Effect of Large Shear Deformation on Rail Steels and Pure Metals

University <strong>of</strong> Leoben
Doctoral Thesis
<strong>Effect</strong> <strong>of</strong> <strong>Large</strong> <strong>Shear</strong> <strong>Deformation</strong>
on Rail Steels and Pure Metals
Florian Wetscher
Leoben, March 2006

March 2 th , 2006
This doctoral thesis was typeset by the use <strong>of</strong> KOMA-Script and L ATEX 2ε.
The template was modified by Dr. Weinhandl and Dr. Vorhauer.
Copyright ©2006 by Florian Wetscher
For information, address:
Erich-Schmid-Institute <strong>of</strong> Materials Science <strong>of</strong> the Austrian Academy <strong>of</strong> Sciences,
Christian Doppler Laboratory <strong>of</strong> Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture,
Jahnstrasse 12, 8700 Leoben, Austria.
Homepage: http://www.oeaw.ac.at/esi

ASTRID, LÄTITIA

Ich erkläre an Eides statt, dass ich die hier
vorliegende Arbeit selbstständig verfasst
habe, und nur die hier angegebene Literatur
verwendet habe.
Erklärung
Leoben, März 2006
V

Danksagung
Die hier vorliegende Arbeit wurde im Rahmen meiner Tätigkeit als wissenschaftlicher Angestellter
des Erich-Schmid Instituts für Materialwissenschaften der Österreichischen Akademie der
Wissenschaften in den Jahren 2002 bis 2006 erstellt.
Bedanken möchte ich mich vor allem bei meinem Betreuer, Pr<strong>of</strong>. Dr. Reinhard Pippan für
den hervorragenden wissenschaftlichen Rückhalt und das in mich gesetzte Vertrauen. Dem
leider inzwischen verstorbenen Herrn Em. Univ. Pr<strong>of</strong>. DDr. Hein-Peter Stüwe danke ich für
viele anregende Diskussionen, die immer wieder geholfen haben, einen neuen Blickwinkel auf
die Ergebnisse zu bekommen. Mein Dank gilt auch unserem neuen Direktor, Univ. Pr<strong>of</strong>. Dr.
Gerhard Dehm, der mir mehrere Aufenthalte am Max Plank Institut in Stuttgart ermöglicht
hat, um wichtige Messungen durchführen zu können.
Besonders bedanken möchte ich mich noch bei unseren Kooperationspartner aus der Industrie,
Herrn Dr. Peter Pointer und Herrn DI Richard Stock von der voestAlpine Schienen GmbH für
die hervorragende Zusammenarbeit und die Bereitschaft, in grundlegende Forschung zu investieren.
Zu besonderen Dank bin ich auch den Herren Günther Aschauer und Franz Hubner von
unserer Werkstatt verpflichtet. Ob Proben oder spezielle Vorrichtungen gefertigt werden sollten,
immer wurden die Arbeiten schnell und sorgfältig durchgeführt. Für die hervorragende
Mithilfe bei zahlreichen Experimenten und der Weiterentwicklung verschiedenster Vorrichtung
gilt mein Dank Herrn Ing. Hannes Schlager. Sein Einfallsreichtum und Engagement
dabei sind besonders hervorzuheben.
Mein besonderer Dank gilt natürlich auch den Damen von der Metallographie, Frau Edeltraud
Haberz und Frau Gabriele Moser. Diese schafften mit großem Erfolg, meine nicht standardmäßigen
Proben regelmäßig und pünklich zu den Messungen zu präparieren. Man glaubt
kaum, wie widerspenstig sich feinstkörnige Materialien einer Präparation zu entziehen versuchen!
Den folgenden Herren danke ich für die Unterstützung bei verschiedensten technischen
Problemen und Aufgabenstellungen: Herwig Felber, Gerald Reiter, Jörg Thomas, und Fritz
Mitter.
Ohne Frau Marianne Fliesser, der Seele des Hauses, und Frau Doris Schrutt hätten viele
administrative Arbeiten sicherlich um vieles länger gedauert, danke auch dafür!
Für die Bereitschaft, mir bei vor allem bei wissenschaftlichen aber auch nichtwissenschaftlichen
Fragen zu helfen, bedanke ich mich bei Christian Motz, Ottmar Kolednik, Werner Prantl,
Balder Ortner, Herbert Weinhandl, Jozef Keckes und Thomas Schöberl.
Ich bedanke mich auch bei allen ehemaligen und aktuellen Bürokollegen, die mir ein angenehmes
Arbeiten ermöglicht haben, besonders bei Gernot Trattnig und Jaroslav Zenisek.
Mein besonderer Dank gilt Andreas Vorhauer, von dem ich viel über SPD gelernt habe. Die
VII

Zusammenarbeit mit ihm war immer ausgezeichnet.
Bedanken muss ich mich auch bei allen student’schen Hilfskräften, die für mich kleinere
und größere Arbeiten verrichtet haben: Beate Wagner, Christoph Kammerh<strong>of</strong>er und Peter
Jessner. Ganz besonders ist aber die Hilfe von Anton Hohenwarter herauszustreichen, der
immer höchst engagiert und verlässlich bei der Sache war.
Allen weiteren Mitarbeitern, Dissertanten, Diplomanten und Werksstudenten danke ich für
das angenhme Arbeitsklima und die freundliche Aufnahme.
Meiner Familie, besonders aber meiner Frau Astrid danke ich für den Rückhalt, die Unterstützung
und das Verständnis für meine Arbeit. Mein Dank geht auch an meine Tochter
Lätitia, die mich mit ihrem sonnigen Gemüt immer wieder auf neue Gedanken bringt.
VIII

Summary
The investigation <strong>of</strong> the effect <strong>of</strong> large shear deformations on rail steels is <strong>of</strong> great technical
importance. On the surface <strong>of</strong> rails a deformation layer evolves, that strongly influences the
further performance <strong>of</strong> the rail. Therefore it is necessary to know fundamental properties <strong>of</strong>
the deformed microstructure to develope tools (simulations, modells) for optimizing service
intervalls or for a purposefully designing <strong>of</strong> new materials.
The deformation <strong>of</strong> materials to very large strains under high hydrostatic pressure is possible
mainly because in the last fifteen year many methods <strong>of</strong> severe plastic deformation were
invented. The aim <strong>of</strong> this work is to deform model materials as well as the technical used
rail steels by these methods <strong>of</strong> severe plastic deformation under controlled conditions. The
changes <strong>of</strong> the microstructure and the resulting changes in the mechanical properties where
determined with this samples by applying various methods.
It can be seen that there are markedly differences in the development <strong>of</strong> the microstructure
between the pure metalls and the multiphase steels. In all pure metalls a similar behaviour can
be observed: At the beginning <strong>of</strong> the deformation by torsion under high pressure a quick fragmentation
<strong>of</strong> the large grains can be seen. This fragmentation is finished at a certain strain, that
is a function <strong>of</strong> the material and the crystall structure. If the sample is deformed to even higher
strains, no further refinement occures. When the microstructure is depicted in the direction
<strong>of</strong> the torsion axis, an equiaxed structure is present. The microstructure in radial direction <strong>of</strong>
the sample consists <strong>of</strong> elongated elements, that have a certain angle between the length axis <strong>of</strong>
the elements and the shear direction. The microstructure that develops during a cyclic form <strong>of</strong>
high pressure torsion is very similar. But it turned out that the size <strong>of</strong> the structural elements
after the onset <strong>of</strong> saturation is strongly determined by the strain per cycle. The smaller this
strain increment is, the larger is the resulting microstructure. Also the developement <strong>of</strong> the
missorientation between the elements as well as the character <strong>of</strong> the boundaries between them
is markedly influence by the deformation mode. Due to this fragmentation and grain refinement,
an enourmous increase <strong>of</strong> the mechanical strength is present, both in monotonic and in
cyclic experiments. Again, the increase in strength is largest for the monotonic deformation,
the smallest increase in the strength was measured for the smallest applied strains per deformation
step.
In contradiction to pure metalls, the development <strong>of</strong> the microstructure in steels is mainly
influence by the behaviour <strong>of</strong> the carbides. Coarse cementite lamellae are severely deformed
and fragmented, the fragments algin parallel to the shear plane. If the lamellae are finer, they
are more easily to deform without fragmentation. A markedly fragmentation and alignment
can only be observed after larger strains. The very fine carbides <strong>of</strong> the bainitic steel are hardly
deformed at the beginning <strong>of</strong> the deformation and just align accoring to the actual shear angle.
Just as in pure metals, also in steels the microstructure resulting from cyclic deformation is
strongly influenced by the strain increment. When this strain increment is small, a very high
total strain is necessary to fragment the lamellae and align the fragments. At higher strains per
deformation step, similar microstructurs as after monotonic deformation are observed. Due
to the high shear, also a markedly decrease both in lamellae spacing down to some 10 nm as
well as in the lamellae thickness down to 1-2 nm occurs. After strains <strong>of</strong> ≈ 800% in all steels
IX

a very fine lamellar structure parallel to the shear plane is present. Indications from X-ray
investigation suggest that a dissolution <strong>of</strong> the carbon from the carbide has occured. This could
be verified by measuring the electronic structure <strong>of</strong> the iron for different strains.
The distinctive decrease <strong>of</strong> the lamellae spacing and the alignment <strong>of</strong> the lamellae naturally
leads to a change in the mechanical properties. Due to the strong anisotropy in the microstructure,
also a strong anisotropy in the mechanical properties is present. In the direction <strong>of</strong> this
aligned lamellar structure a distinctive increase in the mechanical strength can be measured
(tensile strength <strong>of</strong> more than 3 GPa). The strength normal to this prefered direction is also
increased, but much lower than the strength in the direction <strong>of</strong> the aligned lamellae. The mechanical
strength, the fracture toughness and the fatigue crack propagation speed for different
directions and different strains was determined.
X

Kurzfassung
Die Untersuchung der Auswirkung von hohen Scherverformunge auf Schienenstähle ist
von großer technischer Bedeutung. Auf der Oberfläche von Schienen entstehen Verformungsschichten,
die das weitere Verhalten der Schiene wesentlich beeinflussen. Nur wenn grundlegende
Eigenschaften des verformten Gefüge bekannt sind, können Werkzeuge (Simulationen,
Modelle) entwickelt werden, um etwa die Serviceintervalle für die Schienen zu optimieren
oder gezielt Werkst<strong>of</strong>fdesign betreiben zu können.
Das Aufbringen hoher Scherverformungen ist vor allem durch die in den letzten etwa fünfzehn
Jahren entwickelten Methoden der Hochverformung unter hohem hydrostatischen Druck möglich.
Ziel dieser Arbeit ist es, mit Hilfe der Methoden der Hochverformung Modellwerkst<strong>of</strong>fe und
die technisch genutzten Schienestähle bis zu genau definierten Scherungen zu verformen. An
diesen verformten Proben soll die Mikrostrukturentwicklung und die damit einhergehenden
Eigenschaftsänderungen untersucht werden.
Dabei zeigt sich, dass wesentliche Unterschiede in der Strukturentwicklung zwischen Reinmetallen
und den mehrphasigen Stählen bestehen. Bei allen Reinmetallen ist ein ähnliches
Verhalten zu beobachten: Zu Beginn der Verformung durch Torsion unter hohem Druck erfolgt
eine relativ rasche Fragmentierung der großen Körner, die bei Erreichen einer werkst<strong>of</strong>f- und
kristallstrukturabhängigen Scherung scheinbar abgeschlossen ist. Verformt man die Proben zu
noch höheren Scherungen, erfolgt keine weitere Kornfeinung. Bei Betrachtung der Struktur
in Torsionsrichtung erscheint diese gleichachsig, in radialer Richtung aber sind die Elemente
länglich, mit einem bestimmten Winkel zwischen der Längsachse und der Scherebene. Die
Mikrostruktur, die durch eine zyklische Form der Hochdrucktorsion eingestellt wird, ist sehr
ähnlich. Es zeigt sich aber, dass die Größe der Strukturelemente, die nach Erreichen der
Sättigung vorliegt, sehr stark von der Verformung pro Zyklus abhängt. Je kleiner diese Verformung
pro Verformungsschritt ist, desto größer ist die entstehende Mikrostruktur. Auch
die Ent-wicklung der Missorientierung sowie der Charakter der Grenzen zwischen Strukturelementen
wird vom Verformungsmodus stark beeinflusst. Durch diese Fragmentierung
und die Kornfeinung kommt es auch zu einem enormen Anstieg in der mechanischen Festigkeit,
sowohl bei monoton als auch bei zyklisch verformten Proben. Wieder ist der Anstieg
bei monoton verformten Proben am größten, die kleinsten Festigkeitssteigerungen werden bei
den kleinsten Verformungen pro Verformungsschritt gemessen.
Im Gegensatz zu den Reinmetallen ist die Strukturentwicklung bei den Stählen vor allem
durch das Verhalten der Karbide bestimmt. Grobe Zementitlamellen werden verformt und
aufgebrochen, die Bruchstücke richten sich parallel zur Scherebene aus. Sind die Lamellen
feiner, lassen sie sich leichter verformen, ohne zu zerbrechen. Eine ausgeprägte Ausrichtung
und Fragmentierung findet erst bei höheren Verformungen statt. Die sehr feinen Karbide
des bainitischen Stahls werden anfangs kaum verformt und richten sich gemäß des aktuellen
Scherwinkels aus. Durch die hoher Scherung kommt es auch zu einer Verringerung des Lamellenabstandes
auf wenige 10 nm sowie der Lamellendicke auf 1-2 nm. Nach Verformungen von
≈ 800% liegt in allen Stählen eine sehr feinlamellare Struktur fast exakt parallel zur Scherrichtung
vor. Hinweise aus Röntgenuntersuchungen, die auf eine Auflösung der Karbide schließen
lassen, konnten durch Messung der elektronischen Struktur des Eisens bei verschiedenen Ver-
XI

