The weld has its own chemistry - set by the filler, not the base metal. It has its own microstructure - refined dendrites from rapid solidification, not the wrought structure of the plate. It has its own failure modes - sigma phase in the heat-affected zone, nitride precipitation if the cooling rate runs away, ferrite imbalance if the filler chemistry drifts. The weld is a product. And if you treat it like a step in the welding procedure rather than a piece of material you are responsible for, your 2507 pipe will pass a hydrotest and fail in service.
Product Forms
Bar & Rod
Plate & Sheet
Coil & Strip
Pipe & Tube
Fitting: Flange, Tee, Elbow, Reducer etc.
Forging: Ring, Shaft, Circle, Block etc.
1. Why the Weld Is a Different Product
Every welded 2507 pipe has three microstructural zones: the base metal (BM), the heat-affected zone (HAZ), and the weld metal (WM). The base metal is the wrought plate or pipe as-rolled and solution-annealed - fine grain, balanced ferrite-austenite, PREN 42. The weld metal is cast - it solidified from liquid at the weld pool, with dendrites that are 100% ferrite at the liquidus and only re-form austenite as the weld cools. The HAZ is the narrow band beside the weld that did not melt but did see temperatures from 600 °C up to the solidus and back down - and that thermal cycle is what creates the metallurgical problems 2507 welds are famous for.
The practical consequence: the filler metal you choose sets the weld metal chemistry. The base metal chemistry is fixed. And if the weld metal chemistry does not match what the HAZ microstructure needs to stay balanced, the weld passes the WPS qualification and fails in service. The most common version of this failure is weld pitting at 80 °C seawater that the base metal would have handled - same pipe, same service, but the weld is the weak link.
2. The Four Filler Choices and What Each Does to the Weld
For welding 2507 pipe, the filler metal is not optional. It is the difference between a weld that lasts 25 years and a weld that pits in the second overhaul. There are four candidate fillers, and only two of them are correct.
| Filler | AWS Class | PREN_W (vs Base 42) | Ni Content | When It Is Right | When It Is Wrong |
|---|---|---|---|---|---|
| ER2594 | A5.9 | ~ 44 (overmatched) | 9.5 – 10.5% | Default - subsea, sour, all critical service | Nothing; ER2594 is the safe default |
| ER2553 | A5.9 | ~ 40 (matched) | 8.5 – 9.5% | Less aggressive service, shop fabrication only | Subsea, sour H2S above 0.05 psi, chloride above 30,000 ppm |
| ER2209 | A5.9 | ~ 36 (under-matched) | 8.0 – 9.0% | 2205 base metal - NOT 2507 base | 2507 base metal (every time) |
| ER309LMo | A5.9 | ~ 26 (wrong alloy) | 12.0 – 14.0% | Nothing for 2507 | Every time - wrong Cr, wrong Mo, wrong N |
The numbers that matter are PREN_W (calculated the same way as the base metal - %Cr + 3.3 × %Mo + 16 × %N, no tungsten correction for these fillers) and the nickel content. The nickel sets the austenite reformation during cooling - too low, and the weld stays ferritic (brittle, low corrosion resistance); too high, and the weld goes austenitic (loses the strength advantage the ferrite phase provides).
ER2594 is over-matched on purpose. Its nitrogen content is 0.25 – 0.30%, higher than the 0.28% in the base metal, because some nitrogen is lost to the atmosphere during the transfer across the arc. The extra nitrogen compensates for that loss and ensures the final weld deposit retains a ferrite number in the 35–65% range. ER2553 is matched (same nitrogen as the base metal). ER2209 is the correct filler for 2205 base metal but under-matched for 2507 - using it on 2507 gives a weld with ferrite above 65% and Cr2N nitride risk. ER309LMo is a stainless filler for joining stainless to carbon steel - it has no nitrogen, no molybdenum balance for duplex, and no business being used on 2507. Yet it shows up on shop floors because it is in stock.
The single number to verify on delivery: PREN_W of the filler, on the MTC. If it is below 40, you do not have ER2594 or ER2553. You have a wrong filler.
