9HSTFMG*agddei+

ISBN 978-952-60-6334-8 (printed) ISBN 978-952-60-6335-5 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) Aalto University School of Engineering Department of Engineering Design and Production www.aalto.fi

BUSINESS + ECONOMY ART + DESIGN + ARCHITECTURE SCIENCE + TECHNOLOGY CROSSOVER DOCTORAL DISSERTATIONS

Aalto-D

D 115

/2015

Metastable austenitic stainless steels possess an excellent combination of strength and elongation due to formation of strain-induced martensite which enhances work hardening. Low-nickel austenitic stainless steels provide a cost-effective alternative for the conventional austenitic Fe-Cr-Ni grades. Unfortunately, metastable low-nickel austenitic stainless steels can be susceptible to delayed cracking after forming operations. This thesis describes research aiming to clarify the role of hydrogen, residual stresses and strain-induced martensite in delayed cracking, and to find reasons for the high susceptibility of metastable low-Ni austenitic stainless steels. The research includes comprehensive Swift cup forming tests for comparing the delayed cracking susceptibility of seven austenitic stainless steels and finding the critical conditions leading to cracking, and constant load tensile tests for studying the kinetics of the phenomenon.

Suvi Papula D

elayed cracking of metastable low

-nickel austenitic stainless steels A

alto U

nive

rsity

Department of Engineering Design and Production

Delayed cracking of metastable low-nickel austenitic stainless steels

Suvi Papula

DOCTORAL DISSERTATIONS

Aalto University publication series DOCTORAL DISSERTATIONS 115/2015

DelayedαCrackingαofαMetastableαLow-NickelαAusteniticαStainlessαSteelsα

Suvi Papula

A doctoral dissertation completed for the degree of Doctor of Science in Technology to be defended, with the permission of the Aalto University School of Engineering, at a public examination and debate held in auditorium K216 at Aalto University School of Engineering (Espoo, Finland) on the 2nd of October 2015, at 12 noon.

Aalto University School of Engineering Department of Engineering Design and Production Engineering Materials

Supervising professor Hannu Hänninen Thesis advisor Juho Talonen Preliminary examiners Prof. Dan Eliezer, Ben Gurion University of the Negev, Israel Prof. Staffan Hertzman, KTH Royal Institute of Technology, Sweden Opponents Prof. Dan Eliezer, Ben Gurion University of the Negev, Israel Prof. Veli-Tapani Kuokkala, Tampere University of Technology, Finland

Aalto University publication series DOCTORAL DISSERTATIONS 115/2015 © Suvi Papula ISBN 978-952-60-6334-8 (printed) ISBN 978-952-60-6335-5 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) http://urn.fi/URN:ISBN:978-952-60-6335-5 Unigrafia Oy Helsinki 2015 Finland

Abstract Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi

Author SuviαPapulaαName of the doctoral dissertation DelayedαCrackingαofαMetastableαLow-NickelαAusteniticαStainlessαSteelsαPublisher SchoolαofαEngineeringαUnit DepartmentαofαEngineeringαDesignαandαProductionα

Series AaltoαUniversityαpublicationαseriesαDOCTORALαDISSERTATIONSα115/2015α

Field of research EngineeringαMaterialsα

Manuscript submitted 15αAprilα2015α Date of the defence 2αOctoberα2015α

Permission to publish granted (date) 15αJuneα2015α Language Englishα

Monograph Article dissertation (summary + original articles)

Abstract

Keywords Metastableαlow-nickelαausteniticαstainlessαsteels,αdelayedαcracking,αhydrogen,αstrain-inducedαmartensite,αresidualαstressα

ISBN (printed) 978-952-60-6334-8α ISBN (pdf) 978-952-60-6335-5α

ISSN-L 1799-4934α ISSN (printed) 1799-4934α ISSN (pdf) 1799-4942α

Location of publisher Helsinkiα Location of printing Helsinkiα Year 2015α

Pages 165α urn http://urn.fi/URN:ISBN:978-952-60-6335-5α

Metastable austenitic stainless steels can be susceptible to delayed cracking after forming processes. The objective of this research was to explain the high susceptibility of low-Ni high-Mn austenitic stainless steels to delayed cracking, and to clarify the role of the different contributing factors, i.e. solute hydrogen, strain-induced α’-martensite and tensile residual stresses, in the phenomenon. Susceptibility of seven austenitic stainless steels, both conventional Fe-Cr-Ni grades and low-Ni grades, to delayed cracking was investigated by means of deep drawing Swift cup tests. Residual stresses in the cups were measured with X-ray diffraction and a ring slitting method. Volume fraction of strain-induced α’-martensite was determined with a Ferritescope. Hydrogen content of the test materials was analysed with hot extraction, melt extraction and thermal desorption spectroscopy. Hardness of α’-martensite and austenite phases was measured using nanoindentation. Additionally, a constant load tensile testing arrangement was developed and applied for systematic study on the role of different contributing factors in delayed cracking kinetics. The presence of α’-martensite was a necessary prerequisite for delayed cracking to occur in austenitic stainless steels with typical internal hydrogen concentrations (< 5.5 wppm). Cracking proceeded through α’-martensite phase. Martensitic transformation substantially increased the magnitude of residual stresses in deep-drawn cups. Alloying elements of the stainless steels influenced the sensitivity to delayed cracking through their effect on the austenite stability and properties of α’-martensite. According to nanoindentation measurements the hardness of α’-martensite correlated with the level of residual stresses. Critical combinations of residual stress, α´-martensite and hydrogen content for delayed cracking were specified for each test material. Explanations for the high susceptibility of low-Ni grades were their high residual stresses after forming, and high internal hydrogen content. Lowering the hydrogen content by annealing markedly lowered the risk of delayed fracture. Constant load tensile testing results demonstrated the role of α’-martensite as a medium for hydrogen diffusion. Hydrogen content seemed to have the strongest effect on time to fracture, which supports the assumption that cracking requires hydrogen accumulation at regions of high stress. The results of this research work can be utilized in the design of demanding forming applications, to avoid the risk of delayed fracture. Increased understanding of the delayed cracking phenomenon enables the development of novel cost-efficient, high-strength stainless steels and wider usage of current austenitic stainless steels.

Tiivistelmä Aalto-yliopisto, PL 11000, 00076 Aalto www.aalto.fi

Tekijä SuviαPapulaαVäitöskirjan nimi ViivästynytαmurtuminenαmetastabiileissaαmatalanikkelisissäαausteniittisissaαruostumattomissaαteräksissäαJulkaisija InsinööritieteidenαkorkeakouluαYksikkö Koneenrakennustekniikanαlaitosα

Sarja AaltoαUniversityαpublicationαseriesαDOCTORALαDISSERTATIONSα115/2015α

Tutkimusala Koneenrakennuksenαmateriaalitekniikkaα

Käsikirjoituksen pvm 15.04.2015α Väitöspäivä 02.10.2015α

Julkaisuluvan myöntämispäivä 15.06.2015α Kieli Englantiα

Monografia Yhdistelmäväitöskirja (yhteenveto-osa + erillisartikkelit)

Tiivistelmä

Avainsanat Metastabiiliα nikkelinenαausteniittinenαruostumatonαteräs,αviivästynytαmurtuminen,αvety,αvenymänαaiheuttamaαmartensiitti,αjäännösjännitysα

ISBN (painettu) 978-952-60-6334-8α ISBN (pdf) 978-952-60-6335-5α

ISSN-L 1799-4934α ISSN (painettu) 1799-4934α ISSN (pdf) 1799-4942α

Julkaisupaikka Helsinkiα Painopaikka Helsinkiα Vuosi 2015α

Sivumäärä 16 α urn http://urn.fi/URN:ISBN:978-952-60-6335-5α

Metastabiilit austeniittiset ruostumattomat teräkset voivat olla alttiita viivästyneelle murtumalle muovauksen jälkeen. Tämän tutkimuksen tavoitteena oli selittää matalan nikkelipitoisuuden austeniittisten ruostumattomien terästen korkeaa alttiutta viivästyneelle murtumiselle ja selvittää ilmiöön vaikuttavien tekijöiden, kuten materiaaliin liuenneen vedyn, muovauksen aiheuttaman α’-martensiitin ja jäännösjännitysten merkitys. Seitsemän austeniittisen ruostumattoman teräksen, sekä Fe-Cr-Ni- että matalanikkelisten lajien, alttiutta viivästyneelle murtumiselle tutkittiin Swiftin kuppikokeiden avulla. Syvävedetyistä kupeista määritettiin jäännösjännitykset röntgendiffraktiolla ja jännitysten laukeamiseen perustuvalla menetelmällä. Muovauksen aiheuttaman α’-martensiitin määrä mitattiin ferriittimittarilla. Koemateriaalien vetypitoisuus määritettiin kolmella eri menetelmällä. Muovatuista materiaaleista mitattiin α’-martensiitin ja austeniitin kovuus nanoindentaatiolla. Lisäksi kehitettiin koejärjestely vakiovoima-vetokeisiin, joilla tutkittiin systemaattisesti viivästyneeseen murtumiseen vaikuttavien tekijöiden roolia ilmiön kinetiikassa. Viivästynyttä murtumista ei tapahtunut stabiileissa austeniittisissa ruostumattomissa teräksissä, joihin ei muodostunut muovauksessa α’-martensiittia, tyypillisillä valmistuksen jälkeisillä vetypitoisuuksilla (< 5.5 wppm). Viivästyneet murtumat etenivät α’-martensiittifaasia pitkin. Nanoindentaatiomittausten perusteella α’-martensiitin kovuus korreloi jäännösjännitystason kanssa. Tutkittaville materiaaleille määritettiin kriittiset jäännösjännitysten, α’-martensiitin ja vetypitoisuuden tasot, jotka aiheuttavat viivästyneitä murtumia. Matalan nikkelipitoisuuden terästen korkeaa alttiutta viivästyneelle murtumiselle selittävät niiden korkea jäännösjännitystaso ja korkea vetypitoisuus. Vetypitoisuuden alentaminen hehkutuksella alensi merkittävästi murtumisriskiä. Vakiovoima-vetokokeiden tulokset havainnollistivat α’-martensiitin roolia vedyn diffuusiossa. Vetypitoisuudella oli suuri vaikutus murtumisaikaan, mikä tukee käsitystä, että murtuminen edellyttää vedyn kerääntymistä jännityskeskittymien alueelle. Tämän tutkimuksen tuloksia voidaan hyödyntää viivästyneen murtumisen välttämiseksi suunniteltaessa vaativia muovaussovelluksia. Ilmiön syvällinen ymmärtäminen tukee uusien kustannustehokkaiden, lujien ruostumattomien terästen kehittämistä sekä olemassa olevien teräslaatujen laajempaa käyttöä.

5

matala

7

Preface

This research was done as part of the Finnish Metals and Engineering Compe-tence Cluster (FIMECC) Light and Efficient Solutions research program fundedby the Finnish Funding Agency for Technology and Innovation (Tekes) andFinnish industrial companies. I would like to express gratitude for FIMECC Ltd,Tekes, Outokumpu Oyj and Doctoral Program in Concurrent Mechanical Engi-neering financed by the Ministry of Education and the Academy of Finland formaking this research work possible. Financial support enabling the completionof the work was also provided by the Federation of Finnish Technology Indus-tries, the Walter Ahlström Foundation and the KAUTE Foundation.

The research work was carried out in the Laboratory of Engineering Materials atAalto University School of Engineering. I want to express my gratitude to mysupervisor, Prof. Hannu Hänninen, for his patient guidance, support and encour-agement throughout the project. I would also like to express my gratitude to Dr.Juho Talonen from Outokumpu Oyj for providing me this interesting and chal-lenging research topic and for extensive guidance throughout the research pro-ject and valuable comments. I want to address special thanks to my colleagueTimo Kiesi for instructive discussions and for assistance and help in performingthe residual stress measurements. I am deeply grateful to Tapio Saukkonen whoprovided his great expertise in the SEM and EBSD measurements. Warm thanksalso to Olga Todoshchenko for the TDS measurements, to Kim Widell for hisassistance in the tensile testing experiments, and to Janne Peuraniemi and SeppoNurmi for their big work contribution in the machining of the test specimens. Inthe fine-tuning phase of this thesis I also received valuable guidance from Prof.Iikka Virkkunen, which I am grateful of.

Helsinki, 24 August 2015Suvi Kristiina Papula

9

Contents

Preface ........................................................................................................... 7

Original Features .......................................................................................... 11

List of Publications ...................................................................................... 12

Author’s Contribution .................................................................................. 13

1. Introduction ...................................................................................... 17

1.1 Delayed cracking of austenitic stainless steels ............................... 17

1.1.1 Effect of austenite stability ............................................................ 19

1.1.2 Effect of hydrogen ........................................................................ 21

1.1.3 Effect of residual stress ................................................................. 24

1.2 Low-nickel austenitic stainless steels............................................. 26

2. Aims of the Study ............................................................................. 27

3. Experimental Procedures .................................................................. 29

3.1 Test materials................................................................................ 29

3.2 Swift cup testing ........................................................................... 30

3.3 Martensite content measurement ................................................... 31

3.4 Hydrogen content analysis ............................................................ 32

3.5 Residual stress measurements ....................................................... 33

3.5.1 X-ray diffraction ....................................................................... 33

3.5.2 Ring slitting (deflection method) ............................................... 35

3.6 Constant load tensile testing .......................................................... 36

3.7 Scanning electron microscopy ....................................................... 38

3.8 Nanoindentation ............................................................................ 38

4. Results.............................................................................................. 40

4.1 Stability of austenite ..................................................................... 40

4.2 Hydrogen content.......................................................................... 41

4.3 Swift cup testing ........................................................................... 43

4.4 Residual stress measurements ....................................................... 50

4.4.1 X-ray diffraction ....................................................................... 50

10

4.4.2 Ring slitting ............................................................................... 56

4.4.3 Residual stress relaxation and the Villari effect .......................... 60

4.5 Scanning electron microscopy ....................................................... 60

4.5.1 Fractography.............................................................................. 60

4.5.2 EBSD measurements ................................................................. 63

4.6 Nanoindentation ............................................................................ 65

4.6 Constant load tensile testing........................................................... 68

5. Discussion ......................................................................................... 72

6. Conclusions....................................................................................... 81

References .................................................................................................... 83

Publications .................................................................................................. 95

11

Original Features

Delayed cracking of several metastable low-nickel austenitic stainless steels wasinvestigated, in order to clarify the role of the different contributing factors inthe phenomenon. The major part of the research consisted of a large test seriesof deep drawing Swift cup tests and thorough characterization of the cups. Re-sidual stresses present in the formed cups were determined by two different ap-proaches. The marked role of strain-induced martensitic transformation on theresidual stress state was demonstrated. Residual stresses were found to be de-pendent on the chemical composition, which affects both the amount and prop-erties of α’-martensite.

The research concentrated on the effects of internal hydrogen absorbed in stain-less steels during the production operations, and no hydrogen charging was ap-plied. Internal hydrogen contained in as-supplied low-nickel austenitic stainlesssteels was shown to be sufficient to cause delayed cracking, if high enoughamount of α’-martensite and high residual stresses were present. Lowering thehydrogen content in austenitic stainless steels by annealing treatments signifi-cantly improved their resistance to delayed cracking.

High susceptibility of metastable low-Ni Mn-alloyed austenitic stainless steelswas explained by their higher hydrogen content and higher residual stresses afterforming, in comparison to Fe-Cr-Ni grades. Critical combinations of hydrogen,strain-induced α’-martensite and residual tensile stresses for delayed crackingwere determined for the test materials.

A constant load tensile testing arrangement was developed and applied for sys-tematic examination of the effect of applied stress, α’-martensite content andinternal hydrogen content on delayed cracking kinetics. The results of a largetest series demonstrated the role of strain-induced α’-martensite as a medium forhydrogen transport and supported the view that delayed cracking initiation re-quires accumulation of a critical hydrogen concentration to highly stressed re-gions.

According to EBSD examination, localized martensitic transformation occurredahead of crack tip, and cracks propagated predominantly through α’-martensite,irrespective of its bulk volume fraction in the material.

12

List of Publications

This doctoral dissertation consists of a summary and of the following publica-tions which are referred to in the text by their Roman numerals

I Papula, S., Talonen, J., Hänninen, H., Effect of residual stress and strain-induced α’-martensite on delayed cracking of metastable austenitic stain-less steels, Metallurgical and Materials Transactions A, Vol. 45, Issue 3(2014), p. 1238-1246.

II Papula, S., Talonen, J., Todoshchenko, O., Hänninen, H., Effect of inter-nal hydrogen on delayed cracking of metastable low-Ni austenitic stain-less steels, Metallurgical and Materials Transactions A, Vol. 45, Issue 11(2014), p. 5270-5279.

III Papula, S., Talonen, J., Hänninen, H., Delayed cracking of metastablelow-nickel austenitic stainless steel studied with constant load tensiletesting, Fatigue and Fracture of Engineering Materials and Structures,DOI:10.1111/ffe.12301, in press

IV Papula, S., Saukkonen, T., Talonen, J., Hänninen, H., Delayed crackingof metastable austenitic stainless steels after deep drawing, accepted forpublication in ISIJ International, Vol. 55, No. 10 (2015)

V Pulkkinen, H., Papula, S., Todoshchenko, O., Talonen, J., Hänninen, H.,Effect of inclusions and precipitates on hydrogen embrittlement of Mn-alloyed austenitic stainless steels, Steel Research International, Vol. 84(2013), p. 966-974.

VI Ortega, S., Papula, S., Saukkonen, T., Talonen, J. and Hänninen, H.,Characterization of delayed cracking in deep-drawn Swift cups of meta-stable austenitic stainless steels, Fatigue and Fracture of EngineeringMaterials and Structures, Vol. 38, (2015), p. 29-39.

13

Author’s Contribution

Publication I: “Effect of residual stress and strain-induced α’-martensite on de-layed cracking of metastable austenitic stainless steels”

The research was planned and performed in close cooperation with OutokumpuStainless Oy. The Swift cup tests were performed by personnel at OutokumpuResearch Centre, as well as the hydrogen content analyses. The author performedthe examination of the deep drawn cups, the evaluation of the α’-martensite con-tent, the measurement of residual stresses by two separate methods and the spec-imen preparation for scanning electron microscopy. The analysis of the resultswas done in conjunction with the co-outhors. Electron microscopy was carriedout by Mr. Tapio Saukkonen. The significant effect of strain-induced α’-marten-site on the residual stress state of deep drawn cups and on the susceptibility ofmetastable austenitic stainless steels to delayed cracking was demonstrated. Thearticle was mainly written by the author and the co-authors participated in thediscussion of the results.

