Aluminium Embrittlement

Edward ValvesAvoiding Aluminum Nitride Embrittlement in

Steel Castings for Valve ComponentsV-Rep 84-1

2

The Flow Control Division of Rockwell isconcerned that low ductility steel castings,embrittled by aluminum nitride, may not bedetected prior to their becoming valvecomponents. This concern stems from thetrend in recent years away from greensand (clay-bonded) molding in favor of air-set sand molding (using organic binders),notable effect of the latter being to pro-duce castings that cool more slowlythrough the temperature range where alu-minum nitride precipitates around the pri-mary austenite grains of the steel. The lackof provisions in the ASTM specifications toforestall the shipping of castings embrittledby aluminum nitride has promptedRockwell to spearhead a task force withthe objective of removing this deficiency inthe specifications. The accompanying arti-cle discusses the present status of activitiesboth within Rockwell and in ASTM aimedat ensuring that steel valve body and relat-ed pressure-containing castings do not con-tain deleterious amounts of aluminumnitride, regardless of the foundry practicesused to produce them.

IntroductionSteel castings are the mainstay of the steelvalve manufacturing industry. They are ahighly cost-effective means of obtainingmedium- to large-size valve bodies of thedesired shape, wall thickness, and compo-sition in a broad range of valve sizes andtypes. Over the years, the steel foundryindustry has developed modifications insteel melting practices, molding methods,welding processes, and nondestructive test-ing to keep pace with the ever-increasingconcern for quality standards shown by the

users of steel castings. Some of these modi-fications introduce new manufacturing vari-ables that can affect the metallurgical prop-erties and internal quality of the castings;therefore, modifications in practice andcontrol procedures may become necessary.A case in point is a trend in recent yearsfor foundries to discontinue their use ofgreen sand (clay-bonded) molding in favorof air-setting sands bonded with organicbinders. In addition to increasing produc-tivity, the air-set process enables thefoundries to make cleaner, better dimen-sioned castings with more faithful repro-duction of pattern details. Although quanti-tative data is not available, it has beenobserved that castings cool much slower inthe organically bonded sand molds thanthey do in clay-bonded green sand molds.The difference in cooling rate in the air-setmolds is generally attributed to the mildexothermic mold/metal reaction from theorganic binders plus the insulating effect ofthe dead-air spaces created when theorganic binders are volatilized by the beatfrom the casting. Furthermore, on the basisof such studies as reported by T. T. Rick*,molding mixtures with the greatest appar-ent density will have the greatest heatabstracting capacity. By virtue of the differ-ent methods used to compact clay-bondedsand and organically bonded sand againstthe patterns, green sand rams to a some-what greater density than air-set sand andtherefore exhibits a greater capacity forheat abstraction which in turn results infaster cooling of the casting.

Avoiding Aluminum Nitride Embrittlement in Steel Castings for Valve Components

ByWilliam C. Banks,Senior MetallurgicalEngineerMaterials EngineeringDepartmentFlow Control Division,Rockwell InternationalFirst Published 1984

*Heat Abstraction by Molding Materials”, Foundry. 77 1949, p. 96-97, 182, 184, 186,188.

FIGURE 2: Fine-grain fracture appearancetypical of normal cast carbon steel(Grade WCB). Section is from a 1.5 -inch (37 mm) thick test blockrepresentative of a heavy sectionvalve body.

FIGURE 1: Fracture surface of section from butt-weld end of a Grade WCB carbonsteel valve body casting exhibiting a“rock candy” appearance,characteristic of aluminum nitrideembrittlement.

3

Slow cooling rates, in themselves, are nor-mally not considered detrimental in theproduction of steel castings. However, thecommon use of aluminum as a deoxidizerin the molten steel can, under certain cir-cumstances, when combined with a slowcooling rate and the presence of nitrogen,reduce the ductility and toughness of thesteel by a precipitation phenomenon.When this embrittling condition is present,fracture appearance is characterized bycoarse angular facets and the phenome-non is referred to as intergranular or rock-candy fracture. Figure 1 shows the typicalrock-candy fracture exhibited by an embrit-tled weld-end section from a carbon steel(Grade WCB) valve body. For comparison,the normal fine-grain fracture of this gradeof steel is illustrated in Figure 2.Experience indicates that under certainwelding conditions, cracking may beencountered in the heat-affected zone ofcastings susceptible to intergranular frac-ture. Because the cooling rates of castingsproduced by modern air-set molding proce-dures are slower, such castings are likelyto be more sensitive to this problem thansimilar castings produced by older greensand methods.Many years ago it was established that thecompound known as aluminum nitride(AIN) could precipitate in the form of thinfilms on the boundary surfaces of the pri-mary austenite grains and be a cause ofintergranular fracture. The primary vari-ables controlling the susceptibility of a steelto this type of embrittlement are the alu-minum content, nitrogen content, and thecooling rate through the austenite region of

