface initiation | VAR melt history + NDT detection floor |
| Weld toe | Geometry and tint both matter | Toe radius + gas pickup | Colour acceptance and purge record |
| Cut / laser edge | ~235 to 363 MPa after polishing | Edge notches and HAZ | Edge finishing after cutting |
| Corrosion fatigue | No single public design value | Film rupture + cyclic load | Coupon data in your actual medium |
How to use this table: each row is a question your quotation should answer before a price is agreed. A supplier who can discuss all nine without checking is usually the one who has actually made fatigue-critical parts. For the commercial side of the same conversation, see our quote-comparison guide.
Fatigue is a surface-initiated process in most engineering alloys, and titanium is unforgiving here because it does not tolerate a damaged surface the way a ductile low-strength steel might. A crack starts at the roughest groove, the sharpest corner, the hardest inclusion or the most oxygen-rich layer it can find. Everything upstream — the melt, the grade, the heat treatment — matters, but it can be erased by the final machining step.
The practical rule is to treat finish as a specified property, not a cosmetic one. If you buy Grade 5 plate or bar for aerospace fasteners and then leave “machine to drawing” undefined, you have left the largest fatigue variable to whoever runs the last shift. That is why this article is organised around the places cracks actually start rather than around a list of grade properties.
The clearest published demonstration comes from a study compiled by Salamina and co-workers on Ti-6Al-4V. Rotating-bend specimens at R = -1 were given different surface treatments and tested to 107 cycles. The chemistry, the melt and the baseline heat treatment were effectively the same; the surface was not.
| Last operation | Fatigue limit | Reduction factor Kf | Reading |
|---|---|---|---|
| Polished, 320–600 alumina grit | 596 MPa | — | Reference condition |
| Gentle peripheral end milling | 455 MPa | — | Sharp tool, light finish cut |
| Gentle surface grinding | 427 MPa | 0.94 | Controlled wheel and feed |
| Chemical milling | 352 MPa | 0.77 | Etched surface, no mechanical notches |
| Abusive peripheral end milling | 220 MPa | 0.48 | Worn or overloaded cutting action |
| Abusive surface grinding | 90 MPa | 0.20 | Severe surface damage |
| Belted and glass-bead blasted | 610 MPa | 1.02 | Smooth + peening benefit |
| Belted, beaded, shot-peened, grit blasted | 555 MPa | 0.93 | Compressive surface stress |
The spread from 596 MPa to 90 MPa is larger than the difference between most titanium grades. Note also that the reduction factor for abusive grinding was 0.20: four fifths of the smooth-specimen fatigue limit had gone before the part ever entered service. If your supplier cannot state how the final surface will be produced, the number on the certificate describes a different component from the one in your hand.
It is tempting to solve a fatigue problem by moving from Grade 2 to Grade 5. The tensile numbers support the move: Grade 2 has a minimum 0.2 per cent yield of 275 MPa, while annealed Grade 5 is normally specified above 800 MPa. Fatigue does not follow that line, because strength and notch sensitivity tend to rise together.
| Condition | Published fatigue limit | Ratio to tensile | Source and basis |
|---|---|---|---|
| Grade 2, smooth Kt=1 | 232 MPa | 0.56 | TIMET TIMETAL 50A, rotating bend, 107 cycles |
| Grade 2, notched Kt=3 | 154 MPa | 0.37 | Same datasheet, same test basis |
| Grade 5, polished | 596 MPa | — | Salamina et al. Table 4, rotating bend, R = -1, 107 cycles |
| Grade 5, abusively ground | 90 MPa | 0.20 | Same study, same alloy, different last operation |
| Read-out | Finish can outweigh grade | — | Compare only within a stated test basis |
Read the last two rows together. A polished Grade 5 specimen reaches 596 MPa; an abusively ground one reaches 90 MPa, which is below the 232 MPa smooth limit published for Grade 2. The grade did not change — the last operation did. This is why we tell buyers to fix the surface specification before upgrading the alloy, and why the Grade 5 versus Grade 23 selection guide and the equivalence guide should be read as chemistry references rather than fatigue answers.