Publication II: “Effect of internal hydrogen on delayed cracking of metastablelow-Ni austenitic stainless steels”

The research was performed in close cooperation with Outokumpu Stainless Oy.The author performed the heat treatments and preparation of the specimens forSwift cup tests, the examination of the deep drawn cups, the evaluation of theirα’-martensite content and the residual stress analysis. The author also did thespecimen preparation for hydrogen content analyses and for EBSD measure-ments. SEM and EBSD measurements were carried out by Mr. Tapio Saukko-nen. The hydrogen content analyses were done at VTT Technical Research Cen-tre of Finland, the melt extraction measurements at Outokumpu Research Centreand the thermal desorption spectroscopy measurements by the co-author Mrs.Olga Todoshchenko. Examination of the results revealed the effect of internalhydrogen content on delayed cracking. Lowering the internal hydrogen contentby annealing improved the resistance of metastable austenitic stainless steels,particularly of the low-Ni grades, to delayed cracking. The article was writtenby the author and the co-authors participated in the discussion of the results.

14

Publication III: “Delayed cracking of metastable low-nickel austenitic stainlesssteel studied with constant load tensile testing”

The constant load tensile test arrangement for this research was designed in co-operation with Dr. Juho Talonen, and the author was responsible for the imple-mentation. The research work, including heat treatments, specimen preparation,α’-martensite content measurement, constant load tensile testing and interpreta-tion of the results was predominantly carried out by the author. The SEM exam-ination and EBSD analysis were performed by Mr. Tapio Saukkonen. The de-veloped test arrangement allowed a systematical study on the effect of differentfactors on delayed cracking kinetics. The results support the view that delayedcracking initiation requires accumulation of a critical hydrogen concentration tohighly stressed regions. The article was written by the author and the co-authorsparticipated in the discussion of the results.

Publication IV: “Delayed cracking of metastable austenitic stainless steels afterdeep drawing”

The Swift cup tests were performed by personnel in the Outokumpu ResearchCentre, as well as the hydrogen content analyses. The author carried out the ex-amination of the deep drawn cups and measurement of their α’-martensite con-tent, the residual stress measurements by X-ray diffraction, the nanoindentationmeasurements as well as specimen preparation for scanning electron microscopyand EBSD analyses. The author also performed the tensile testing for the deter-mination of the anisotropy coefficients of the test materials. It was found that themagnitude of residual stresses and the susceptibility to delayed cracking dependson the chemical composition and hardness of the materials. The article was writ-ten by the author and the co-authors participated in the discussion of the results.

Publication V: “Effect of Inclusions and Precipitates on Hydrogen Embrittle-ment of Mn-Alloyed Austenitic Stainless Steels”

The experiments were mostly done by Mr. Heikki Pulkkinen. The INCA anal-yses were performed at the Outokumpu Research Centre, and the TDS measure-ments by Mrs. Olga Todoshchenko. The author was responsible for specimenpreparation for the hydrogen content analyses and the TDS measurements, andparticipated in the analysis of the results and the writing of the article.

15

Publication VI: “Characterization of delayed cracking in deep-drawn Swiftcups of metastable austenitic stainless steels”

The experiments and calculations based on macroscopy and optical metallog-raphy were performed by Ms. Sandra Ortega. Scanning electron microscopy ex-amination and EBSD analysis was performed by Mr. Tapio Saukkonen. The au-thor was responsible for the determination of residual stresses and α’-martensitecontent of the Swift cups. The article was jointly written by Ms. Sandra Ortegaand the author.

17

1. Introduction

Novel high-strength stainless steels can provide means to reach substantialweight savings and increased performance in various applications of movingequipment and machinery. Metastable austenitic stainless steels, in which aus-tenite is partly transformed to martensite under plastic strain, possess an excel-lent combination of strength and elongation, as formation of strain-induced α’-martensite increases work hardening of the material. Because of their high en-ergy-absorbing capacity, they have become attractive materials to be used e.g.in automotive crash-relevant structures. Generally, however, high strengthmeans increased susceptibility to hydrogen embrittlement phenomena, due togreater number of potential crack initiation sites and increased stress level actingat these sites (Pérez Escobar et al., 2012). Metastable austenitic stainless steels,particularly low-Ni grades, may be susceptible to delayed cracking after formingoperations (Schaller et al., 1972; Hoshino, 1980; Sumitomo, 1987; Frehn et al.,2003). Cracks may appear in successfully formed components after days or evenmonths from forming, such as deep drawing for example. Deep drawing is acommon sheet metal forming process used in the production of beverage cans,sinks, cooking pots and different kinds of automotive parts and panels. Delayedcracking phenomenon is related to several factors, including hydrogen containedin the material, strain-induced martensitic transformation, residual stresses, edgequality and chemical composition of the steel. The significance of the variousfeatures, as well as the exact mechanisms and kinetics of the cracking phenom-enon still remain unclear.

Delayed cracking is a problem especially in high-strength steels and certain aus-tenitic stainless steels. Failures caused by delayed cracking have been observed,for example, in bolts and fasteners made of high strength steel, pre-stressed con-crete steel bars, reinforcing steel rods and stainless steel clamps after a bendingforming process (Akiyama, 2012; Zinbi et al., 2010).

1.1 Delayedcrackingofausteniticstainlesssteels

Small amounts of hydrogen dissolved in stainless steel can cause embrittlementand loss of tensile ductility. Even when the quantity of hydrogen in solid solutionis too small to reduce tensile test ductility, hydrogen-induced delayed cracking

Introduction

18

may occur. Delayed cracking, also referred to as internal hydrogen embrittle-ment, occurs under static stress, applied or residual, below the yield stress of thematerial, without hydrogen undergoing any type of chemical reactions (Beckeret al., 2002; Kovalev et al., 2002). It can occur after a small average concentra-tion of hydrogen has been absorbed from the environment, for example duringthe manufacturing process. Hydrogen may enter steel from water contained inthe raw materials or in the furnace gases, during pickling in mineral acids orcathodic electrolytical cleaning, or during bright annealing.

Delayed cracking is considered a subcritical crack growth mechanism that pro-duces time-delayed fractures in formed components even with no externally ap-plied stress. The phenomenon is controlled by hydrogen diffusion to regions ofhigh tensile stress, and that is why there is a delay time, or incubation time, be-fore cracking occurs. The rate of crack advance is controlled by the supply andaccumulation of hydrogen in that region. Hydrogen embrittlement is enhancedby slow strain rates, suggesting that time-dependent diffusion is a controllingfactor (Becker et al., 2002).

Delayed cracking in austenitic stainless steels is related to coexistence of solutehydrogen, strain-induced α’-martensite and residual tensile stresses (Fig. 1). Theembrittling effect of hydrogen is strongly influenced by other variables, such asstrength (or hardness) level of the alloy, microstructure, magnitude of residualand/or applied stress, presence of a localized triaxial tensile stress, and amountof prior cold work. Applied strain should also be considered as an importantfactor in hydrogen embrittlement fracture of high-strength steels (Takagi et al.,2012).

Figure 1. Schematic representation showing a region for delayed cracking (modified from Tojiet al., 2010).

Residualstress

Regionofdelayedcracking Hydrogenα’-martensite

Introduction

19

1.1.1 Effectofaustenitestability

In metastable austenitic stainless steels martensitic transformation occurs as aresponse to plastic deformation, in which austenite is partially transformed tothermodynamically more stable α’-martensite. Deformation-induced α’-marten-site enhances the work-hardening rate of the material, which is desirable for goodformability, because the onset of necking is delayed and uniform elongation en-hanced, allowing greater strains (Talonen, 2007; Post et al., 2008). Conse-quently, the formation of strain-induced α’-martensite increases the tensilestrength of the stainless steel, and in some cases also the ductility (Lebedev etal., 2000; da Rocha et al., 2009). This strengthening mechanism, often called theTransformation-Induced Plasticity (TRIP) -effect, produces an excellent combi-nation of strength and elongation.

Two different martensite phases may exist in austenitic stainless steels: hexago-nal close-packed (hcp) ε-martensite and body-centered cubic (bcc) α’-marten-site. ε-martensite is often considered to be an intermediate (precursor) phase, thatreaches its highest volume fraction at relatively low strains and eventually trans-forms to α’-martensite at higher strains, so that at higher strains only α’-marten-site is observed (Lo et al., 2009; Das et al., 2009).

Hydrogen embrittlement sensitivity of austenitic stainless steels has generallybeen considered to be a function of austenite stability – the larger the strain-induced α’-martensite content the higher the risk for fracture (Schaller et al.,1972; Hoshino, 1980; Singh et al., 1982; Frehn et al., 2003). The existence ofα’-martensite, which is a harder and intrinsically more brittle phase than austen-ite, can increase markedly the embrittling effect of hydrogen (Hänninen et al.,1980; Chu et al., 1984). The martensitic structure is more sensitive to the pres-ence of hydrogen than the austenitic structure. When tested in hydrogen gas at-mosphere, stable austenitic stainless steels can also suffer from hydrogen em-brittlement, so that the existence of α’-martensite is not always a necessary con-dition for hydrogen-induced slow crack growth to occur (Pan et al., 2003; Mich-ler et al., 2010). But it is generally accepted that, if present, α’-martensite in-creases the susceptibility to embrittlement because it serves as a fast diffusionpath for hydrogen to potential crack initiation sites.

Hydrogen diffusion limits the kinetics of hydrogen embrittlement, and due to thelarge hydrogen diffusivity in the bcc phase, α’-martensite may act as a suitablemedium for entry into and transport of hydrogen within the solid and may alsooffer a suitable medium for crack propagation (Perng et al., 1986; Singh et al.,1982; Kanezaki et al., 2008). The diffusivity of hydrogen in bcc phases, such asα’-martensite or ferrite, is significantly greater, even up to eight orders of mag-nitude greater than in fcc phases, such as austenite (San Marchi et al., 2008).Hydrogen solubility in the martensite is, however, low so that hydrogen tends togather in the phase boundary between martensite and austenite making delayed

Introduction

20

cracking easier (Berrahmoune et al., 2006 A). When the interphase surface areais increased, the number of potential trapping sites for hydrogen accumulation isincreased as well.

Alloying elements of austenitic stainless steels affect the intensity of strain-in-duced austenite-martensite transformation due to their influence on stackingfault energy (SFE). Stacking fault energy describes the plastic deformation be-haviour of the material, the ability of dislocations in a crystal to glide onto anintersecting slip plane. It controls the formation of shear bands, and thus the for-mation of nucleation sites for the α’-martensite. The lower the SFE of the aus-tenitic stainless steel, the smaller is the austenite stability (Ratte, 2007; Kar-jalainen et al., 2008). SFE of austenitic stainless steels is a function of alloy com-position and temperature. In general, the SFE of austenitic stainless steels tendsto increase with increasing alloying. The influence of different alloying elementson the SFE is quite complicated and significant uncertainty concerning individ-ual elements still exists. Ni, Cu and Al tend to increase SFE of austenitic stainlesssteels, and Cr, Mn, Si, C and N to decrease it (Gonzalez et al., 2003; Ratte, 2007;Milititsky et al., 2008). The SFE decreases with increasing Cr content up to lev-els of 10-15% Cr. Nickel strongly increases the SFE up to about 12 % Ni content,while at higher contents the increase is comparatively small (Vitos et al., 2006).The effect of Ni is dependent on Cr content. Manganese decreases the SFE up tolevels of 10-15% Mn, but at higher levels there is a sharp increase in SFE (Ratte,2007). Recently it has been shown that the relative effect of an individual soluteelement significantly depends on the initial chemical composition of the solvent;the same alloying element can cause totally opposite changes in the SFE of al-loys with different host composition (Vitos et al., 2006). Therefore it is impos-sible to establish universal composition equations for the stacking fault energy.Increase in the amount of dissolved hydrogen also decreases the SFE of austeniteand increases the volume fraction of strain-induced α’-martensite produced informing operations (Sumitomo, 1987; Ferreira et al., 1996).

Alloying elements not only affect the volume fraction of α’-martensite, but alsoits properties. Carbon and nitrogen, in particular, increase the hardness and brit-tleness of the α’-phase (Sumitomo, 1987; Berveiller et al., 2009). According toa US-patent (Fujioka et al., 1975), delayed cracking in metastable austeniticstainless steels can be avoided by limiting the total content of carbon and nitro-gen to less than 0.04 wt-%. Phosphorous has also been reported to increase thebrittleness of α’-martensite (Hoshino, 1980) and low nickel content has a similareffect (Sumitomo, 1987).

Susceptibility of austenitic stainless steel to internal hydrogen embrittlement hasbeen shown to depend strongly on the nickel content (Zhang et al., 2012). Nickelplays an important role in deformation processes that affect hydrogen-assistedfracture (San Marchi et al., 2008). Nickel increases the stability and the stackingfault energy of austenite, which promotes cross slip and thus reduces planar slip

Introduction

21

participation in fracture to favour a macroscopically more ductile appearance(Teus et al., 2008). Presumably, sufficiently high Ni content can mitigate thenegative effect of hydrogen; the combined effect of austenite stabilisation andpromotion of cross slip leads to the high hydrogen embrittlement resistance andmacroscopically ductile fracture behaviour of stable austenitic stainless steels(Michler et al., 2009). However, nickel has been shown to segregate in Cr-Niaustenitic stainless steels, leading to a heterogeneous microstructure and varia-tion in the distribution of α’-martensite (Michler et al., 2009). Carbon and nitro-gen, in particular, co-segregate with nickel (Weber et al., 2010).

According to some published studies, nitrogen may have a negative influence onhydrogen embrittlement of austenitic stainless steels (Gavriljuk et al., 2005;Michler et al., 2010; Weber et al., 2011). The effect of nitrogen is related to itsinfluence on the electronic structure and increase of the concentration of freeelectrons. Hydrogen and nitrogen have a similar effect of enhancing the mobilityof dislocations leading to localized plasticity.

1.1.2 Effectofhydrogen

Internal hydrogen can be considered to be the principal cause of delayed crack-ing. The severity of the phenomenon, indicated by the number of cracks appear-ing in deep drawn cups, has been found to depend directly on the hydrogen con-tent of the material (Sumitomo, 1987).

Several possible mechanisms have been proposed to explain hydrogen embrit-tlement of high-strength alloys. The oldest of the HE mechanisms is hydrogen-enhanced decohesion (HEDE), which is often considered the basic damagemechanism for internal-hydrogen-assisted cracking. The model proposes thathigh, localized concentrations of hydrogen can weaken the interatomic cohesiveforces between metal atoms at or ahead of crack tips (Gangloff, 2003). Themodel provides the basic notion that hydrogen damage occurs in the fractureprocess zone when the local crack-tip-opening tensile stress exceeds the maxi-mum local atomic cohesion strength, lowered by the presence of hydrogen (Ori-ani, 1972). This mechanism, however, has been considered unlikely to be re-sponsible for hydrogen embrittlement in austenitic stainless steels (Gavriljuk etal., 2003).

According to another proposed mechanism, the hydrogen enhanced localizedplasticity (HELP), dissolved hydrogen in stainless steels enhances dislocationmobility and slip planarity, which can lead to heterogeneously localized plasticstrain and stress concentrations, presumably sufficient to enable subcritical crackgrowth. In the regions of high hydrogen concentration the flow stress is de-creased and slip occurs at stresses well below those required for plastic defor-mation in other parts of the specimen, i.e., slip localization occurs in the vicinity

Introduction

22

of the crack tip (Pan et al., 2003; Martin et al., 2011 A). This leads to highlylocalized failure by ductile processes, while the macroscopic deformation re-mains small, because of reduced macroscopic ductility (Birnbaum et al., 1994;Sofronis et al., 2001; Robertson, 2001; Lai et el. 2013). The locally enhancedplasticity at the crack tip has been observed in careful fractographic examina-tions with high resolution techniques (Nibur et al., 2009). The influence of hy-drogen on the behaviour of dislocations has been demonstrated by in situ TEMdeformation experiments (Robertron et al., 2015). Hydrogen affects the stressfield of dislocations, allowing them to move at a lower stress level in certaindirections. Hydrogen has been shown to increase the concentration of free elec-trons, i.e. to enhance the metallic character of atomic interactions, which assiststhe increased plasticity (Gavriljuk et al., 2003). Therefore, the relaxation ofstresses at the crack tip due to dislocation slip, i.e. crack blunting, is expectedearlier than brittle fracture. The local increase in the concentration of free elec-trons results in enhanced mobility of dislocations, dislocation multiplication andincreased number of dislocations in the pile-ups, promoting the opening of mi-crocracks under lower applied stresses (Gavriljuk et al., 2003). The hydrogenenhanced localized plasticity model has been considered the most importantmechanism of hydrogen degradation in austenitic stainless steels (Rozenak et al.,1990; Shivanyuk et al., 2001; Nibur et al., 2006; Lo et al., 2009).

Different proposed mechanisms of hydrogen embrittlement are often interre-lated. According to a detailed review (Lynch, 1988), both mechanisms, hydro-gen-enhanced decohesion and localized plasticity may be part of the fracturemechanism. Adsorption of hydrogen weakens the interatomic bonds at the cracktip, thereby facilitating the nucleation of dislocations so that crack-tip sourcesare activated before extensive dislocation activity occurs ahead of the crack.

In addition to the effect of hydrogen, the degree of localized deformation andmacroscopic ductility in austenitic stainless steels is strongly dependent on thenickel content (San Marchi et al., 2008; Michler et al., 2012). Due to enhancedheterogeneous planar slip with hydrogen present, the macroscopic fracture ap-pearance is “brittle” and involves either slip band interface cracking or mi-crocrack formation at slip band intersections and grain boundary or α’-γ inter-sections (Michler et al., 2009; Martin et al., 2011 A). Hydrogen embrittlementin metastable stainless steels causes typically transgranular fracture along strain-induced martensite structure (Zhang et el. 2008; Zhang et el. 2012). Elongatedvoids, shallow microvoids and cleavage facets with visible slip traces have beenobserved along the fracture surfaces (Michler et al., 2009; Berrahmoune et al.,2006 A).

Delayed cracking involves hydrogen transport to highly stressed areas (stressconcentrations) in the material microstructure under the influence of a stress gra-dient. Hydrogen can be transported by diffusion and by mobile dislocations. In-ternal hydrogen tends to accumulate at the crack tip area and thus enhance crack

Introduction

23

propagation (Fig. 2). The application of a stress causes a redistribution of dis-solved hydrogen from the surrounding microstructure to the crack-tip fractureprocess zone, which can result in strong gradients in hydrogen concentration andstress at the crack tip. Hydrogen accumulates in the crack-tip fracture processzone (FPZ) under the influence of two driving forces. First, the concentration ofhydrogen in interstitial lattice sites is increased proportional to an exponentialdependence on hydrostatic stress that dilates the lattice. Secondly, hydrogen alsoaccumulates at the crack tip due to trapping associated with the high density ofdislocations present owing to intense localized plastic deformation. This lattercontribution is dominant provided that crack-tip blunting is significant, typicalof lower strength alloys with limited gradient plasticity enhancement of the hy-drostatic stress. In high-strength alloys the hydrostatic enhancement of hydrogenaccumulation is emphasized. (Kovalev et al., 2002; Gangloff, 2003).