formungsgraden bestätigt werden. Ähnlich wie bei den Reinmetallen ist auch die durch zyklische
Verformung entstehende Mikrostruktur von Schienenstählen stark von der Verformung
pro Zyklus abhängig. Ist diese klein, so gibt es vorwiegend eine Fragmentierung der Karbide,
die sich anschließend ausrichten. Bei einer hohen Verformung pro Verformungsschritt ist die
Strukur sehr änlich der Struktur nach monotoner Verformung.
Die ausgeprägte Abnahme der Lamellenabstände und die Ausrichtung der Lamellen führt
natürlich auch zu einer Änderung der mechanischen Eigenschaften. Aufgrund der Anisotropie
der Mikrostrukur gibt es auch eine starke Anisotropie in den mechanischen Eigenschaften.
In Richtung dieser ausgerichteten Lamellenstruktur kommt es zu einer starken Festigkeitszunahme
(Zugfestigkeiten von über 3 GPa). Die Festigkeit normal zu dieser Vorzugsrichtung
steigt zwar auch an, der Anstieg ist aber geringer als der Anstieg der Festigkeit in Richtung
der ausgerichteten Lamellenstruktur. Die mechanische Festigkeit, die Bruchzähigkeit und
das Ermüdungsrißwachstum wurden für unterschiedliche Richtungen und für unterschiedliche
Verformungsgrade bestimmt.
XII

Sobald jemand in einer Sache Meister geworden ist,
sollte er in einer neuen Sache Schüler werden
Gerhard Hauptmann
XIII

XIV

Contents
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Methods <strong>of</strong> Severe Plastic <strong>Deformation</strong> 5
2.1 High Pressure Torsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Cyclic High Pressure Torsion . . . . . . . . . . . . . . . . . . . . . 7
2.2 Equal Channel Angular Pressing . . . . . . . . . . . . . . . . . . . . . . . . 7
3 Results and Discussion 11
3.1 Structural Evolution <strong>of</strong> Model Materials during Severe Plastic <strong>Deformation</strong> . 11
3.1.1 High Pressure Torsion . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1.2 Cyclic High Pressure Torsion . . . . . . . . . . . . . . . . . . . . . 13
3.2 Structural Evolution <strong>of</strong> Rail Steels during Severe Plastic <strong>Deformation</strong> . . . . 16
3.2.1 High Pressure Torsion . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.2 Cyclic High Pressure Torison . . . . . . . . . . . . . . . . . . . . . 21
3.2.3 Microstructure after additional heat treatment . . . . . . . . . . . . . 23
3.3 Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3.1 In-situ Measurement <strong>of</strong> the Torque . . . . . . . . . . . . . . . . . . . 26
3.3.2 Tensile Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3.3 Microhardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3.4 Fracture Toughness . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3.5 Fatigue Crack Progagation . . . . . . . . . . . . . . . . . . . . . . . 32
4 Conclusions 35
5 List <strong>of</strong> appended papers 41
A Structural Refinement <strong>of</strong> Low Alloyed Steels during Severe Plastic <strong>Deformation</strong>
A–1
A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A–3
A.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A–3
A.2.1 Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A–3
A.2.2 Microstructural investigations . . . . . . . . . . . . . . . . . . . . . A–3
A.2.3 Microhardness Measurements . . . . . . . . . . . . . . . . . . . . . A–4
A.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A–4
XV

Contents
A.3.1 Microstructural evolution . . . . . . . . . . . . . . . . . . . . . . . . A–4
A.3.2 Microhardness and microstructural features . . . . . . . . . . . . . . A–6
A.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A–6
A.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A–8
B Structural Changes <strong>of</strong> Severely Plastic Deformed Rail Steel B–1
B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B–3
B.2 Experimental Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . B–3
B.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B–4
B.3.1 X-Ray investigations . . . . . . . . . . . . . . . . . . . . . . . . . . B–4
B.3.2 TEM-investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . B–6
B.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B–9
C Strain Hardening during High Pressure Torsion <strong>Deformation</strong> C–1
C.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–3
C.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–3
C.2.1 Test equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–3
C.2.2 Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–3
C.2.3 Evaluation <strong>of</strong> the torque curves . . . . . . . . . . . . . . . . . . . . C–4
C.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–5
C.3.1 Torque versus number <strong>of</strong> turns . . . . . . . . . . . . . . . . . . . . . C–5
C.3.2 Tensile tests and microhardness measurements . . . . . . . . . . . . C–6
C.3.3 Microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–6
C.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–6
C.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C–8
D Formation <strong>of</strong> surface layers: the effect <strong>of</strong> the strain path D–1
D.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D–3
D.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D–3
D.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D–5
D.3.1 Monotonic <strong>Deformation</strong> . . . . . . . . . . . . . . . . . . . . . . . . D–5
D.3.2 Cyclic deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . D–5
D.3.3 <strong>Deformation</strong> <strong>of</strong> a surface due to friction . . . . . . . . . . . . . . . . D–6
D.3.4 Microhardness measurements . . . . . . . . . . . . . . . . . . . . . D–7
D.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D–8
D.4.1 Microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D–8
D.4.2 Microhardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D–11
D.4.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D–11
E High Pressure Torsion <strong>of</strong> Rail Steels E–1
E.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E–3
E.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E–3
E.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E–4
E.3.1 Microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E–4
XVI

Contents
E.3.2 Mechanical strengths . . . . . . . . . . . . . . . . . . . . . . . . . . E–7
E.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E–7
E.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E–8
F TEM Investigation <strong>of</strong> the Structural Evolution in a Pearlitic Steel Deformed
by High Pressure Torsion F–1
F.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F–3
F.2 Experimental Details and Material . . . . . . . . . . . . . . . . . . . . . . . F–3
F.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F–4
F.3.1 Microstructure and Elemental Maps . . . . . . . . . . . . . . . . . . F–4
F.3.2 ELNES Measurements . . . . . . . . . . . . . . . . . . . . . . . . . F–5
F.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F–5
F.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F–10
F.6 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F–11
F.6.1 Fitting <strong>of</strong> the EELS-spectra . . . . . . . . . . . . . . . . . . . . . . F–11
F.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F–12
G Fracture Processes in Severe Plastic Deformed Rail Steels G–1
G.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–1
G.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–3
G.3 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–3
G.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–4
G.4.1 Microstructure and crack path . . . . . . . . . . . . . . . . . . . . . G–4
G.4.2 Fracture toughness . . . . . . . . . . . . . . . . . . . . . . . . . . . G–5
G.4.3 Fracture surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–5
G.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–7
G.5.1 Microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–7
G.5.2 Fracture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–9
G.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G–10
H Cyclic High Pressure Torsion <strong>of</strong>
Nickel and Armco Iron H–1
H.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–3
H.2 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–3
H.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–4
H.3.1 Flow stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–4
H.3.2 Microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–5
H.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–12
H.4.1 Severe Plastic <strong>Deformation</strong> . . . . . . . . . . . . . . . . . . . . . . . H–12
H.4.2 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–14
H.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . H–15
XVII

Contents
I Structural Evolution during Cyclic Severe Plastic <strong>Deformation</strong> I–1
I.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I–3
I.2 Experimental Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I–3
I.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I–3
I.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I–7
I.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I–9
XVIII

1.1 Motivation
1
Introduction
Today, the demands on rails are higher than ever: Increasing axle loads, increasing train speeds
as well as increasing traffic add up to an enormous exposure <strong>of</strong> the material. This inevitable
leads to the formation <strong>of</strong> a severely deformed surface layer. 1–5 In this layer, cracks are nucleated
and may grow, 6–9 an extreme result from this shear deformation is the formation <strong>of</strong>
a so-called white etching layer (WEL), 7, 10, 11 a feature <strong>of</strong> the microstructure that is <strong>of</strong>ten observed
where there is slip between surfaces. 12–19 WEL are very hard and brittle and are <strong>of</strong>ten
Figure 1.1: (a) optical micrograph <strong>of</strong> a WEL on a rail (350 LHT) after service (b) SEM micrograph <strong>of</strong> an area that
is white in an optical micrograph after etching.
1
1

1
1 Introduction
the origin <strong>of</strong> cracks, too. Figure 1.1 depicts two micrographs <strong>of</strong> the gauge corner <strong>of</strong> a rail after
service. The deformation layer with cracks inside is well pronounced and also a WEL can be
seen. Generally it is assumed that if the rate <strong>of</strong> crack growth is larger than the wear rate, these
cracks can grow to a critical length and result in spalling damage or transverse rail fracture as
can be seen in Figure 1.2 ∗ . This becomes especially important for the modern rail steels, that
are much more wear restistant, but also have a somewhat lower fracture toughness. Therefore,
Figure 1.2: Damages <strong>of</strong> rail tracks. (a) Head check cracking on the gauge corner (b) Transverse rail cracking from
a head check crack. (c) Spalling damage on the gauge corner in a curve track.
it is <strong>of</strong> utmost importance to be able to predict the occurrence and growth <strong>of</strong> cracks for different
rail steels and different loading conditions. One possibility to do this are field tests20 or tests
on a testing rig for rails. These tests are very time consuming and cost-intensive. Hence, great
affords are made to simulate the rail-wheel contacts in order to be able to study the influence
<strong>of</strong> different geometries or to develop tools to predict the occurrence <strong>of</strong> critical cracks.
8, 21–23
At the moment such calculations suffer especially from two shortcomings: Firstly, there exists
almost no fundamental knowledge <strong>of</strong> the microstructural changes <strong>of</strong> the material due to large
shear deformation. Secondly, virtually nothing is known about the mechanical properties <strong>of</strong>
this deformed material as a function <strong>of</strong> the shear deformation, especially in terms <strong>of</strong> fracture
toughness and anisotropy.
2
∗ taken from a presentation held by Jonas Rinsberg at the CHARMEC Meeting, 14.-18.01.2004, Leoben

1.2 Objectives
The aim <strong>of</strong> this work is to understand the processes occurring in pearlitic and bainitic steels
used as rail steels during severe plastic shear deformation and to relate these to the resulting
changes in the mechanical properties. Together with steels, the behavior <strong>of</strong> pure metals shall
be investigated in order to gain fundamental insight in the processes during severe plastic
deformation. The investigated materials comprise the coarse-pearlitic steel 900A, the finepearlitic
steel 350 LHT (or HSH-S), the bainitic steel Dobain 430, pure iron, pure nickel and
pure copper. Figure 1.3 depicts the initial microstructure <strong>of</strong> the investigated rail steels. The
composition <strong>of</strong> the investigated steels is given in Table 1.1.
C Si Mn Cr Pmax Smax
900A 0,76 0,35 1,0 0,014 0,017 0,04
350 LHT 0,78 0,46 1,18 0,23 0,012 0,015
Dobain 430 0,70-0,82 0,4-1,0 0,7-1,1 0,4-0,7 0,02 0,02
Table 1.1: Composition <strong>of</strong> the used materials
In the last years, in the scientific field <strong>of</strong> severe plastic deformation many methods were
developed to severely plastically deform samples to strains not reachable with conventional
processes, see for example the proceedings <strong>of</strong> the many conferences held in this area. 24–26
In the present work, some <strong>of</strong> this methods are applied to the investigated materials to produce
specimens with a defined deformation for further investigations. With samples produced
by High Pressure Torsion (HPT), the microstructural evolution and the mechanical strength
as a function <strong>of</strong> the monotonic shear strain are determined. In order to evaluate the influence
<strong>of</strong> cyclic severe plastic deformation a new method <strong>of</strong> SPD, Cyclic High Pressure Torsion
(CHPT), was developed and applied to different materials. Investigation methods for characterizing
the microstructure comprised scanning electron microscopy (SEM) using secondary
(SE) electrons or backscattered electrons (BSE) for depicting or obtaining orientation image
maps (OIM), transmission electron microscopy (TEM), analytical transmission electron microscopy,
ion microscopy and different X-ray diffraction techniques. The mechanical strength
was determined by means <strong>of</strong> microhardness, in-situ measurement <strong>of</strong> the torque and subsize
tensile tests. To determine the fracture toughness and the crack growth properties, samples <strong>of</strong>
the rail steel 900A were deformed by Equal Channel Angular Pressing (ECAP) using Route A.
With these samples, also the resulting anisotropy (in terms <strong>of</strong> fracture toughness, crack growth
and mechanical strength) was studied. To investigate the formation <strong>of</strong> WEL, combinations
<strong>of</strong> severe plastic deformation and different heat treatments were performed for the rail steel
900A.
3
1