the iron-carbon diagram. In the past, thisproblem was manifested only in veryheavy wall castings produced in greensand molds. However, long-establishedmelting and deoxidation practices provensatisfactory for steel cast in green sandmolds may not necessarily provide assur-ance of avoiding aluminum nitride embrit-tlement in steel cast in air-set molds. Theslower cooling rate in air-set molds intro-

duces a new variable. The purpose of thisarticle Is to describe some technical activi-ties aimed at ensuring that steel valve bodyand related pressure-containing castingsare free of deleterious amounts of alu-minum nitride, regardless of the foundrypractices used to produce them.


CURVE X–1" SECTION COOLING IN GREEN SANDCURVE Y–21⁄2" SECTION COOLING IN GREEN SANDCURVE Z–21⁄2 SECTION COOLING IN AIR-SET SAND

A: 0.127% AIB: 0.09% AIC: 0.06% AID: 0.03% AIE: 0.01% AI

0.018% NCURVE Y

CURVE Z

CURVE X

CBA

D E

TIME T IN SECONDS

TEM

PERA

TURE

IN °

C

TEM

PERA

TURE

IN °

F

1800

1600

1400

1200

1000

800

600

400

3272

2912

2552

2192

1832

1472

1112

752101 102 103 104 105 106 107

(10 SEC.) (1 MIN. 40 SEC.) (16 MIN. 40 SEC.) (2.78 HR.) (27.78 HR.) (277.78 HR.) (2777.78 HR.)

FIGURE 3: Schematic representation of the interrelation of aluminum content, nitrogen content, andcooling rate on the formation of aluminum nitride in cast steel. lntersection of a coolingcurve with any of the curves (A, B, C, etc.) representing the start of AIN precipitationresults in the presence of AIN in the steel at room temperature.

Factors Affecting the Formation ofAluminum NitrideAlthough the amount of aluminum addedas a deoxidizer to the steel can normallybe closely controlled, the amount recov-ered and reported as a residual variesconsiderably, This variation is largely dueto the different states of oxidation the steelis in at the time of the aluminum additionand also due to physical mixing problemsassociated with a lightweight, relativelylow-melting-point, reactive metal beingadded to molten steel.Because most steel foundries use electricarc melting, the nature of this process canintroduce unknown amounts of nitrogeninto the steel, and despite efforts to flush itout, maintaining consistently low levels isdifficult. An additional problem with nitro-gen control is that the analytical equipmentfor nitrogen determinations is costly andthe time required for analysis may be toolong for the melting department’s needs.Few foundries have in-house capability formaking nitrogen analyzes. Reliance musttherefore be placed on standardizing melt-ing practices as much as possible by usingtechniques associated with the attainmentof low nitrogen levels as determined subse-quent to the casting of the steel.Valve body castings are produced in awide variety of shapes and sizes, so thefoundry must contend with a considerablerange of section thicknesses and conse-quently a wide range of cooling rates. Asthe schematic graph of Figure 3 illustrates,the foundryman must control his ALU-MINUM ) and nitrogen levels so that the

thickest section in any given heat of steelhe pours will spend a minimum amount oftime in the temperature range where thealuminum nitride precipitation is occurring,For a given nitrogen Content, it is impor-tant that the residual aluminum content bedecreased as the section thickness beingcast increases, The same principle appliesif the cooling rate of a given casting isdecreased as a result of making the cast-ing in an air-set mold. In other words, thecritical thickness for avoiding aluminumnitride formation with a given aluminumand nitrogen content in the steel decreasesas the cooling rate decreases. Thus, for agiven thickness, aluminum levels must bereduced and the nitrogen content carefullycontrolled when air-set molding practicesare employed,

Methods for Detecting AluminumNitride Embrittled CastingsIn view of the interacting factors contribut-ing to the formation of aluminum nitride,evaluation methods are needed to deter-mine whether conditions were such thatharmful amounts of aluminum nitride mayhave formed in a given casting. The prob-lem primarily encompasses castings withthick cross sections and those of a compactdesign having a relatively low ratio of sur-face area to volume. In the latter case,such castings cool slowly because lessarea is available to effect heat transfer tothe mold. It is fortunate that a relativelywide range of casting conditions existwherein aluminum nitride embrittlement isnot a problem. In his paper presented to