Figure 2. Schematic illustration of hydrogen diffusion path from surface (a) and near a crack tip(b). Martensite acts as a pathway for hydrogen diffusion. (Kanezaki et al., 2008)

Generally, absorbed hydrogen is not homogeneously distributed inside the ma-terial. Hydrogen may reside either at interstitial lattice sites or it can be trappedat various microstructural defects. Hydrogen prefers to occupy crystal lattice de-fects with increased specific free volume, such as vacancies, vacancy clusters,dislocation cores and grain boundaries. The solid solubility of hydrogen in steelis very low at ambient temperature, and most hydrogen is in trapped states atdislocations, grain boundaries, inclusions, voids, interfaces or impurity atoms(Perng et al., 1989; Nagumo et al., 1999, Nagumo et al., 2003). Hydrogen canbe effectively trapped at defects introduced by deformation. According to a pro-posed vacancy agglomeration model, increased density of vacancies in thecourse of plastic straining and their agglomeration promoted by hydrogen leads

Introduction

24

to promotion of failure (Nagumo, 2001; Nagumo et al., 2003; Hatano et al.,2014).

High tensile strength and strong hydrogen trapping frequently correlate becausenanoscale features/defects that strengthen an alloy often provide effective sitesfor hydrogen segregation (Gangloff, 2003). Additionally, in multi-phase high-strength materials phase transformations cause distortions in the crystal latticethat are potential hydrogen trapping sites (Pérez Escobar et al., 2012). Reversi-ble, weak hydrogen trapping sites with low activation energy for detrapping, e.g.grain boundaries, dislocations and coherent carbide interfaces are primarily con-trolling HE resistance in high-strength steels (Chun et al., 2012). Large, unfa-vourably located particles may accumulate hydrogen to preferential crackingsites in the regions which are intrinsically more prone to brittleness. Manganesesulphides are one of the most effective traps for hydrogen (Kovalev et al., 2002).On the other hand, hydrogen trapping in strong, irreversible traps, e.g. incoherentsmall precipitates that are homogeneously distributed in the matrix, can be uti-lized to reduce the sensitivity of steel to hydrogen-induced delayed cracking(Mohrbacher, 2008). Reversible traps can affect the kinetics of hydrogentransport, and sometimes reduce the susceptibility to delayed cracking by in-creasing the time necessary to reach a critical local hydrogen concentration.

Hydrogen diffusivity and solubility are key parameters in hydrogen embrittle-ment and delayed cracking. Parameters of hydrogen transport, e.g. diffusivityand permeability, in stainless steels at room temperature have been found to berelatively insensitive to chemical composition (Perng et al., 1986; Perng et al.,1990; Kumar et al., 1997). It has been reported by several researchers, that in-creasing Mn alloying in steels increases hydrogen solubility in austenite(Nagumo, 2001; San Marchi et al., 2007; Lob et al., 2011; Todoshchenko et al.,2012) and this may explain the higher susceptibility of high-Mn steels to hydro-gen embrittlement phenomena. Mn alloying has also been found to increase hy-drogen mobility (Ismer et al., 2010). However, there is also evidence of lowerhydrogen diffusivity in high-Mn low-Ni grades like AISI 201 in comparison toCr-Ni grades (Yagodzinskyy et al., 2009; Yagodzinskyy et al., 2011).

1.1.3 Effectofresidualstress

Residual stresses play an important role in delayed cracking (Yuying et al., 1992;Hong et al., 2013 B). Residual stress is an internal stress which remains in a bodythat is stationary and at equilibrium with its surroundings. Residual stresses arisefrom the elastic response of the material to an inhomogeneous distribution ofnonelastic strains. In all metal forming processes, residual stresses are generatedin the material. Residual stresses, which are a consequence of interactions be-tween deformation, temperature and microstructure, are caused by inhomogene-ous plastic strain levels at different regions of the product (Wang Z. et al., 2002).

Introduction

25

The difference in the actual strain level in different locations may be caused byseveral reasons, including a difference in strength between the co-existent phasesin the material, due to die/mold shape or constraints from the gripping force onthe workpiece, or by temperature gradients (Wang Z. et al., 2002).

Residual stresses are also introduced to metallic materials by phase transfor-mations. Plastic strains are inhomogeneously partitioned between the phases dueto the constraining effect of the stronger phases on the weaker ones. In metasta-ble austenitic stainless steels martensitic transformation occurs as a response toplastic deformation, in which austenite is partially transformed to thermodynam-ically more stable α’-martensite. The strain-induced martensitic transformationcauses local strain fields and therefore creates varying internal stresses in boththe austenite and the α’-martensite phase (Withers et al., 2001 B; Berrahmouneet al., 2006 B). Martensitic transformation is a diffusionless shear process, whichis accompanied by shear strains and a volumetric change. The specific volumeof α’-martensite is about 2% larger than that of austenite (Taran et al., 2004).This martensite transformation induced volume dilation, in addition to the elasticmismatch and the plastic misfit between α’-martensite and austenite, are all caus-ing considerable stresses throughout the material (Withers et al., 2001 B). Theα’-martensite phase having higher hardness and yield strength has a higher stresslevel than the austenite phase (Talonen, 2007).

Residual stresses consist of macro- and microstress components. Macrostresses,caused primarily by inhomogeneous plastic deformation at different parts of aformed component, vary slowly on a scale that is large compared to the mate-rial’s microstructure. Microstresses, which vary on the scale of the material’smicrostructure, are caused by imperfections in the crystal lattice and inhomoge-neous partitioning of plastic strain between different phases, due to the constrain-ing effect of the stronger phases on the weaker ones, and local differences indislocation density. The degree of plastic deformation in the phases is differentdue to differences in yield strength: the softer phase begins to yield earlier andexperiences larger plastic strains. In previous residual stress studies of plasticallydeformed dual-phase materials it has been found that the residual microstressesare tensile in the harder phase and compressive in the softer phase (Johansson etal., 2000; Taran et al., 2004; Berrahmoune et al., 2004; Che et al., 2007). Mi-crostresses in polycrystalline materials are also grain orientation dependent andresidual stresses within each grain of a deformed material can have large varia-tion depending on the orientation of the grains (Wang Y. et al., 2002). In mate-rials that have undergone large deformation, crystallographic texture introducesan additional stress component caused by stress/strain incompatibility betweengrains of different orientations.

Delayed cracking is a time-dependent hydrogen redistribution process driven bystress (Guo et al., 2011). Stresses promote nucleation of new reversible hydrogen

Introduction

26

traps. Concentration of tensile stress and dissolved hydrogen initiates nucleationof crack and promotes its propagation.

1.2 Low-nickelausteniticstainlesssteels

High and volatile price of nickel has during the recent years made producers andend-users of stainless steels to seek low-nickel alternatives for conventionalnickel containing stainless steel grades. Substantial cost savings can be achievedby reducing nickel alloying in stainless steels. Nickel accounts for a significantpercentage of the raw material cost and the energy used in the production ofaustenitic stainless steels (Norgate et al., 2004; Johnson et al., 2008). The globalmarket share of low-nickel Cr-Mn stainless steels has been constantly growing,being currently about 15-20 %, according to ISSF Annual Review 2013.

The most commonly used substitutes for nickel, for achieving austenitic micro-structure, are manganese, carbon and nitrogen. Although Mn is considered anaustenite stabilizer, the addition of Mn alone is not sufficient to stabilize the aus-tenite phase at room temperature, especially in the presence of Cr which is astrong ferrite stabilizer (Milititsky et al., 2008). Therefore, the chromium contentin low-Ni austenitic stainless steels has to be reduced, which has a negative effecton corrosion resistance (Cutler et al., 2008). Nevertheless, Mn addition effec-tively increases the solubility of nitrogen in the liquid steel, increasing the frac-tion of austenite formed during solidification. Nitrogen is a strong austenite sta-bilizer and solid solution strengthener, and also has a positive effect on the pit-ting corrosion resistance (Simmons, 1996). Copper is also added in some low-nickel grades to stabilize austenite and to improve their formability (Sibanda etal., 1994; Gonzalez et al., 2003).

Cr-Mn-N alloyed austenitic stainless steels provide a cost-effective alternativefor the conventional austenitic grades. Their mechanical properties are similar,yield strength even higher, in comparison to the 300-series grades. Some low-nickel grades have poorer corrosion resistance and/or formability. Metastablelow-nickel austenitic stainless steels may be susceptible to delayed cracking afterforming operations (Schaller et al., 1972; Hoshino, 1980; Sumitomo, 1987;Frehn et al., 2003). This has been one of the most important factors limitingincreasing use of low-nickel austenitic stainless steels.

27

2. Aims of the Study

This study was part of a research project that focused on the development of newhigh-strength carbon steels and stainless steels. There is growing demand forthese steels for developing lightweight structures in many application areas, forinstance, in the transport industry. Metastable austenitic stainless steels possessan excellent combination of strength and elongation, and a good energy absorp-tion capacity, which is important in a situation of collision. Generally, however,high strength means increased susceptibility to hydrogen embrittlement phenom-ena, and exploitation of novel high-strength stainless steels presumes detailedexamination of the relationships between chemical composition, microstructure,deformation mechanisms and hydrogen-induced delayed cracking.

The objective of this research was to clarify the reasons for high susceptibilityof metastable low-Ni austenitic stainless steels to delayed cracking and to exam-ine the role of the different factors, namely hydrogen, austenite stability and re-sidual stresses, in the phenomenon. The research included comprehensive Swiftcup forming tests and constant load tensile tests enabling a systematic study ofthe different influencing factors. The research concentrated on the effects of in-ternal hydrogen present in stainless steels as an inevitable impurity after produc-tion operations (typically < 5 wppm), and no hydrogen charging was applied.Several different austenitic stainless steels, both low-Ni high-Mn grades andconventional Cr-Ni grades, were studied in this work. The aim was to determinethe limiting conditions for delayed cracking, i.e. combinations of hydrogen con-tent, stress level and α’-martensite volume fraction, for the test materials. Thisknowledge can be utilized in the design of demanding forming applications, toavoid the risk of delayed fracture. Increased understanding of the delayed crack-ing phenomenon in metastable low-nickel austenitic stainless steels enables thedevelopment of novel cost-efficient, high-strength stainless steels and the widerusage of current austenitic stainless steels.

29

3. Experimental Procedures

3.1 Testmaterials

In this research, seven industrially processed austenitic stainless steels with dif-ferent chemical compositions and thus different austenite stability were investi-gated. The chemical compositions of the steels are presented in Table 1. Nickelcontent of the steels varied over a wide range, from 1.1 to 8.2 wt-%. Manganesecontent varied from 1.22 to 9.0 wt-%. Manganese is an austenite stabilizer usedfor replacing nickel. Nitrogen is a strong austenite stabilizer and solid solutionstrengthener. Copper is also used for stabilizing austenite and for improvingformability. The surface finish in all the test materials was 2B: cold rolled, an-nealed, pickled and skin passed.

Table 1. Chemical compositions of the test materials in weight percents.Grade C Cr Ni Mn Si Cu Mo N P S Fe301 0.093 16.7 6.4 1.22 1.12 0.25 0.67 0.074 0.024 0.001 bal301LN 0.030 17.4 6.6 1.23 0.50 0.17 0.19 0.168 0.028 0.001 bal304 0.049 18.2 8.2 1.49 0.43 0.43 0.13 0.047 0.032 0.002 bal201(a) 0.045 17.6 4.5 6.97 0.35 0.25 0.09 0.198 0.041 0.003 bal201(b) 0.047 17.0 3.7 7.16 0.29 0.21 0.11 0.217 0.034 0.002 bal201Cu 0.047 17.3 4.7 5.7 0.29 2.39 0.21 0.107 0.028 0.002 bal204Cu 0.079 15.2 1.1 9.00 0.40 1.68 0.03 0.115 0.032 0.004 bal

Mechanical properties of the test materials, provided by the producer, are pre-sented in Table 2. The ultimate tensile strength is highest in 301, 201(b) and204Cu. Grade 201(b) has the highest yield strength, which is probably due to itshigh nitrogen content and small grain size.

Table 2. Mechanical properties and grain size of the test materials.Grade Thickness

(mm)Rp0.2

(MPa)Rm (MPa) Ag (%) A80 (%) Grain size

(μm)301 1.3 370 823 49.5 54 13.1301LN 1.5 374 764 48.2 52 15.2304 1.0 313 651 50.8 56 10.9201(a) 0.7 423 773 42.1 49 9.5201(b) 0.8 488 835 41.4 47 6.5201Cu 1.0 330 632 44.7 54 14.3204Cu 1.0 367 821 51 54 16.4

Experimental Procedures

30

3.2 SwiftcuptestingSwift cup test is a common test for simulating drawing conditions that exist inpress-forming operations. It is used to evaluate the formability of sheet metalsby deep drawing a series of blanks of increasing diameter. It is also used forevaluating the susceptibility of certain materials to delayed cracking. In Swiftcup test, a flat circular blank cut from the material to be tested, is clamped intoa blank holder under a certain pressure and the central part of the sheet is forcedwith a cylindrical punch through a die to form a cup (Fig. 3). Deep drawing ofcircular cups can be viewed as two processes; stretching a sheet over a circularpunch and drawing an annulus inwards (Marciniak et al., 2002). Cylindrical cupwall transmits the force between these two regions. The maximum blank sizethat can be successfully drawn is used to calculate the limiting drawing ratio(LDR), which is defined as the maximum blank diameter divided by the punchdiameter.

Figure 3. Standard tooling for flat-bottomed Swift cup test (Miles, 2006).

When studying the susceptibility of materials to delayed cracking, appearanceof cracks in the cups after the drawing process is visually examined (Fig. 4). Thelargest cup without any cracks, or more precisely with no cracks in any of severalparallel cups of the same drawing ratio, after sufficiently long observation timedefines the limiting drawing ratio for delayed cracking (LDR-DC).

Experimental Procedures

31

Figure 4. Deep drawn cups of stainless steel 301 at drawing ratios 1.8, 1.9, 2.0 and 2.12.

During a deep drawing operation, the metal sheet is subjected to different typesof stresses. In the flange there are radial tensile stresses due to the blank beingpulled into the die cavity and a compressive stress normal to the sheet which isdue to the blank-holder pressure (Hussaini et al., 2014). The radial tensilestresses lead to compressive hoop stresses in the flange, because of the reductionin the circumferential direction. Cup wall is primarily experiencing a longitudi-nal tensile stress, as the punch transmits the drawing force through the walls ofthe cup and through the flange as it is drawn into the die cavity. The tensile stressin the cup wall is dependent on the blank holding force and the friction betweenthe metal sheet and the blank holder. Additionally, there is a tensile hoop stresscaused by the cup being held tightly over the punch (Colgan et el., 2003).

High levels of residual stresses may exist in Swift cups after the deep drawingoperation. With finite element modelling (FEM) simulations it has been shownthat residual stresses introduced to the deep drawn cup are mainly caused by theunbending of the material when it leaves the draw die profile (Danckert, 1995;Fereshteh-Saniee et al., 2003).

Swift cup tests were carried out at Outokumpu Research Centre with Erichsen142/40 equipment. Laser cut circular steel blanks of varying diameter were deepdrawn into cups with a flat-bottomed cylindrical punch with 50 mm diameter.Produced drawing ratios of the cups ranged from 1.4 to 2.12. Novacel lubricantfilm and blank holder force of 25 kN were used. The tests were performed atroom temperature.3.3 Martensitecontentmeasurement

Volume fraction of ferromagnetic phase in the walls of the deep drawn cups andin the pre-strained tensile specimens was measured using a Ferritescope (FischerMP3C). The measurement is based on magnetic induction: a low frequency al-ternating current in the probe coil induces an eddy current in the studied metal.A magnetic field opposite to the field induced by the probe is built up in themetal and the relative magnetic permeability is measured. Ferritescope

Experimental Procedures

32

measurement can be performed in situ during mechanical testing to monitor themartensite content evolution.

The measured values of the ferromagnetic phase content were converted to α’-martensite contents by multiplying with a correction factor of 1.7 defined byTalonen et al. in a comprehensive study comparing the results of Satmaganmeasurement, magnetic balance, X-ray diffraction, density measurement andquantitative optical metallography (Talonen et al., 2004). This correction factorhas been verified by other researchers by comparing the Ferritescope readings tomagnetic saturation measurements (Beese et al., 2011). Effect of sheet thicknessand specimen curvature on the results was compensated by means of correctioncurves provided by the device manufacturer.

Ferritescope measurement is affected by the stress state of the material. The Vil-lari effect, also known as inverse magnetostriction, characterizes the change ofmagnetic permeability of a material under the influence of change in dimensionscaused by mechanical stress (Tumanski, 2011). The Villari effect is explainedby mechanically induced rotation of the domains of uniform polarization withina ferromagnetic material. Thus, magnetic permeability is a function of not onlymartensite volume fraction but also of elastic and plastic deformation. The effectof tensile residual stress on the magnetic properties of the stainless steels and,thus, on the measured ferromagnetic phase contents was taken into account bycorrection factors experimentally defined for each material. This was done bypre-straining tensile test specimens to different strain levels, measuring their fer-romagnetic phase contents after unloading and then performing in situ Fer-ritescope measurements during stepwise tensile reloading of the specimens. Dur-ing the test, the applied force was rised 1 kN at a time with a constant strain rateof 3x10-4 1/s, and held constant for 50 s after each increment, during which timefive Ferritescope measurements were taken in the middle part of the test speci-men. From the collected data, i.e. the average Ferritescope reading vs. appliedtensile stress, the correction factors were defined by dividing each average read-ing with the reading at zero load (initial state).

3.4 Hydrogencontentanalysis

The hydrogen contents of the studied stainless steels in as-supplied conditionand after different annealing treatments were first measured by inert gas fusionmethod with Leco TCH 600 NOH melt extraction analyser. In this method thesteel sample is heated up to melting point and the amount of desorbed gases isanalysed by infra-red detection. By this method the total (bulk) hydrogen contentin the material is determined.

Experimental Procedures

33

The hydrogen contents of the studied materials were also analysed with hot ex-traction using a Leybold-Heraeus H2A 2002 hydrogen analyser. In this analysismethod the sample is heated up to 1100°C and the amount of evolving gases isanalysed by a thermal conductivity detector.

Additionally, thermal desorption spectroscopy (TDS) was applied, using an ap-paratus designed and assembled at Laboratory of Engineering Materials at AaltoUniversity School of Engineering. The TDS apparatus measures hydrogen de-sorption during heating in the temperature range 25-900°C, with backgroundvacuum level down to 10-8 mbar. A constant heating rate of 6 K/min was used.

