Guangdong Juyao Intelligent Technology Co., Ltd. · Laser Software & Process R&D
23°07′N 113°15′E · GUANGZHOU · PURE STATIC MONOGRAPH
← Back to research
Material & Process2026-11-16· 9 min read

Adjustable-ring-mode scanning laser welding of TA1 titanium to 304 stainless steel: joints reach 93.7% of the filler strength at 800–1000 W ring power — then fall off a cliff

Butt-welding 0.5 mm TA1 titanium to 304 stainless steel with a 1.2 mm AZ61S magnesium filler wire: at 800–1000 W ring power or 100–200 Hz scanning frequency, every joint averaged over 80% of the filler strength, peaking at 281.2 MPa; at 1400 W ring power the strength fell to 117.4 MPa, and at 400 Hz scanning frequency to 106.5 MPa. The good process is not “bigger” — it is “just right”.

Bar charts of joint tensile strength versus process parameters (Fig. 5 of Li et al., Materials 2026, CC BY 4.0; resized only, white padding added to fit 16:9): (a) ring powers of 800, 1000, 1200 and 1400 W give 270.5, 244.1, 172.9 and 117.4 MPa; (b) scanning frequencies of 100, 200, 300 and 400 Hz give 273.4, 281.2, 172.9 and 106.5 MPa — strength falls off a cliff beyond the sweet spot
COVERBar charts of joint tensile strength versus process parameters (Fig. 5 of Li et al., Materials 2026, CC BY 4.0; resized only, white padding added to fit 16:9): (a) ring powers of 800, 1000, 1200 and 1400 W give 270.5, 244.1, 172.9 and 117.4 MPa; (b) scanning frequencies of 100, 200, 300 and 400 Hz give 273.4, 281.2, 172.9 and 106.5 MPa — strength falls off a cliff beyond the sweet spot
CORE RESEARCH QUESTION

Core question: for 0.5 mm titanium–steel dissimilar thin sheets, can a ring-mode laser with circular scanning produce a usable joint — and how wide is the “good window” for ring power and scanning frequency, and what happens beyond it?

ABSTRACT

Li and colleagues (Hubei Engineering University, Wuhan National Laboratory for Optoelectronics at Huazhong University of Science and Technology, Pangang Group Research Institute, Hubei University of Technology) butt-welded 0.5 mm TA1 titanium alloy to 304 stainless steel thin sheets with an adjustable-ring-mode (ARM) scanning laser, using AZ61S magnesium-alloy wire as filler under argon shielding. Central power was fixed at 800 W and one parameter was varied at a time: ring power from 800 to 1400 W, scanning frequency from 100 to 400 Hz. Weld-zone hardness stayed below both base metals, with no weld hardening; horizontal and axial bending tests all reached 108° without cracking. Tensile strength followed a single-peak shape — stable, then falling: inside the sweet spot every joint exceeded 80% of the filler tensile strength (about 300 MPa), peaking at an average of 281.2 MPa (93.7% of the filler); beyond the sweet spot it dropped sharply. Seven of eight tensile specimen sets fractured at the 304 stainless steel / filler interface. Mechanism analysis shows the ARM beam makes energy distribution along the weld path more uniform (trough-to-crest ratio up from 39.3% to 68.0%, weld width up 33.3%), but excessive ring power lets the trailing edge remelt and break the freshly formed FeAl interface layer, dissolving iron into the pool and thickening the layer; above about 275 Hz the molten-pool flow turns from laminar to turbulent, scouring the interface and weakening the joint the same way. This note checks every process parameter and key figure, reconstructs the “narrow sweet spot, excess backfires” conclusion, and states where it stops applying. Cover: Fig. 5 of the paper (CC BY 4.0, resized only).