FIGURE 4: Reference photographs of macroetched cast steel showing 10 levels of severity of inter-granular network structures indicative of the presence of aluminum nitride precipitation inthe primary austenitic grain boundaries. 4

5

the 17th Annual (British) Steel CastingResearch and Trade Association (SCRATA)Conference in 1971, H. M. Kuhn* report-ed that in unalloyed steel castings withwall thicknesses less than 4 inches (100mm), restricting the aluminum content to0.08% maximum prevents the occurrenceof intergranular fracture. Presumably thisconclusion was based on experience withcastings made in clay-bonded green sandmolds because the newer organicallybonded molding sands were not yet ingeneral use at that time.As part of its overall quality assurance pro-gram, the Rockwell Edward Valve organi-zation (in the early 1950s) began evaluat-ing the steel in its large valve bodies byconducting metallurgical tests on speci-mens from lugs or bars integrally cast onthe bodies. For this purpose, a 2 inch by 2inch by 6 inch long (50x50x150 mm) barwas used on all body castings with a 1.5inch (37 mm) thick wall or greater in com-bination with a weight of more than 1000lb (455 kg). This extension bar wasenlarged to 3 inches by 3 inches by 6inches (75x75x150 mm) on castings witha wall thickness of 4 inches (100 mm) andgreater. This testing was done in additionto that required by the basic ASTM specifi-cations (such as A216 and A217) for pres-sure-containing steel castings. Evaluation ofthe steel in accordance with these specifi-cations requires, as a minimum, the tensiletesting of specimens from ASTM A370 sep-arately cast keel blocks whose couponcross-sections are 1 inch by 1.25 inch (25by 32 mm). The Rockwell Edward programthus provided the opportunity to gain data

on the tensile properties of the cost steel Inthe heavy sections of valve bodies throughtesting of specimens that were representa-tive of the actual products.

TABLE 1Descriptive Data Applicable

to Network Structures Shown In Figure 4Rating Delineation % Boundary

Width Outline1 Fine – .001” 202 Fine – .001” 403 Fine – .001” 604 Fine – .002” 505 Fine – .002” 1006 Medium – .005” 1007 Heavy – .010” 1008 0.020” 1009 1/32” 100

10 1/16” 100Note: These ratings are based on the physical width

and continuity of the precipitate pattern devel-oped by the acid etchant on the primaryaustenitic grain boundaries of the cast steel. A 0rating indicates that no delineation of the grainboundaries has been observed. Supplementarytesting is normally conducted to determine thefinal disposition of castings with ratings of 5 orgreater.

In the course of this year’s long programinvolving close cooperation with the steelfoundries supplying the castings, it wasdetermined that a major cause of deficientlevels of ductility in the cast-on coupons (asindicated by low values of percent elonga-tion and percent reduction of area in thetensile test) was poor melting and deoxida-tion practices that promoted the formation

of aluminum nitride, Etching of smooth-ground cross sections of these heavy sec-tion coupons in a hot (160–180°F)(71–82°C) 1:1 hydrochloric acid solutionrevealed significantly different macro-etched structures depending upon theamount of aluminum nitride precipitationthat was present in the primary austeniticgrain boundaries. From the accumulatedbody of data it was possible to establish aseries of reference photographs of theetched steel indicating various degrees orseverity levels of network (“chicken-wire”)structures created by increasing widths ofthe grain boundaries and greater degreesof their continuity. The graded series ofthese reference photographs are illustratedin Figure 4, and Table 1 is used in associa-tion with the photographs to give addition-al data for establishing a network ratingfor a given steel specimen. Analysis of net-work ratings for carbon steel (GradeWCB) indicated that severity ratings belowa 3 had a negligible effect on tensile duc-tility values whereas ratings of 3 or 4 hada statistically noticeable effect on thesevalues. However, ratings of 4 or less weregenerally considered acceptable.To illustrate how deleterious an aluminumnitride network rating of 8 can be, thedata in the following table are presented.Specimens 1 and 2 were from a 2.5 inch(62 mm)-thick buttweld end section ofGrade WCB steel similar to the one shownin Figure 1. The steel had residual alu-minum and nitrogen levels of 0.10% and0.012%, respectively, and exhibited a veryobvious rock-candy fracture with a networkrating of 8.


*Electric-Arc Furnace Steelmaking Practice With Regard to Special Foundry Problems”, proceedings of the above SCRATA conference,Paper No. 2 or Foundry Trade Journal 1972 Vol. 132 (2876) 81-88.