The fastest way to make suppliers quote on one basis is to send the same surface and notch fields to every one of them. These are the lines we ask for; copy them into your enquiry.
|
What to send us for an exact titanium quote Copy these nine lines into your enquiry. A complete spec means a firm price on the first reply, not a second email asking for the grade. |
||||||||||||||||||||
|
||||||||||||||||||||
|
Ready? Copy the nine lines above into the Requirement Details box on our Request a Quotation page and we will come back with a firm price and a dispatch date within 24 hours. |
The TIMET datasheet for TIMETAL 50A, the producer’s equivalent of ASTM Grade 2, publishes two rotating-bend fatigue limits at 107 cycles: about 232 MPa for a smooth specimen with Kt = 1 and about 154 MPa with Kt = 3. The fatigue ratio against tensile strength falls from 0.56 to 0.37. A shoulder, a thread root or a badly blended weld toe produces exactly that kind of local concentration.
The design consequence is uncomfortable but simple. You cannot buy your way out of a sharp corner with a better grade, because the stress multiplier acts on whatever material you install. If a shaft keeps failing at the same step, change the radius or add a transition before you change the alloy. Our fastener receiving guide and weld qualification page both treat geometry as an inspection item for this reason.
Titanium’s affinity for oxygen is useful in service and dangerous in the furnace. Two different layers get called alpha case in general conversation, and separating them is worth doing because the controls differ.
| Layer or study | Condition | What was observed | Buyer action |
| Furnace alpha case | Processing heat, roughly 850–1000°C | Beta-denuded, hard oxygen-stabilised alpha | Remove by pickling or grinding |
| In-service oxygen-rich layer | Lower service temperature, up to about 600°C | Hard and oxygen-rich without microstructural change | Control time, temperature and atmosphere |
| Near-alpha TIMETAL 834 | Oxygen diffusion reported above ~480°C | Fatigue crack formation over thousands of hours near 600°C | Alloy-specific; do not generalise to CP grades |
| Measured alpha-case effect | ~70 ±3 µm layer in L-PBF Ti-6Al-4V | Promoted crack initiation and reduced fatigue life | Thinner layers showed a smaller effect |
| Weld heat tint | Same oxygen/nitrogen reaction at smaller scale | Surface initiation site in cyclic service | Purge pool, HAZ and back side; accept by colour |
A 2024 study on laser powder bed fusion Ti-6Al-4V found that a 70 ±3 µm alpha-case layer promoted crack initiation and reduced fatigue life, while thinner layers had much less effect. A separate study of the near-alpha alloy TIMETAL 834, used in compressor discs, reports oxygen diffusion above roughly 480°C and enough embrittlement near 600°C to promote cracks under cyclic loading over thousands of hours. That second number is alloy-specific: do not paste it onto a Grade 2 vessel. Our failure casebook covers the fabrication side of the same problem; this section is about what happens afterwards under load.
Hard alpha, also called a low-density inclusion, is an oxygen- or nitrogen-enriched hard zone formed during melting. It is especially associated with rotor-grade Ti-6Al-4V because a sub-surface initiation site is exactly what a surface finish specification cannot fix.
| Observation | What it means | How to respond |
|---|---|---|
| Hard alpha / low-density inclusion | Oxygen- or nitrogen-enriched hard zone from melting | Fracture surface, not bulk certificate |
| FAA/SwRI study, 1999 | Surface-touching synthetic defects cracked at ~5–10 ksi (35–70 MPa) static | Do not treat as a design number; it is a defect-sensitivity observation |
| Internal anomaly | Transducers saw little at 100 ksi; sectioning showed extensive cracking | NDT has a detection floor |
| L-DED repair study, 2026 | Oxygen-rich defects missed by bulk chemistry and X-ray CT | Found only by fractography |
| Buyer action | Trace melt route and agree NDT method and rejection size | Certificate alone is not evidence of a clean local volume |
FAA-sponsored work at Southwest Research Institute in 1999 made synthetic hard-alpha defects of two sizes in Ti-6Al-4V and loaded them statically and in fatigue. Surface-touching defects cracked at relatively low applied stress, typically 5 to 10 ksi, about 35 to 70 MPa. That figure is a defect-sensitivity observation, not a design value. More recent work on laser-directed energy deposition repair found oxygen-rich inclusions that bulk chemistry and X-ray computed tomography both missed; fractography identified them. The message for a buyer is that a compliant EN 10204 3.1 certificate is necessary but not sufficient for a fatigue-critical lot. Ask about melt route and the detection floor of the inspection.