3. The Heat Input Window - Why "Weld It Like 316L" Breaks the Weld
The 2507 heat input window is 0.5 to 2.5 kJ/mm, and both ends of that range matter. The low end prevents Cr2N nitride precipitation in the HAZ (a risk when the weld cools through 800 °C too quickly for austenite to reform). The high end prevents sigma phase precipitation in the HAZ (a risk when the weld cools through 600–1000 °C too slowly). The 0.5–2.5 kJ/mm band is the one that hits both targets - fast enough to skip the sigma nose, slow enough to reform austenite.
A fabricator who is used to welding 316L will push the heat input up to 1.5–2.0 kJ/mm and call it a good weld. For 316L, that is fine. For 2507, that is the top of the safe band - and any small variation (a long arc, a slow travel speed, a thick section that holds heat) pushes the weld into the sigma-phase territory. The WPS-qualified band is 0.5 to 2.5 kJ/mm, but the production set-point should be 0.8–1.5 kJ/mm to leave margin for the inevitable small variations.
The other window that matters is interpass temperature. For 2507, max 150 °C - and ideally below 100 °C for subsea service. Above 150 °C, the previous bead stays hot long enough for the HAZ to accumulate thermal cycles, and the cumulative heat input shifts the ferrite balance.
The practical rule: write the WPS to a tight production window, not the wide qualified range. If the WPS says "0.5–2.5 kJ/mm," the welder will hit 2.5 when conditions are bad. If the WPS says "0.8–1.5 kJ/mm," the welder has to actively work to leave the band.
The HAZ Microstructure Map - What Happens in the Three Zones
Inside every 2507 weld, the HAZ is not one thing. It is a continuous gradient of microstructures, indexed by peak temperature. The three sub-zones that matter:
| HAZ Sub-zone | Peak Temperature | Microstructure | Failure Risk |
|---|---|---|---|
| High-temperature HAZ (HT-HAZ) | 1250 – 1350 °C (near solidus) | Excessive grain growth; ferrite near 100% on rapid cool; some austenite reformation | Lowest corrosion resistance in the weld joint; ferrite-PREN effectively 28–32 here |
| Mid-temperature HAZ | 950 – 1200 °C | Partial grain refinement; ferrite-austenite balance closer to 50/50 | Acceptable for most service |
| Low-temperature HAZ (LT-HAZ) | 600 – 950 °C | Sigma-phase precipitation risk if held 60 – 600 seconds | Embrittlement; loss of Charpy impact; pitting susceptibility |
The HT-HAZ is the silent failure zone. After welding, it looks balanced under the microscope - but its corrosion resistance is measurably lower than the base metal because the large-grained ferrite phase has a lower effective PREN than the wrought base. In chloride service above 40 °C, the HT-HAZ is where pitting starts.
The LT-HAZ is the slow-embrittlement zone. Sigma phase takes 60 seconds to nucleate in 2507 at 850 °C - which sounds fast, but it is slow compared to weld cooling rates. If the interpass temperature runs above 150 °C and the section is thick enough to hold heat, the LT-HAZ can accumulate sigma over multiple thermal cycles. The Charpy impact at –46 °C drops from 200 J in the base metal to 60–80 J in the sigma-bearing LT-HAZ. That is the difference between passing and failing NORSOK M-630 subsea qualification.
The weld metal itself (WM) starts at 100% ferrite at the liquidus (1600 °C) and reforms austenite during cooling through 1200 – 800 °C. If the cooling rate through this range is fast (high heat input + thick section = slow cool), austenite does not get enough time to reform and the weld stays tooferritic. If the cooling rate is very fast (low heat input + thin section = rapid cool), austenite reformation is suppressed and Cr2N nitrides form instead. Both extremes take the ferrite number out of the 35–65% target.
What the WPS controls: heat input (sets the cooling rate), interpass temperature (sets the cumulative thermal cycle), and preheat (usually not needed for 2507 below 50 mm wall; above 50 mm, max 100 °C preheat to drive off moisture).
What the WPS does not control: the HAZ microstructure that results. That is what the procedure qualification test (PQR) and ASTM A923 verification are for.