Prior to the hydrogen content analysis, the specimens were cut mechanically,and the surfaces were grinded up to 1200 grit finish. Before the measurementsthe specimens were cleaned in ethanol and dried with air flow. Due to varyingequipment specifications, the specimen size varied from less than 1 g in weightin TDS and inert gas fusion to about 5 g in hot extraction analysis.3.5 Residualstressmeasurements

3.5.1 X-raydiffraction

By using X-ray diffraction, interplanar lattice spacings can be measured in finegrained crystalline materials. Spacing between atomic planes in a metal ischanged under stress, either applied or residual. Stress induces an interatomicstrain, which is measured from the crystal lattice, and the residual stress produc-ing the strain is calculated assuming a linear elastic distortion of the crystal lat-tice. Although residual stresses result from non-uniform plastic deformation, allresidual macrostresses remaining after deformation are elastic. Changes in inter-planar spacing are measured as a shift in the diffraction pattern. The width of thediffraction peaks is an indication of the amount of plastic strain in the material(Noyan et al., 1987).

The principle of X-ray diffraction measurement of residual stress is illustrated inFig. 5. Lattice spacing is determined for multiple Ψ tilts, a straight line is fittedin d-sin2Ψ plot by least squares regression, and stress is calculated from the slopeof the best fit line using Eq (1).

𝜎 = ( ) (1)Because the spacings of lattice planes are extremely small they are affected byboth micro and macro stresses, and the X-ray method measures the sum of these

Experimental Procedures

34

stresses (Noyan et al., 1987). Diffraction is inherently selective and biased to-wards a particular set of grains, and this can be exploited for multiphase materi-als to provide information about the stressing of individual phases separately(Withers et al., 2001 A). The stress determined using X-ray diffraction is an av-erage stress in the irradiated area, typically from a surface layer less than 10 μmdeep.

Figure 5. (a) Schematic presentation of X-ray diffraction measurement of residual stress. Theincident beam diffracts X-rays of wavelength λ from planes that satisfy Bragg’s law in crystalswith these planes parallel to the surface of the sample. After the specimen is tilted (b), thediffraction peak occurs at higher 2θ angles. The stress is measured in a direction which is theintersection of the circle of tilt and the specimen surface (Noyan et al., 1987).

Residual stresses in the deep-drawn cups were measured with X-ray diffractionusing the sin2Ψ (multi-angle) method, in which the shift of a diffraction peakposition (2θ angle) is recorded as a function of the specimen tilt angle Ψ. Numberof inclination angles in the measurement was 10 (±45°). Measurements weremade with an X-ray diffractometer XStress3000. The measurements were doneseparately for each phase, employing Cr-Kα -radiation for the α’-martensite

Experimental Procedures

35

phase and Mn-Kα -radiation for the austenite phase. The collimator used was 3mm in diameter. The peak middle positions at each Ψ tilt were determined bythe peak fit (Pearson VII) method. Residual stress profiles were measured alongthe cup walls, from the outer surface of the cups. Measurements were done atthree different sides of each cup, both in tangential and axial directions. Tangen-tial stresses were, however, considered the most significant ones in this case, asthe delayed cracks propagate vertically down the cup walls.

3.5.2 Ringslitting(deflectionmethod)

Macroscopic hoop residual stresses, i.e. through-thickness mean residualstresses, in the deep-drawn cups were also determined by a ring slitting methodbased on residual stress relaxation. Cups were sectioned with electrical dischargemachining (EDM) into 3-4 mm wide rings, which were then slitted in the rollingdirection of the original blank. After slitting, ring opening and change in theouter diameter of the rings were measured (Fig. 6).

Figure 6. Determination of circumferential residual stresses in thin-walled tube by deflectionmethod (Walton, 2002).

The residual stresses were then calculated using the following equations(Garstka et al., 2006; Walton, 2002):

𝜎 = 𝐸 ∙ 𝑡 ∙ − (2)𝜎 = 𝐸 ∙ 𝑡 ∙ − (3)In the equations, E is the modulus of elasticity, t is the material thickness, D0 isthe diameter before slitting, d is the measured ring opening and s is the thicknessof the EDM wire and D1 is the diameter after slitting. Value of the modulus ofelasticity used in the calculations was 200 GPa.

Experimental Procedures

36

Residual stresses that had existed in the cups prior to cracking were also esti-mated in the cracked areas of the cups, by calculating the radius of curvature andfinal diameter D1 from the sectioned pieces (Fig. 7 and eq. 4)

Figure 7. Definition of parameters used in Eq. 4 to calculate the radius of curvature R.

4)(12

222

1

Lhtth

RD (4)

3.6 Constantloadtensiletesting

Constant load tensile testing was used for studying systematically the effect ofα’-martensite, stress and hydrogen content on delayed cracking kinetics. A largetest matrix was implemented with hundreds of specimens having different α’-martensite and hydrogen contents, tested at various stress levels. The size andgeometry of the test specimens is presented in Fig. 8.

Figure 8. The size and geometry of the test specimens prior to pre-straining. The measures arein millimetres.

Experimental Procedures

37

Tensile test specimens were first pre-strained to certain strain levels. In order toproduce a multi-axial stress state in the flat constant load test specimens, notchedspecimens were used. The notches were produced in the pre-strained specimensusing electric discharge machining. Specimen geometry was chosen accordingto standard ASTM E 292–09, “Standard test methods for conducting time-for-rupture notch tension tests of materials”. The specimens had 60° double-edgenotches with a root radius 0.3 mm, resulting in a stress concentration factor kt =4.5.

The applied stress ratio for each constant load tensile test was defined as theapplied stress divided by the ultimate tensile strength, or the notched tensilestrength, of a notched tensile test specimen (the maximum load in a tensile testdivided by the original cross-sectional area between the opposing notches). Asthe constant load tensile tests were performed on specimens with varying α’-martensite content and, thus, varying yield and ultimate tensile strength, thenotched tensile strength of the studied material was first measured as a functionof its α’-martensite content. The results are presented in Fig. 9 for steel 204Cu.Because of plastic constraint at the notch, i.e. a triaxial stress field acting to plas-tically constrain the material from deforming, yield strength and ultimate rupturestrength of a notched specimen can be higher than those of a smooth specimenof the same material. This phenomenon, occurring in ductile materials, is re-ferred to as notch strengthening. It means that a notched tensile specimen canwithstand higher nominal stresses than a smooth specimen.

Figure 9. Notched tensile strength of 204Cu steel as a function of α’-martensite content.

In all constant load tensile tests, the machining of the notches was done shortly,within one hour, after pre-straining of the specimen, and the constant load test

Experimental Procedures

38

was started shortly after preparation of the notches. From each constant load test,the time to fracture was recorded. The maximum duration of the tests was set to400 hours.

3.7 ScanningelectronmicroscopyFracture surfaces of delayed cracks in the deep drawn Swift cups and in the con-stant load tensile test specimens were examined with FEG-SEM Zeiss Ultra 55field emission gun scanning electron microscope (FEG-SEM) and some of themwith a Zeiss Merlin Compact VP field emission gun scanning electronmicroscope. Electron backscattering diffraction (EBSD) measurements toidentify the microstructure and phases in the cracked areas were done with aNordlys II digital EBSD detector and with a Bruker e-FlashHR high resolutionEBSD camera. HKL Technologies Channel 5 software and Bruker‘s QUAN-TAX CrystAlign EBSD analysis system were used for data acquisition and anal-ysis. Specimen preparation for EBSD measurements was done by first grindingthe specimen surface with SiC water grinding papers up to 2400 grit with StruersLaboPol-21 equipment, then polishing the surface with 3 μm, 1 μm and ¼ μmdiamond pastes, and finally polishing with Buehler VibraMet2 vibratory polisherin a colloidal silica polishing suspension for 5 h.

3.8 Nanoindentation

Mechanical properties of austenite and strain-induced α’-martensite phases inthe formed test materials were determined by nanoindentation. Specimens forthe measurements were cut from deep drawn cups with drawing ratio 2.0 for 301,301LN and 204Cu and with drawing ratio 2.12 for the more stable grades 304and 201(b), at about 5-10 mm from the upper edge of the cups. Specimenpreparation for nanoindentation meaurements was done by first grinding thespecimen surface with SiC water grinding papers up to 2400 grit with StruersLaboPol-21 equipment, then polishing the surface with 3 μm, 1 μm and ¼ μmdiamond pastes, and finally polishing with Buehler VibraMet2 vibratory polisherin a colloidal silica polishing suspension for 5 h.

Nanoindentation experiments were performed using a CSM instrumented inden-tation tester. A three-sided pyramidal Berkovich diamond indenter tip with nom-inal angle of 65.3° was employed. Analyses for the calculation of hardness wereconducted by the method used by Oliver and Pharr (Oliver et al., 1992). Theapplied maximum force was 1.5 mN, which produced an average indentationdepth of 100 nm. A constant loading rate of 10 mN/min was used. Holding timeat the maximum load was 15 s and unloading rate was 10 mN/min.

Experimental Procedures

39

Based on visual examination of the indented area under optical microscope, therepresentative indentations of each phase were selected. Indentations locatedclose to grain or phase boundaries were ignored, as well as unsuccessful inden-tations with anomalies in the shape of the load-displacement curve or located tooclose to neighboring indentations. Electron backscattering diffraction (EBSD)was utilized for identifying the phases present in the indented areas and forstudying the dependence of indentation hardness and modulus on grain orienta-tion.

40

4. Results

4.1 Stabilityofaustenite

Austenite stability in each test material was studied by making interrupted tensiletests to various strain levels and measuring α’-martensite content after the testsusing Ferritescope. The tensile tests were performed at room temperature withstrain rate 3*10-4 1/s. Results of the measurements are shown in Fig. 10, togetherwith fitted sigmoidal curves based on the Olson-Cohen model. Three of the testmaterials, namely 301, 301LN and 204Cu were the most unstable ones. 201Cuwas stable with almost neglible α’-martensite content even at high strains. In201(a) the volume fraction of α’-martensite was very low except at the highestexamined strain level.

Figure 10. Volume fraction of strain-induced α’-martensite in the test materials as a function oftrue strain.

The effect of tensile residual stress on the magnetic properties (the Villari effect)of the test materials and, thus, on the measured ferromagnetic phase contentswas taken into account by correction factors experimentally defined for eachmaterial. In Fig. 11 the measured Ferritescope readings as a function of applied

Results

41

stress are shown (a), as well as the defined correction factors as a function ofstress (b).

Figure 11. a) Ferritescope readings and b) correction factors as a function of applied tensilestress for 201(b), 301, 301LN and 204Cu.

Some results of measured α’-martensite contents in deep drawn Swift cups arepresented in Publications I, II, IV and VI. In all test materials the maximum α’-martensite content in the Swift cups was higher than in tensile straining, becauseof triaxial stress state and higher plastic strains in the Swift cups.

4.2 Hydrogencontent

The hydrogen contents of the studied stainless steels in as-supplied condition,measured by melt extraction, hot extraction and thermal desorption spectroscopy

a)

b)

Results

42

(TDS) are shown in Table 3 and Fig. 12. Shown contents are mean values of 3to 6 analyzed specimens (more repetitions were done for the materials showinghigh variation in analysis results). The measured values were different depend-ing on the analysis method, but the trend was similar with all used methods: theMn-alloyed grades had clearly higher hydrogen contents than 304, 301 and301LN. Hydrogen content was the highest in 204Cu and 201(b). Standard devi-ation of the results was also highest for 201(b) and 204Cu.

Table 3. Hydrogen contents of the studied stainless steels in wppm.Grade 304 301 301LN 201(a) 201(b) 201Cu 204CuHydrogen, meltextraction

2.7 ±0.14

2.5 ±0.35

2.2 ±0.12

3.7 ±0.57

5.3 ±2.25

3.3 ±0.25

5.2 ±1.55

Hydrogen, hotextraction

1.2 ±0.20

1.6 ±0.18

1.2 ±0.17

1.6 ±0.18

2.3 ±0.31

2.1 ±0.05

2.6 ±0.38

Hydrogen,TDS

- 1.8 ±0.25

1.3 ±0.16

- 2.2 ±0.53

- 2.9 ±0.61

a) b)Figure 12. a) Hydrogen contents of the test materials analyzed with (a) melt extraction and (b)hot extraction.

The content of hydrogen in the studied stainless steels was effectively reducedby heat treatments at 300-400°C. Results of hydrogen content as a function ofannealing time at 400°C, as well as discussion about the effect of hydrogen ondelayed cracking are presented in Publication II.

Annealing of the studied stainless steels before forming reduced the volume frac-tion of strain-induced α’-martensite transformed during the forming process, i.e.reduction of hydrogen content slightly increased the stability of austenite. Truestress - true strain tensile curves for stainless steels 301 and 204Cu in as-suppliedand annealed (400°C/24 h) conditions are presented in Fig. 13. The ultimate ten-sile strength and the maximum volume fraction of α’-martensite were higher inthe as-supplied state.

Results

43

Figure 13. Tensile true stress - true strain behaviour of stainless steels 301 (a) and 204Cu (b) inas-supplied and annealed conditions.

4.3 Swiftcuptesting

Several series of Swift cups, with 320 cups altogether, were deep drawn at Ou-tokumpu Research Centre during this research project. The limiting drawing ra-tios for delayed cracking (LDR-DC) in the Swift tests for each tested materialare shown in Table 4. The criterion for the LDR-DC was no cracks observed inany of the cups at that drawing ratio during 1500 hours after the forming process.Generally no cracks initiated in the cups after about 1000 hours from forming.Stainless steels 304 and 201Cu did not suffer from delayed cracking even at thehighest applied drawing ratio 2.12. The grade 204Cu was the most prone to de-layed cracking: it had the lowest LDR-DC and the highest amount of cracks at

a)

b)

Results

44

each drawing ratio. The grade 201(b) was more prone to delayed cracking thanthe grades 201(a), 301 and 301LN.

Table 4. Limiting drawing ratios for delayed cracking defined by Swift cup tests.

Material 304 301LN 301 201(a) 201(b) 201Cu 204CuLDR-DC - 2.1 1.8 1.9 1.7 - 1.4

Number of cracks in all the Swift cups drawn from as-supplied materials withdifferent drawing ratios, together with mean values and standard deviation, arepresented in Table 5. Test materials 304 and 201Cu are not included in the table,as they did not suffer from delayed cracking. Number of cracks in the cups re-flects the severity of delayed cracking. The higher is the number of cracks in thecups, the lower is the residual stress level required to initiate cracks, becauseeach crack in part relaxes tangential residual stresses in the cups.

Table 5. Number of cracks in Swift cups drawn from as-supplied materials.Material Drawing

ratioFinal number of cracks in parallel cups(after 60 days from drawing)

Meanvalue

Standarddeviation

204Cu 1.4 0 0 0 01.5 2 0 0 0.67 1.151.6 3 3 6 4.0 1.731.7 10 13 15 12.67 2.521.8 15 14 15 14.67 0.582.0 16 16 15 16 15.75 0.5

201(a) 1.8 0 0 0 01.9 0 0 0 02.0 0 4 1 1.67 2.082.1 0 1 0.5 0.71

201(b) 1.6 0 0 0 01.7 0 0 0 0 01.8 2 0 2 1 0 1.0 1.01.9 2 0 3 1 1.5 1.292.0 11 12 5 12 7 9.4 3.212.12 18 12 8 12.67 5.03

301 1.6 0 0 01.8 0 0 0 0 01.9 2 0 6 0 2.0 2.832.0 4 3 7 0 3 3.4 2.512.1 8 6 7.0 1.412.12 9 2 1 13 5 6.0 5.0

301LN 2.0 0 0 0 02.1 0 0 0 02.12 3 0 0 1.0 1.73

Results

45

Annealing the stainless steels at 400°C prior to deep drawing had a distinct effecton the severity of delayed cracking. The limiting drawing ratios for delayedcracking for the three most susceptible grades 204Cu, 201(b) and 301 both in as-supplied state and annealed states are shown in Table 6. After annealing for 3 to72 h at 400°C, the limiting drawing ratios for delayed cracking were improvedfor each grade. Hydrogen present in the materials after annealing is predomi-nantly nondiffusible, located at strong trapping sites. The LDR-DC as a functionof total hydrogen content is presented in Fig. 14. In 301 delayed cracking wascompletely eliminated after 72 h annealing and it did not show any cracking atthe highest DR 2.12. Most of the 301 cups at DR 2.12, drawn from 24 h annealedmaterial, did not show any cracks either, but the criterion for LDR-DC was thatthere were no cracks in any of the parallel cups. In 204Cu and 201(b) delayedcracking was detected also in 72 h annealed material.

Table 6. Limiting drawing ratios for delayed cracking as a function of annealing time at 400°C.

Material 204Cu 201(b) 301

LDR-DC (as-supplied) 1.4 1.7 1.8

LDR-DC (400°C/1 h) 1.6 1.8

LDR-DC (400°C/3 h) 1.6 1.9 1.9

LDR-DC (400°C/24 h) 1.6 2.0 2.1

LDR-DC (400°C/72 h) 1.7 2.0 -

Figure 14. Limiting drawing ratio for delayed cracking as a function of hydrogen content in steels204Cu, 201(b) and 301.

Consistent with the tensile test results presented in section 4.2, reduction of hy-drogen content of the stainless steels by annealing prior to forming increased the

Results

46

stability of austenite also during deep drawing. The difference between the max-imum α’-martensite content in Swift cups drawn from as-supplied and annealedmaterials was about 5-7 % at highest.

Number of cracks in all the Swift cups drawn from annealed materials with dif-ferent drawing ratios are presented in Table 7.

Table 7. Number of cracks in Swift cups drawn from annealed materials.Material Annealing Drawing

ratioFinal number of cracks in parallelcups

Meanvalue

Stddev

204Cu 400°C/1 h 1.6 0 0 0 01.7 3 3 3.0 01.8 2 3 5 3.33 1.532.0 6 2 2 3 3.25 1.89

400°C/3 h 1.6 0 0 0 01.8 4 1 2 1 2.0 1.412.0 12 3 2 1 3 4.2 4.44

400°C/24 h 1.7 0 3 0 1.0 1.731.8 2 0 0 2 1 1.0 1.02.0 7 5 2 2 4.0 2.45

400°C/72 h 1.7 0 0 0 01.8 1 0 0.5 0.712.0 4 1 1 2.0 1.73

201(b) 400°C/1 h 1.9 0 1 0.5 0.712.0 2 0 1 1.0 1.0

2.12 7 7 7.0 0400°C/3 h 1.9 0 0 0 0 0

2.0 2 0 2 1.33 1.152.12 5 0 2.5 3.54

400°C/24 h 2.0 0 0 0 0 02.12 2 1 0 1.0 1.0

400°C/72 h 2.0 2 0 0 0.67 1.152.12 2 0 1.0 1.41

301 400°C/3 h 1.8 0 0 02.0 3 0 0 1.0 1.73

2.12 3 0 0 0 0.75 1.5400°C/24 h 2.0 0 0 0 0

2.1 0 0 0 02.12 0 5 0 0 1.25 2.5

400°C/72 h 2.0 0 0 02.12 0 0 0 0 0

The initiation time for the first crack to appear in the Swift cups, as a function ofdrawing ratio is presented in Fig. 15(a-c) for 204Cu, 201(b) and 301 in as-sup-plied and in annealed conditions. The lower is the drawing ratio, the lower is thevolume fraction of α’-martensite and the lower are the residual stresses in thecups. Both of these factors, as well as the hydrogen content being dependent on

Results

47

annealing parameters, affect the crack initiation time. The crack initiation timesare much shorter in 204Cu, especially in as-supplied condition, in comparison to201(b) and 301.