1
1 Introduction
Figure 1.3: Micrographs <strong>of</strong> the initial microstructures <strong>of</strong> investigated materials (a) 900A, SEM, (b) 350 LHT,
SEM, (c) Dobain 430, SEM and (d) Armco Iron by means <strong>of</strong> ion microscopy.
4

2
Methods <strong>of</strong> Severe Plastic <strong>Deformation</strong>
In the following sections, the SPD-methods used in this work will be described shortly. Of
course, beside these methods, a large number <strong>of</strong> further techniques exist, for instance accumulative
roll bonding, 27–29 cyclic extrusion compression, 30, 31 continuous equal channel angular
pressing, 32 torsion extrusion33 or cyclic channel die compression. 34, 35
2.1 High Pressure Torsion
To quickly produce samples with a known, very large strain, High Pressure Torsion (HPT) is
the easiest method. Numerous papers prove the capability <strong>of</strong> HPT to achieve ultra-fine grained
material. 36–40 For this method, a coin-shaped samples is pressed between two anvils under a
high hydrostatic pressure (The maximum pressure is ≈ 7.5GP a for our tool). During the
build-up <strong>of</strong> the pressure, the sample is pressed into the cavities in the anvil. In this process, a
burr is formed at the edge <strong>of</strong> the sample. Then one anvil is rotated with respect to the other
anvil, the rotation speed can be varied over a large range. This leads to a deformation <strong>of</strong> the
sample by almost simple shear. The burr prevents a contact between the two anvils and upholds
the hydrostatic pressure. Due to the high pressure, in most metals the formation <strong>of</strong> cracks is
suppressed, therefore it is possible to apply very high strains without failure <strong>of</strong> the material.
Another function <strong>of</strong> the high pressure is to supply enough friction to prevent the occurrence
<strong>of</strong> slip. To support this mechanism, both the anvils and the sample are sandblasted before the
experiment. To be sure whether the experiment is useful or not, in most cases two checks are
made: Firstly, a line is drawn on both sides <strong>of</strong> the sample. After the experiment, these lines
are still visible and are rotated with respect to the other according the number <strong>of</strong> rotations <strong>of</strong>
the anvils if no slip has occurred. Secondly, the torque during the experiment is measured
by means <strong>of</strong> strain gauges. In a valid experiment, the torque is a monotonic function <strong>of</strong> the
strain/time. If the torque suddenly decreases markedly, this indicates that slip or a formation
<strong>of</strong> fatal cracks inside the sample is present. The reached shear strain γ is a function <strong>of</strong> the twist
5
2

2
2 Methods <strong>of</strong> Severe Plastic <strong>Deformation</strong>
angle φ, the radius r (<strong>of</strong> the site <strong>of</strong> investigation, not the sample radius) and the thickness t. γ
can be calculated according to Equation 2.1.
γ = φr
t
(2.1)
This strain can be expressed in terms <strong>of</strong> an equivalent von Mises strain by dividing the shear
strain by √ 3 as it was shown by Stüwe. 41 The equivalent von Mises strain ɛeq as a function <strong>of</strong>
the number <strong>of</strong> turns n is then given by Equation 2.2.
ɛeq = 2πrn
t √ 3
(2.2)
A photograph <strong>of</strong> the used tool can be seen in Figure 2.1, for a sketch <strong>of</strong> the principle <strong>of</strong> HPT
seen, e.g. Paper C - Paper E.
Figure 2.1: Photograph <strong>of</strong> the used HPT facility.
A detailed analysis <strong>of</strong> the homogeneity <strong>of</strong> the deformation and the accuracy <strong>of</strong> this tool is
given by Vorhauer and Pippan. 42 In the existing HPT facility it is also possible to vary the temperature
between -196°C (liquid nitrogen) and ≈ 450°to study the influence <strong>of</strong> the processing
temperature. The processing parameters used for this work are given in the corresponding papers.
Since the deformation by HPT is similar to a pure torsional deformation, it is important to
be aware that the microstructure may be different in different directions. To avoid ambiguity,
the investigated direction should always be indicated. A simple way is to define the directions
in respect <strong>of</strong> the sample according to Figure 2.2.
6

2.2 Equal Channel Angular Pressing
Figure 2.2: Definition <strong>of</strong> the sample directions for microstructural investigations
2.1.1 Cyclic High Pressure Torsion
In order to study the effect <strong>of</strong> large cyclic plastic strains, the existing HPT facility has been
enhanced to allow for a cyclic mode. Of course, this can be combined with all the other
possibilities <strong>of</strong> the equipment, e.g. torque measurement, variation <strong>of</strong> temperature and variation
<strong>of</strong> the strain rate. The strain per cycle ∆ɛ (calculated according to Equation 2.3, equivalent to
Equation 2.2) can be varied almost without limits. For the current work, the strain increment
∆ɛ was varied between 0.25 and 4. The total equivalent strain can be calculated according to
Equation 2.4, where N is the number <strong>of</strong> cycles.
∆ɛeq = 2πrn
t √ 3
(2.3)
ɛeq,total = N∆ɛ (2.4)
This method is a useful tool to easily study the influence <strong>of</strong> the strain path and the strain
per cycle on the material. It could be shown that cyclic deformation considerably influences
the deformation <strong>of</strong> carbides in pearlitic steels. The results from these investigations where
published in Paper D and E. With Paper H, for the first time a detailed study <strong>of</strong> the development
<strong>of</strong> the microstructure and the mechanical properties <strong>of</strong> pure metals deformed by CHPT was
done and the results where compared to results from HPT. In Paper I, the behaviour <strong>of</strong> CHPT
deformed material is compared to conventional fatigue. It can be shown that many features that
are well known in fatigue have similarities with the observed properties in CHPT deformed
materials.
2.2 Equal Channel Angular Pressing
At the moment, the most frequently used method in the severe plastic deformation community
is Equal Channel Angular Pressing (ECAP). 43–46 For the present work, ECAP was used to obtain
deformed samples large enough to machine compact tension (CT) specimens to measure
7
2

2
2 Methods <strong>of</strong> Severe Plastic <strong>Deformation</strong>
fracture toughness and fatigue crack growth. In ECAP, a sample with a round or square cross
section is pressed through two intersecting channels that are tilted by a intersection angle Φ by
means <strong>of</strong> a plunger. The intersection angle lies in most cases between 90°and 150°, in some
tools the corner is rounded introducing a gap angle Ψ. For such a tool, the strain per pass can
be calculated 47 according to Equation 3:
ɛ = 1
Φ
√ 2 cot
3 2
Ψ
Φ
+ + Ψcosec
2
2
Ψ
+
2
(2.5)
If there is no radius at the gap (ψ = 0), Equation 2.5 simplifies to Equation 2.6, the strain
per pass is therefore only a function <strong>of</strong> the intersection angle Φ.
ɛ = 2
√ 3 cot Φ
2
(2.6)
Figure 2.3: Photograph <strong>of</strong> the used ECAP tool. Inside the channel, a halfway deformed sample <strong>of</strong> 900A can be
seen.
A photograph <strong>of</strong> our tool can be seen in Figure 2.3, for a sketch <strong>of</strong> the principle <strong>of</strong> ECAP
see, e.g. Paper G. In contradiction to the continuous HPT, ECAP is a stepwise process, that
allows to apply different deformation paths. There exist four possible routes (at least, when
8

2.2 Equal Channel Angular Pressing
only rotations <strong>of</strong> 90°are considered), depending on how the sample is inserted in the tool. In
Route A, the sample is inserted exactly in the same way, there is no rotation <strong>of</strong> the sample.
Although the shear in each pass occurs on different shear planes, the overall effect is that <strong>of</strong> a
monotonic shear in the longitudinal direction <strong>of</strong> the sample. In Route C, the sample is rotated
by 180°after each pass. This leads to a shear on the same shear plane in each pass, but with
alternating shear directions. This can be considered as a cyclic shear deformation. In Route
B, the rotation is only 90°, and therefore there exist two subgroups, Route BA and Route BC,
depending on whether in the next pass the rotation is done in the same direction or not. 48
Our tool has a intersection angle <strong>of</strong> 120°and a cross section <strong>of</strong> 10 x 10 mm. The pressing
speed is ≈ 1mm/sec, therefore almost no increase <strong>of</strong> the temperature during the deformation
occurs. In order to minimize the friction and to prevent fusing <strong>of</strong> the tool and the sample, the
tool was coated with TiN. With this configuration it is possible to even deform pearlitic steels
at room temperature.
9
2

2
10

3
Results and Discussion
In the following chapter, the results <strong>of</strong> the various investigations shall only be summarized. A
more detailed discussion <strong>of</strong> most <strong>of</strong> the results can be found in the appended papers.
3.1 Structural Evolution <strong>of</strong> Model Materials during Severe
Plastic <strong>Deformation</strong>
3.1.1 High Pressure Torsion
When pure metals are deformed by methods <strong>of</strong> severe plastic deformation, a saturation in the
decrease <strong>of</strong> the structure size can be seen when the strain reachs a certain value. The reason
for this saturation in structure refinement is not clear at the moment, explanations given for
this in the literature are a kind <strong>of</strong> dynamic recrystallisation 36 or changes in the deformation
mechanisms 49 due to smaller grain sizes.
Figure 3.1 demonstrates the structural refinement in pure iron. The micrographs are taken
in axial direction. It can be seen that the higher the strain ɛeq was, the smaller and the more
equiaxed the structure becomes. After strain between 16 and 32 (for Armco iron) no further
refinement <strong>of</strong> the structure can be observed. The microstructure consists <strong>of</strong> equiaxed structural
elements with a size <strong>of</strong> ≈ 200 − 300nm with blurred boundaries. It has to be noted that
the structure <strong>of</strong> pure metals after HPT is not a pure grain structure (at least, when the HPT
experiment is carried out at room temperature and the homologue temperature <strong>of</strong> the material
is low), although a large fraction <strong>of</strong> the boundaries between the structural elements are high
angle boundaries.
The microstructure <strong>of</strong> nickel deformed by HPT in radial direction is depicted in Figure 3.2.
In this direction, no equiaxed elements are formed. Instead, elongated elements can be seen.
At quite small strains, the angle between the length axis <strong>of</strong> these elements and the normal to
the shear plane is in good agreement to the shear angle. When the strain gets higher, this angle
11
3

3
3 Results and Discussion
Figure 3.1: SEM micrographs (BSE) <strong>of</strong> HPT-deformed Armco iron in axial direction (a) ɛeq = 2, (b) ɛeq = 4, (c)
ɛeq = 8 and (d) ɛeq = 32.
should almost be 90°. But this is not observed, giving rise to the assumption that new elements
are continiously created by a process similar to dynamic recrystallisation 50
In Figure 3.3 the structure size <strong>of</strong> nickel as well as the missorientation distribution as a
function <strong>of</strong> the strain is presented. This structure size was calculated from OIM maps using
5°missorientation as a criterium to define a grain. In the OIM images, the process <strong>of</strong> fragmentation
and structure refinement can be seen. At the beginning <strong>of</strong> the deformation, the grains
are quickly subdivided into subgrains, after strains ≥ 4, the original grains are no longer recognizable.
It can be seen that the higher the strain gets, the more similar the missorientation
distribution becomes to the random Mackenzie distribution. The results found in this study for
nickel and Armco iron are comparable to published results for other pure metals 36
12

3.1 Structural Evolution <strong>of</strong> Model Materials during Severe Plastic <strong>Deformation</strong>
Figure 3.2: Orientation image maps <strong>of</strong> HPT-deformed nickel in radial direction (a) ɛeq = 0.5, (b) ɛeq = 1, (c)
ɛeq = 2 and (d) ɛeq = 4.
3.1.2 Cyclic High Pressure Torsion
The development <strong>of</strong> the microstructure in CHPT is very similar to HPT. But as can be seen
in Figure 3.4 for nickel depicted in tangential direction, the minimum structure size strongly
depends on the strain per cycle ∆ɛ. The smaller this strain increment is, the larger is the
structure size after saturation. The onset <strong>of</strong> saturation is also a function <strong>of</strong> the applied ∆ɛ, the
13
3