6

It is interesting to note that the minimumtensile and yield strength requirements forGrade WCB steel are met by the alu-minum-nitride-embrittled steel despite Itsexhibited sharply lower ductility values inthe relatively thick weld end section. Suchresults are typical of steel embrittled by alu-minum nitride. No clue as to the tendencyof the heat toward aluminum nitride precip-itation is observable from the results of thekeel block specimen.Should castings with harmful amounts ofaluminum nitride precipitates fail to bedetected and enter the field, the concernswill be the same as those for any productdepending upon a steel with an expectedlevel of ductility but which it may not have.The ability of a steel to accommodate acci-dental overloads in components with stressconcentration areas or sudden impactloads is certainly decreased when theinherent ductility is impaired by the pres-ence of embrittling, intergranular films. Asmentioned earlier, evidence exists that suchfilms can contribute to the formation ofcracks in the heat-affected zone of highlyrestrained welds even in a carbon steel.

Provision for Aluminum NitrideControl In ASTM SpecificationsDespite the fact that it has been known forover 30 years that aluminum-nitride-embrit-tled steel castings could be produced, theASTM specifications for steel castings havebeen strangely silent on the subject.Evidently field failures attributable to thisembrittlement phenomenon have been rela-tively infrequent and the difficulty of obtain-ing a consensus among the foundries as towhat specific control measures would beeffective have inhibited initiatives towarddeveloping an ASTM standard practice,Just as in the Rockwell Edward approach,other responsible companies dependentupon reliable steel castings have protectedthemselves through appropriate provisionsfor AIN control in private-type material pur-chase specifications.Although Rockwell has encountered only afew isolated instances of aluminum-nitride-embrittled steel valve bodies, the FlowControl Division believed that the potentialhazard for its type of products—especiallythose used in gas transmission pipelinesand steam power systems-merited bringing

the issue to the attention of ASTMSubcommittee A01. 18 on Steel Castings.In this way the entire steel valve manufac-turing industry, as well as all users of steelcastings for pressure containment, in par-ticular, would be alerted to the hazard andavailable for supporting ASTM specifica-tion action. As a result of the Rockwell ini-tiative, a task group was formed in May,1982, under the chairmanship of theauthor to develop appropriate wording forinsertion in ASTM Specification A703“General Requirements Applicable to SteelCastings for Pressure-Containing Parts”.Initially it was proposed that the aluminumnitride control provision should be made apart of the main body of A703 and there-by automatically apply to all the specifica-tions for pressure-containing castings linkedto A703 namely, A216, A217, A351,A352, A389, A487, and A643. However,because these specifications are widelyused for purchasing both pressure-contain-ing and nonpressure-containing castings,many of which have small cross sectionsand are relatively lightweight, adding amandatory provision for aluminum nitridecontrol across the board would place anunjustifiable economic burden on the steelcasting industry. Therefore, the task groupdecided to make its proposal as aSupplementary Requirement in A703 foruse by anyone concerned about possiblealuminum nitride embrittlement in his steelcastings.Although at this writing precise wordingfor the control provisions under the title“Macroetch Test” has been approved byASTM Subcommittee A01.18, and is to be


Tensile Yield Elongation ReductionStrength, Strength, in 2” (50 mm) of Area

Specimen No. ksi (MPa) ksi (MPa) % %1 75.2 (518) 63.6 (438) 11.5 14.12 75.5 (520) 71.5 (493) 9.5 10.8Foundry Keel Block from Heat 76.2 (528) 42.8 (295) 28.0 51.0ASTM A216 GradeWCB Requirements, min. 70.0 (485) 36.0 (250) 22.0 35.0

7

balloted in the Main Committee A01 dur-ing 1984 it would be improper to includethe balloted item in this technical article atthis time, However, the key features of theproposal include:

(1) The total residual aluminum contentof the heat is to be reported to thepurchaser

(2) When the aluminum content exceeds0.08%, the foundry is to conduct amacroetch test on a sample repre-sentative of the heaviest section of acasting in the heat by the proceduregiven in ASTM E340.

(3) The resulting etched sample is to becompared and rated with the refer-ence photographs similar to thoseshown in Figure 4 of this article. Ifthe etched network rating does notexceed Level 4, the heat is accept-able.

(4) Provision is made for the dispositionof castings with etched ratingsabove Level 4 by using supplemen-tary testing, for example, fractureexamination, bend tests, and tensiletests, to confirm the actual degree ofembrittlement. High-temperature solu-tion annealing for possible salvagingof rejected castings is to be permit-ted.