01 / The difficulty first: titanium–steel thin sheet is hard at the interface and at the keyhole

Titanium alloys offer high specific strength, corrosion resistance and high-temperature performance; stainless steel is cheap and easy to process — joining the two into one structural part is a standing wish in aerospace, automotive and marine engineering. But direct fusion welding of titanium to steel forms brittle Fe–Ti intermetallics, and the joint fractures at the interface under load; worse, the sheets here are only 0.5 mm thick, and the keyhole effect of a conventional Gaussian laser is inherently deep-penetration welding, with energy concentrated in a single central spot — on thin sheet that readily produces penetration fluctuations, undercut, burn-through and porosity, leaving a narrow process window. Solid-phase welding and brazing each carry their own cost: explosion welding is hard to control and leaves large residual stress; diffusion welding is slow and environmentally demanding; friction stir welding is inefficient, and the stirring pin drags iron into the titanium side to form alternating Fe/Ti layers; brazing is constrained by gap width and vacuum environment, with strength sensitive to the gap.

The paper’s combination stacks two mature techniques: an adjustable-ring-mode (ARM) laser — a Gaussian central spot wrapped in an independently adjustable annular beam, giving a far more uniform energy distribution than a single Gaussian beam — plus circular scanning, which swings the spot in small circles along the weld path to homogenize energy further and stir the molten pool. ARM has already been shown in aluminium welding to suppress spatter and porosity and reduce deformation, but its use on titanium–steel dissimilar thin sheet, combined with scanning, had barely been reported. This paper’s contribution is to put the two together and answer systematically which parameters decide the joint’s mechanical properties, and where the window lies.

One design choice deserves a note: the filler is AZ61S magnesium-alloy wire (1.2 mm diameter), not titanium or steel. That is deliberate — magnesium melts at a low temperature and forms no high-melting-point brittle phases with either titanium or steel, acting as a “buffer layer” between two mutually incompatible metals. The price is direct: magnesium alloy is itself weak (about 300 MPa tensile strength), so the joint’s strength ceiling is fixed by the filler from the start. Every “percentage of the filler” figure that follows must be read on that basis: it measures “how well the interface is joined”, not “how strong the joint is in absolute terms”.

02 / The experiment: a ring-mode beam, circular scanning, and one magnesium wire

The base metals were 0.5 mm thick TA1 titanium alloy and 304 stainless steel, cut to 100 mm × 50 mm, butt-joined with a 0.4 mm gap and wire-filled under argon shielding. The laser was a DGLaser DGMM2000/2000-APP ARM system, with 2000 W maximum central power, 2000 W maximum ring power and 4000 W maximum total power; the wire feeder was a Fronius VR7000-CMT (0.5–22 m/min); motion came from a Fanuc six-degree-of-freedom robot with a scanner head, producing the circular scanning motion. Fixed parameters were central power 800 W, scanning amplitude 0.6 mm, welding speed 2.5 m/min, wire feed rate 6.0 m/min and defocus distance 0 mm; only two variables moved — ring power from 800 to 1400 W and scanning frequency from 100 to 400 Hz — one at a time, against baselines of 1400 W ring power and 300 Hz scanning frequency.

Mechanical testing followed standards, three specimens averaged per condition: Vickers hardness per ISO 6507-1:2023 at 200 g load, 15 s dwell and 0.2 mm point spacing, with a complete uncertainty evaluation per the ISO 6507 series; bending per ISO 5173:2023 with a 30 mm indenter on 100 mm × 8 mm specimens, horizontal and axial weld bending each tested, at 0.5 mm/min loading; tensile testing per ISO 4136:2022, also at 0.5 mm/min. Tensile fracture surfaces were examined by scanning electron microscopy with energy-dispersive spectroscopy for composition. With n = 3 the sample size is small, so the error bars must be read together with the values — the strength drops reported below are far larger than the error bars, so the conclusions do not hinge on sample size, but the specific numbers should still be read as “average behaviour inside this window”.

The clean variable control is the basis of this paper’s credibility: one laser, one fixed recipe, with ring power and scanning frequency as the only systematic changes. Table 3 lists the full variation ranges and baseline values, so the processing is reproducible. One caveat: the study covers only one sheet thickness (0.5 mm) and one filler (AZ61S), and that bounds where the conclusions apply; central power was fixed at 800 W, and how changing it would shift the sweet spot is not answered here.