A welded joint brings two fatigue initiators together in one place: a geometric toe that behaves like a notch, and a heat-affected surface that can pick up oxygen and nitrogen if the shielding was poor. Neither shows up in a room-temperature tensile test of the parent plate.
| Weld factor | Why it matters in cyclic service | Control |
|---|---|---|
| Weld toe geometry | Sharp toe acts as a notch | Blend and inspect the toe on fatigue-critical joints |
| Heat tint | Oxygen and nitrogen pickup embrittles the surface | Shield pool, HAZ and back side until cool |
| Colour acceptance | Silver / light straw acceptable; blue or grey rejected | Write the acceptance tint into the procedure |
| Back purging | Root side is exposed without it | Record purge gas purity and flow |
| Post-weld machining | Removes tint and restores a defined surface | Sequence before final inspection |
The colour of a titanium weld is a chemistry signal, not a cosmetic detail. Silver or light straw is normally accepted; blue, purple or grey means the surface picked up oxygen and nitrogen and should be removed and re-welded under proper purge. Our failure casebook sets out the colour scale and the back-purge rule in detail, and the AWS A5.16 welding wire page covers filler selection. For a fatigue-loaded weld, add toe blending and tint acceptance to the drawing rather than leaving them to the shop.
Laser cutting, waterjet and punching all leave an edge with a different roughness and, for thermal processes, a heat-affected zone. In a study of non-optimised laser-cut Ti-6Al-4V ELI sheet for medical use, as-cut specimens gave a fatigue strength of 235 MPa at 2 × 106 cycles. Barrel grinding raised that to 351 MPa and mechanical polishing to 363 MPa, roughly a 50 per cent improvement. Every crack initiated at an edge or corner.
The study also found that polishing removed the heat-affected zone almost completely, while barrel grinding left micro-notches in less accessible regions. The lesson transfers directly to thin Grade 5 sheet, spring strip and formed parts: specify the edge condition after cutting, not just the sheet condition before it. Our cold-drawn spring wire page and cut-to-size service both start from a declared stock surface so the starting point is known.
Titanium’s corrosion resistance comes from a thin, fast-reforming oxide film. Under cyclic stress, a surface notch or a tight crevice can rupture that film locally; if the medium and temperature allow the exposed metal to react before it repassivates, corrosion and fatigue accelerate each other. This is why a part that survives a static corrosion test can still fail early in a pumping or diving application.
There is no single public corrosion-fatigue design value for Grade 2 in seawater that we are willing to quote as a universal number. Published results depend on stress ratio, frequency, crevice geometry, dissolved oxidisers and temperature. Ask your supplier for coupon data in your actual medium and at your actual stress ratio, or write a fatigue test into the purchase order. The failure casebook explains the temperature and pH boundaries for crevice attack; the tube selection guide covers product form. For a general introduction to the corrosion behaviour itself, the China sourcing guide lists the medium fields a quotation should contain.
If your drawing says only “Grade 5, annealed” for a cyclic-loaded part, send it through our request a quotation page. A named engineer will return the finish, notch, weld and inspection wording in writing — within 24 hours — so the suppliers on your list are quoting the same component.