Figure 15. Fracture initiation times in the Swift cups of (a) 204Cu, (b) 201(b) and (c) 301 steels.

a)

b)

c)

Results

48

Length of the growing cracks in the Swift cups was monitored after the deepdrawing tests. Results about the propagation and final length of the cracks in theSwift cups of each test material are presented in Publication II. It was noticed,that in the cups drawn from annealed materials, the final length of the cracks wasshorter. As the α’-martensite content decreased down the cup wall, shorter cracklength in the annealed materials means that the cracks stopped at a higher α’-martensite level and, thus, the resistance of the test materials to crack growthwas improved by reduction of the hydrogen content.

From the slope of the graphs of crack length as a function of time, the crackgrowth rates were determined. In Publication I, growth rates of first appearingcracks in the cups of 204Cu, 201(b) and 301 at drawing ratio 2.0 are presented.The growth rate as a function of time since the crack initiation was highest in204Cu and lowest in 301.

In Fig. 16(a) the growth rate of first cracks in the cups of 201(b) at differentdrawing ratios are presented. The growth rate is fastest right after initiation andit slows down as the cracks proceed. The initial crack growth rate is the higherthe higher the drawing ratio. In Fig. 16(b) the growth rates of first cracks in thecups of 204Cu with drawing ratio 2.0 in as-supplied state and after different an-nealing times at 400°C are presented. In Fig. 6(c) the growth rates of first cracksin the cups of 201(b) with drawing ratio 2.12 in as-supplied state and after dif-ferent annealing times at 400°C are presented. The growth rates are the lowerthe longer the annealing time and, thus, the lower the hydrogen content of thematerial.

Results

49

Figure 16. Growth rate of first cracks initiating in Swift cups of 201(b) at different drawing ratios(a), in cups of 204Cu with drawing ratio 2.0 in as-supplied state and after different annealingtimes at 400°C (b) and in cups of 201(b) with drawing ratio 2.12 in as-supplied state and afterdifferent annealing times at 400°C (c).

a)

b)

c)

Results

50

4.4 Residualstressmeasurements

4.4.1 X-raydiffraction

All the residual stresses measured with X-ray diffraction from the outer surfacesof the Swift cups were tensile. Stress maximum was located at about 50-60 % ofcup wall height. The magnitude of the tangential residual stresses in each studiedmaterial depended on the drawing ratio, increasing with the amount of defor-mation. In the studied metastable austenitic stainless steels, the magnitude of theinternal residual stresses in the α’-martensite phase was markedly higher thanthat in the austenite phase.

Some results of the residual stress measurements of the Swift cups are presentedin Publication I, Publication IV and Publication VI. Tangential residual stressmeasured from the austenite phase of DR 2.12 Swift cup drawn from the moststable test material, 201Cu, is presented in Fig. 17(a). Tangential residualstresses measured separately from the austenite and martensite phases of theLDR-DC (highest drawing ratio without delayed cracking) Swift cups of 201(a),304, 301LN, 301 and 201(b) are presented in Fig. 17(b-f), together with the α’-martensite content in the cup walls. The α’-martensite content rises up the cupwall and reaches a maximum value near the cup edge, where the strain is at max-imum. Therefore, the evolution of residual stress along the cup wall does notcorrespond to the evolution of α’-martensite content, which is in agreement withprevious results (Berrahmoune et al., 2006 A).

In the stable stainless steel 201Cu the overall residual stress state was consider-ably lower than in the other test materials. It is also noteworthy that residualstresses in austenite and α’-martensite phases were relatively low in the Swiftcup of 304 at drawing ratio 2.12. Grade 304 did not suffer from delayed cracking,although it is metastable and α’-martensite content reached almost 50 % near thecup edge at DR 2.12. Tangential residual stress in the α’-martensite phase wasmuch higher in 301, 201(a) and 201(b). It was about the same magnitude also inthe LDR-DR cup of 301LN (Fig. 17d), but in that material the limiting drawingratio is higher. The difference between the maximum value of residual stressesin austenite and α’-martensite phases was about 150 to 270 MPa in 304, 301 and301LN and about 250 to 400 MPa in 201(a) and 201(b). Estimated error in theXRD residual stress measurements was on the order of ± 30 to 60 MPa.

Results

51

a)

b)

c)Figure 17 (to be continued). Tangential residual stresses in LDR-DC Swift cups of a) 201Cu, b)201(a), c) 304, d) 301LN, e) 301 and f) 201(b) stainless steels measured with X-ray diffraction,as a function of cup wall height starting from cup bottom.

Results

52

d)

e)

f)Figure 17. Tangential residual stresses in LDR-DC Swift cups of a) 201Cu, b) 201(a), c) 304, d)301LN, e) 301 and f) 201(b) stainless steels measured with X-ray diffraction, as a function ofcup wall height starting from cup bottom.

Tangential residual stresses measured from the martensite phase of 301LN Swiftcups with different drawing ratios are presented in Fig. 18. As the degree of de-

Results

53

formation and the content of α’-martensite increase with the drawing ratio, re-sidual stresses also increase. Tangential residual stresses measured from the α’-martensite phase of LDR-DC Swift cups (the highest cups without cracking) of301LN, 301, 201(b) and 204Cu, and the maximum α’-martensite contents in thecups, are compared in Fig. 19. The limiting residual stress of α’-martensite notcausing cracking is similar in 301LN, 301 and 201(b). 204Cu suffers from de-layed cracking at much lower drawing ratio and lower stress level.

Figure 18. Tangential residual stresses measured from the α’-martensite phase of 301LN Swiftcups with X-ray diffraction.

Figure 19. Tangential residual stresses measured from the α’-martensite phase of LDR-DCSwift cups with X-ray diffraction.

Plastic deformation causes broadening of the X-ray diffraction peaks, and thewidth of the diffraction peaks is an indication of the amount of plastic strain inthe material (Noyan et al., 1987). Peak broadening is caused by lattice distortionsdue to the presence of crystal defects, such as dislocations, and can be utilizedin the estimation of dislocation densities. The full width at half maximum inten-sity (FWHM) of the measured X-ray diffraction peaks is presented in Fig. 20(a-c) as a function of the height of cup wall for 201Cu (DR 2.12), 304 (DR 2.12)

Results

54

and 301LN (DR 2.1). The profiles of FWHM along the cup walls demonstratethe distribution of plastic strain and the associated dislocation density in the cups.FWHM of the diffraction peaks measured from the austenite phase increasescontinuously as a function of cup wall height. Instead, FWHM values measuredfrom the α’-martensite phase show a more shallow gradient, and reach a maxi-mum at about 60% of cup height.

Figure 20. Full width of diffraction peak at half maximum intensity (FWHM) as a function of cupheight in (a) 201Cu DR 2.12, (b) 304 DR 2.12 and (c) 301LN DR 2.1.

a)

b)

c)

Results

55

Effect of annealing (400°C/24 h) on the residual stresses of deep drawn cups (ofstainless steel 301 as an example) is demonstrated in Fig. 21(a-b). The α’-mar-tensite phase stress is slightly lower in the annealed state, but the maximum ofaustenite phase stress is almost similar in both states.

Figure 21. Effect of annealing (400°C/24 h) on the residual stresses in the α’-martensite phase(a) and austenite phase (b) of 301 steel Swift cup with drawing ratio 1.8.

Due to annealing of the stainless steels before deep drawing, hydrogen contentwas reduced and the limiting drawing ratios for delayed cracking were increased.In the annealed state it was therefore possible to compare the level of residualstresses in stainless steel 204Cu with the stress levels in 201(b) and 301 at con-stant drawing ratio. In Fig. 22, tangential residual stresses measured from the α’-martensite phase of 204Cu, 201(b) and 301 Swift cups at DR 1.8, drawn fromannealed material (400°C/24 h), are presented. Maximum residual stress is thehighest in 204Cu.

a)

b)

Results

56

Figure 22. Tangential residual stresses measured from the α’-martensite phase of 204Cu,201(b) and 301 Swift cups at DR 1.8, drawn from annealed material (400°C/24 h).

4.4.2 Ringslitting

Circumferential hoop residual stresses in deep-drawn cups were also measuredby a ring slitting method based on residual stress relaxation. An advantage ofthis approach in comparison to X-ray diffraction is that residual stresses that hadexisted in the cups prior to cracking can also be determined from the crackedareas of the cups. In Fig. 23 rings cut from a Swift cup, both in noncracked andcracked areas, are shown.

Figure 23. Rings cut from a Swift cup.

Some results on the hoop residual stress measurements are presented in Publica-tion I. Macroscopic hoop residual stress measured by the ring slitting methodwas between the XRD phase stresses of austenite and α’-martensite, closer to thelatter. Macroscopic hoop residual stresses measured from Swift cups with dif-ferent drawing ratios are presented in Fig. 24(a-d) for stainless steels 204Cu,201(b), 301 and 301LN. The stress of the smallest cup with cracking is presented

Results

57

with bright red colour in each figure. Stainless steel 204Cu showed delayedcracking at much lower stress level than the other materials. Cups of 204Cu atdrawing ratio 1.8 or higher broke into small pieces (~ 5 mm) during ring cutting,and it was not possible to measure the radius of curvature from such short sec-tions.

a)

b)

c)

Results

58

d)Figure 24. Hoop residual stresses measured with ring slitting method from Swift cups of (a)204Cu, (b) 201(b), (c) 301 and (d) 301LN stainless steels with different drawing ratios.

Effect of austenite stability on the circumferential hoop residual stresses in thematerial is demonstrated in Fig. 25, showing the hoop stress near the upper edgeof the cups of 201Cu, 201(a) and 201(b) at two drawing ratios as a function ofthe α’-martensite content at the same location. Material 201Cu with 4.7 wt-%nickel is the most stable. Residual stresses in the cups of 201(a) (4.5 wt-% Ni)and 201(b) (3.7 wt-% Ni) are higher than those in 201Cu, because of the mar-tensitic transformation.

Figure 25. Hoop residual stress as a function of martensite content in 201 steels.

Macroscopic hoop residual stresses measured from the LDR-DC Swift cups of301LN, 301, 201(b) and 204Cu are presented in Publication I. The results are

Results

59

consistent with the XRD residual stress results presented in Fig. 19, i.e. accord-ing to both methods the highest level of residual stress not causing cracking isalmost similar in 301, 301LN and 201(b).

Effect of annealing (400°C/24 h) on the hoop residual stresses of deep drawncups of 301 and 201(b) with DR 1.8 is demonstrated in Fig. 26(a-b). Hoop stressis lower in the cups drawn from annealed materials with lower hydrogen content.

a)

b)Figure 26. Effect of annealing on the hoop residual stresses in the Swift cups of (a) 301 and (b)201(b) at drawing ratio 1.8.

By plotting the residual stress data along the cup walls, measured both fromcracked and uncracked areas, together with α’-martensite volume fractions at thecorresponding cup height locations, the critical combinations of residual stressand α’-martensite content for delayed cracking were specified for each stainlesssteel grade. These graphs for 301, 201(b) and 204Cu are presented in PublicationII.

Results

60

4.4.3 ResidualstressrelaxationandtheVillarieffect

Cracking of the cups relaxes residual stresses in the cups, which affects the Fer-ritescope readings. Ferritescope reading and the calculated α’-martensite contentis higher when measured from cracked cups than from cups without cracks. Ifthe α’-martensite contents measured from uncracked cups are multiplied withthe Villari effect correction factors defined from tensile tests (section 4.1), theresulting corrected α’-martensite contents are almost equal to the values meas-ured from cracked cups (Fig. 27). This verifies the applicability of the used cor-rection factors.

Figure 27. Effect of residual stress relaxation in the cups compared to the experimentally de-fined Villari effect correction.

4.5 Scanningelectronmicroscopy

4.5.1 Fractography

Some fractography of the Swift cups is presented in Publication I, PublicationIV and Publication VI. Fracture appearance in all the test materials waspredominantly quasi-cleavage type, with varying areas of transgranular cleavagealong the martensitic stucture and ductile fracture. Quasi-cleavage does notrepresent a separate fracture mode, but is a localized feature on hydrogen-induced fracture surfaces that exhibits characteristics of both cleavage andplastic deformation. Flat facets are a typical feature, but they are often finer thanin traditional cleavage fracture. Quasi-cleavage fracture surfaces are frequentlydecorated with ridges that can be correlated with highly localized deformationbands (Martin et al., 2011 B).

Typical fractography of 204Cu, 201(b) and 301 is presented in Figure 28 and athigher magnification in Figure 29 (301LN included). Examples of dimpled areas

Results

61

are marked with white arrows and quasi-cleavage facets with black arrows. Theobservations are consistent with previous results of other researchers(Berrahmoune et al., 2006 B; Zhang et el., 2008; Guo et al., 2011; Michler et al.,2012). Middle regions of the fracture surfaces were more ductile, and closer tothe edges there were larger areas with cleavage facets. An example of a facetwith fine tear ridges is presented in Figure 30 a. In the ductile regions the dimpleswere generally very small and shallow (Figure 29 c-d and Figure 30 b). Narrowshear lips with slant fracture appearance, in 45° angle to the thickness, werefound at the edges of the fracture surfaces. The width of the shear lips can bethought of being an indication of the plasticity of the material associated withthe extending crack. The average width of the shear lips, measured from thefracture surfaces, was largest in 301 and smallest in 204Cu (Publication VI). Thisindicates the lowest toughness and resistance to crack propagation in 204Cu.

a) b)

c) d)Figure 28. Typical fracture surfaces in 204Cu (a-b), (c) 201(b), and (d) 301.

Results

62

a) b)

c) d)Figure 29. Typical fracture surfaces in (a) 204Cu, (b) 201(b), (c) 301 and (d) 301LN.

a) b)Figure 30. (a) Fracture facet in 204Cu and (b) small, shallow dimples in 201(b).

The overall fracture appearance was most brittle type in 204Cu and 201(b).Largest fraction of dimpled fracture surface was found in 301LN, which ispresumably explained by the higher nickel content and low carbon content. Dif-ferences between as-supplied and annealed conditions for the same grade weresmall, but on average, fracture in annealed state seemed to be slightly more duc-tile than in as-supplied state. The main fracture mode was quasi-cleavage in bothconditions, as-supplied and annealed. Heat treatment at 300-400°C does notchange the fracture mechanism, but it reduces the hydrogen content and affectsthe cracking kinetics.

Results

63

In constant load tensile test specimens (Publication III) fractures initiated at thenotch root and a flat triangular area associated with high triaxiality was visibleon the fracture surface. In this region of hydrogen-induced cracking the fracturemechanism was predominantly transgranular quasi-cleavage. The triangular re-gion was followed by slant shear fracture with dimpled fracture surface. Fractureappearance was very similar in the annealed specimens in comparison to as-sup-plied specimens. Variation in the hydrogen content (in the studied range), alt-hough having a big effect on the crack initiation time, did not seem to affect themechanism of fracture.

Possible effect of bulk α’-martensite content on the fracture mechanism in con-stant load tensile testing was also examined. The macroscopic appearance of thefracture surface in the shear fracture region was more flat in the specimens withhigher α’-martensite content. In specimens with lower α’-martensite there wasnoticeable necking and features indicating higher ductility, but the fracture sur-face in the triangular crack initiation region was similar transgranular quasi-cleavage type.

4.5.2 EBSDmeasurements

EBSD analyses were performed in the cracked areas of the Swift cups and theconstant load tensile test specimens. Results of EBSD analyses are presented inPublication III, IV and VI. Because of the high degree of deformation in the cupwalls, the percentage of non-indexed points in the analysis was rather high. InFig. 31(a-b), EBSD phase maps of the cracked area in a Swift cup of 301 steelat drawing ratio 2.0 are presented. Austenite phase is coloured with red and α’-martensite with blue colour. The average bulk α’-martensite content in the ex-amined area was about 60 %. In Fig. 32 an EBSD phase map around a crack ina Swift cup of 204Cu steel at drawing ratio 1.8 is presented. The average bulkα’-martensite content in the examined area was about 54 %. In Fig. 33(a-b)EBSD phase maps along the crack path and in the crack tip area in a Swift cupof 201(a) steel at drawing ratio 2.0 are presented. The average bulk α’-martensitecontent in the examined area was about 33 %.

Results

64

a) b)Figure 31. EBSD phase maps along the crack path (a) and in the vicinity of the crack wake (b) ina Swift cup of 301 steel at drawing ratio 2.0.

a) b)Figure 32. EBSD phase map along the crack path (a) and around the crack tip (b) in a Swift cupof 204Cu steel at drawing ratio 1.8.

a) b)Figure 33. EBSD phase map along the crack path (a) and in the crack tip area (b) in a Swift cupof 201(a) at drawing ratio 2.0.

According to EBSD investigation, the microstructure in the close vicinity of thecracks and ahead of the crack tips consisted predominantly of α’-martensite ineach material, irrespective of the bulk α’-martensite volume fraction. This indi-cates that localized transformation of austenite to α’-martensite occurs ahead ofthe crack as a response to intense plastic strain, and the crack propagates alongα’-martensite phase. Similar observations have been reported in hydrogen em-brittlement studies of stainless steels (Narita et al., 1980; Lai et al., 2013) and

Results

65

also in investigations on the effect of hydrogen on fatigue crack growth in aus-tenitic stainless steels (Kanezaki et al., 2008; Chen et al., 2014).

4.6 Nanoindentation

Mechanical properties of austenite and strain-induced α’-martensite in 304, 301,301LN, 201(b) and 204Cu steels were determined by nanoindentation. Resultsof nanoindentation experiments for some of the materials are presented in Pub-lication IV. An optical micrograph of a nanoindentation matrix measured onsteel 304 is shown in Fig. 34(a). Martensite phase is shown as dark grey areasand the light grey areas represent austenite. An EBSD phase map over the samespecimen area is presented in Fig. 34(b). Martensite phase is marked with bluecolour and austenite with red. An inverse pole figure map, indicating the grainorientations, measured in the normal direction to the specimen surface is pre-sented in Fig. 34(c).

a) b)

c)Figure 34. (a) An optical micrograph of a nanoindentation matrix measured on stainless steel304, (b) an EBSD phase map over the same specimen area and (c) an inverse pole figure map.