3
3 Results and Discussion
Figure 3.3: (a) The structure size <strong>of</strong> nickel after deformation by high pressure torsion (b) missorientation angle
distribution <strong>of</strong> nickel after deformation by high pressure torsion to different strains.
lower this value is, the lower is the total equivalent strain necessary for this onset.
Experiments with a larger ∆ɛ than 4 were not performed because there is almost no difference
between the structure size after HPT and CHPT with a ∆ɛ = 4. In radial direction,
the same elongated elements as in HPT are visible. Here also after large ɛeq,total, the angle
between the length axis <strong>of</strong> the elongated elements and the normal <strong>of</strong> the shear plane is in good
agreement with the shear angle corresponding to the strain increment. It is interesting to notice
that many features that are well known for conventional fatigue are also present in these CHPT
deformed samples. A detailed description <strong>of</strong> the processes and a comparison <strong>of</strong> the results for
HPT with other areas are given in Papers H and I.
14

3.1 Structural Evolution <strong>of</strong> Model Materials during Severe Plastic <strong>Deformation</strong>
Figure 3.4: SEM micrographs (BSE) <strong>of</strong> CHPT-deformed nickel in tangential direction, ɛeq,total = 64, (a) ∆ɛ =
0.5, (b) ∆ɛ = 1, (c) ∆ɛ = 2 and (d) ∆ɛ = 4.
15
3

3
3 Results and Discussion
3.2 Structural Evolution <strong>of</strong> Rail Steels during Severe Plastic
<strong>Deformation</strong>
3.2.1 High Pressure Torsion
As it was shown for pure metals, the structure size in these materials is decreased very fast
until a saturation in the structure size is reached. The microstructure in saturation is then
controlled by an equilibrium between grain refinement and a kind <strong>of</strong> dynamic recrystallisation.
In contradiction to this, the dominating mechanism during HPT <strong>of</strong> pearlitic and bainitic steels
is the deformation and the alignment <strong>of</strong> the cementite lamellae. Figure 3.5 and Figure 3.6
depict the microstructure <strong>of</strong> the rail steel 900A after HPT as a function <strong>of</strong> the strain in radial
and tangential direction.
Figure 3.5: SEM micrographs <strong>of</strong> HPT-deformed 900A in radial direction (a) ɛeq = 2, (b) ɛeq = 4, (c) ɛeq = 8
and (d) ɛeq = 16.
It can be seen that after a deformation to strains <strong>of</strong> ɛeq = 1 the cementite lamellae are no
longer straight, they begin to bend and align parallel to the shear plane. When the strain gets
16

3.2 Structural Evolution <strong>of</strong> Rail Steels during Severe Plastic <strong>Deformation</strong>
Figure 3.6: SEM micrographs <strong>of</strong> HPT-deformed 900A in tangential direction (a)ɛeq = 2, (b) ɛeq = 4, (c)ɛeq = 8
and (d) ɛeq = 16.
higher, the lamellae break up and are severely deformed. After strains ≥ 4, the original pearlite
colonies are no longer recognisable and most <strong>of</strong> the lamellae are fragmented and aligned parallel
to the shear plane. Due to a quite homogeneous deformation, the lamellae distance as well
as the lamellae spacing decreases markedly. After strains ≥ 8, SEM micrographs show a very
fine lamellar structure consisting <strong>of</strong> fragments <strong>of</strong> the former cementite lamellae (with a maximum
(length) dimension <strong>of</strong> approximately 1 µm) and ferrite. The maximum dimension <strong>of</strong> the
ferrite structural elements in axial direction is limited by the lamellae (fragment) spacing, as
can be seen in Figure 3.7. The size <strong>of</strong> the elements normal to this direction is approximately 2
- 5 times larger. Almost the same behavior was observed for the second investigated pearlitic
rail steel, 350 LHT, after HPT deformation to large strains. The difference between these two
steels is mainly the lamellae spacing (≈ 300nm for 900A and ≈ 100nm for 350 LHT) and the
size <strong>of</strong> the pearlite colonies. The microstructure after HPT deformation <strong>of</strong> the rail steel 350
LHT is depicted in Figure 3.8. From comparison <strong>of</strong> the SEM-micrographs it can be seen that
the significantly thinner cementite lamellae and the smaller colony size <strong>of</strong> the steel 350 LHT
17
3

3
3 Results and Discussion
Figure 3.7: Ion microscopy micrographs <strong>of</strong> HPT-deformed 900A in tangential direction (a) ɛeq = 1, (b) ɛeq = 4.
are more easily deformed without fragmentation than the thick lamellae <strong>of</strong> the steel 900A.
Due to this differences, especially at quite low strains (≈ smaller than 4), the alignment and
the fragmentation <strong>of</strong> the lamellae is much less pronounced. Only after deformations larger
than 4, a markedly alignment can be observed. Figure 3.8d shows that at higher deformation
the fragmentation is now very pronounced and again, the fragments are mostly aligned parallel
to the shear direction.
In many areas, the structure after large shear strains is to fine to be clearly resolved by
SEM. Therefore, electron transparent areas where made by means <strong>of</strong> a Focused Ion Beam
Workstation (FIB), for details see Paper B and Paper F.
TEM-investigations <strong>of</strong> the deformed steels 900A and 350LHT revealed that the lamellae
spacing was markedly decreased, in some areas to 20 nm or even less, see Figure 3.9. Together
with this, the lamellae thickness was also decreased from ≈ 20nm to ≈ 2nm. Dark
field images and selected area diffraction (SAD) pattern confirmed that the size <strong>of</strong> the structural
elements <strong>of</strong> the ferritic phase is governed by the lamellae distance and also that most boundaries
are large angle boundaries, see Figure 3.10. In order to determine a possible dissolution
<strong>of</strong> cementite due to severe plastic deformation, a detailed study <strong>of</strong> the electronic structure <strong>of</strong>
the cementite lamellae in the steel 900A was performed. By means <strong>of</strong> electron energy-loss it
can be shown that after a deformation by HPT to strain <strong>of</strong> ɛeq = 8, the carbon rich areas in the
material do no longer have the electronic fingerprint <strong>of</strong> the cementite as measured by electron
energy-loss spectroscopy (EEL) in the undeformed microstructure. The results from various
measurements in the undeformed material as well as the measured EEL spectra are presented
in Figure 3.11a - c. The changes in the properties due to deformation can be seen in Figure
3.11d, where the measurements in deformed samples are compared to the results from the
initial microstructure. It can be seen that the area ratio between the Fe-L2 and the Fe-L3 peak
ratio is significantly different for the ferritic matrix and cementite in the undeformed sample.
Also in the sample deformed to ɛeq = 2, this difference is observable. In the highest deformed
ratio were near the value for cementite,
sample for this study (ɛeq = 2), no values <strong>of</strong> the Fe- L3
L2
18

3.2 Structural Evolution <strong>of</strong> Rail Steels during Severe Plastic <strong>Deformation</strong>
Figure 3.8: SEM micrographs <strong>of</strong> HPT-deformed 350 LHT in tangential direction (a) ɛeq = 1, (b) ɛeq = 2, (c)
ɛeq = 4 and (d) ɛeq = 8.
for details see Paper F.
The deformation <strong>of</strong> the bainitic rail steel Dobain 430 is mainly controlled by the fine carbides
present in the initial microstructure. Due to their small size, they are not so severely
deformed as the cementite lamellae in the pearlitic steels, they just align according to the
shear angle when viewed in radial direction. This can be seen in Figure 3.12 where SEM micrographs
<strong>of</strong> Dobain 430 for different strains are presented. After a deformation to a strain <strong>of</strong>
ɛeq = 2, almost no fragmentation and only little deformation <strong>of</strong> the carbides can be observed.
A markedly fragmentation and indications <strong>of</strong> a dissolution <strong>of</strong> the carbon from the carbides can
be seen after strains larger than ɛeq = 8. The microstructure viewed in tangential direction
after relatively small strains is less affected than in radial direction. After a strain <strong>of</strong> ɛeq = 8,
there is no obvious difference in the microstructure when viewed in these two directions.
The present results are comparable with results from the literature for wire drawing and
HPT. The process <strong>of</strong> wire drawing is extensively investigated. Embury and Fisher 51 and Lang-
19
3

3
3 Results and Discussion
Figure 3.9: TEM bright field micrographs <strong>of</strong> (a) initial microstructure <strong>of</strong> 900A, (b) HPT deformed 900A, ɛeq = 8
and (c) HPT deformed HSH, ɛeq = 8.
Figure 3.10: TEM micrographs <strong>of</strong> (a)HPT deformed HSH, ɛeq = 2, bright field (b) corresponding dark field
image and (c)HPT deformed HSH, ɛeq = 8, dark field.
ford 52, 53 investigated the microstructural evolution in wire drawing <strong>of</strong> pearlitic steels and the
resulting changes in strength. The results presented in these papers are very similar to the
present findings. Also in wire drawing an intensive alignment <strong>of</strong> the lamellae in the drawing
direction is observed. The main mechanism for the increase in strength was found to be the
decreasing lamellae distance, this was confirmed by many studies, see for instance. 54, 55 Later
investigations 56–60 revealed that also in drawn pearlitic wires dissolution <strong>of</strong> cementite can be
observed. Two reasons for this dissolution are discussed in the literature: Firstly, it is assumed
that the binding energy <strong>of</strong> carbon atom to a dislocation is larger than the binding energy <strong>of</strong>
the carbon in the carbide. Therefore, carbon atoms are dragged out <strong>of</strong> the carbide by crossing
dislocations. Secondly, it is assumed that the very fine carbides resulting from this deformation
become thermodynamical unstable due to the Gibbs Thomson effect. A comprehensive
discussion about the different mechanisms is given by Gavriljuk. 61 There is even less agree-
20

3.2 Structural Evolution <strong>of</strong> Rail Steels during Severe Plastic <strong>Deformation</strong>
Figure 3.11: (a) Measured EEL spectra <strong>of</strong> ferrite and carbide in the initial microstructure (b) Fe-L2/L3 area and
peak high ratio for various measurements in the initial microstructure (c) FWHM for various measurements in
the initial microstructure and (d) Comparison <strong>of</strong> the Fe-L3/Fe-L2 ratio <strong>of</strong> undeformed and deformed samples.
The black lines mark the average values in the initial microstructure, the gray lines mark the highest and lowest
observed values in the initial microstructure.
ment about the question, where the carbon is situated after the dissolution. Ivanisenko et al. 62
and Sauvage 63 reported that a dissolution <strong>of</strong> the cementite lamellae can occure after HPT. The
microstructures, the mechanical properties and the observed dissolution <strong>of</strong> carbide seem to be
in good agreement with the present finding. Nevertheless it has to be noted that the strains
needed to reach this results are much higher in their investigations.
3.2.2 Cyclic High Pressure Torison
The microstructure resulting from Cyclic High Pressure Torsion is depicted in Figure 3.13.
Similar as in pure metals, also the microstructure <strong>of</strong> the rail steels is strongly influenced by the
strain per cycle. After CHPT with a small ∆ɛ the main feature are severely fragmented carbides
that are aligned parallel to the shear plane. There are no indications that something like
a dissolution <strong>of</strong> the carbon from the carbides has occured. After CHPT with a large ∆ɛ the microstructure
is quite similar like in a monotonic deformed sample, but the strains needed for a
21
3

3
3 Results and Discussion
Figure 3.12: SEM micrographs <strong>of</strong> HPT-deformed Dobain 430 (a) ɛeq = 1, (b) ɛeq = 2, (c) ɛeq = 4 and (d)
ɛeq = 8, all in radial direction, (e) ɛeq = 1, tangential direction and (f) ɛeq = 8, tangential direction.
similar microstructure (and a similar mechanical strength) are much higher. Already after two
cycles with ∆ɛ = 4 (ɛeq,total = 8) the lamellae are extremely deformed and fanfold structures
22

3.2 Structural Evolution <strong>of</strong> Rail Steels during Severe Plastic <strong>Deformation</strong>
are present. After higher total strains, the lamellae are no longer clearly recognisable.
Figure 3.13: SEM micrographs <strong>of</strong> CHPT deformed 900A (a) ∆ɛ = 0.25, ɛeq,total = 8 (b) ∆ɛ = 0.25, ɛeq,total =
32 (c) ∆ɛ = 4, ɛeq,total = 8 and (d) ∆ɛ = 4, ɛeq,total = 64.
3.2.3 Microstructure after additional heat treatment
Deformed samples <strong>of</strong> the three steels (900A with ɛeq = 16, 350 LHT with ɛeq = 8 and Dobain
430 with ɛeq = 8) were head treated a different temperatures and for different times. As
can be seen in Figure 3.14, only after temperatures larger than 600°, a markedly change in
the microstructure has occurred. In etched samples round carbides with a size <strong>of</strong> ≈ 10nm are
clearly visible. SEM micrographs using back scatterd electrons reveal that the ferritic matrix is
recrystallized, the grain size grows with increasing temperature and increasing heat treatment
time. In order to obtain ultra-short heat treatment times, samples <strong>of</strong> the rail steel 900A were
heated by by a laser pulse. The pulse time was 10 - 50 µs, the temperature (average value,
measured by a spectrometer) was varied between 600 and 750 degree. This heat treatment by
a laser pulse resulted in a heat affected zone part <strong>of</strong> which is white in optical microscopy after
23
3