ASTM A923 - Three Methods to Verify the Weld You Got
ASTM A923 is the standard test method for detecting detrimental intermetallic phases in duplex stainless steels. It has three methods (A, B, C), and each one looks for a different problem:
| Method | Test | What It Catches | Pass Criterion | When Required |
|---|---|---|---|---|
| Method A | Etch in 40% NaOH, electrolytic at 1 – 3 V; examine at 10 – 30× for grain-boundary attack | Sigma and chi phase in the weld and HAZ | No continuous grain-boundary attack on any sample | All 2507 welds for sour service or chloride above 40 °C |
| Method B | Charpy V-notch impact at –46 °C on weld and HAZ specimens | Embrittlement from sigma, chi, or alpha-prime (475 °C) | ≥ 54 J average of three specimens on weld metal; ≥ 54 J on HAZ specimens | Subsea, cryogenic, sour service |
| Method C | ASTM G48 Method A pitting test at 40 °C, 24 hours, on weld coupon | Weld metal pitting resistance (PREN loss) | No pitting at 20× magnification; weight loss < 0.001 g/cm² | Sour H2S service, high-chloride subsea |
Method A is the cheapest and fastest - it can be done in a shop lab in two hours. It catches gross sigma-phase problems. If Method A shows grain-boundary attack, the weld is rejected and the WPS needs to be re-qualified.
Method B is the impact test - and it is the one most often cited as the failure mode. A weld that passes Method A may still fail Method B because the impact energy is sensitive to alpha-prime precipitation (the 475 °C embrittlement that forms over months in service). For subsea service, Method B is the non-negotiable check.
Method C is the corrosion test - it is the most expensive and slowest (24-hour soak + examination), but it is the only one that directly measures what the weld will do in chloride service. If the PREN_W of your filler was 36 (because you used ER2209 by mistake), Method C will catch it. Method A might miss it. Method B might miss it.
The practical rule for subsea and sour service: require all three Methods on every procedure qualification coupon. Require Method A and Method B on every production weld (or at least on a sampling basis per ASME Section IX). Skip Method C only if the service is non-corrosive and the weld is shop fabrication only.
PWHT - The One Treatment You Do Not Do on 2507
This is short, and it is the most violated rule in 2507 fabrication. Post-weld heat treatment (PWHT) in the 600 – 950 °C range precipitates sigma phase in 2507 welds and HAZ. It is the opposite of what you want. The weld needs to be cooled rapidly through that range, not held in it.
If your fabricator proposes PWHT for a 2507 weld - to relieve residual stress, to "stabilize" the microstructure, or because the same furnace is used for carbon steel and they have a cycle running - stop the order and re-qualify the WPS. PWHT on 2507 is not optional. It is destructive.
The only heat treatment that is acceptable on 2507 welds is a re-solution anneal at 1050 – 1100 °C followed by rapid water quench. This is done at the mill on the plate or pipe, not in the field after welding. A field re-solution anneal is impractical (you cannot put a 6-inch field weld in a furnace). So the WPS is designed to produce the correct microstructure on the first cool - and that is why the heat input and interpass windows matter so much.
The Filler Decision Tree - Choose Right in 30 Seconds
Use this tree to pick the filler for a 2507 weld before you open a filler catalog:
| Your Service | Filler to Use | Why |
|---|---|---|
| Subsea flowline or jumper, seawater, 5 – 50 °C | ER2594 | Overmatched PREN_W 44; nitrogen-rich; passes A923 Methods A+B+C |
| Subsea flowline with sour H2S above 0.05 psi partial pressure | ER2594 (mandatory) | Only filler that consistently passes NACE MR0175 hardness limit on HAZ |
| Topside piping, seawater cooling, ambient temperature | ER2553 | Matched PREN_W 40; cheaper than ER2594; acceptable for less critical service |
| Shop fabrication of pressure vessel or tank, atmospheric exposure | ER2553 | Matched filler, qualified for shop-only use |
| Joining 2507 to 316L (dissimilar weld) | ER309LMo (with restrictions) or ERNiCrMo-3 (Inconel 625 filler) | 309LMo for shop-only; 625 filler for any field service |
| Joining 2507 to carbon steel | ERNiCrMo-3 (Inconel 625 filler) | Standard dissimilar weld practice; nickel filler manages the dilution |
If your answer to "which filler" is "ER2209 because the shop has it in stock" or "ER309LMo because the fabricator always uses it" - you have the wrong filler, and the weld will fail ASTM A923 Method C if you test it.