Typical load-displacement curves for austenite and α’-martensite phases in 304and 301 steels are illustrated in Fig. 35(a-b). At the applied maximum force of1.5 mN, the maximum indentation depth varied between approximately 88 and100 nm in martensite and between 102-116 nm in austenite. The indentationhardness of the studied material is calculated as the maximum load divided by

Results

66

the projected contact area of the indent. Indentation modulus presents the elasticmodulus of the material that is calculated from the slope of the unloading curve.

a)

b)Figure 35. Typical load-displacement graphs measured from austenite and α’-martensite in (a)304 and (b) 301.

The results of nanoindentation measurements, i.e. the average values of the suc-cessful and representative indentations on both phases, for each test material arepresented in Table 8. The indentation hardness of the α’-martensite phase ishighest in grades 301, 201(b) and 204Cu and lowest in grade 304. Hardness ofmartensite in steels is generally controlled by the interstitial elements carbon andnitrogen. In the test materials, carbon content is highest in 301, which also hasthe highest hardness of both the phases. Nitrogen alloying in steels results ininterstitial solid solution strengthening and grain refinement, both of which in-crease hardness. Nitrogen is a more effective solid-solution strengthener thancarbon and also enhances grain size (Hall-Petch) strengthening. Nitrogen contentis highest in grade 201(b), which has high indentation hardness in both phases,

Results

67

the highest yield strength and smallest grain size among the test materials. Ni-trogen content is lowest in grade 304, which has the lowest hardness values andthe lowest yield strength. Nitrogen content also contributes to the hardness of α’-martensite in 301LN, which has the lowest carbon content.

Table 8. Measured indentation hardness and modulus of α’-martensite and austenite phases.304 301LN 301 201(b) 204Cu

HIT α’ (GPa) 7.04 ± 0.59 7.46 ± 0.42 8.27 ± 0.54 8.11 ± 0.77 7.86 ± 0.73HIT γ (GPa) 5.56 ± 0.44 5.64 ± 0.34 5.80 ± 0.66 5.73 ± 0.78 5.56 ± 0.51EIT α’ (GPa) 224.6 ± 17.3 230.6 ± 17.8 235.7 ± 14.4 220.6 ± 19.3 231.1 ± 16.8EIT γ (GPa) 218.7 ± 18.2 224.5 ± 19.1 223.6 ± 18.8 213.1 ± 29.5 214.6 ± 17.4

The indentation hardness of the austenite phase is almost similar in 304, 301LNand 204Cu and higher in 301 and 201(b). The indentation hardness of the fullyaustenitic base material in as-supplied state was between 4.85-5.30 GPa for allthe test materials. The measured indentation modulus values for α’-martensiteand austenite phases are quite similar in all the test materials. In each stainlesssteel grade the indentation modulus of austenite is lower than that of α’-marten-site. Indentation modulus values of metallic materials are often higher than ten-sile-test elastic modulus values. This difference can be attributed to indentationpile-up (meaning the formation of a “crater” around the contact impression, af-fecting the determination of the contact area), indent size effect and elastic re-covery (Sargent et al., 1981). The displacement recovery during unloading maynot be entirely elastic, but can also include a plasticity effect due to internalstresses in the material (Shuman et al., 2007).

The variation in the measured indentation hardness and modulus data is pre-sented in Table 8 as standard deviation. Variation is highest in material 201(b).Histograms representing the distribution of measured data for 301 and 204Cu,as an example, are shown in Fig. 36 (a-b). The indentation hardness values below6900 MPa for 301 and below 6600 MPa for 204Cu represent data measured fromthe austenite phase and hardness values above 7900 MPa for 301 and above 7300MPa for 204Cu represent data from the α’-martensite phase.

Results

68

a)

b)Figure 36. Distribution of the measured indentation hardness values in 301 (a) and 204Cu (b).

Dependence of indentation hardness and modulus on grain orientation causesvariation in the measured phase properties. The orientation dependence of meas-ured indentation properties was studied by performing EBSD analysis of the in-dented areas in 304 and 201(b). According to the analysis, in both materials thehighest values of indentation hardness and modulus of α’-martensite corre-sponded to the [111] direction and the lowest values to the [001] direction. Thisis in agreement with previous results (Hausild et al., 2014; Stinville et al., 2011).

4.6 Constantloadtensiletesting

Results of the constant load tensile tests are presented in Publication III. Effectof applied stress level and α’-martensite content on time to fracture for stainlesssteel 204Cu is shown in Fig. 37. It is seen that at high (0.975) applied stress ratio,

Results

69

i.e. the ratio of applied stress to the notched tensile strength, α’-martensite con-tent has very little effect on time to fracture. When the applied stress level islower the effect of α’-martensite content increases. The lower the stress level andthe α’-martensite content, the longer the time to fracture. Data points markedwith arrows indicate that no cracking occurred during that time. Fracture was notobserved in any of the test specimens after 200 hours. The results demonstratethe role of strain-induced α’-martensite as a medium for hydrogen transport. Atthe highest applied stress levels cracking occurred even at or below 30% α’-martensite, but the limiting content of α’-martensite, below which no crackingwas detected, increased as the stress level was lowered.

Figure 37. Time to fracture as a function of α’-martensite content of 204Cu steel at severalapplied stress ratios.

Effect of hydrogen content on time to fracture was also studied by constant loadtests. The results of these tests are presented in Fig. 38(a-d). The hydrogen con-tent in these graphs is according to the hot extraction analysis. Time to fracturedepends on the hydrogen content in a power law manner. The lower is the hy-drogen content the longer is the time necessary for accumulating a sufficientamount of hydrogen at a crack initiation site. Even at the highest applied stressratio (0.975) hydrogen content had a marked effect on time to fracture, althoughα’-martensite content did not. Reduction of the hydrogen content is apparentlyan efficient way to suppress the kinetics of delayed cracking.

Results

70

a)

b)

c)

Results

71

d)Figure 38. Time to fracture as a function of hydrogen content of 204Cu steel at several appliedstress ratios.

Delayed cracking susceptibility of stainless steel 204Cu, according to the resultsof the constant load tensile tests, as a combination of stress level and α’-martensite content leading to delayed fracture for the as supplied state and an-nealed (400°C/3 h) state is presented in Publication III. Lowering the hydrogencontent by annealing significantly reduces the susceptibility to delayed cracking,as the critical range for cracking is moved to much higher stress levels and alsoto higher α’-martensite content. According to these results, delayed crackingdoes not occur in 204Cu steel at stress ratio 0.84 or lower after annealing at400°C for 3 hours.

EBSD results measured in the fractured areas of constant load test specimens arepresented in Publication III. Similarly to the EBSD results of fractures in theSwift cups, the microstructure in the close vicinity of the fracture surfaces con-sisted predominantly of α’-martensite.

72

5. Discussion

Metastable austenitic stainless steels, in which austenite is partly transformed tomartensite under plastic strain, possess an excellent combination of strength andelongation, as formation of strain-induced α’-martensite increases work harden-ing of the material. Generally, however, high strength means increased suscep-tibility to hydrogen embrittlement, and α’-martensite phase is more sensitive tothe effect of hydrogen than austenite. In this study, the presence of α’-martensitewas shown to be a necessary prerequisite for delayed cracking to occur in as-produced low-nickel austenitic stainless steels with typical internal hydrogenconcentrations (< 5.5 wppm). No delayed cracking was observed in the stablelow-nickel stainless steel 201Cu. The studied metastable low-Ni grades showedmore severe cracking at lower degree of deformation and lower volume fractionof α’-martensite than that of the metastable 300-series grades. The limiting draw-ing ratio for delayed cracking in Swift cup testing decreased with the nickel con-tent of the stainless steels. In the studied metastable low-Ni grades 204Cu,201(b) and 201(a) delayed cracking occurred when α’-martensite content ex-ceeded about 30 vol% and the severity of cracking increased strongly with thedegree of deformation. Stainless steel 204Cu with very low (1%) nickel contentwas the weakest material with respect to delayed cracking and its resistance tocrack propagation was very low, particularly above 45 vol% α’-martensite. Forthe 300-series metastable austenitic stainless steels the limiting α’-martensitecontent for delayed cracking was much higher, 55-60 vol% for 301 and 70 vol%for 301LN. Stainless steel 304 with the highest Ni content did not suffer of de-layed cracking even at the highest studied drawing ratio (2.12) in Swift cup test-ing, although the maximum α’-martensite content in the cups was almost 50 %.

Volume fraction of α’-martensite in the studied materials was measured withmagnetic induction method (Ferritescope). In another study within the same re-search project, α’-martensite phase fractions of the same test materials in un-stressed state after pre-straining to certain levels was determined by three meth-ods: Ferritescope, EBSD analysis and quatitative optical metallography (imageanalysis), and a good correlation between all the methods was found (Oikari,2010). Ferritescope measurement was therefore considered to be a reliablemethod. In the determination of α’-martensite content of the Swift cups, the Vil-

Discussion

73

lari effect, i.e. the dependency of magnetic permeability and thus the Fer-ritescope reading on prevailing stress, was taken into account by experimentallydefined correction factors for each material. These correction factors, deter-mined with tensile tests, were also confirmed by the measurements of α’-mar-tensite content from uncracked and cracked Swift cups. Residual stress state inthe cups had a marked effect on the measured Ferritescope readings. If the Villarieffect was not taken into account, it would lead to large (up to 35 %) underesti-mation of the true α’-martensite content in the Swift cups.

The analysis depth of Ferritescope is on the order of a few millimetres, so that itgave an average through-thickness value of the ferromagnetic phase content inthe walls of the studied Swift cups. According to simulation of deep drawingprocess of metastable austenitic stainless steel (Gallée et al., 2010), there existsa gradient in the volume fraction of α’-martensite between the outer and innersurfaces of the cup. The volume fraction of α’-martensite is higher in the outersurface, and the through-thickness gradient is higher at the middle sections ofthe cup, in comparison to section close to the upper edge of the cup.

High residual stresses existed in the deep-drawn Swift cups. With FEM simula-tions it has been shown that residual stresses introduced to a deep drawn cup aremainly caused by the unbending of the material when it leaves the draw die pro-file (Danckert, 1995; Fereshteh-Saniee et al., 2003). Tensile residual stress in thestudied Swift cups reached a maximum value at about 50-60% of cup height,which is in agreement with previous results (Berrahmoune et al., 2006 A).

Strain-induced martensitic transformation substantially increased the magnitudeof residual stresses in deep-drawn cups. Macroscopic hoop residual stresses wereseveral hundred MPa lower in the Swift cups of stable stainless steel 201Cu incomparison to the cups of metastable austenitic stainless steels 201(a) and 201(b)at a constant drawing ratio. According to the two residual stress measurementmethods used, the magnitude of residual stresses was highest in 204Cu, 201(b)and 301, in comparison to the other test materials at a constant drawing ratio.High residual stresses, particularly in the α’-martensite phase, are evidently oneexplanatory factor for the high susceptibility of these metastable low-Ni stainlesssteels to delayed cracking. Although the limiting drawing ratios for delayedcracking were different in each material, the limiting level of tangential residualstress not causing delayed cracking was similar in 201(b), 301 and 301LN steels,according to both the X-ray diffraction and ring slitting methods. In contrast,grade 204Cu suffered from delayed cracking already at considerably lower re-sidual stress levels.

In each studied material, the macroscopic hoop residual stress measured by thering slitting method was between the XRD phase stresses of austenite and α’-

Discussion

74

martensite at the same height of the cup. The results obtained with the two meth-ods are in good agreement with each other. A major difference between the twoapplied methods is that with XRD the stress is measured from the outer surfacelayer and in ring slitting the result is an average stress through the cup wall thick-ness. The ring opening characterizes springback caused by non-uniform residualstresses through the thickness of the material that create a bending moment. Ac-cording to former experimental measurements and FEM simulation about thethrough thickness variation of the residual stresses, tensile residual stress is thehighest at the outer surface of the cups, where the XRD measurements were per-formed (Gnaeupel-Herold et al., 2004; Hong et al., 2013 A). Residual stresses atthe inner surface of the cup are compressive. According to the results measuredwith both methods, tangential residual stresses reach a maximum at about middleheight of the Swift cups. This indicates that the difference between the tensileresidual stresses at the outer surface and the compressive stresses at the innersurface is highest at the middle part of the cup.

Near the upper edge of the cups, tangential residual stress values measured withX-ray diffraction were relatively low in comparison to the macroscopic tangen-tial stress values measured by the slitting method. Close to the edge the linearityof the d-sin2Ψ plots in XRD results was not as good as in other parts of the cups,which may have caused inaccuracy in the calculated stress values near the edge.In anisotropic materials having a strong deformation texture and high density oflattice defects, the measured diffraction peaks show relatively low intensity andbroadening, which makes exact peak positioning difficult affecting the precisionof the residual stress calculation (Kapoor et al., 2002; Luo et al., 2010).

Strain-induced α’-martensite has an important role in the delayed cracking mech-anism, as it provides a fast diffusion path for hydrogen. α’-martensite with highhardness, dislocation density and internal stresses is inherently more susceptibleto embrittling effect of hydrogen than austenite. Alloying elements of the stain-less steels influence the sensitivity to delayed cracking through their effect onthe austenite stability and the mechanical properties of the material. Accordingto nanoindentation measurements the indentation hardness of α’-martensite washighest in 301 and 201(b). Stainless steel 301 had the highest carbon contentamong the test materials and 201(b) the highest nitrogen content. Hardness ofthe phases, particularly of α’-martensite, correlated with the level of residualstresses in the materials and their susceptibility to delayed cracking. Stainlesssteel 304, which was not susceptible to delayed cracking, had the lowest inden-tation hardness and relatively low residual stresses.

Austenite stability against strain-induced martensitic transformation has beenshown to be a necessary but not a sufficient criterion for hydrogen embrittlement

Discussion

75

(Michler et al., 2010). It has been recently indicated, that strain hardening in-creases the susceptibility of austenitic stainless steels to hydrogen embrittlementmainly by creating a microstructure with a “critical” number of dislocations(Michler et al., 2015). According to tensile tests performed in hydrogen gas at-mosphere, the effect of strain-induced α’-martensite on the relative reduction ofarea seemed to be smaller compared to the effect of dislocation substructure.

The interaction of hydrogen with dislocations in the course of deformation hasbeen suggested to be mainly responsible for the macroscopic embrittlement ofaustenitic steels (Weber et al., 2011). Hydrogen trapped and transported by dis-locations increases the generation rate and velocity of dislocations, reduces therepulsion between dislocations and enhances the localization of deformation, sothe dislocation configuration is likely to play an important role in hydrogen em-brittlement processes (Michler et al., 2015).

From the measured X-ray diffraction data, useful information can be attained byobserving the width of the diffraction peaks. Full width at half maximum(FWHM) is an indication of plastic strain and dislocation density in the material.FWHM of the diffraction peaks measured from the austenite phase increasedcontinuously as a function of cup wall height. Instead, FWHM values measuredfrom the α’-martensite phase were higher, showed a more shallow gradient, andreached a maximum at about 60% of cup height. According to a former X-raydiffraction examination and dislocation density analysis of metastable austeniticstainless steels under tensile straining (Talonen, 2007), the dislocation density ofaustenite phase increases with increasing plastic strain and volume fraction ofα’-martensite, whereas the dislocation density of the α’-martensite phase re-mains relatively constant. The dislocation density is significantly higher in theα’-martensite phase than in the austenite phase, and plastic deformation occursmainly in the austenite phase.

The hydrogen content in the studied stainless steels in as-supplied state variedfrom 1.2 to 5.3 wppm. The measured values of hydrogen content were differentdepending on the analysis method, melt extraction in which the sample washeated up to melting point giving the largest values. The difference in the meas-ured values between the methods may also be a matter of equipment calibration.The variation in the results was considerably higher for the melt extractionmethod than for the other two methods. This may be explained by the smallerspecimen size in melt extraction analysis. The order between the grades was,however, similar with all used methods: the low-nickel Mn-alloyed grades hadconsiderably higher hydrogen contents than 301, 301LN and 304. It has beenreported previously, that increasing Mn alloying in steels increases hydrogensolubility in austenite (San Marchi et al., 2007; Ismer et al., 2010; Lob et al.,2011; Todoschenko et al., 2012) and this may explain their higher susceptibility

Discussion

76

to hydrogen embrittlement phenomena. Indication of this was also found in theTDS measuremens, as reported in Publication V. Hydrogen content was thehighest in 204Cu and 201(b) steels and standard deviation of the results was alsohighest in these two grades. This probably indicates that hydrogen distributionis more inhomogeneous in these grades. In metallographical examination of thetest materials it was noticed that especially in grade 201(b) there were separateareas highly concentrated of oxide, nitride and sulfide inclusions (Publication V;Pulkkinen, 2011). Inclusions may act as hydrogen traps.

It was evident that a few wppm of residual hydrogen present in stainless steelsin as-supplied state was sufficient to cause delayed cracking. Generally thistrapped “nondiffusible” hydrogen has been ignored as being a cause for hydro-gen embrittlement phenomena, and a majority of former studies have concen-trated on the effects of diffusible hydrogen introduced into the material by hy-drogen charging. In the case of fatigue loading, the typical 2-3 wppm of nondif-fusible hydrogen present in austenitic stainless steels has been shown to increasefatigue crack growth rates, when the loading frequency is small enough to allowsufficient time for the residual hydrogen to concentrate toward the crack tip (Mu-rakami et al., 2008). Reducing the level of nondiffusible hydrogen present inaustenitic stainless steels by special heat treatments (in 300-450°C) has beenshown to decrease fatigue crack growth rates in fatigue tests and to eliminatedelayed cracking in deep drawn cups (Sumitomo, 1987; Murakami et al., 2008).In the studied case, as a result of forming, strain-induced martensitic transfor-mation causes redistribution of hydrogen in the material and the presence of α’-martensite enhances hydrogen diffusivity. Therefore, after forming the internalhydrogen is not entirely nondiffusible.

Annealing of the stainless steels in 300-400°C for 3-24 hours reduced their hy-drogen content by 1-3 wppm and markedly lowered the risk of delayed fracture.Volume fraction of strain-induced α’-martensite produced during formingslightly decreased with hydrogen content. Hydrogen has been reported to affectthe SFE and stability of austenite (Ferreira et al., 1996; Michler at al., 2009).Residual stresses were also lower in cups drawn from annealed materials. Evensmall reductions in internal hydrogen content lowered the susceptibility for de-layed fracture by increasing the limiting levels of residual stress and α’-martensite sustained without fracture (Fig. 39). Resistance to crack propagationwas also improved by low hydrogen content. With lower hydrogen content, lim-iting α’-martensite content was less dependent on stress. The reduction of thehydrogen content of metastable stainless steels seems to be an effective way ofdecreasing their susceptibility to delayed cracking.