3
3 Results and Discussion
etching with Nital. The microstructure <strong>of</strong> this zone in the SEM has various appearances, that
can also vary depending on the distance from the center <strong>of</strong> the zone. Figure 3.14 f shows a
very fine structure that does not look recrystallized, no signs <strong>of</strong> the carbon can be seen. In
other areas, a very fine, martensite-like structure is visible. Microhardness measurements in
the zones showed no significant increase or decrease <strong>of</strong> the microhardness compared to the
deformed microstructure. As can be seen in Figure 3.14 e, the microstructure next to the white
etching zone is significantly s<strong>of</strong>ter and it consists <strong>of</strong> recrystallized ferrite and carbides. It is
similar to the microstructure after the heat treatment for two seconds.
The structure <strong>of</strong> WELs were investigated in many studies, see, for instance.
2, 3, 7, 10, 64, 65
Unfortunately, the results are not unambiguous and sometimes quite conflictive. Martensite,
nanocrystalline ferrite or the occurrence <strong>of</strong> special carbides are reported. In a recent study,
Zhang et al. 66 thoroughly investigated a WEL and found most <strong>of</strong> the reported kinds <strong>of</strong> microstructure
in the same WEL in different areas. The Vickers hardness presented in their
paper is in the same range as the measured hardness for the HPT deformed steels, see Figure
3.19. They concluded that all the observed different microstructure lead to the same etching
behaviour. Due to a decrease in the difference in chemical composition, also the difference
in the chemical potential is decreased. Therefore, these structures are hardly affected by the
etching agent and appear white in optical microscope micrographs. Different structures where
also observed in the samples heat treated by a laser impulse for very short times. The sharp
boundary in many WELs indicate that a significant increase in temperature must have occurred,
this increase may be large enough for an austenitisation <strong>of</strong> the material. 67 The process
<strong>of</strong> austenitisation may even be enhanced by the deformation dissolution <strong>of</strong> the carbides and the
hydrostatic pressure. 68 The present results show that the deformation <strong>of</strong> steels by HPT leads
to similar hardness like in actual WELs, after additional heat treatment, many microstructural
features <strong>of</strong> WELs are also present in the HPT deformed sample. Therefore it can be concluded
that WELs are formed due to a simultane occurrence <strong>of</strong> high (accumulated) shear strains and
very short heat fluxes that are present due to slip between the rail and the wheel. It has to
be noted that an actual WEL is formed during millions <strong>of</strong> cycles and is therefore a dynamic
system. WELs from different rails may have different deformation and heat flux histories.
Hence, some microstructural features may be more pronounced than others. This can explain
the varying results from different studies. The s<strong>of</strong>ter area below the obviously white area in
the laser heat treated HPT samples is not observed in WELs because this region is constantly
deformed in a rail.
24

3.2 Structural Evolution <strong>of</strong> Rail Steels during Severe Plastic <strong>Deformation</strong>
Figure 3.14: Micrographs <strong>of</strong> HPT deformed samples, ɛeq = 16, after heat treatment (a) SE-micrograph, 2 seconds,
700°, (b) BSE-micrograph, 2 seconds, 700°, (c) BSE-micrograph, 60 seconds, 700°, (d) BSE-micrograph, 60
seconds, 600°, (e) optical micrograph showing the heat-affected zone after laser treatment (900A, ɛeq = 16,
710°for 10ms) and the influence on the hardness and (f) BSE-micrograph <strong>of</strong> the heat affected zone after laser
treatment.
25
3

3
3 Results and Discussion
3.3 Mechanical Properties
In order to characterize the changes in the mechanical properties due to the large shear deformation,
various methods were applied either after or during the deformation. In the following
section, the results <strong>of</strong> these measurements are shortly described.
3.3.1 In-situ Measurement <strong>of</strong> the Torque
The development <strong>of</strong> a reliable and accurate possibility to measure the torque during deformation
42, 69 allowed it to quickly determine changes in the mechanical strength without applying
other methods afterwards. The torque that is measured contains next to the torque necessary to
deform the sample MD solely a contribution from the region <strong>of</strong> the burr, MB. From the torque
MD, the shear stress can be calculated according to Equation 3.1:
Md = 2πτ(r)r 2 dr (3.1)
However, the contribution <strong>of</strong> the burr is not easily evaluated, the in-situ measurement <strong>of</strong> the
torque shall only be used to measure relative changes <strong>of</strong> the strength for different strains, different
materials or different regimes <strong>of</strong> deformation (HPT and CHPT). When the experimental
setup is kept constant (especially the hydrostatic pressure and the tool geometry), also MB will
be quite constant and the results <strong>of</strong> the measurement are very accurate and reproducible. Figure
3.15a shows typical in-situ measured torque curves <strong>of</strong> the investigated pure metals. At the
beginning <strong>of</strong> the deformation, a region <strong>of</strong> intense strain hardening is present until the increase
<strong>of</strong> the strength reachs zero and a kind <strong>of</strong> steady state deformation occures. 36 The onset <strong>of</strong> this
steady state deformation coincidences with the saturation in the decrease <strong>of</strong> the structure size
and can also be seen in the microhardness and tensile tests. The strain necessary to reach this
steady state deformation depends on the material and seems to be influenced by the crystalline
structure (fcc vs. bcc).
26
Figure 3.15: Comparison <strong>of</strong> the in-situ measured torque curves for (a) pure metals and (b) rail steels.

3.3 Mechanical Properties
In contradiction to pure metals, in the rail steels no steady state deformation could be
reached. Figure 3.15b depicts the measured torque curves for the rail steels. It can be seen
that the increase in the strength is highest for Dobain 430, and lowest for 900A. When the
measured torques for the steels are compared to the measured torque for the pure metals, the
difference in tensile strength can be estimated (almost 100 percent between Armco iron in
saturation and 900A after ɛeq = 16).
The torque measured in CHPT experiments shows the same characteristic as in monotonic
HPT experiments. For pure metals, the torque saturates after high strains, while in the steels
there is still an increase, although this is much smaller than in the monotonic HPT. The differences
between CHPT and HPT are the levels, at which the torque saturates and the necessary
values <strong>of</strong> strain to reach this steady state deformation (pure metal). The smaller the ∆ɛ was,
the lower is the torque and the earlyer (in terms <strong>of</strong> total equivalent strain) the saturation is
reached. A detailed investigation <strong>of</strong> these observations is given in Papers D, H and I. Figure
3.16 depicts what happens when HPT and CHPT experiments are combined. It can be seen
the the different strength levels (and therefore necessarily the microstructures) can at least be
partial converted into one another. This is especially surprising when a HPT experiment is
performed first, by applying <strong>of</strong> cyclic loading conditions, the torque decreases almost down to
the level that is reached when only a cyclic loading (with the same ∆ɛ) is applied. This process
is even repeatable for some times, although it can be seen that it is not totally reversible. This
area <strong>of</strong> SPD is at the moment completely unexplored. The first results <strong>of</strong> our experiments will
be published in a forthcoming paper 70
Figure 3.16: in-situ measured torque curves for combined HPT and CHPT experiments (a) Armco iron, ∆ɛ = 0.5,
one block → ɛeq,total = 32 and (b) nickel, ∆ɛ = 2, one block → ɛeq,total = 32.
3.3.2 Tensile Tests
To obtain absolute values <strong>of</strong> the mechanical strength, specimens for tensile tests were machined
out <strong>of</strong> HPT-samples. 42, 69 From each sample, two specimens are obtained. With this
method, the validity <strong>of</strong> the in-situ torque measurement as well as the influence <strong>of</strong> the hydrostatic
pressure was investigated, see Paper C. Figure 3.17 show the tensile strength for the
27
3

3
3 Results and Discussion
three rail steels as a function <strong>of</strong> the strain. As expected the development <strong>of</strong> the ultimate tensile
strength is similar to the in-situ measured torque (Figure 3.15b). Ultimate tensile strengths
<strong>of</strong> more than 1900 MPa were measured in radial direction in respect <strong>of</strong> the HPT-sample after
quite low HPT-deformation. The measurement <strong>of</strong> the tensile strength after higher strains did
not give useful results due a formation <strong>of</strong> cracks during the HPT-deformation. But a comparison
between the torque and the tensile tests allow to estimate the tensile strength to be more
than 3 GPa after a deformation <strong>of</strong> ɛeq ≥ 16 for 900A and ɛeq ≥ 8 for 350 LHT and Dobain
430. These values are in good agreement with the strength <strong>of</strong> drawn pearlitic wires in drawing
direction. 51, 52, 60 The strength <strong>of</strong> drawn pearlitic wires in a direction normal to the drawing
direction was never measured, because the diameter <strong>of</strong> wires drawn to such high strains is
markedly smaller than 1 mm.
Figure 3.17: Tensile strength in radial direction <strong>of</strong> rail steels as a function <strong>of</strong> the HPT deformation.
Specimens with the same geometry as in the case <strong>of</strong> HPT samples were machined out
<strong>of</strong> ECAP deformed samples <strong>of</strong> the steel 900A after one and three passes. To determine
anisotropy, two different orientations <strong>of</strong> the samples in respect to the ECAP sample were
measured. Figure 3.18 shows the definition <strong>of</strong> the sample directions and the result <strong>of</strong> these
tensile tests. As can be seen, after one pass the differences in the two directions are already
present. The tensile strength in samples with the orientation A (parallel to the direction <strong>of</strong>
the alignment) increases markedly with the number <strong>of</strong> passes and is almost identical with the
measured tensile strength after HPT when the samples are deformed to the same strain. In
the samples with orientation B the tensile strength is also increased, but the increase is much
smaller. In Figure 3.18d, typical stress-displacement curves for these samples are presented.
Significant differences in the deformation behaviour in samples with different orientations are
recognisable. While in samples with orientation B a distinctive elongation can be seen, this
can not be observed in samples with orientation A after three passes.
28

3.3 Mechanical Properties
Figure 3.18: (a) Definition <strong>of</strong> the loading directions and the sample orientations for tensile tests and fracture
mechanic tests (b) SEM micrograph <strong>of</strong> the cross section <strong>of</strong> 900 A after three ECAP passes (c) Results <strong>of</strong> the
tensile tests <strong>of</strong> ECAP deformed 900A as a function <strong>of</strong> the number <strong>of</strong> ECAP passes and (d) typical stress-crosshead
displacement-curves for the different samples.
3.3.3 Microhardness
Before the development <strong>of</strong> a reliable method to measure the torque during the HPT deformation,
microhardness measurements after the deformation where the quickest and most easily
available method the estimate the mechanical strength and are <strong>of</strong>ten applied in the literature.
39, 71–73 Nevertheless, the problem <strong>of</strong> the microhardness measurement is that it is not
very accurate and only relative values <strong>of</strong> the strength are measurable. In the present work,
this method was mainly used to determine the change <strong>of</strong> the mechanical strength after the heat
treatment <strong>of</strong> deformed samples. Figure 3.19a shows the development <strong>of</strong> the microhardness
29
3

3
3 Results and Discussion
as a function <strong>of</strong> the temperature. It can be seen that already after 400°a little decrease in the
strength is present. After a heat treatment at 600°, the reduction in the strength is almost 50 %
<strong>of</strong> the increase due to HPT. The microhardness as a function <strong>of</strong> the time <strong>of</strong> the head treatment
for the pearlitic rail steels can be seen in Figure 3.19b. After a heat treatment time <strong>of</strong> 60 seconds,
the microhardness has almost reached the value <strong>of</strong> the undeformed material or is even
less than this value although the microstructures are now quite different.
Figure 3.19: (a) The microhardness <strong>of</strong> 900A (ɛeq = 16) as a function <strong>of</strong> the heat treatment temperature (time =
constant = 2s) and (b) the microhardness <strong>of</strong> 900A (ɛeq = 16) and 350 LHT (ɛeq = 8) as a function <strong>of</strong> the heat
treatment time.
3.3.4 Fracture Toughness
Figure 3.20: Fracture toughness <strong>of</strong> ECAP deformed 900A as a function <strong>of</strong> the number <strong>of</strong> ECAP passes.
The fracture toughness in two different directions in respect to the ECAP sample (see Figure
3.18a) as a function <strong>of</strong> the shear strain applied by ECAP was determined. For doing this
30

3.3 Mechanical Properties
Figure 3.21: SEM micrographs <strong>of</strong> the fracture surface <strong>of</strong> (a) as-received and (b - d) ECAP pressed material
with orientation A (b) without side notch, (c) with side notch, (d) with side notch, larger magnification (crack
propagation direction from bottom to top).
slices with a thickness <strong>of</strong> 2 mm were cut form the ECAP sample to machine compact tension
specimens. Figure 3.20 summarizes the results <strong>of</strong> these measurements. In samples with an
orientation A, the crack has to cross the aligned lamellae. After strains larger than 2 (3 ECAP
passes), the crack in suchwise oriented samples is deflected by 90°and runs along the shear
direction <strong>of</strong> the ECAP sample. Therefore, the values measured in these samples are just lower
boundaries for the fracture toughness in this direction. In samples with the orientation B, the
crack can propagate parallel to the aligned lamellae, hence the fracture toughness decreases
with the increasing number <strong>of</strong> ECAP passes, after three passes <strong>of</strong> ECAP (ɛeq = 2), the fracture
toughness has almost decreased by fifty percent. In order to estimate also the fracture
toughness in samples with orientation A after three passes <strong>of</strong> ECAP, specimens with a side
notch were machined. In these samples the fracture toughness increases with increasing strain
31
3