03 / Results: a narrow sweet spot and a steep cliff

Hardness first: from the TA1 side to the 304 side, hardness runs “high on both sides, low in the middle” — TA1 about 149.6 HV, the heat-affected zone about 131 HV, the weld filler zone (AZ61S) about 96.7 HV, the other heat-affected zone about 297.5 HV, and 304 stainless steel about 310.4 HV (ring-power series; the scanning-frequency series is similar). Weld-zone hardness below both base metals means no weld hardening occurred. The literature comparison is telling: with copper filler the weld zone is harder than the base metals (Cu–Fe and Cu–Ti compounds), and with no filler it is harder still (Fe–Ti compounds, 310.2 ± 81.4 HV); with vanadium or magnesium filler the weld zone is softer. That is the point of choosing AZ61S — it avoids weld embrittlement and hardening, at the price of the weld itself becoming the softest link.

Bending performance was surprisingly good: with a 30 mm indenter radius, every specimen in both horizontal and axial weld bending reached 108° without cracking — the machine stopped automatically because the indenter interfered with the fixture, not because the specimen cracked. 108° is not the material’s bending limit but the machine’s travel limit; the real information is “nothing cracked under any parameter set”, which says the joint has ample plastic reserve and no visible brittle weakness at the weld or the interfaces.

Tensile strength is the paper’s main thread, and it is a “single-peak” picture (the cover, Fig. 5 of the paper). Ring-power series (scanning frequency fixed at 300 Hz): 270.5 MPa at 800 W, 244.1 MPa at 1000 W, 172.9 MPa at 1200 W, 117.4 MPa at 1400 W. Scanning-frequency series (ring power fixed at 1400 W): 273.4 MPa at 100 Hz, 281.2 MPa at 200 Hz, 172.9 MPa at 300 Hz, 106.5 MPa at 400 Hz. The sweet spot is ring power 800–1000 W or scanning frequency 100–200 Hz: inside it, every group averaged over 80% of the filler tensile strength (about 300 MPa, so about 240 MPa), peaking at 281.2 MPa — 93.7% of the filler. But just outside it comes the cliff: raising ring power from 1000 W to 1400 W more than halves the strength, and raising frequency from 200 Hz to 400 Hz cuts it by about 60%. Seven of eight tensile specimen sets fractured at the 304 stainless steel / filler interface, only one at the filler / TA1 interface — the latter an occasional incomplete-fusion case whose fracture was cleavage-dominated; the former were dimple-dominated with some cleavage, showing fracture picks the weakest interface rather than a random location.

04 / Mechanism: the FeAl interface layer decides everything

Why does fracture pick the 304 / filler interface? The paper gives a quantitative answer with a Poisson’s-ratio criterion: FeAl tends to form between 304 stainless steel and the magnesium filler, and Mg17Al12 between the magnesium and the titanium; computed from bulk and shear moduli, FeAl has a Poisson’s ratio of 0.20 and Mg17Al12 of 0.23, both below the 0.26 brittleness threshold — both are brittle phases, and FeAl is the more brittle. Under tension, cracks naturally start from the more brittle FeAl layer, matching the 7-of-8 fracture locations. The FeAl layer’s thickness therefore becomes a proxy for joint strength: the thicker the layer, the weaker the joint.

The ARM beam’s contribution shows first in the energy distribution. Using energy-density models for the central and ring beams, the authors computed the distribution along the weld path (Fig. 8 of the paper): with the central beam alone, the trough-to-crest energy-density ratio is only 39.3% (trough 11.2, crest 28.5 J/mm²) and the weld width 1.35 mm; adding the ring beam raises the ratio to 68.0% (trough 150, crest 220.7 J/mm²) and the width to 1.80 mm — uniformity up about 74% and a substantially larger molten pool. The ring beam’s leading edge preheats (raising the temperature and laser absorptivity of the base metal ahead), while its trailing edge holds the pool temperature and remelts. The problem lies exactly in that trailing-edge remelting: at excessive ring power it repeatedly remelts and breaks the just-solidified FeAl interface layer, opening micro-gaps through which iron atoms from the stainless surface keep dissolving into the pool; the pool’s iron concentration rises, the FeAl layer grows thicker, and strength falls. That is the direct cause of the cliff beyond 1000 W ring power — more ring power is not more energy, it is more interface damage.