Discussion

77

Figure 39. Limiting combinations of residual stress and α’-martensite for delayed cracking in204Cu Swift cups in as-supplied and annealed conditions. In the left side of the curves therewas no delayed cracking observed.

Strain-induced α’-martensite has an important role in delayed cracking kinetics,as it provides a diffusion path for hydrogen and affects the crack initiation timeand crack growth rate. Cracking initiation times in the Swift cups, especially inas-supplied condition, were the shortest in stainless steel 204Cu. This may beexplained by the high volume fraction of α’-martensite combined with high hy-drogen content and residual stresses in the cups of 204Cu, all of which facilitatethe accumulation of a critical hydrogen content at a local stress concentrationsite. At a constant drawing ratio, the growth rate of cracks was highest in 204Cuin comparison to 201(b) and 301.

According to fractography, the fracture mechanism was essentially similar,mainly transgranular quasi-cleavage, in all the test materials and with all studiedα’-martensite volume fractions and hydrogen contents. According to EBSDanalyses, microstructure along the crack wakes and ahead of the crack tip waspredominantly α’-martensite and cracks mainly propagated through α’-martensite phase. This observation proves the important role of α’-martensite indelayed cracking of metastable austenitic stainless steels. Localized transfor-mation of austenite to α’-martensite occurs ahead of the crack as a response tointense plastic strain and the presence of hydrogen causes embrittlement of themartensitic structure. Hydrogen accumulates at the crack tip area under the in-fluence of hydrostatic stress that dilates the lattice. Additionally, the intense plas-tic straining at the crack tip area results in martensitic transformation and highdislocation density, which enhances hydrogen entry and trapping in the region(Lufrano et al., 1998). Increase in the amount of dissolved hydrogen decreases

CRACKING

Discussion

78

the stacking fault energy of austenite and in turn increases the volume fractionof strain-induced α’-martensite (Sumitomo et al., 1987; Ferreira et al., 1996). Ithas been suggested that the embrittlement of unstable stainless steels may beautocatalytic – martensite enhances hydrogen entry and cracking, which in turnaids martensite formation (Eliezer et al., 1979).

Crack initiation and propagation requires accumulation of a critical concentra-tion of hydrogen at local sites. Hydrogen diffuses to highly stressed areas underthe influence of hydrostatic stresses and trapping phenomena. When a criticalhydrogen concentration is reached locally, fracture occurs at the weakest elementof the microstructure. Crack propagation is due to multiple reoccurrence of thisprocess, consisting of a sequence of incubation periods, followed by a restrictedadvance of the crack (Kovalev et al., 2002). Thus, the requirement of hydrogenaccumulation by diffusion leads to stepwise crack growth.

High density of small, shallow dimples and facet-like features in the fracturesurfaces are an indication of localized deformation, according to the HELPmechanism (Tiwari et al., 2000; San Marchi et al., 2008; Hatano et al., 2014).Low nickel content promotes the localization of deformation in austenitic stain-less steels, which increases the susceptibility to hydrogen-assisted fracture.Cracking along α’-martensite may also be a hydrogen induced decohesion re-lated mechanism, and in internal hydrogen embrittlement the two mechanisms,HELP and HEDE, act in cooperative manner (Koyama et al., 2014). The differ-ent mechanisms are often interrelated. Hydrogen assisted fracture of a marten-sitic steel has recently been concluded to follow “a hydrogen-enhanced and plas-ticity mediated decohesion mechanism” (Nagao et al., 2012)

Determining the deformation processes that had preceded fracture, as well as theorigin of the features on the fracture surface cannot be made simply based on themorphological features. Furthermore, some of the fracture surface features maynot be related to the hydrogen embrittlement mechanism, but arise during thestages of final separation. Observations beneath hydrogen embrittlement fracturesurfaces, by using focused-ion beam machining and transmission electron mi-croscopy, have revealed intense slip bands and significant plasticity in close vi-cinity of the fracture surfaces (Nagao et al., 2012).

Swift cup tests provide a simple way to compare the performance of differentmaterials. However, exactly the same stress state is not usually reproducible inrepeated cup tests. Thus, a constant load tensile testing arrangement was set upfor making systematic experiments about the role of stress level, α’-martensitecontent and hydrogen content on delayed cracking susceptibility and kinetics.Time to fracture in stainless steel 204Cu, studied by constant load tensile tests,depended on applied stress level, α’-martensite content and hydrogen content,

Discussion

79

being the longer the lower those are. At high applied stress levels, α’-martensitecontent had very little effect on time to fracture, but hydrogen content had amarked effect. Hydrogen content seemed to have the strongest effect on crackinginitiation time at all tested stress levels. This supports the view that delayedcracking requires accumulation of a critical hydrogen concentration to sites ofhigh triaxial stress (Wang et al., 2006; Kim et al.,2009; Akiyama et al., 2011).At lower stress levels α’-martensite content had a clear effect on the crackinginitiation time, as the martensite phase offers the path for fast hydrogen diffu-sion. Limiting α’-martensite content of 204Cu steel for delayed cracking wasquite similar, about 30-35 %, with both approaches, Swift cup testing and con-stant load tensile tests. This limiting value is consistent with the theoretical per-colation threshold of α’-martensite, 0.312, in a three-dimensional cubic lattice(Talonen, 2007; Stauffer et al., 1994). Above the percolation threshold α’-mar-tensite forms percolating clusters and provides a continuous diffusion path forhydrogen.

Constant load tests were only performed for the steel 204Cu. Delayed fracturedid not take place in constant load tensile testing trials for 201(b) or 301 in theset time limit of 400 hours even at high stress levels. This was because the max-imum α’-martensite volume fraction attainable in uniaxial pre-straining of thetest specimens was lower than the limiting value for delayed cracking in 201(b)and 301.

Internal hydrogen seems to be the primary cause of delayed cracking in the stud-ied metastable austenitic stainless steels, and cracking is preceded by hydrogenaccumulation by diffusion. Strain-induced α’-martensite also has an importantrole, as it provides a diffusion path for hydrogen, and markedly increases theresidual stress state of a formed material. The presence of hydrogen causes em-brittlement of the martensitic structure. Cracking susceptibility is not directly afunction the volume fraction of α’-martensite, but depends additionally on themechanical properties of the phases. Delayed cracking in low-nickel austeniticstainless steels could be avoided by designing the alloy to be stable against mar-tensitic transformation, but that would mean considerably lower strength levels.Alternatively, the susceptibility of metastable austenitic stainless steels to de-layed cracking can be substantially decreased by reducing the internal hydrogencontent.

In possible further studies, it would be very interesting to combine the presentedexperimental results with modelling of hydrogen diffusion and delayed crackingbehavior in these materials. More detailed examination on the effect of specificalloying elements on hydrogen solubility, diffusibility and delayed cracking sus-ceptibility is suggested as well. Deeper analysis of hydrogen trapping sites in

Discussion

80

each test material could be done using thermal desorption spectroscopy. Analy-sis of the binding energies and microstructural nature of the TDS peaks was notin the scope of this study.

81

6. Conclusions

Based on the results of this research, the following conclusions can be drawn:- Internal hydrogen contained in as-supplied low-nickel austenitic stainless steelsis sufficient to cause delayed cracking, if high enough amount of α’-martensiteand high residual stresses are present.

- The presence of strain-induced α’-martensite is a necessary prerequisite fordelayed cracking. Delayed cracking did not occur in stable austenitic stainlesssteel 201Cu, and not below 30% α’-martensite content in metastable grades201(a), 201(b) and 204Cu. At lower volume fractions the α’-martensite phaseapparently does not form a continuous network to allow fast hydrogen diffusionto potential cracking sites.

- Strain-induced martensitic transformation substantially increases the magni-tude of residual stresses in a formed material. Residual stress is higher in α’-martensite than in austenite. High residual stresses increase the susceptibility todelayed cracking.

- Chemical composition of stainless steels influences their susceptibility to de-layed cracking through its effects on austenite stability and mechanical proper-ties of the phases. Hardness of the phases, α’-martensite in particular, correlateswith residual stress level and delayed cracking susceptibility.

- Crack initiation time and crack growth rate in Swift cups depend on α’-marten-site volume fraction and hydrogen content, as they both affect the rate of accu-mulation of hydrogen in the crack tip region.

- Time to fracture in the constant load tensile tests depends on applied stresslevel, α’-martensite content and hydrogen content. At high applied stress levels,α’-martensite content has only a minor effect on time to fracture, but hydrogencontent has a marked effect. Delayed cracking requires accumulation of hydro-gen at highly stressed areas.

Conclusions

82

- Cracking mainly propagates through α’-martensite phase, because local mar-tensitic transformation occurs ahead of the crack tip. Volume fraction of α’-mar-tensite in the surrounding material affects cracking kinetics.

- In the studied low-Ni high-Mn austenitic stainless steels the internal hydrogencontent in as-supplied state was considerably higher than in the studied Fe-Cr-Ni grades. High hydrogen content, combined with high hardness of α’-marten-site and high level of residual stresses after forming, explain the increased sus-ceptibility of the metastable low-Ni grades to delayed cracking.

- Lowering the internal hydrogen content of the stainless steels by heat treatmentat 300-400°C can effectively reduce the risk of delayed cracking. At lower hy-drogen content a higher stress is required to initiate cracking.

83

References

Akiyama, E., Matsukado, K., Li, S. and Tsuzaki, K., Constant-load delayed fracture testof atmospherically corroded high stregth steels. Applied Surface Science, 257 (2011),p. 8275-8281.

Akiyama, E., Evaluation of delayed fracture property of high strength bolt steels, ISIJInternational, 52 (2012), p. 307-315.

Becker, W.T. and Shipley, R.J. (editors), Hydrogen damage and embrittlement, in: ASMHandbook, vol. 11, 2002, ASM International, p. 809-822.

Beese, A.M. and Mohr, D., Identification of the direction-dependency of the martensitictransformation in stainless steel using in situ magnetic permeability measurements, Ex-perimental Mechanics, 51 (2011), p. 667-676.

Berrahmoune, M.R., Berveiller, S., Inal, K., Moulin, A. and Patoor, E., Analysis of themartensitic transformation at various scales in TRIP steel, Materials Science and Engi-neering A, 378 (2004), p. 304-307.

Berrahmoune, M. R., Berveiller. S., Inal. K. and Patoor, E., Delayed cracking in 301LNaustenitic steel after deep drawing: martensitic transformation and residual stress anal-ysis, Materials Science and Engineering A, 438-440 (2006), p. 262-266.

Berrahmoune, M. R., Berveiller, S., Inal, K. and Patoor, E., Residual stress state at dif-ferent scales in deep drawn cup of unstable austenitic steel, Materials Science Forum,524-525 (2006), p. 95-100.

Berveiller, S., Kemdehoundja, M., Patoor, E., Bouscaud, D. and Berrahmoune, M.R.,ESOMAT 2009 - 8th European Symposium on Martensitic Transformations, Prague,September 7-11, 2009.

Birnbaum, H.K. and Sofronis, P., Hydrogen-enhanced localized plasticity—a mecha-nism for hydrogen-related fracture, Materials Science and Engineering A, 176 (1994),p. 191-202.

Che, L., Gotoh, M., Horimoto, Y. and Hirose, Y., Effect of microstructure of cementiteon interphase stress state in carbon steel, Journal of Iron and Steel Research Interna-tional, 14 (2007), p. 31-38.

Chen, T.C., Chen, S.T. and Tsay, L.W., The role of induced α’-martensite on the hydro-gen-assisted fatigue crack growth of austenitic stainless steels, International Journal ofHydrogen Energy, 39 (2014), p. 10293-10302.

References

84

Chu, W.-Y., Yao, J. and Hsiao, C.-M., Hydrogen induced slow crack growth in stableaustenitic stainless steels, Metallurgical Transactions A, 15A (1984), p. 729-733.

Chun, Y.S., Park, K.-T. and Lee, C.S., Delayed static failure of twinning-induced plas-ticity steels, Scripta Materialia, 66 (2012), p. 960-965.

Colgan, M. and Monaghan, J., Deep drawing process: analysis and experiment, Journalof Materials Processing Technology, 132 (2003), p. 35-41.

Cutler, C.P., Coates, G.E. and Jenkinson D.C., Nickel in stainless steels, Proceedings ofthe Sixth European Stainless Steel Science and Market Conference, June 10-13, 2008,Helsinki, Finland, p. 421-426.

Danckert, J., Reduction of the residual stresses in a deep-drawn cup by modifying thedraw die profile, CIRP Annals – Manufacturing Technology, 44 (1995), p. 259-262.

Das, A. and Tarafder, S., Experimental investigation on martensitic transformation andfracture morphologies of austenitic stainless steel, International Journal of Plasticity, 25(2009), p. 2222-2247.

Eliezer, D., Chakrapani, D.G., Altstetter, C.J. and Pugh, E.N., The influence of austenitestability on the hydrogen embrittlement and stress-corrosion cracking of stainless steels,Met. Trans. A, 10 (1979), p. 935-941.

Fereshteh-Saniee, F. and Montazeran, M.H., A comparative estimation of the formingload in the deep drawing process, Journal of Materials Processing Technology, 140(2003), p. 555-561.

Ferreira, P.J., Robertson, I.M. and Birnbaum., H.K., Influence of hydrogen on the stack-ing-fault energy of an austenitic stainless steel, Materials Science Forum, 207-209(1996), p. 93-96.

Frehn, A. and Bleck, W., Effect of austenite stability on the mechanical properties anddelayed cracking in metastable austenitic steels, Stainless Steel World, 2003, Jan/Feb,p. 40-45.

Fujioka, T., Hoshino, K., Igawa, T. and Ito, K., Austenitic stainless steel, US3910788A,Nisshin Steel Co Ltd, 1975.

Gallée, S. and Pilvin, P. Deep drawing simulation of a metastable austenitic stainlesssteel using a two-phase model, Journal of Materials Processing Technology, 210 (2010),p. 835-843.

Gangloff, R. P., Hydrogen-assisted cracking, in: Comprehensive structural integrity,Vol. 6, ed. Petit, J and Scott, P. M., Oxford, 2003, Elsevier Ltd.

References

85

Garstka, T. and Dyl, T., Circumferential residual stresses in tubes estimated by meansof deflection method, Archives of Metallurgy and Materials, 51 (2006), p. 199-203.

Gavriljuk, V.G., Shivanyuk, V.N. and Foct, J., Diagnostic experimental results on thehydrogen embrittlement of austenitic steels, Acta Materialia, 51 (2003), p. 1293-1305.

Gavriljuk, V.G., Shivanyuk, V.N. and Shanina, B.D., Change in the electron structurecaused by C, N and H atoms in iron and its effect on their interaction with dislocations,Acta Materialia, 53 (2005), p. 5017-5024.

Gnaeupel-Herold, T., Prask, H., Fields, R., Foecke, T., Xia, C. and Lienert, U., A syn-chrotron study of residual stresses in a Al6022 deep drawn cup, Materials Science andEngineering A, 366 (2004), p. 104-113.

Gonzalez, B.M., Castro, C.S.B., Buono, V.T.L., Vilela, J.M.C., Andrade, M.S. Moraes,J.M.D. and Mantel, M.J., The influence of copper addition on the formability of AISI304 stainless steel, Materials Science and Engineering A, 343 (2003), p. 51-56.

Guo, X., Post, J., Groen, M. and Bleck, W., Stress oriented delayed cracking inducedby dynamic martensitic transformation in meta-stable austenitic stainless steels, SteelResearch International, 82 (2011), p. 6-13.

Hatano, M., Fujinami, M., Arai, K., Fujii, H. and Nagumo, M., Hydrogen embrittlementof austenitic stainless steels revealed by deformation microstructures and strain-inducedcreation of vacancies, Acta Materialia, 67 (2014), p. 342-353.

Hausild, P., Materna, A. and Nohava, J., Characterization of anisotropy in hardness andindentation modulus by nanoindentation, Metallography, Microstructure and Analysis,3 (2014), p. 5-10.

Hong, S., Lee, J., Lee, B.-J., Kim, H., Kim, S.-K., Chin, K.-G. and Lee, S., Effects ofintergranular carbide precipitation on delayed fracture behavior of three twinning in-duced plasticity (TWIP) steels, Materials Science and Engineering A, 587 (2013), p. 85-99.

Hong, S., Shin, S. Y., Kim, H. S., Lee, S., Kim, S.-K., Chin, K.-G. and Kim, N. J.,Effects of inclusions on delayed fracture properties of three twinning induced plasticity(TWIP) steels, Metallurgical and Materials Transactions A, 44 (2013), p. 776-786.

Hoshino, K., Effect of chemical compositions on season cracking of metastable austen-itic stainless steel, Transactions ISIJ, 20 (1980), p. 147-153.

Hussaini, S. M., Singh, S. K. and Gupta, A. K., Formability and fracture studies of aus-tenitic stainless steel 316 at different temperatures, Journal of King Saud University –Engineering Sciences, 26 (2014), p. 184-190.

References

86

Hänninen, H. and Hakkarainen, T., On the effects of α’ martensite in hydrogen embrit-tlement of a cathodically charged AISI type 304 austenitic stainless steel, Corrosion, 36(1980), p. 47-51.

Ismer, L., Hickel, T. and Neugebauer, J., Ab initio study of the solubility and kineticsof hydrogen in austenitic high Mn steels, Physical Review B, 81 (2010), p. 094111 1-9.

Johansson, J. and Odén, M., Load sharing between austenite and ferrite in a duplexstainless steel during cyclic loading, Metallurgical and Materials Transactions A, 31A(2000), p. 1557-1570.

Johnson, J., Reck, B.K., Wang, T. and Graedel, T.E., The energy benefit of stainlesssteel recycling, Energy Policy, 36 (2008), p. 181-192.

Kanezaki, T., Narazaki, C., Mine, Y., Matsuoka, S. and Murakami, Y., Effects of hy-drogen on fatigue crack growth behavior of austenitic stainless steels, InternationalJournal of Hydrogen Energy, 33 (2008), p. 2604-2619.

Kapoor, K., Lahiri, D., Padmaprabu, C. and Sanyal, T., X-ray measurement of near sur-face residual stress in textured cold-worked stress-relieved Zr-2.5%Nb pressure tubematerial, Journal of Nuclear Materials, 303 (2002), p. 147-155.

Karjalainen, L. P., Taulavuori, T., Sellman, M. and Kyröläinen, A., Some strengtheningmethods for austenitic stainless steels, Steel Research International, 79 (2008), p. 404-412.

Kim, J. S., Lee, Y. H., Lee, D. L., Park, K.-T. and Lee, C. S., Microstructural influenceson hydrogen delayed fracture of high strength steels, Materials Science and EngineeringA, 505 (2009), p. 105-110.

Kovalev, A. I., Wainstein, V. P., Mishina, V. P. and Zabilsky, V. V., Effect of residualstress on hydrogen embrittlement and stress corrosion cracking, in: Handbook of resid-ual stress and deformation of steel, Materials Park, Ohio, 2002, ASM International, p.70-88.

