Welding metallurgy: the heat-affected zone (HAZ)

The weakest link in a weld is often not the molten metal itself. It is the narrow strip right beside it — the metal that never melted, only heated up and cooled down. The heat-affected zone (HAZ) is the transition band between the fusion line and the thermally unaffected parent metal. It is only a few millimeters wide, yet metallurgically it is the most complex part of the joint. A large share of service fractures, hydrogen-induced cold cracks and toughness losses originates exactly there, in the strip the welder never sees. This article walks through the HAZ sub-zones, how peak temperature and cooling time (t_8/5) write themselves into the microstructure, why the coarse-grained zone loses toughness and hardens, what heat input and preheat actually control, and which hardness limits apply in the field and in service — all from an inspector's point of view.

1. What the HAZ is, and why it matters

The HAZ is the region of parent metal that stays below the melting temperature and therefore never melts, but gets hot enough to undergo solid-state phase transformation, grain growth or tempering. A welded joint has three metallurgical regions: the weld metal (melted and re-solidified), the fusion line, and the HAZ. What makes the HAZ critical is this: unlike the weld metal, you cannot tune it with filler metal chemistry. It depends entirely on the parent metal composition and on the thermal cycle it sees. As an inspector or welding engineer, you can manage the HAZ only by controlling the thermal cycle — peak temperature and cooling time.

In procedure qualification this shows up directly in specimen locations. Under EN ISO 15614-1, the notch in impact test specimens sits on the fusion line and at defined distances from it (typically FL, FL+2 mm and FL+5 mm). The intent is to sample the different HAZ sub-zones separately, not the weld metal. ASME BPVC Section IX works differently. Section IX by itself requires neither impact testing nor hardness measurement; it only manages qualification variables (essential, supplementary essential, non-essential). The impact and hardness requirements themselves come from the applicable construction code (Section VIII Div. 1/Div. 2, B31.1, B31.3, NACE/ISO 15156 and so on), and specimen locations follow the provisions that code invokes. Treating the two systems as equivalent in the same sentence leads straight to missing tests in the inspection plan.

Think of the HAZ as a region you cannot reach with chemistry — only with heat. That is precisely why the specification pins down heat input and preheat parameters so tightly. Welder qualification (EN ISO 9606-1) touches the HAZ indirectly in the same way: the welder is the person who actually has to hit the heat input and interpass temperature range of the qualified WPS on site. The qualification certificate covers manual skill, not compliance with those parameters, so thermal parameter conformity must be verified separately.

2. Peak temperature and the thermal cycle: mapping the HAZ

The HAZ is not uniform. Every point reaches a different peak temperature depending on its distance from the fusion line. Points adjacent to the fusion line end up between solidus and liquidus (in carbon steels the solidus is typically ~1450–1500 °C; pure iron melts at 1538 °C, and the solidus drops as carbon increases). In that narrow strip, local liquation can occur at the grain boundaries. Moving toward the unaffected parent metal, the peak temperature falls below the A₁ temperature (the equilibrium eutectoid temperature, ≈ 727 °C; alloying shifts it — Mn and Ni lower it, while Cr, Mo, Si and V raise it).

This peak temperature gradient turns the HAZ into a working cross-section of the iron–carbon phase diagram. Each distance from the fusion line determines which phase field that point reached, and therefore which microstructure it will produce on cooling. The key to understanding the HAZ is to stop thinking of it as "one zone" and start thinking of it as a sequence of sub-zones ordered by peak temperature. Within a single millimeter of the same weld, the microstructure can run from coarse-grained and hard near the fusion line to virtually unchanged near the parent metal.