3
3 Results and Discussion
compared to the initial microstructure. That is an unusual behaviour ∗ . In Figure 3.21 the
fracture surface <strong>of</strong> undeformed and deformed samples is presented. The fracture surface <strong>of</strong>
the undeformed samples is quite rough and is typical for a cleavage-like fracture. In samples
predeformed by three ECAP passes, the fracture surface is quite flat for both sample orientations.
If the crack is forced to run normal to the loading direction in samples with orientation
A due to a side notch, the fracture surface becomes very rough. The crack wants to follow
the aligned microstructure, hence branching <strong>of</strong> the crack can be observed sometimes. This
crack branching may be one reason for the increase in the fracture toughness. The intrinsic
anisotropy <strong>of</strong> the fracture toughness <strong>of</strong> a lamellar microstructure may be a second reason. A
detailed description <strong>of</strong> the processes occurring in the material during fracture is given in Paper
G.
Similar behavior <strong>of</strong> drawn pearlitic wires were observed by Toribio et al. 74–76 They found a
crack deflection in notched tensile specimens depending on the drawing strain.
Figure 3.22: Fatigue crack propagation <strong>of</strong> ECAP deformed 900A.
3.3.5 Fatigue Crack Progagation
With the same type <strong>of</strong> samples as used for the determination <strong>of</strong> the fracture toughness, the
fatigue crack propagation rate in different orientations was measured. The material after a different
number <strong>of</strong> ECAP passes was studied. The experiments were performed with an R-value
(R = Kmin ) <strong>of</strong> 0.5 to reduce the influence <strong>of</strong> internal stresses generated by the ECAP pro-
Kmax
cess. The actual crack length was measured using a potential drop technique. In contradiction
to the fracture toughness tests, the crack propagates in all samples according to the applied
loading. Figure 3.22 exhibits the results <strong>of</strong> the fatigue crack propagation measurements. It
∗ An increase in strength causes usually a decrease in fracture toughness
32

3.3 Mechanical Properties
can be seen that after one pass, the difference in the fatigue crack propagation rate is quite
small. However, even after one pass it is obviously that an influence <strong>of</strong> the sample or loading
direction is present. This influence is even more pronounced after three passes <strong>of</strong> ECAP. The
crack propagation rate is quite small when the crack has to cross the aligned lamellae which
are a severe obstacle, see Figure 3.18b. In samples where the crack can propagate mostly
within the ferrite, the crack propagation speed is markedly increased compared to the samples
with one ECAP pass. The fracture surface after the fatigue tests depicted in Figure 3.23. As
expected, the fatigue fracture surface is very smooth in both samples. Nevertheless, significant
differences resulting from the different path <strong>of</strong> the crack through the aligned microstructure
can be seen. The lamellae structure that can be seen in higher magnification is much finer and
more pronounced in samples with orientation A. In samples with orientation B, a wide-spaced
lamellae structure with large areas <strong>of</strong> a very flat surface is present. It seems that in samples
with orientation B the fatigue crack propagates partly by a quasi-cleavage mechanism, which
is responsible for the significant larger fatigue crack propagation rate.
Figure 3.23: SEM micrographs <strong>of</strong> the fracture surface <strong>of</strong> 900A, three passes <strong>of</strong> ECAP, after fatigue tests (a) and
(b) sample with orientation A, (c) and (d) sample with orientation B.
33
3

3
34

4
Conclusions
The methods <strong>of</strong> Severe Plastic <strong>Deformation</strong> were applied to rail steels and model material to
study the influence <strong>of</strong> large monotonic and cyclic shear deformation on the microstructure and
the mechanical properties <strong>of</strong> these materials.
High Pressure Torsion <strong>of</strong> pure metals leads to a grain refinement until an equilibrium between
the fragmentation <strong>of</strong> large structural elements and grain restoration processes leads to a
saturation <strong>of</strong> the refinement process. Parallel to the refinement <strong>of</strong> the structure, the mechanical
strength increases until it saturates, too. Ultimate Tensile Strengths <strong>of</strong> ≈ 1500 MPa for pure
iron and ≈ 450 - 500 MPa for pure copper were measured after High Pressure Torsion. From
comparison <strong>of</strong> the in-situ measured torque <strong>of</strong> nickels with iron and copper, the ultimate tensile
strength <strong>of</strong> nickel can be estimated to be ≈ 1300 MPa.
A similar grain fragmentation process is present during Cyclic High Pressure Torsion. The
structure size is determined by the strain increment per cycle. The larger the strain per cycle
is, the smaller becomes the structure size and the higher becomes the in-situ measured torque
and therefore the tensile strength. The total strain necessary to reach the steady-state regime is
larger for monoton High Pressure Torsion and decreases with a decreasing strain per cycle.
<strong>Large</strong> shear deformation in pearlitic and bainitic steels leads to an alignment <strong>of</strong> the carbides.
Depending on the size and the thickness <strong>of</strong> the carbides, they are severely deformed,
fragmented and even dissolution <strong>of</strong> the cementite during this process takes place. After an
equivalent shear strain <strong>of</strong> ≈8, in all investigated steels an almost perfectly aligned lamellae
structure parallel to the shear plane can be observed.
During the alignment, both the carbide thickness and the lamellae spacing is decreased significantly.
The size <strong>of</strong> the structural elements <strong>of</strong> the ferrite phase in direction <strong>of</strong> the shear plane
normal is determined by the lamellae thickness. This refinement both in the lamellae structure
35
4

4
4 Conclusions
as well as in the structure size <strong>of</strong> the ferrite lead to an increase in the mechanical strength.
The increase in the mechanical strength perpendicular to this direction is much smaller than
the increase parallel to the aligned structure, where tensile strengths <strong>of</strong> 1900 MPa (for 900A
at ɛeq = 4 were measured. In contradiction to pure metals, no saturation in the increase in the
mechanical strength could be observed in the investigated strain regime.
The microstructure after Cyclic High Pressure Torsion <strong>of</strong> the rail steels is distinctly influenced
by the strain increment per cycle. If this increment is small (≈ smaller than 2) the
carbides are mainly broken up after a larger number <strong>of</strong> cycles and align very slowly parallel
to the shear direction. At higher strains per cycle, the microstructure is similar in appearance
to the monotonic deformed samples, but only after higher total strains. In Cyclic High Pressure
Torsion <strong>of</strong> rail steels, no saturation in the increase <strong>of</strong> the strength could be observed, too,
although the increase is much slower in Cyclic High Pressure Torsion compared to High Pressure
Torsion.
The aligned microstructure after large shear deformations also influences the mechanisms <strong>of</strong>
crack propagation. Both the fracture toughness and the fatigue crack propagation are markedly
different in the two investigated sample orientations. In samples with orientation A, the crack
has to cross the aligned lamellae. This and the branching <strong>of</strong> the crack leads to an increase in
the fracture toughness with increasing strain. The opposite occures in samples with orientation
B. The crack can propagate along the aligned microstructure and hardly has to cross the strong
cementite lamellae. Therefore, the fracture toughness decreases significantly with increasing
strain. After three passes <strong>of</strong> ECAP, the fracture toughness in samples with orientation A is
more than twice as high than in samples with orientation B. The fatigue crack propagation rate
is influenced, too. Here a markedly decrease <strong>of</strong> the fatigue propagation speed is measured for
samples with orientation A, while it is increased for samples with orientation B. Again, the
effect is amplified with increasing predeformation.
The affected zone in a High Pressure Torsion deformed steel after a heat treatment with a
Laser for very short times contain many microstructural features <strong>of</strong> WELs. Therefore, it may
be concluded that WELs on rails are formed when large shear deformations occure together
with short heat pulses.
36

[1] G. Baumann, H.-J. Fecht, and S. Liebelt. Wear, 191:133, 1991.
Bibliography
[2] W. Lojkowski, Y. Millman, S.I. Chugunova, I.V. Goncharova, M. Djahanbakhsh,
G. Buerkle, and H.-J. Fecht. Mater. Sci. Eng.A, 203:209–215, 2001.
[3] A. Pyzalla, L. Wang, E. Wild, and T. Wroblewski. Wear, 251:901–907, 2001.
[4] U. Ol<strong>of</strong>sson and T. Telliskivi. Wear, 254:80–93, 2003.
[5] J.-K. Kim and C.-S. Kim. Mater. Sci. Eng.A, 338:191–201, 2002.
[6] G. Baumann, K. Knothe, and H. J. Fecht. Nanostructured Materials, 9(1-8):751–754,
1997.
[7] L. Wang, A. Pyzalla, W. Stadlbauer, and E. A. Werner. Mater. Sci. Eng.A, 359:31–43,
2003.
[8] J. W. Ringsberg, M. Loo-Morrey, B. L. Josefson, A. Kapoor, and J. H. Beynon. Int. J.
Fatigue, 22:205–215, 2003.
[9] J. H. Beynon, A. Kapoor, and W. R. Tyfour. Wear, 197:255–265, 1996.
[10] G. Baumann, H. J. Fecht, and S. Liebelt. Wear, 191:133–140, 1996.
[11] W. Oesterle, H. Rooch, A. Pyzalla, and L. Wang. Mater. Sci. Eng.A, 303:150–157, 2001.
[12] Y. K. Chou and H. Song. Journal <strong>of</strong> Materials Processing Technology, 148:259–268,
2004.
[13] G. Poulachon, A. Albert, M. Schluraff, and I.S. Jawahir. International Journal <strong>of</strong> Machine
Tools and Manufacture, in press, 2005.
[14] J. Rech and A. Moisan. International Journal <strong>of</strong> Machine Tools and Manufacture,
43:543–550, 2003.
[15] J. Palmers, M. Van Stappen, J. D’Haen, and M. D’Olieslaeger. Surface and Coatings
Technology, 74-75:162–167, 1995.
[16] O. D. Tasbaz, R. J. K. Wood, M. Browne, H. E. G. Powrie, and G. Denuault. Wear,
230:86–97, 1999.
37
BIB

BIB
Bibliography
[17] E. Jisheng and D. T. Gawne. Wear, 211:1–8, 1997.
[18] L. Xu, S. Clough, P. Howard, and D. Stjohn. Wear, 181-183:112–117, 1995.
[19] O. Barrau, C. Boher, R. Gras, and F. Rezai-Aria. Wear, 255:1444–1454, 2003.
[20] R. Heyder and G. Girsch. Wear, 258:1014–1021, 2005.
[21] W. Daves and F. D. Fischer. Wear, 253:241–246, 2002.
[22] J. W. Ringsberg and T. Lindbäck. Int. J. Fatigue, 25:547–558, 2003.
[23] J. W. Ringsberg, B. L. Josefson F. J. Franklin, A. Kapoor, and J. C.O. Nielsen. Int. J.
Fatigue, 27:680–694, 2005.
[24] Y. T. Zhu, T. G. Langdon, R. S. Mishra, S. L. Semiatin, M. J. Saran, and T. C. Lowe,
editors. Ultrafine Grained Materials II. TMS Publications, Warrendale, Pennsylvania,
2002.
[25] Y. T. Zhu, T. G. Langdon, R. Z. Valiev, S. L. Semiatin, D. H. Shin, and T. C. Lowe,
editors. Ultrafine Grained Materials III. TMS Publications, Warrendale, Pennsylvania,
2004.
[26] M. J. Zehetbauer and R. Z. Valiev, editors. Nanomaterials by Severe Plastic <strong>Deformation</strong>.
J. Wiley, VCH Weinheim (Germany), 2002.
[27] Y. Saito, N. Tsuji, H. Utsunomiya, T. Sakai, and R. G. Hong. Scripta Mater., 39:1221–
1227, 1998.
[28] N. Tsuji, Y. Saito, H. Utsunomiya, and S. Tanigawa. Scripta Mater., 40:795–800, 1999.
[29] I. Salvatori2006. Mater. Sci. Forum, 503-504:311–316, 2006.
[30] H. J. Mcqueen M. Richert and J. Richert. Canadian Metallurgical Quarterly, 37:449–
457, 1998.
[31] M. Richert, Q. Liu, and N. Hansen. Mater. Sci. Eng.A, 260:275–283, 1999.
[32] J. C. Lee, H. K. Seok, and J. Y. Suh. Acta mater., 50:4005–4019, 2002.
[33] S. Mizunuma. Mater. Sci. Forum, 503-504:185–190, 2006.
[34] T. Hebesberger, A. Vorhauer, H. P. Stüwe, and R. Pippan. In M. J. Zehetbauer and R. Z.
Valiev, editors, Nanomaterials by Severe Plastic <strong>Deformation</strong>, pages 447–452. J. Wiley,
VCH Weinheim (Germany), 2002.
[35] A. Vorhauer and R. Pippan. In M. J. Zehetbauer and R. Z. Valiev, editors, Nanomaterials
by Severe Plastic <strong>Deformation</strong>, pages 648–653. J. Wiley, VCH Weinheim (Germany),
2003.
38