Koyama, M., Tasan, C., Akiyama, E., Tsuzaki, K. and Raabe, D., Hydrogen-assisteddecohesion and localized plasticity in dual-phase steel, Acta Materialia, 70 (2014), p.174-187.

Kumar, P. and Balasubramaniam, R., Determination of hydrogen diffusivity in austen-itic stainless steels by subscale microhardness profiling, Journal of Alloys and Com-pounds, 255 (1997), p. 130-134.

Lai, C.L., Tsay, L.W. and Chen, C., Effect of microstructure on hydrogen embrittlementof various stainless steels, Materials Science and Engineering A, 584 (2013), p. 14-20.

References

87

Lebedev, A.A. and Kosarchuk, V.V., Influence of phase transformations on the me-chanical properties of austenitic stainless steels, International Journal of Plasticity, 16(2000), p. 749-767.

Lo, K.H., Shek, C.H. and Lai, J.K.L., Recent developments in stainless steels, MaterialsScience and Engineering R, 65 (2009), p. 39-104.

Lob, A., Senk, D. and Hallstedt, B., Determination of hydrogen solubility in Fe-Mn-Cmelts, Steel Research International, 82 (2011), p. 108-113.

Lufrano, J. and Sofronis, P., Enhanced hydrogen concentrations ahead of roundednotches and cracks – competition between plastic strain and hydrostatic stress, ActaMaterialia, 46 (1998), p. 1519-1526.

Luo, Q. and Jones, A.H., High-precision determination of residual stress of polycrystal-line coatings using optimized XRD-sin2Ψ technique, Surface and Coatings Technology,205 (2010), p. 1403-1408.

Lynch, S.P., Environmentally assisted cracking: Overview of evidence for an adsorp-tion-induced localized-slip process, Acta Metallurgica, 20 (1988), p. 2639-2661.

Marciniak, Z., Duncan, J. L. and Hu, S. J. (editors), Cylindrical deep drawing, in: Me-chanics of sheet metal forming, 2nd ed., Oxford, 2002, Butterworth-Heinemann, p. 117-127.

Martin, M., Weber, S., Izawa, C., Wagner, S., Pundt, A. and Theisen, W., Influence ofmachining-induced martensite on hydrogen-assisted fracture of AISI type 304 austeniticstainless steel, International Journal of Hydrogen Energy, 36 (2011), p. 11195-11206.

Martin, M., Fenske, J., Liu, G., Sofronis, P. and Robertson, I., On the formation andnature of quasi-cleavage fracture surfaces in hydrogen embrittled steels, Acta Materi-alia, 59 (2011), p. 1601-1606.

Michler, T., Lee, Y., Gangloff, R. P. and Naumann, J., Influence of macro segregationon hydrogen environment embrittlement of SUS 316L stainless steel, International Jour-nal of Hydrogen Energy, 34 (2009), p. 3201-3209.

Michler, T. and Naumann, J., Hydrogen embrittlement of Cr-Mn-N austenitic stainlesssteels, International Journal of Hydrogen Energy, 35 (2010), p. 1485-1492.

Michler, T., San Marchi, C., Naumann, J., Weber, S. and Martin, M., Hydrogen envi-ronment embrittlement of stable austenitic steels, International Journal of Hydrogen En-ergy, 37 (2012), p. 16231-16246.

Michler, T., Naumann, J., Hock, M., Berreth, K., Balogh, M.P. and Sattler, E., Micro-structural properties controlling hydrogen environment embrittlement of cold worked

References

88

316 type austenitic stainless steels, Materials Science and Engineering A, 628 (2015),p. 252-261.

Miles, M., Formability testing of sheet metals, in: ASM Handbook, vol. 14B, MaterialsPark, OHIO, 2006, ASM International, p. 673-696.

Milititsky, M., De Wispelaere, N., Petrov, R., Ramos, J.E., Reguly, A. and Hänninen,H., Characterization of the mechanical properties of low-nickel austenitic stainlesssteels, Materials Science and Engineering A, 498 (2008), p. 289-295.

Mohrbacher, H., Delayed cracking in ultra-high strength automotive steels: Damagemechanisms and remedies by microstructural engineering, Materials Science and Tech-nology, 10 (2008), p. 1744-1755.

Murakami, Y., Kanezaki, T., Mine, Y. and Matsuoka, S., Hydrogen embrittlementmechanism in fatigue of austenitic stainless steels, Metallurgical and Materials Trans-actions A, 39 (2008), p. 1327-1339.

Nagao, A., Smith, C., Dadfarnia, M., Sofronis, P. and Robertson, I., The role of hydro-gen in hydrogen embrittlement fracture of lath martensitic steel, Acta Materialia, 60(2012), p. 5182-5189.

Nagumo, M., Takai, K. and Okuda, N., Nature of hydrogen trapping sites in steels in-duced by plastic deformation, Journal of Alloys and Compounds, 293-295 (1999), p.310-316.

Nagumo, M., Function of hydrogen in embrittlement of high-strength steels, ISIJ Inter-national, 41 (2001), p. 590-598.

Nagumo, M., Sekiguchi, S., Hayashi, H. and Takai, K., Enhanced susceptibility to de-layed fracture in pre-fatigued martensitic steel, Materials Science and Engineering A,344 (2003), p. 86-91.

Narita, N. and Birnbaum, H.K., On the role of phase transitions in the hydrogen embrit-tlement of stainless steels, Scripta Metallurgica, 14 (1980) p. 1355-1358.

Nibur, K.A., Bahr, D.F. and Somerday, B.P., Hydrogen effects on dislocation activityin austenitic stainless steel, Acta Materialia, 54 (2006), p. 2677-2684.

Nibur, K.A., Somerday, B.P., Balch, D.K. and San Marchi, C., The role of localizeddeformation in hydrogen-assisted crack propagation in 21Cr-6Ni-9Mn stainless steel,Acta Materiala, 57 (2009), p. 3795-3809.

Norgate, T.E., Jahanshahi, S. and Rankin, W.J., Alternative routes to stainless steel – alife cycle approach, Proceedings of the Tenth International Ferroalloys Congress, Feb-ruary 1-4, 2004, Cape Town, South Africa, p. 693-704.

References

89

Noyan, I. C. and Cohen, J. B., Residual stress – measurement by diffraction and inter-pretation, New York, 1987, Springer-Verlag, 276 p. ISBN 0-387-96378-2.

Oikari, A., Deformation mechanisms and strain hardening of chromium-manganese-nickel alloyed austenitic stainless steels, Master’s thesis, Aalto University, 2010.

Oliver, W.C. and Pharr, G.M., An improved technique for determining hardness andelastic modulus using load and displacement sensing indentation experiments, Journalof Materials Research, 7 (1992), p. 1564-1583.

Oriani, A., The mechanistic theory of hydrogen embrittlement of steels, Berichte derBunsengesellschaft für Physikalische Chemie, 76 (1972), p. 848-857.

Pan, C., Chu, W. Y., Li, Z. B., Liang, D. T., Su, Y. J., Gao, K. W. and Qiao, L. J.,Hydrogen embrittlement induced by atomic hydrogen and hydrogen-induced marten-sites in type 304L stainless steel, Materials Science and Engineering A, 351 (2003), p.293-298.

Pérez Escobar, D., Verbeken, K., Duprez, L. and Verhaege, M., Evaluation of hydrogentrapping in high strength steels by thermal desorption spectroscopy, Materials Scienceand Engineering A, 551 (2012), p. 50-58.

Perng, T-P. and Altstetter, C.J., Effects of deformation on hydrogen permeation in aus-tenitic stainless steels, Acta Metallurgica, 34 (1986), p. 1771-1781.

Perng, T-P., Johnson, M. and Altstetter, C.J., Influence of plastic deformation on hy-drogen diffusion and permeation in stainless steels, Acta Metallurgica, 37 (1989), p.3393-3397.

Perng, T-P. and Altstetter, C.J., Hydrogen effects in austenitic stainless steels, MaterialsScience and Engineering A, 129 (1990), p. 99-107.

Post, J., Datta, K. and Beyer, J., A macroscopic constitutive model for a metastableaustenitic stainless steel, Materials Science and Engineering A, 485 (2008), p. 290-298.

Pulkkinen, H., Effect of inclusions and precipitates on pitting corrosion and hydrogenembrittlement of Mn-alloyed austenitic stainless steels, Master’s thesis, Aalto Univer-sity, 2011.

Ratte, E., Hydrogen induced delayed cracking of austenitic CrNi(Mn)- and Mn-basedsteels, Doctoral Thesis, Technical University of Aachen, May 8, 2007.

Robertson, I.M., The effect of hydrogen on dislocation dynamics, Engineering FractureMechanics, 68 (2001), p. 671-692.

References

90

Robertson, I.M., Sofronis, P., Nagao, A., Martin, M.L., Wang, S., Gross, D.W. and Ny-gren, K.E., Hydrogen embrittlement understood, Metallurgical and Materials Transac-tions B, 46 (2015), p. 1085-1103.

da Rocha, M. R. and de Oliviera, C. A., Evaluation of the martensitic transformationsin austenitic stainless steels, Materials Science and Engineering A, 517 (2009), p. 281-285

Rozenak, P., Robertson, I.M. and Birnbaum, H.K., HVEM studies of the effects of hy-drogen on the deformation and fracture of AISI type 316 austenitic stainless steel, ActaMetallurgica et Materialia, 38 (1990), p. 2031-2040.

San Marchi, C., Somerday, B.P. and Robinson, S.L., Permeability, solubility and dif-fusivity of hydrogen isotopes in stainless steels at high gas pressures, International Jour-nal of Hydrogen Energy, 32 (2007), p. 100-116.

San Marchi, C., Somerday, B.P., Tang, X. and Schiroky, G.H., Effects of alloy compo-sition and strain hardening on tensile fracture of hydrogen-precharged type 316 stainlesssteels, International Journal of Hydrogen Energy, 33 (2008), p. 889-904.

Sargent, P.M. and Page, T.F., The possible effects of elastic recovery on the microhard-ness of anisotropic materials, Scripta Metallurgica, 15 (1981), p. 245-250.

Schaller, F.W., Schmid, T.E. and Snape, E., Delayed cracking of deep-drawn stainlesssteel, Sheet Metal Industries, 49 (1972), p. 621-624.

Shivanyuk, V.N., Foct, J. and Gavriljuk, V.G., Hydrogen-enhanced microplasticity ofaustenitic steels studied by means of internal friction, Materials Science and Engineer-ing A, 300 (2001), p. 284-290.

Shuman, D., Costa A. and Andrade, M., Calculating the elastic modulus fromnanoindentation and microindentation reload curves, Materials Characterization, 58(2007), p. 380-389.

Sibanda, M., Vismer, S.L. and Knutsen, R.D., Consideration of reduced nickel contain-ing austenitic stainless steels for forming applications, Materials Letters, 21 (1994), p.203-207.

Simmons, J.W., Overview: high-nitrogen alloying of stainless steels, Materials Scienceand Engineering A, 207 (1996), p. 159-169.

Singh, S. and Altstetter, C., Effects of hydrogen concentration on slow crack growth instainless steels, Metallurgical Transactions A, 13A (1982), p. 1799-1808.

Sofronis, P., Liang, Y. and Aravas, N., Hydrogen induced shear localization of theplastic flow in metals and alloys, European Journal of Mechanics - A/Solids, 20(2001), p. 857-872.

References

91

Stauffer, D. and Aharony, A., Introduction to percolation theory, 2nd ed., London, 1994,Taylor & Francis Ltd.

Stinville, J. C., Thomas, C., Villechaise, P. and Templier C., Anisotropy changes inhardness and indentation modulus induced by plasma nitriding of 316L polycrystallinestainless steel, Scripta Materialia, 64 (2011), p. 37-40.

Sumitomo, H., Earing and delayed cracking of deep-drawn cup of austenitic stainlesssteel sheets, Advanced Technology of Plasticity, vol. II, Proceedings of the 2nd Interna-tional Conference on Technology of Plasticity, Stuttgart, August 24-28, 1987, p. 1289-1296.

Takagi, S., Toji, Y., Yoshino, M. and Hasegawa, K., Hydrogen embrittlement resistanceevaluation of ultra high strength steel sheets for automobiles, ISIJ International, 52(2012), p. 316-322.

Talonen, J., Effect of strain-induced α’-martensite transformation on mechanical prop-erties of metastable austenitic stainless steels, Doctoral Thesis, 2007, Helsinki Univer-sity of Technology, Espoo, Finland.

Talonen, J., Aspegren, P. and Hänninen, H., Comparison of different methods for meas-uring strain-induced α’-martensite content in austenitic steels, Materials Science andTechnology, 20 (2004), p. 1506-1512.

Taran, Y.V., Daymond, M.R. and Schreiber, J., Interplay of stresses induced by phasetransformation and plastic deformation during cyclic load of austenitic stainless steel,Physica B, 350 (2004), p. 98-101.

Teus, S. M., Shyvanyuk, V.N. and Gavriljuk, V.G., Hydrogen-induced γ→ε transfor-mation and the role of ε-martensite in hydrogen embrittlement of austenitic steels, Ma-terials Science and Engineering A, 497 (2008), p. 290-294.

Tiwari, G.P., Bose, A., Chakravartty, J.K., Wadekar, S.L., Totlani, M.K., Arya, R.N.and Fotedar, R.K., A study of internal hydrogen embrittlement of steels, Materials Sci-ence and Engineering A, 286 (2000), p. 269-281.

Todoshchenko, O., Yagodzinskyy, Y., Papula, S. and Hänninen, H., Hydrogen solubil-ity and diffusion in metastable austenitic stainless steels studied with thermal desorptionspectroscopy, Proceedings of the International Hydrogen Conference, Sept. 9-12, 2012,Jackson Lake Lodge, Moran, Wyoming, USA, p. 615-623.

Toji, Y., Takagi, S., Yoshino, M., Hasegawa, K. and Tanaka, Y., Evaluation of hydro-gen embrittlement for high strength steel sheets, Materials Science Forum, 638-642(2010), p. 3537-3542.

Tumanski, S., Handbook of magnetic measurements, Boca Raton, FL, 2011, CRC Press,p. 240.

References

92

Vitos, L., Nilsson, J. O. and Johansson, B., Alloying effects on the stacking fault energyin austenitic stainless steels from first-principles theory, Acta Materialia, 54 (2006), p.3821-3826.

Walton, H. W., Deflection methods to estimate residual stress, in: Handbook of residualstress and deformation of steel, Materials Park, Ohio, 2002, ASM International, p. 89-98.

Wang, M., Akiyama, E. and Tsuzaki, K., Determination of the critical hydrogen con-centration for delayed fracture of high strength steel by constant load test and numericalcalculation, Corrosion Science, 48 (2006), p. 2189-2202.

Wang, Y., Lin Peng, R., Wang, X.-L. and McGreevy, R., Grain-orientation-dependentresidual stress and the effect of annealing in cold-rolled stainless steel, Acta Materialia,50 (2002), p. 1717-1734.

Wang, Z. and Gong, B., Residual stress in the forming of materials, in: Handbook ofresidual stress and deformation of steel, ASM International, Materials Park, Ohio, 2002,p. 141-149.

Weber, S., Martin, M. and Theisen, W., Computer assisted development of high alloyedsteels for hydrogen applications, HTM Journal of Heat Treatment and Materials, 65(2010), p. 230-234.

Weber, S., Martin, M. and Theisen, W., Lean-alloyed austenitic stainless steel with highresistance against hydrogen environment embrittlement, Materials Science and Engi-neering A, 528 (2011), p. 7688-7695.

Withers, P. J. and Bhadeshia, H. K. D. H., Residual stress, Part I – Measurement tech-niques, Materials Science and Technology, 17 (2001), p. 355-365.

Withers, P. J. and Bhadeshia, H. K. D. H., Residual stress, Part II – Nature and origins,Materials Science and Technology, 17 (2001), p. 366-375.

Yagodzinskyy, Y., Andronova, E., Ivanchenko, M. and Hänninen, H., Anelastic me-chanical loss spectrometry of hydrogen in austenitic stainless steels, Materials Scienceand Engineering A, 521-522 (2009), p. 159-162.

Yagodzinskyy, Y., Todoshchenko, O., Papula, S. and Hänninen, H., Hydrogen solubil-ity and diffusion in austenitic stainless steels studied with thermal desorption spectros-copy, Steel Research International., 82 (2011), p. 20-25.

Yuying, Y., Chunfeng, L. and Hongzhi, X., Study of longitudinal cracking and the form-ing technology for deep-drawn austenitic stainless steel cups, Journal of Materials Pro-cessing Technology, 30 (1992), p. 167-172.

References

93

Zhang L., Wen, M., Imade, M., Fukuyama, S. and Yokogawa, K., Effect of nickel equiv-alent on hydrogen gas embrittlement of austenitic stainless steels based on type 316 atlow temperatures, Acta Materialia, 56 (2008), p. 3414-3421.

Zhang, L., Imade, M., Wen, M., Iijima, T., Fukuyama, S., and Yokogawa, K., Internalreversible hydrogen embrittlement of austenitic stainless steels based on type 316 at lowtemperatures, ISIJ International, 52 (2012), p. 240-246.

Zinbi, A., Bouchou, A., Delayed cracking in 301 austenitic steel after bending process:martensitic transformation and hydrogen embrittlement analysis, Engineering FailureAnalysis, 17 (2010), p. 1028-1037.

9HSTFMG*agddei+

ISBN 978-952-60-6334-8 (printed) ISBN 978-952-60-6335-5 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) Aalto University School of Engineering Department of Engineering Design and Production www.aalto.fi

BUSINESS + ECONOMY ART + DESIGN + ARCHITECTURE SCIENCE + TECHNOLOGY CROSSOVER DOCTORAL DISSERTATIONS

Aalto-D

D 115

/2015

Metastable austenitic stainless steels possess an excellent combination of strength and elongation due to formation of strain-induced martensite which enhances work hardening. Low-nickel austenitic stainless steels provide a cost-effective alternative for the conventional austenitic Fe-Cr-Ni grades. Unfortunately, metastable low-nickel austenitic stainless steels can be susceptible to delayed cracking after forming operations. This thesis describes research aiming to clarify the role of hydrogen, residual stresses and strain-induced martensite in delayed cracking, and to find reasons for the high susceptibility of metastable low-Ni austenitic stainless steels. The research includes comprehensive Swift cup forming tests for comparing the delayed cracking susceptibility of seven austenitic stainless steels and finding the critical conditions leading to cracking, and constant load tensile tests for studying the kinetics of the phenomenon.

Suvi Papula D

elayed cracking of metastable low

-nickel austenitic stainless steels A

alto U

nive

rsity

Department of Engineering Design and Production

Delayed cracking of metastable low-nickel austenitic stainless steels

Suvi Papula

DOCTORAL DISSERTATIONS