The scanning-frequency cliff runs on a different, fluid-mechanical mechanism. Treating the molten pool as a stirred tank and the fast-scanning keyhole as the agitator, the agitation Reynolds number Re scales with the square of the scanning frequency; using the magnesium pool’s density and viscosity, Re stays at or below 1000 (laminar flow) up to 275 Hz and turns turbulent above it. A turbulent pool scours the freshly formed solid–liquid interface far more violently, again breaking the FeAl layer, dissolving iron into the pool and thickening the interface — which shows up in tensile strength as a sharp drop at frequencies of 300 Hz and above. Both cliffs converge on the interface layer: whether ring power or frequency is too high, excessive agitation damages the interface and thickens the brittle layer. The physical meaning of the sweet spot is therefore clear: energy uniform enough and the pool calm enough to join the interface fully, without breaking it apart again.

05 / Lessons for process software — and the boundaries

Three lessons carry directly into laser-processing software design. First, ring power and scanning frequency belong in the parameter space as first-class process parameters, and their process windows have upper bounds — neither parameter here is “more is better”, and performance falls off a cliff past a threshold. A parameter recommender that only sets lower bounds, or assumes “more power means a stronger weld”, will be systematically wrong here; a process window must state its upper bound and its failure mechanism together. Second, interface-layer thickness is an earlier signal than strength: the FeAl layer thickens before strength falls, so a quality criterion in software can use intermediate quantities such as interface-layer thickness or iron diffusion as an online-monitoring proxy, instead of learning about it only after a tensile test breaks. Third, energy-distribution uniformity itself is worth quantifying: the trough-to-crest ratio rising from 39.3% to 68.0% is ARM’s hardest gain over a conventional Gaussian beam, and software comparing beam modes can use such distribution metrics rather than total power alone.

The cross-literature comparison is also worth keeping for filler selection in software: vanadium filler (2 mm sheet) averaged 355 MPa, 83.5% of TA1; copper filler (0.6 mm copper sheet) reached 330 MPa, 89.2% of copper; AZ31B magnesium wire with a conventional laser reached 221 MPa, 92.1% of the wire; this study’s AZ61S with ARM scanning laser reached 281.2 MPa, 93.7% of the wire. Measured as “joint strength over weakest-link strength”, the ARM scanning laser does improve on the conventional laser — but the absolute value must be seen alongside: 281.2 MPa is only about 66% of TA1 (425 MPa), because the ceiling is the magnesium filler itself. Choosing a filler trades “interface friendliness” against “absolute strength”, and software recommending a filler should show the user that trade-off rather than reporting a percentage alone.

Boundary statement (consistent with the paper; please preserve it when citing). Licence and version: the paper is open access under CC BY 4.0; this note is an original interpretation with short quotations and does not reproduce the paper’s data tables; the cover is Fig. 5 of the paper, resized only (white padding added to fit 16:9), with source and licence in the caption. Experimental boundaries: 0.5 mm TA1/304 sheet only, AZ61S as the only filler, and only inside the window with 800 W central power and a fixed welding speed of 2.5 m/min; no fatigue, corrosion, impact or long-service testing; other thicknesses, other fillers and no-filler conditions were not tested. Measurement boundaries: n = 3 per tensile and bending group, with strength as the three-specimen average and error bars in Fig. 5 of the paper; the 108° bending angle is the testing machine’s travel limit, not the material’s bending limit; FeAl layer thickness is a mechanistic interpretation, not a directly measured layer thickness; the agitation Reynolds number is estimated from magnesium-pool properties and is a model inference. JUYAO has not replicated this on its own equipment. This is a research interpretation, not a process-recipe commitment.

PRIMARY SOURCE DOSSIER

Paper Title
Study on Mechanical Properties of Adjustable-Ring-Mode Laser Scanning Welding of TA1 Titanium Alloy to 304 Stainless Steel Dissimilar Thin Sheets
Authors
Geng Li, Tengyi Yu, Peiqing Yang, Suning Zhao, Shuai Zhang, Honghua Ma, Shang Wu, Ji Li, Ming Gao
Journal
Materials, 2026, Vol. 19, 230
Published
2026-01-07
DOI
10.3390/ma19020230
Licence
Creative Commons Attribution 4.0 International (CC BY 4.0)
Original URL
Open original paper ↗