3. The HAZ sub-zones

In a ferritic-pearlitic carbon-manganese steel, the following sub-zones run from the fusion line toward the parent metal:

Partially melted zone (PMZ — partially melted / liquation zone): A very narrow strip between the fusion line and the coarse-grained HAZ, with a peak temperature between solidus and liquidus. Low-melting-point films and segregations at the grain boundaries (S, P, low-melting eutectics) liquate locally and leave weak boundaries behind on solidification. This is the classic site of liquation cracking, and it matters most in austenitic stainless steels, nickel-based alloys and steels with high S/P content. Metallography often reveals it only at high magnification, so routine macro examination misses it.
Coarse-grained HAZ (CGHAZ): Extends from the grain coarsening temperature up to the solidus. That threshold is typically ~1100 °C in C-Mn steels; in microalloyed steels it shifts to ~1150–1250 °C, depending on the dissolution temperature of the TiN/NbC precipitates that pin the grain boundaries. Austenite grains grow rapidly and coarsen heavily. Coarse austenite grains carry high hardenability, so on cooling they tend to transform to bainite or martensite. Mechanically, this is the most problematic zone.
Grain-refined / normalized HAZ (FGHAZ): Peak temperature between ~900 °C and the grain coarsening temperature, just above A₃. Full austenitization occurs, but grain growth is limited; on cooling, fine-grained ferrite-pearlite forms. This zone is usually the toughest part of the joint — effectively an in-situ normalization.
Partial (intercritical) austenitization HAZ (ICHAZ): Peak temperature between A₁ and A₃ (~727–900 °C; in low-carbon C-Mn steel A₃ typically sits around 850–900 °C and falls as carbon increases in the hypoeutectoid range — beyond the eutectoid composition the rising branch is A_cm). Only part of the structure reverts to austenite. On cooling, these carbon-enriched islands harden and can form locally hard and brittle regions, particularly M-A (martensite-austenite) islands.
Tempered / subcritical HAZ (SCHAZ): Peak temperature below A₁. No phase transformation occurs, but in quenched and tempered (Q&T) or cold-formed materials you can see tempering, excessive softening and loss of strength. In high-strength steels, do not dismiss this "soft zone" at the design stage.

In multi-pass welds, each pass reheats the HAZ of the previous one, so these zones overlap. The following pass partially refines or tempers the coarse-grained zone left behind by the pass before it. This "intercritically reheated coarse-grained HAZ" (ICCGHAZ) is exactly where toughness tests return their lowest values, and where the hard, brittle M-A islands known as local brittle zones (LBZ) concentrate. That is why the LBZ phenomenon is associated not with the single-pass ICHAZ but essentially with the reheated coarse-grained zone in multi-pass welding.

4. Cooling rate and the t_8/5 concept

Peak temperature decides which phase field a point reached; cooling rate decides which microstructure it freezes into. In welding metallurgy, cooling rate is usually expressed as t_8/5: the time in seconds for the material to cool from 800 °C to 500 °C. This window is critical because most of the ferrite, pearlite, bainite and martensite transformations happen inside it.

A common mistake is to treat t_8/5 as a function of heat input alone. In the EN 1011-2 relationships, t_8/5 is set jointly by heat input (Q), the initial/preheat temperature (T₀), section thickness, the dimensionality of heat flow (two-dimensional or three-dimensional) and joint geometry correction factors (the F2 and F3 shape factors for plate and fillet joints). As thickness increases, heat flow moves from two-dimensional to three-dimensional; the component acts as its own heat sink and t_8/5 shortens. The thickness that separates the two regimes is called the transition thickness: in the three-dimensional regime t_8/5 is practically independent of thickness, while in the two-dimensional regime it scales roughly with (Q/d)².

Short t_8/5 (fast cooling): Thick section / three-dimensional heat flow, low heat input, no preheat, high restraint. The cooling curve runs down the left-hand side of the transformation curves on the CCT diagram; in hardenable steels this gives martensite/upper bainite and high hardness. The risk of hydrogen-induced cold cracking is high. This is exactly why preheat becomes mandatory in thick sections.
Long t_8/5 (slow cooling): Thin to medium section / two-dimensional heat flow, high heat input and/or preheat. Softer ferritic-bainitic structures form and hardness drops. To reach the same t_8/5 in a thick section, you have to raise heat input and preheat together.