Provision for Aluminum NitrideControl In Flow Control DivisionSpecificationsIn the interim, until the ASTM steel castingspecifications are revised, the internalRockwell Material Specifications used byits Flow Control Division plants contain alu-minum nitride control provisions based onthe proposals now moving through theASTM approval process. The Rockwellspecifications contain one additional pre-caution not yet included in the ASTM pro-posal but which will be suggested as a fur-ther revision once the basic control provi-sion is in place in A703. This further con-trol feature requires the checking of so-called “heavy wall” castings defined ashaving section thicknesses exceeding 1.5inches (37 mm) and a weight of at least1000 lb (455 kg) by the macroetch testwhen the residual aluminum content of theheat exceeds 0.06%. As an alternative tothe macroetch test for these heavy-wallcastings, tensile testing of a couponremoved from a heavy section of the cast-ing may be conducted after heat treatment,the results of which are required to meetthe minimum valves specified by ASTM forthe grade of steel involved.

SummaryIt is the intent of this article to alert theusers of steel castings, particularly those inthe valve industry, to the possibility that anold but troublesome phenomenon, alu-minum nitride embrittlement, may occurwith greater frequency as a result offoundries replacing the old clay-bondedgreen sand molding methods with new

organically bonded (air-set) molding meth-ods. This concern is based on the observa-tion that castings made with aluminum-deoxidized steel in air-set molds cool moreslowly than those made in green sandmolds; therefore, they may be more proneto the formation of thin, embrittling films ofaluminum nitride around the primaryaustenite grains. Such films markedlyreduce the ductility of the steel as mea-sured in the conventional tension tests.Since the critical thickness for avoiding alu-minum nitride with a given aluminum andnitrogen content in the steel decreases asthe cooling rate decreases, the significanceof this change in the molding method mustbe recognized.A level of residual aluminum of 0.08%max. in a heat is proposed, above whichvarious evaluation methods are employedto detect any possible harmful level of alu-minum nitride embrittlement.Rockwell has initiated actions through itsparticipation in ASTM committee activitiesto focus national attention on the aluminumnitride threat to the integrity of valve bod-ies and other related pressure-containingcastings. Rather than wait for the provi-sions for aluminum nitride control to clearthe ASTM approval process, Rockwell hasimplemented its own Flow Control Divisionmaterial specifications to ensure that cur-rently supplied cast steel valve bodies arenot embrittled by aluminum nitride, regard-less of the molding methods used to pro-duce them.


Flowserve Corporation has established industry leadership in the design and manufacture of its products. When properly selected, this Flowserve product is designed to perform its intended function safely during its useful life. However,the purchaser or user of Flowserve products should be aware that Flowserve products might be used in numerous applications under a wide variety of industrial service conditions. Although Flowserve can (and often does) provide generalguidelines, it cannot provide specific data and warnings for all possible applications. The purchaser/user must therefore assume the ultimate responsibility for the proper sizing and selection, installation, operation, and maintenance ofFlowserve products. The purchaser/user should read and understand the Installation Operation Maintenance (IOM) instructions included with the product, and train its employees and contractors in the safe use of Flowserve products inconnection with the specific application.

While the information and specifications contained in this literature are believed to be accurate, they are supplied for informative purposes only and should not be considered certified or as a guarantee of satisfactory results by reliancethereon. Nothing contained herein is to be construed as a warranty or guarantee, express or implied, regarding any matter with respect to this product. Because Flowserve is continually improving and upgrading its product design, thespecifications, dimensions and information contained herein are subject to change without notice. Should any question arise concerning these provisions, the purchaser/user should contact Flowserve Corporation at any one of its world-wide operations or offices.

© 2003 Flowserve Corporation, Irving, Texas, USA. Flowserve and Edward Valves are registered trademarks of Flowserve Corporation. V-Rep 84-1 3/03 Printed in USA

FLOWSERVE CORPORATIONFLOW CONTROL DIVISIONEdward Valves1900 South Saunders StreetRaleigh, NC 27603 USA

Toll- Free Telephone Service(U. S. and Canada)Day: 1-800-225-6989

After Hours Customer Service1-800-543-3927

US Sales OfficesPhone: 919-832-0525Facsimile: 919-831-3369Facsimile: 919-831-3376

Visit Our Websitewww.edwardvalves.com

For more information about Flowserve Corporation, contact www.flowserve.com or call USA 1-800-225-6989.

Flow Control Division

Edward Valves

Documents

Aluminium Embrittlement