Bibliography
[36] T. Hebesberger, H. P. Stüwe, A. Vorhauer, F. Wetscher, and R. Pippan. Acta Mater.,
53:393–402, 2005.
[37] M. Furukawa, Z. Horita, M. Nemoto, and T. G. Langdon. Mater. Sci. Eng.A, 324(1-
2):82–89, 2002.
[38] A. P. Zhilyaev, S. Lee, G. V. Nurislamova, R. Z. Valiev, and T. G. Langdon. Scripta
Mater., 44:2753–2758, 2001.
[39] H. Jiang, Y. T. Zhu, D. P. Butt, I. V. Alexandrov, and T. C. Lowe. Mater. Sci. Eng.A,
290:128–138, 2000.
[40] A. P. Zhilyaev, G. V. Nurislomova, B. G. Kim, M. D. Baro, J. A. Szpunar, and T. G.
Langdon. Acta Mater., 51:753–765, 2003.
[41] H. P. Stüwe. In M. J. Zehetbauer and R. Z. Valiev, editors, Nanomaterials by Severe
Plastic <strong>Deformation</strong>, pages 55–64. J. Wiley, VCH Weinheim (Germany), 2002.
[42] A. Vorhauer and R. Pippan. Scripta Mater., 51:921–925, 2004.
[43] V. M. Segal. Mater. Sci. Eng.A, 271:322–333, 1999.
[44] D. H. Shin, I. Kim, J. Kim, and K.-T. Park. Acta Mater., 49:1285–1292, 2001.
[45] L. Dupuy and E. F. Rauch. Mat. Sci. Eng.A, 337:241–247, 2002.
[46] M. Kamachi, M. Furukawa, Z. Horita, and T. G. Langdon. Mater. Sci. Eng.A, 300:1–8,
2002.
[47] Y. Iwahashi, J. Wang, Z. Horita, M. Nemoto, and T. G. Langdon. Scripta mater., 35:143–
146, 1996.
[48] M. Furukawa, Y. Iwahashi, Z. Horita, M. Nomoto, and T. G. Langdon. Mater. Sci. Eng.A,
257:328–332, 1998.
[49] A. Vorhauer. PhD thesis, Leoben, 2005.
[50] H. P. Stüwe. Mater. Sci. Forum, 503-504:175–178, 2006.
[51] J. D. Embury and R. M. Fisher. Acta Met., 14:147–159, 1966.
[52] G. Langford. Met. Trans., 1:465–477, 1970.
[53] G. Langford. Met. Trans., 8A:861–875, 1977.
[54] M. Dollar, I. M. Bernstein, and A. W. Thompson. Acta Metall., 36:311–320, 1988.
[55] J. Toribio. Mater. Sci. Eng.A, 387-389:227–230, 2004.
[56] H. G. Read, W. T. Reynolds Jr., K. Hono, and T. Tarui. Scripta Mater., 37:1221–1230,
1997.
39
BIB

4
Bibliography
[57] J. Languillaume, G. Kapelski, and B. Baudelet. Materials Letters, 33(3-4):241–245,
1997.
[58] F. Danoix, D. Julien, X. Sauvage, and J. Copreaux. Mater. Sci. Eng.A, 250:8–13, 1998.
[59] W. J. Nam, C. M. Bae, S. J. Oh, and S.-J. Kwon. Scripta Mater., 42:457–463, 2000.
[60] K. Hono, M. Ohnuma, M. Murayama, S. Nishida, A. Yoshie, and T. Takahashi. Scripta
Mater., 44:977–983, 2001.
[61] V. G. Gavriljuk. Mater. Sci. Eng.A, 345:81–89, 2002.
[62] Y. Ivanisenko, W. Lojkowski, R. Z. Valiev, and H.-J. Fecht. Acta Mater., 51:5555–5570,
2003.
[63] X. Sauvage. Mater. Sci. Forum, 503-504:433–438, 2006.
[64] E. Wild, L. Wang, B. Hasse, T. Wroblewski, G. Goerigk, and A. Pyzalla. Wear, 254:876–
883, 2003.
[65] Y. Jirásková, J. Svoboda, O. Schneeweiss, W. Daves, and F. D. Fischer. Appl. Surf. Sci.,
239:132–141, 2005.
[66] H. W. Zhang, S. Ohsaki, S. Mitao, M. Ohnuma, and K. Hono. to be published in Mater.
Sci. Eng.A, 2006.
[67] F. D. Fischer, E. Werner, and K. Knothe. Z. Angew. Math. Mech, 81:75–81, 2001.
[68] L. D. Blackburn, L. Kaufman, and M. Cohen. Acta Metal., 13:533–541, 1965.
[69] F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan. Mater. Sci. Eng.A, 387-389:809–816,
2004.
[70] F. Wetscher and R. Pippan. to be submitted, 2006.
[71] R. K. Islamgaliev, W. Buchgraber, Y. R. Kolobov, N. M. Amirkhanov, A. V. Sergueeva,
K. V. Ivanov, and G. P. Grabovetskaya. Mater. Sci. Eng.A, 319-321:872–876, 2000.
[72] A. Dubravina, M. J. Zehetbauer, E. Schafler, and I. V. Alexandrov. Mater. Sci. Eng.A,
387-389:817–821, 2004.
[73] Z. Yang and U. Welzel. Materials Letters, 59:3406–3409, 2005.
[74] J. Toribio and E. Ovejero. Mater. Sci. Eng.A, 234-236:579, 1997.
[75] J. Toribio and J. Ayaso. Int. J. Fracture, 115:L29–L34, 2002.
[76] J. Toribio and J. Ayaso. Mater. Sci. Eng.A, 343:265–272, 2003.
40

5
List <strong>of</strong> appended papers
Paper A
F. Wetscher, A. Vorhauer, R. Stock and R. Pippan
Structural Refinement <strong>of</strong> Low Alloyed Steels during Severe Plastic <strong>Deformation</strong>
Materials Science and Engineering A387-389 (2004) 809-816
Paper B
F. Wetscher, R. Stock, B. Tian and R. Pippan
Structural Changes <strong>of</strong> Severely Plastic Deformed Rail Steel
Proceedings: Symposium <strong>of</strong> Ultrafine Grained Materials III, TMS Annual Meeting (2004)
315-320
Paper C
F. Wetscher, A. Vorhauer and R. Pippan
Strain Hardening during High Pressure Torsion <strong>Deformation</strong>
Materials Science and Engineering A410-411 (2005) 213-216
Paper D
F. Wetscher, R. Stock and R. Pippan
Formation <strong>of</strong> Surface Layers: The <strong>Effect</strong> <strong>of</strong> the Strain Path
Proceedings: 1 st Vienna Conference Micro- and Nano-Technology, Vienna, Austria, (2005)
357-365
Paper E
F. Wetscher, B. Tian, R. Stock and R. Pippan
High Pressure Torsion <strong>of</strong> Rail Steels
Materials Science Forum, 503-504 (2006) 455-460
41
5

5
5 List <strong>of</strong> appended papers
Paper F
F. Wetscher, R. Pippan, S. Sturm, F. Kauffmann, C. Scheu and G. Dehm
Microstructural Evolution <strong>of</strong> a Pearlitic Steel during Severe Plastic <strong>Deformation</strong>
accepted for publication in Metallurgical Transactions A
Paper G
F. Wetscher, R. Stock and R. Pippan
Fracture Processes in Severe Plastic Deformed Rail Steels
accepted for publication in: Proceedings <strong>of</strong> the 16 th European Conference <strong>of</strong> Fracture (2006)
Paper H
F. Wetscher and R. Pippan
Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
submitted for publication in Philosophical Magazine
Paper I
F. Wetscher and R. Pippan
Structural Evolution during Cyclic Severe Plastic <strong>Deformation</strong>
accepted for publication in: Proceedings <strong>of</strong> the 9 th International Fatigue Congress (2006)
42

A
Structural Refinement <strong>of</strong> Low Alloyed
Steels during Severe Plastic <strong>Deformation</strong>
F. Wetscher 1,2 , A. Vorhauer 1,2 , R. Stock 3 and R. Pippan 1,2
1 Erich Schmid Institute for Materials Science, Austrian Academy <strong>of</strong> Sciences, A-8700
Leoben, Austria
2 CD-Laboratory for local analysis <strong>of</strong> deformation and fracture, Jahnstr. 12, A-8700 Leoben
voestalpine SCHIENEN GMBH, Kerpelystr. 199, A-8700 Leoben, Austria
Abstract
The rail steel S900A and Armco-iron have been deformed by severe plastic deformation,
the microstructural evolution and the changes in microhardness have been investigated. In
both materials the deformation leads to a decrease in structure size. In the case <strong>of</strong> the steel an
alignment <strong>of</strong> the lamellae parallel to the shear plane and a decrease <strong>of</strong> the lamellae spacing
is observed. Lamellae which have been originally not favourably aligned are heavily bent
and broken into small fragments. This refinement is also reflected in the evolution <strong>of</strong> the
microhardness.
A–1
A

A
A–2

A.1 Introduction
A.1 Introduction
Knowledge on Severe Plastic <strong>Deformation</strong> (SPD) not only is important for producing nanostructured
materials, but can also be used to solve problems where high deformation <strong>of</strong> material
is involved. An industrial relevant area is the deformation layer on the surface <strong>of</strong> rails. 1–3 This
region is the origin <strong>of</strong> all phenomena like headcecks, ratcheting, formation <strong>of</strong> white-etching
layers and wear, and the understanding <strong>of</strong> the features <strong>of</strong> the severely distorted material is
vitally important. By means <strong>of</strong> High Pressure Torsion (HPT) the production <strong>of</strong> samples with
a reproducible and well defined deformation history is possible. The aim <strong>of</strong> this work is to
gain a better understanding <strong>of</strong> the behaviour <strong>of</strong> the ferrite-cementite structure <strong>of</strong> a rail-steel
under a shear deformation and at high hydrostatic pressure. To understand the influence <strong>of</strong> the
cementite on the refinement for comparison Armco-iron has been deformed and analysed in
the same way.
A.2 Experimental
A.2.1 Samples
Two materials have been investigated: Armco-iron and the pearlitic rail steel S900A. Samples
have been deformed by HPT, a meanwhile well established method in the SPD community. 4–6
This method simulates both the high hydrostatic pressure and the shear deformation as occuring
on the surface <strong>of</strong> rails. The grain size <strong>of</strong> the undeformed material in the case <strong>of</strong> Armco-iron
was approximately 50µm, the size <strong>of</strong> the pearlite colonies in the rail steel was 10-20µm. The
samples had a diameter <strong>of</strong> 8 mm and a thickness t <strong>of</strong> 0.7 mm. They were deformed at a pressure
<strong>of</strong> 5 GPa, the number <strong>of</strong> turns n was selected to obtain equivalent strains <strong>of</strong> 1, 2, 4, 8,
16 and 32 (only for Armco-iron) at a radius r <strong>of</strong> 3 mm. The equivalent von Mises-strain was
calculated according to the relation
εeq = 2πnr
t √ 3
A.2.2 Microstructural investigations
(A.1)
To investigate the microstructure <strong>of</strong> the steel S900A different cross sections <strong>of</strong> the samples
were prepared: (i) A transversal section with a plane normal SP parallel to the torsion axis
representing the shear plane; (ii) a longitudinal section parallel to the torsion axis at the sample
periphery containing the shear direction SD, with plane normal P, and (iii) a longitudinal
section containing the torsion axis and the sample centre, with plane normal SD as depicted
in Figure A.1. Three different methods have been used. For a view <strong>of</strong> the cementite lamellae,
the samples have been grinded, polished and etched with picrin acid. Then pictures were
taken using the secondary electron detector <strong>of</strong> a scanning electron microscope (SEM). For
analysis <strong>of</strong> the fragmentation <strong>of</strong> the ferrite, the samples were grinded and alternatingly polished
and etched. After a final polishing step the surface was depicted both with a SEM
using a backscatter electron detector (BSE) and with a focused ion beam workstation LEO
XB1540 (FIB), detecting secondary electrons induced by the gallium ion beam. Grain size
A–3
A