So t_8/5 has both a lower limit (for hardness and cold cracking control) and an upper limit (for toughness). A good WPS is written to keep heat input and preheat inside that window. Another frequent misconception is that EN 1011-2 provides "t_8/5 tables by steel type". EN 1011-2 mainly gives the calculation method for t_8/5 (2D/3D relationships, shape factors) and the methods for determining preheat. The acceptable t_8/5 window comes largely from the steel producer's product data sheet — especially for TMCP and quenched and tempered high-strength steels — and from the project specification.

5. The coarse-grained HAZ: toughness loss and hardening

The problem child of the HAZ is the coarse-grained zone, for two separate but overlapping reasons. The first is grain coarsening: austenite grain growth does not happen inside the t_8/5 window but much higher up, during the time spent above the grain coarsening temperature (typically ~1100 °C). High heat input holds the material above 1100 °C for longer and coarsens the grain. The second is the transformation product: the t_8/5 window does not set grain size, it sets what that coarse austenite turns into. A long t_8/5 gives low-toughness products such as coarse ferrite side plates (Widmanstätten), upper bainite and M-A islands; a short t_8/5 gives hard martensite/lower bainite. Because a coarse prior austenite grain also raises hardenability, the tendency to form martensite increases even at the same cooling rate. The result is a narrow strip that can be both hard and brittle.

It is worth adding that CGHAZ toughness is not governed by welding parameters alone. The real driver is the microalloy design of the parent metal. Grain boundary pinning by TiN and Ti-oxide precipitates, low-carbon + Nb compositions and oxide metallurgy (for example HTUFF-type plates developed for high heat input) limit CGHAZ grain coarsening even at high heat input. So the blanket claim that "TMCP steels are sensitive to high heat input" is incomplete: sensitivity varies by product family, and plates designed specifically for high heat input exist.

Under creep conditions — high-temperature service such as Grade 91 / CrMo(V) steels — do not confuse the roles of the sub-zones. The coarse-grained HAZ is the classic site of reheat (stress relief) cracking and of "Type III" creep damage. By contrast, in creep-resistant ferritic steels the weakest link is "Type IV" creep cracking, which nucleates and propagates not in the coarse-grained zone but in the fine-grained / intercritical HAZ (FGHAZ–ICHAZ) at the outer edge of the HAZ, facing the parent metal. Carbide coarsening and softening there create a zone of low creep strength. Type IV susceptibility depends on the PWHT temperature and the service temperature. Because joint life can fall well below the plain parent metal life, design applies a weld strength reduction factor (WSRF) — in Grade 91 piping systems that factor feeds directly into the allowable stress. Since the two mechanisms develop in different sub-zones, life assessment and inspection planning must be built accordingly. For the inspector the lesson is clear: impact testing and hardness measurement must cover the fusion line and the CGHAZ as well. Testing the weld metal alone is misleading.

6. Hardness: why it is a proxy parameter

HAZ hardness is the quantity most often measured and most often specified, because it is cheap, fast, and indicative both of hardening and — indirectly — of the risk of hydrogen-induced cold cracking and sulfide stress cracking (SSC). Despite the word "corrosion" in the family name, SSC is a cracking mechanism driven by hydrogen embrittlement. Do not confuse it with classic stress corrosion cracking (SCC), which propagates by anodic dissolution.

High hardness indicates a high martensite fraction, which means low toughness plus susceptibility to hydrogen embrittlement and to sulfide stress cracking in H₂S environments (NACE conditions). But you cannot draw that link from martensite fraction alone: the hardness of martensite depends largely on carbon content. In a steel with 0.05% C, even 100% martensite stays around 300 HV and can show reasonable toughness; at 0.25% C the same martensite fraction exceeds 450 HV and is brittle. The correct reading is "hardness ≈ martensite fraction × carbon content".

Hardness can be predicted from carbon equivalent and t_8/5, but the two carbon equivalent formulas are not interchangeable. The CEV (IIW) formula is meaningful for traditional, relatively high-carbon steels (typically C > 0.18%); Pcm was developed for low-carbon modern TMCP/HSLA steels. Choosing the wrong formula produces a seriously wrong preheat temperature. Checking the right carbon equivalent during material selection is therefore the first gate of weldability.