FAQ
What is the difference between ER2594 and ER2553 for welding 2507?
Both are nitrogen-enhanced duplex fillers designed for 2507. ER2594 is overmatched (PREN_W ~44 vs base metal 42), with higher nitrogen (0.25 – 0.30%) to compensate for arc losses. ER2553 is matched (PREN_W ~40, nitrogen 0.20 – 0.25%). Use ER2594 for subsea, sour H2S, and any critical chloride service. Use ER2553 for shop fabrication of less critical equipment where the cost saving matters.
Can I use ER2209 to weld 2507?
No. ER2209 is the correct filler for duplex 2205 (S32205), with PREN_W ~36. Using it on 2507 base metal gives a weld with ferrite content above 65% and Cr2N nitride risk. The weld will likely fail ASTM A923 Method A (grain-boundary attack visible after etch). Use ER2594 or ER2553 on 2507.
What is the heat input range for 2507 welding?
0.5 to 2.5 kJ/mm per ASME Section IX qualified range, with a tight production set-point of 0.8 to 1.5 kJ/mm. Below 0.5 kJ/mm, ferrite excess and Cr2N nitride risk. Above 2.5 kJ/mm, sigma phase risk in HAZ. Use the production set-point, not the qualified range, in the WPS.
Does 2507 weld require PWHT?
No. PWHT in the 600 – 950 °C range precipitates sigma phase in 2507 welds and HAZ. The weld must be cooled rapidly through that range. If your fabricator proposes PWHT for 2507, stop the order. The only acceptable heat treatment is a re-solution anneal at 1050 – 1100 °C with rapid water quench, done at the mill.
What does ASTM A923 Method A actually test?
Method A electrolytically etches a polished weld cross-section in 40% NaOH at 1 – 3 V for 5 – 10 seconds, then examines it at 10 – 30× magnification for grain-boundary attack. Continuous attack along grain boundaries indicates sigma or chi phase precipitation in the weld or HAZ. A passing coupon shows clean grain boundaries with no preferential attack.
What ferrite number is acceptable in a 2507 weld?
35 – 65% ferrite by ASTM E562 (Feritscope or magnetic method) or by WRC-1992 diagram calculation. Below 35%, the weld loses the strength and corrosion resistance the ferrite phase provides. Above 65%, the weld becomes brittle and Cr2N nitride precipitation is a risk. The 50/50 split is the target.
How long does a properly welded 2507 pipe last in subsea service?
For Marjan-class offshore service (chlorinated seawater injection at 40 °C, design life 25 years), a properly welded 2507 flowline with ER2594 filler, heat input 0.8 – 1.5 kJ/mm, ferrite 35 – 65%, and passing ASTM A923 Methods A+B+C is qualified for the full design life. Improperly welded 2507 (wrong filler, high heat input, no A923 verification) has been observed to pit in 3 – 7 years in the same service.
Need a 2507 pipe or prefabricated spool with the weld chemistry, ferrite balance, and ASTM A923 verification that actually pass the third-party inspection?
HUITONG supplies UNS S32750 pipe and prefabricated spools to ASTM A790 / A928 with ASME Section IX-qualified WPS, ER2594 / ER2553 filler (no 309LMo or ER2209), heat input 0.8 – 1.5 kJ/mm production set-point, ferrite 35 – 65% per ASTM E562, and ASTM A923 Methods A + B + C on every procedure qualification coupon. Send your pipe schedule, service conditions (chloride, H2S partial pressure, temperature), and required NDE scope to market@htpipe.com or WhatsApp +86-19339900201 - and if you are not sure whether 2507 is the right grade or whether ER2594 vs ER2553 is the right filler for your service, tell us the service and we will tell you the minimum-cost combination that passes.