Naming the cases where a fatigue limit is the wrong instrument is part of a straight technical answer. A supplier who sells a fatigue number for every application is selling the wrong thing in several of them.
| Situation | Why fatigue is not the main lever | What to specify instead |
|---|---|---|
| Static or one-off load | Fatigue limit is the wrong criterion | Specify tensile, yield and proof instead |
| Low-cycle plastic strain | High-cycle limit does not apply | Request low-cycle data for your strain range |
| Elevated temperature | Creep and oxygen ingress may control | State temperature and hours; see the heat-treatment guide |
| General corrosion thinning | Corrosion rate sets the life | Choose by medium with coupon data |
| Fixed sharp geometry | A better grade cannot remove Kt | Change the radius or add a transition |
| Medical implant | Fatigue is only one of several release tests | Quote to the named ASTM / ISO route and revision |
Outside this table, surface integrity is a normal and important specification field. Inside it, another property controls and the fatigue conversation follows. The same distinction applies across our nickel alloy family, where a fatigue question is sometimes a corrosion question wearing different clothes.
Twelve lines make a fatigue-critical part quotable and inspectable. Send them together so the answers come back on the same basis. The last line is the one most often omitted, and it is the one that tells you whether a quoted fatigue number means anything.
| # | Field | Example entry |
|---|---|---|
| 1 | Grade, UNS and product standard | Gr5 UNS R56400 / ASTM B348, revision stated |
| 2 | Condition | Mill annealed / STA / cold-worked |
| 3 | Final machining route | Mill, grind, EDM, laser — name the last operation |
| 4 | Surface roughness | Ra 0.8 µm max, measured on the loaded face |
| 5 | Tooling condition | Fresh cutting edge; no abusive grinding |
| 6 | Residual stress treatment | Shot peening with agreed Almen intensity and coverage |
| 7 | Notch geometry | Fillet radius, thread profile, UNJ where required |
| 8 | Weld acceptance | Colour scale, toe blend, purge record |
| 9 | Alpha-case control | Removal method and maximum remaining depth |
| 10 | Melt route | VAR cycles, heat number, ingot traceability |
| 11 | NDT method and rejection size | Method, coverage and smallest reportable indication |
| 12 | Fatigue data basis | Stress ratio, cycles, surface, medium; coupon on request |
Two lines deserve emphasis. First, name the last operation: “finish grind, Ra 0.8 µm max, fresh wheel” is a requirement; “machine to drawing” is not. Second, agree the NDT detection floor. A method with a stated rejection size is a control; a method with no stated floor is a reassurance. These are the same fields we ask for on flange, fittings and fastener enquiries.
Choose commercially pure Grade 2 if the medium is the hard part: seawater, chlorides, oxidising acids and most process piping. It is the material to beat on corrosion, it welds readily, and a smooth finish gives a published rotating-bend limit of about 232 MPa at 107 cycles. Start from our Grade 2 plate page or welded versus seamless guide.
Choose Grade 5 if the load is the hard part and you are prepared to control the surface: a polished specimen reaches 596 MPa in the same test basis, but only if the machining route, notch radius and inspection are written down. Aerospace and medical routes also bring melt traceability and specification revisions that CP stock does not normally carry. See Grade 5 bar stock and Ti-6Al-4V stock.
Choose neither on fatigue grounds alone. If the part has a sharp shoulder, an abusive grind or an uninspected weld toe, the grade decision will not save it. Fix the geometry and the surface first, then pick the material for the medium and the load. That sequence is cheaper than a redesign after the first field failure.
We are a stockholder plus light fabricator in Baoji, the centre of China’s Titanium Valley. Fatigue-critical questions are mostly stock and surface questions for us:
Start with plate, tube, bar, wire, flange or fittings; each page lists the forms, sizes and condition options that matter for that product.
|
Get a firm titanium quote within 24 hours Same-day dispatch from ~800 t of Baoji stock · cut-to-size and waterjet blanks · free drawing review by our engineering team · EN 10204 3.1 MTC with every lot. |
Request my quote →
No obligation. A named engineer replies, not an auto-responder. |
|
EN 10204 3.1 MTC on every lot · ISO 9001 quality system · ASTM / ASME / GB standard literacy · Third-party inspection (SGS / BV) on request · Mill heat & lot traceability · Export packing & full documentation |
Not by itself. Published rotating-bend data at 10 to the power of 7 cycles puts polished Grade 5 far above smooth Grade 2, but the same Grade 5 study records a fatigue limit of only 90 MPa after abusive grinding, below the 232 MPa smooth limit published for Grade 2. Surface condition, notch geometry and inclusion cleanliness can overturn the grade ranking. Specify the finish and the notch before you pay for a higher grade.