The measurement method standard matters just as much. In welded joints, the position of hardness traverses and the number of indentations are defined by EN ISO 9015-1 (9015-2 for microhardness), and the indentation itself by EN ISO 6507-1. Ignore the spacing and edge distance rules and you will systematically miss the peak CGHAZ value. The diagonal of an HV10 indentation is around 0.4 mm, so in narrow CGHAZ strips it can spill outside the zone; use HV5 or HV1 in that case. Draw the traverse so that it crosses the fusion line, and expect the peak value in the CGHAZ.

7. The role of heat input and preheat in HAZ control

You have two main levers over the HAZ: heat input and preheat/interpass temperature.

Heat input acts directly on t_8/5. Two different quantities must be kept apart here, and confusing them is the single most common nonconformity in PQR/WPS records:

Arc energy: E = (U × I) / (1000 × v) [kJ/mm] — no efficiency factor.
Heat input: Q = k × E, where k is the thermal efficiency factor (η).

Typical k values in EN 1011-1 and ISO/TR 18491: submerged arc welding (SAW) 1.0; covered electrode (MMA/SMAW), MIG/MAG (GMAW) and flux-cored (FCAW) 0.8; TIG (GTAW) 0.6. In the formula, U is voltage (V), I is current (A) and v is travel speed (mm/s); the factor of 1000 converts J/mm to kJ/mm. The critical point: ASME BPVC Section IX and AWS D1.1 apply no efficiency factor in the calculation, so the ISO and ASME numbers for the same weld come out different. The WPS and PQR must state clearly which definition is being limited — arc energy, or heat input multiplied by k. Note also that EN ISO 15614-1 allows control by run-out length instead of heat input.

Raising heat input slows cooling and lowers hardness and cold cracking risk — but too much of it widens the coarse-grained HAZ, drops toughness, and increases HAZ softening in thermomechanically rolled (TMCP) steels sensitive to high heat input. That is why heat input is recorded in the PQR and controlled with lower and upper limits in the WPS. In the ASME system, heat input becomes a supplementary essential variable when impact testing is required.

Preheat brings the component to a defined temperature before and during welding, and does three jobs at once: it lengthens the cooling time (t_8/5) and so reduces hard martensite formation; it gives diffusible hydrogen (HD) time to escape from the structure, which is the basis of preventing hydrogen-induced cold cracking; and it lowers thermal stresses. Hydrogen-induced cold cracking typically occurs below ~200 °C, mostly below 100 °C, and appears with a delay at room temperature. That is why preheat must be maintained throughout welding, with post-heating applied where required. Post-heating is not a PWHT: its purpose is not stress relief but giving hydrogen time to diffuse out.

Determine the preheat temperature from carbon equivalent, section thickness, hydrogen level and degree of restraint, using the methods of EN 1011-2 or AWS D1.1. The two EN 1011-2 methods are not interchangeable:

Method A: Based on CE (IIW) — CE = C + Mn/6 + (Cr+Mo+V)/5 + (Ni+Cu)/15 — and intended essentially for traditional C-Mn steels.
Method B: Based on CET — CET = C + (Mn+Mo)/10 + (Cr+Cu)/20 + Ni/40 — and intended for low-alloy high-strength (quenched and tempered / TMCP) steels.

Picking the wrong method means insufficient preheat, plain and simple. Both methods take as input the combined thickness — the sum of the heat-flow contributions of all parts entering the joint — and the hydrogen class of the filler metal. Hydrogen level is classified on the EN 1011-2 scale A–E, running from scale A (above 15 ml per 100 g of deposited metal) down to scale E (no more than 3 ml per 100 g); consumable classifications use the equivalent H15/H10/H5 designations of EN ISO 2560. Measure HD to EN ISO 3690. On the AWS D1.1 side, the table-based method and the supplementary method based on hydrogen control can give different answers; state which one was applied in the WPS.

Do not forget that preheat also has an upper limit. In quenched and tempered and TMCP steels in particular, excessive preheat stretches t_8/5 further than needed and drops CGHAZ toughness. The upper limit on interpass temperature follows the same logic: in ferritic steels the reason is loss of toughness; in austenitic stainless steels it is sensitization from Cr carbide precipitation at the grain boundaries; in duplex steels it is disruption of the ferrite/austenite phase balance and sigma phase precipitation.