The TIMET TIMETAL 50A datasheet, equivalent to ASTM Grade 2, lists a rotating-bend fatigue limit at 10 to the power of 7 cycles of about 232 MPa for a smooth specimen with Kt = 1 and about 154 MPa for a notched specimen with Kt = 3. Those are laboratory results at one stress ratio and one surface state. They are not a design allowable for a different geometry, temperature or environment.
A peer-reviewed study on Ti-6Al-4V measured 596 MPa after polishing with 320 to 600 alumina grit and 90 MPa after abusive grinding, both at 10 to the power of 7 cycles under rotating bend with R = -1. Gentle milling and grinding sat in between, and bead blasting with peening preserved or slightly improved the polished value. The mechanism is surface roughness acting as notches plus the residual stress state left by the last operation.
Alpha case is a hard, oxygen-enriched surface layer formed when titanium is held at high processing temperature. A 2024 study on laser powder bed fusion Ti-6Al-4V found that a layer about 70 micrometres thick promoted crack initiation and reduced fatigue life, while thinner layers had a smaller effect. It is different from the oxygen-rich layer that can form slowly in service. Both are controlled by atmosphere, time and temperature, and furnace alpha case is removed by pickling or grinding before the part enters fatigue service.
No. A mill certificate reports bulk chemistry, and hard alpha is a local oxygen- or nitrogen-enriched inclusion. FAA-sponsored work in 1999 showed surface-touching synthetic hard-alpha defects cracking at relatively low applied stress, and later work has reported oxygen-rich defects that bulk composition analysis and X-ray computed tomography both missed, found only by fractography. For fatigue-critical parts, trace the melt route and agree the NDT method, coverage and rejection size with your supplier.
It can. In the same Ti-6Al-4V surface study, a belted and glass-bead blasted condition reached 610 MPa against 596 MPa for the polished reference, and a peened and grit-blasted condition reached 555 MPa. Peening helps by putting the surface into compression and by smoothing asperities, but it is not a repair for deep notches, alpha case or internal inclusions. Specify the process by Almen intensity and coverage rather than by the word peening alone.
Name the grade, UNS number and product standard with revision, then the condition. Add the last machining operation, the required roughness on the loaded face, the fillet radius or thread profile, and how the weld toe will be finished if there is a weld. For critical parts add alpha-case removal, melt-route traceability, the NDT method with its rejection size, and the basis on which any fatigue number was measured. Ask for coupon data in your own medium rather than accepting a generic catalogue value.
The last operation on a titanium part can move its fatigue limit by more than the grade change can. The same Ti-6Al-4V measured 596 MPa polished and 90 MPa after abusive grinding; a Grade 2 notch reduced the published rotating-bend limit from 232 MPa to 154 MPa. Alpha case, hard-alpha inclusions, weld toes and cut edges are initiation sites that a bulk certificate cannot rule out. Write the finishing route, roughness, notch radius, weld acceptance, melt traceability and NDT detection floor into the RFQ, and ask for coupon data in your own medium. Send the twelve-line checklist through our request a quotation page; a named engineer will return the reconciled wording within 24 hours.
Need a quote on this specification?
Click below and your email app opens with the 6-point RFQ already typed out. Just fill in your grade, size and quantity - no form, no captcha.
Prefer the web form? Open the inquiry page and paste the 6 points into the message box.
Baoji Titanium Valley · ~800 t stock · cut-to-size · drawing-based machining · ASTM / ASME / GB · export-ready · same-day quote.
Contact Person: Wang Xiaoning
Tel: +86-917-355-7966
Email: [email protected]
WhatsApp: +86 188 2967 9311
Company: Shaanxi Qianyi Tuoda Technology Co., Ltd
Address: Floor 4, Building 4, No. 188 Gaoxin Avenue, High-tech Zone, Baoji City, Shaanxi Province, China