8. PWHT and its effect on the HAZ

Post-weld heat treatment (PWHT — stress relief) does more than reduce residual stresses. Metallurgically it also helps the HAZ: it tempers hard martensite/bainite structures, lowering hardness and improving toughness, and it drives off diffusible hydrogen.

Whether PWHT is mandatory does not tolerate generalization. It depends on the combination of thickness, material group (ASME P-No / ISO/TR 15608 group) and service conditions, and it is settled by the PWHT table and exemption provisions of the applicable construction code — for example ASME BPVC Section VIII Div. 1 UCS-56 and its exemption conditions, or EN 13445-4. Saying "it is usually mandatory in thick sections" pushes the reader either into unnecessary PWHT or into a wrong exemption.

PWHT has its own limits too:

The Ac1 limit: The PWHT temperature must never exceed the Ac1 of the material. Exceed it and you get local re-austenitization, leaving untempered martensite behind on cooling — the treatment itself produces the very damage it was supposed to correct. In Grade 91 the Ac1 is typically ~800 °C and shifts with chemistry (falling as Ni + Mn increases), which is why the typical PWHT window is quoted as 730–780 °C and why a specification limit is placed on the Ni + Mn sum.
Reheat (stress relief) cracking: Susceptibility is linked to precipitation hardening in the CGHAZ by secondary carbide formers (V, Nb, Ti, Cr, Mo). The classic susceptible families are Cr-Mo-V steels, 2¼Cr-1Mo, some Ni-Cr-Mo steels and stabilized austenitic stainless steels (types 321/347). The mechanism: during PWHT heating, fine carbides precipitate within the grains in the ~500–700 °C band, intragranular strength rises, and relaxation strain concentrates at the prior austenite grain boundaries, opening a creep-type crack along them.
Over-tempering: Long holding times or high temperatures cause loss of strength, and in quenched and tempered steels can take properties below the design values.

So PWHT parameters, exactly like heat input, must be held inside a window — one bounded from below (effective tempering) and from above (Ac1 and loss of strength).

9. HAZ hardness limits: service and NACE

Typical maximum HAZ hardness limits by application are as follows:

General structural / pressure service: There is no single universal threshold. EN ISO 15614-1 gives hardness limits by steel group and heat treatment condition (Table 2). For example, for ISO/TR 15608 group 1 and 2 steels the limits are 380 HV10 as-welded and 320 HV10 after heat treatment; for group 3 (quenched and tempered / high-strength) steels, 450 HV10 as-welded and 380 HV10 after heat treatment. For the Cr-Mo groups (4 and 5), the limit was raised to 350 HV10 in the ISO 15614-1:2017 revision. Always check the standard's own hardness table for the exact value — and note that the project specification often sets a tighter limit than the standard, which is the one that binds contractually.
Sour service / H₂S environment (ANSI/NACE MR0175 / ISO 15156): There are two separate criteria against sulfide stress cracking (SSC), and they are not conversions of one another. For the base metal in carbon and low-alloy steels, max. 22 HRC applies (≈ 248 HV ≈ 237 HBW by ASTM E140 conversion). For the weld metal and HAZ, a separate acceptance criterion of max. 250 HV10 applies to the hardness survey in procedure qualification (ISO 15156-2 Annex A); some line pipe/HSLA applications include provisions allowing 275 HV10 with supplementary qualification conditions. HAZ hardness is not measured in HRC: in a strip a few millimeters wide, the size of a Rockwell C indentation and the edge distance rules cannot give a valid result. That is the classic reason a field report gets thrown out.
Choosing the standard: MR0175/ISO 15156 is for upstream (production/field) equipment; for refinery and process equipment a separate standard applies, NACE MR0103 / ISO 17945 (typically a 248 HBW limit for carbon steel welds). Picking the wrong standard means judging a correctly measured hardness against the wrong criterion.
Quenched and tempered high-strength steels: The material-specific specification may set different limits — for example separate values for the fusion line in offshore projects.

Verify these limits with a hardness map at the PQR stage, and in production with production test pieces plus, where required, field metallography (replication) and portable hardness measurement. Set up the measurement layout and indentation positions to EN ISO 9015-1, and in sour service work to the measurement layout given in ISO 15156-2 itself.

From the field

The correction I make most often on site is this: crews hunt for cracks in the weld metal and nowhere else. On classic C-Mn work with a high CEV, delayed cold cracking really does favor the coarse-grained HAZ next to the fusion line. But on modern low-CEV TMCP steels welded with high-strength filler metal, the dominant location has moved into the weld metal — longitudinal cracks and chevron cracking in particular. The inspection plan has to cover both. Lock onto one zone and you send the inspector looking in the wrong place.

Timing is a code matter, not a preference. Hydrogen-induced cracking can appear hours — up to 48 hours — after welding, so codes make delayed inspection mandatory: 48 hours for A514/A517 type quenched and tempered steels in AWS D1.1, and 24 hours for certain materials in ASME B31.3. Never run the final NDT before the waiting period the code prescribes (typically 24–48 hours, depending on material) has elapsed.

Second practical point: check preheat not only at arc strike but between passes as well, with a contact thermometer or a thermocouple. When preheat drops, HAZ t_8/5 shortens and hardness climbs fast. For hardness, never rely on a single point — run a traverse perpendicular to the fusion line and respect the indentation spacing and edge distance rules. Miss the peak and the report reads clean while the component stays wide open to cracking.

Related standards

ASME BPVC Section IX — Welding/brazing procedure and performance qualification (WPS/PQR); manages the essential variables. Impact and hardness requirements themselves come not from Section IX but from the applicable construction code (e.g. Section VIII Div. 1, B31.1, B31.3).
EN ISO 15614-1 — Welding procedure qualification testing for steels and nickel alloys (including heat input range, hardness and impact testing).
ISO/TR 15608 — Grouping system for metallic materials for welding purposes; the source of the "steel group" concept used in this article.
EN ISO 15609-1 — Content of the welding procedure specification (WPS), including preheat and interpass temperature.
EN 1011-1 — General guidance for arc welding; arc energy vs heat input and thermal efficiency (k) factors.
EN 1011-2 — Guidance for arc welding of ferritic steels; preheat (Method A: CE(IIW), Method B: CET), combined thickness and t_8/5 calculation.
EN ISO 9015-1 / -2 — Hardness testing of welded joints (HV10 traverse positions) and microhardness testing.
EN ISO 9016 — Impact testing of welded joints; specimen location and notch designation.
EN ISO 3690 — Determination of diffusible hydrogen (HD) in arc welding.
EN ISO 2560 — Classification of covered electrodes; hydrogen classes (H15/H10/H5).
EN ISO 9606-1 — Welder qualification testing (steels); certification of the personnel responsible for applying the qualified thermal parameters on site.
EN ISO 5817 — Quality (acceptance) levels for weld imperfections.
EN ISO 6520-1 — Classification and terminology of weld imperfections.
AWS D1.1 — Structural welding code, steel; preheat (table method and hydrogen control method), qualification and delayed inspection requirements.
ANSI/NACE MR0175 / ISO 15156 — Materials and weld metal/HAZ hardness limits for sour (H₂S) service, upstream.
NACE MR0103 / ISO 17945 — Materials and hardness requirements for sour service in refinery and process equipment.
ASME BPVC Section VIII Div. 1 (UCS-56) and EN 13445-4 — Pressure vessels; PWHT tables, exemptions and construction requirements.

Take it to the field: To keep the welding standards, procedure logic and field control steps in this article in your pocket — completely offline and free — take a look at the Doawise Welding Eng Guide app.

At Doawise we deliver welding engineering and inspection services to international standards — ASME BPVC Section IX, EN ISO 15614 and NACE MR0175 among them — and we document every step, from WPS/PQR review to HAZ hardness mapping, from preheat verification to impact test evaluation, so the results stay recordable, traceable and auditable. Our aim is simple: the weld should meet the standard not only where you can see it, but also in that critical strip right beside the fusion line.