A
A Structural Refinement <strong>of</strong> Low Alloyed Steels during Severe Plastic <strong>Deformation</strong>
and other microstructural parameters are determined by the computer program analySIS from
S<strong>of</strong>t Imagine System.
Figure A.1: Sample preparation for microscopy
A.2.3 Microhardness Measurements
To follow the evolution <strong>of</strong> mechanical strength, microhardness measurements were performed
at different radii, at least one millimeter away both from the centre and from the edge <strong>of</strong>
the sample. At least three indentations were made for three different loads at each site, the
hardness value being given for a diagonal <strong>of</strong> 20 µm.
A.3 Results
A.3.1 Microstructural evolution
For Armco-iron, micrographs were taken at the shear plane (SP). After a deformation to strains
smaller than 4, the original grains are detectable and a banded structure can be observed (Figure
A.2). At larger strains an equiaxed structure becomes visible, most <strong>of</strong> the refinement seems
to be already finished. The same behaviour is visible in the micrographs taken with the FIBworkstation.
In comparison with the BSE-micrographs the structure appears coarser at higher
strains. The most apparent features <strong>of</strong> deformed pearlitic steel are the fragmentation and the
alignment <strong>of</strong> the cementite lamellae. Figure A.3a and b are micrographs taken from direction
SD, while Figure A.3c shows a view in the direction P. There is no significant difference to notice.
It shows that at a small strain level in the range <strong>of</strong> 0-2 (Figure A.3a) the cementite lamellae
are heavily bent and partly broken, but the cementite colonies are still recognizable. At higher
strains (Figure A.3b and c) it can be clearly seen that some lamellae are aligned parallel to
the shear plane, some are broken into pieces smaller than 1 µm. Because <strong>of</strong> this alignment <strong>of</strong>
the favourable oriented lamellae it is quite useless to take micrographs <strong>of</strong> the plane SP; etched
A–4

A.3 Results
Figure A.2: Micrograph <strong>of</strong> severely deformed Armco-iron, (a) BSE, ɛeq = 0, 5 (b) BSE, ɛeq = 4 (c) BSE,
ɛeq = 32 (d) FIB, ɛeq = 32
A–5
[h!]
A

A
A Structural Refinement <strong>of</strong> Low Alloyed Steels during Severe Plastic <strong>Deformation</strong>
samples show only a lot <strong>of</strong> flat cementite particles. When following the structural evolution <strong>of</strong>
the ferrite (Figure A.3d - f), one in principle observes the same behaviour as in Armco-iron,
only the size <strong>of</strong> the structural elements is significantly smaller. A quick decrease <strong>of</strong> the structure
size, and a somewhat larger size in the FIB-micrograph compared to the BSE-pictures can
be observed.
A.3.2 Microhardness and microstructural features
The deformation leads to a decrease <strong>of</strong> the microstructure for both materials. This implies
an increase in microhardness as it is shown in Figure A.4a. The change <strong>of</strong> the grain size <strong>of</strong>
Armco-iron and the decrease <strong>of</strong> the mean cementite lamellae distance can be seen in Figure
A.4b. Due to the heavy fragmentation <strong>of</strong> the lamellae, it was not possible to obtain significant
values for the lamellae spacing at larger strains.
A.4 Discussion
Armco-iron is a well investigated material and the present results are comparable to the results
<strong>of</strong> Valiev et al. 7 After a small strain the original grains with a size <strong>of</strong> 50 µm have vanished,
the new structure appears with a mean size <strong>of</strong> 500nm which decreases slowly to 250 nm. The
difference in the structure size observed with BSE or FIB comes from the different contrast
sources. While in the FIB-workstation the contrast arises from the different penetration depths
<strong>of</strong> the gallium ions depending on the relative angle between the beam and the lattice, 8 the contrast
in BSE micrographs is dominated by both the presence <strong>of</strong> lattice defects, i.e. dislocation
walls and crystal orientation with a higher sensitivity in respect <strong>of</strong> orientation differences. This
indicates that the misoriented structural elements are subdivided by dislocation walls or small
angle grain boundaries.
<strong>Deformation</strong> <strong>of</strong> pearlitic steel has been studied in various contexts. Toribio et al. 9 and Won
Jon Nam et al. 10 investigated the pearlite colony evolution and the void initiation during wire
drawing. They reported an alignment <strong>of</strong> the cementite lamellae along the drawing axis. Such
an alignment, together with a severe distortion and a heavy fragmentation <strong>of</strong> the majority <strong>of</strong>
the lamellae has also been observed in this study. After a strain <strong>of</strong> 4 almost all remaining
cementite particles are parallel to the surface, i.e. they lay in the shear plane. Shabashov et
al. 11 and Ivanisenko et al. 12 report that the cementite lamellae can be totally dissolved during
SPD. This could not be observed in this study; it may be that the deformation reached is too
small. The deformation <strong>of</strong> the ferrite in the pearlite is dictated by the lamellae spacing. The
structure size after SPD is limited to this distance, so the refinement <strong>of</strong> the ferrite structure in
the steel occurs faster than in the Armco-iron.
In Armco-iron the increase <strong>of</strong> microhardness markedly changes after small strains. This is
due to the fast fragmentation <strong>of</strong> the grains into subgrains. In contradiction to this behaviour,
in the rail steel S900A the change is much smaller here (Figure A.4a), the microhardness is
determined mostly by the cementite lamellae distance, which changes quite slow, too (Figure
A.4b).
A–6

A.4 Discussion
Figure A.3: Micrograph <strong>of</strong> severely deformed rail steel S900A (a) SE-image, etched sample (SD), ɛeq = 1 (b) SE,
etched sample (SD), ɛeq = 8 (c) SE, etched sample (P), ɛeq = 8 (d) BSE (SD), ɛeq = 4 (e) BSE (SD), ɛeq = 16
(f) FIB micrograph (SD) ɛeq = 16
A–7
A

A
A Structural Refinement <strong>of</strong> Low Alloyed Steels during Severe Plastic <strong>Deformation</strong>
Figure A.4: (a) Changes in the microhardness and (b) Evolution <strong>of</strong> the microstructure as a function <strong>of</strong> strain
A.5 Conclusions
A–8
• In Armco-iron the structure size decreases until saturation takes place, after an von
Mises-strain <strong>of</strong> 16 most <strong>of</strong> the refinement seems to have finished.
• The deformation <strong>of</strong> pearlitic steel is characterized by the rearrangement and fragmentation
<strong>of</strong> the cementite, the size <strong>of</strong> the structural elements <strong>of</strong> the ferrite is limited to the
decreasing lamellae spacing.
• The determining factor for the change <strong>of</strong> the microhardness is the change in structure
size in Armco-iron and the decrease <strong>of</strong> the lamellae spacing in the pearlitic steel.

Bibliography to paper A
[1] W. Lojkowski, Y. Millman, S.I. Chugunova, I.V. Goncharova, M. Djahanbakhsh,
G. Buerkle, and H.-J. Fecht. Mater. Sci. Eng.A, 203:209, 2001.
[2] G. Baumann, H.-J. Fecht, and S. Liebelt. Wear, 191:133, 1991.
[3] W. Oesterle, H. Rooch, A. Pyzalla, and L. Wang. Mater. Sci. Eng.A, 303:150–157, 2001.
[4] R. Z. Valiev, R. K. Islamgaliev, and I. V. Alexandrov. Prog. Mat. Sci., 45:103, 2000.
[5] A. P. Zhilyaev, G. V. Nurislamova, B.-K. Kim, M. D. Baro, J. A. Szpunar, and T. G.
Langdon. Acta Mater., 51:753, 2003.
[6] H. Jiang, Y. T. Zhu, D. P. Butt, I. V. Alexandrov, and T. C. Lowe. Mater. Sci. Eng.A,
290:128, 2000.
[7] R. Z. Valiev, Y. V. Ivanisenko, E. F. Rauch, and B. Baudelet. Acta Mater., 44:4702, 1996.
[8] M. W. Phaneuf. micron, 30:277, 1999.
[9] J. Toribio and E. Ovejero. Mater. Sci. Eng.A, 234-236:579, 1997.
[10] W. J. Nam and C. M. Bae. Mater. Sci. Eng.A, 203:278, 1995.
[11] V. A. Shabashov, L. G. Korshunov amd A. G. Mukoseev, V. V. Sagaradze, A. V. Makarov,
V. P. Pilyugin, S. I. Novikov, and N. F. Vildanova. Mater. Sci. Eng.A, 346:196, 2003.
[12] Yu. V. Ivanisenko, R.Z. Valiev, W. Lojkowski, A. Grob, and H.-J. Fecht. In Ultrafine
Grained Materials II, pages 47–54. Warrendale (USA), 2000.
A–9
A

A
A–10

B
Structural Changes <strong>of</strong> Severely Plastic
Deformed Rail Steel
F. Wetscher 1,2 , R.Stock 3 , B.Tian 1 , and R. Pippan 1,2
1 Department <strong>of</strong> Material Erich Schmid Institute <strong>of</strong> Materials Science, Austrian Academy <strong>of</strong>
Sciences, Jahnstr. 12 A-8700 Leoben, Austria
2 CD-Laboratory for local analysis <strong>of</strong> deformation and fracture, Jahnstr. 12, A-8700 Leoben
3 voestAlpine SCHIENEN GMBH, Kerpelystr. 199, A-8700 Leoben, Austria
Abstract
Severe plastic deformation has been applied to the pearlitic rail steel UIC 900A and Armcoiron.
The development <strong>of</strong> the texture and the peak broadening has been determined. In both
cases a characteristic shear texture evolves, the analysis <strong>of</strong> peak pr<strong>of</strong>iles indicates a structural
refinement during the accumulation <strong>of</strong> strain. To assert further the refinement and to investigate
the behavior <strong>of</strong> the cementite lamellae, transmission electron microscope micrographs
were taken. A decrease <strong>of</strong> the lamellae spacing, a heavy fragmentation <strong>of</strong> the lamellae and
increasing missorientation both in the cementite lamellae and the ferrite were observed.
B–1
B

B
B–2

B.1 Introduction
B.1 Introduction
Due to an increasing exposure in respect to both load and frequency the demands on rails
steels are very high. The most critical point during application is the formation <strong>of</strong> a severely
deformed layer on the surface, which is the origin <strong>of</strong> most phenomena related to wear and
rolling contact fatigue. 1–3 Therefore a better knowledge <strong>of</strong> the behavior <strong>of</strong> this deformed
material is <strong>of</strong> immense importance, allowing a purposeful enhancement <strong>of</strong> the used materials
and a better scheduling <strong>of</strong> the maintenance <strong>of</strong> the rail tracks. The methods <strong>of</strong> severe plastic
deformation (SPD) have shown not only to be excellent to produce nanostructured material; 4–6
they can also be a powerful tool to produce highly deformed samples. The advantage <strong>of</strong> SPDsamples
compared to samples directly obtained from rails is the reproducibility and the exact
knowledge <strong>of</strong> the deformation history.
B.2 Experimental Procedures
In the present investigation the rail steel UIC 900A and Armco-iron have been deformed by
high pressure torsion (HPT), a meanwhile well established procedure. Samples with a diameter
<strong>of</strong> 8 mm and a thickness, t, <strong>of</strong> 0.8 mm were deformed at a hydrostatic pressure <strong>of</strong> 5 GPa to
equivalent strains <strong>of</strong> 1, 2, 4, 8, 16 and 32 at a radius, r, <strong>of</strong> 3mm. The equivalent strain, ɛeq, has
been calculated according to Equation B.1, where n is the number <strong>of</strong> revolutions.
ɛeq = 2πrn
t √ 3
(B.1)
Transmission electron microscopy (TEM) investigations were performed for the steel UIC
900A using a Philips CM12 TEM at 120KV. For sample preparation thin slices were cut,
polished to a thickness smaller than 50µm and finally an area, thin enough for transmission,
was prepared using a focused ion beam workstation (FIB), as schematically depicted in Figure
B.1.
Figure B.1: Sample preparation for TEM.
For Armco-iron, X-ray analyses were performed on a Seifert PTS 3000, using Cr-radiation
with a wavelength <strong>of</strong> 2.2897 Angstrom. For a determination <strong>of</strong> the texture pole figures were
B–3
B

B
B Structural Changes <strong>of</strong> Severely Plastic Deformed Rail Steel
measured, to follow the microstructural changes, peak pr<strong>of</strong>ile analysis <strong>of</strong> 2Θ scans were carried
out. On the same equipment, cuts through the pole figure (Psi-scans) were measured for
the steel 900A. 2Θ scans for the steel were made on a D8Advance from Bruker AXS, again
using Cr-radiation.
B.3 Results and Discussion
B.3.1 X-Ray investigations
Figure B.2: Pole figures <strong>of</strong> Armco-iron (a) ɛeq = 1 (b) ɛeq = 4 (c) and (d) ɛeq = 8 (e) and (f) ɛeq = 32.
The evolution <strong>of</strong> the texture <strong>of</strong> Armco-iron and the rail steel 900A has been investigated
by measuring pole figures and cuts through the pole figure. In the undeformed Armco-iron,
a fibre texture is present, the [100] direction parallel to the rotation axis for the following
SPD-treatment. This texture is already destroyed after small strains. In Figure B.2a-f it can
be seen that the maxima <strong>of</strong> the 〈200〉 peak are changing the positions during the deformation,
B–4

B.3 Results and Discussion
after a strain <strong>of</strong> 8, the maxima lie perpendicular to the shear direction, the main maximum <strong>of</strong>
the 〈110〉 peak is in the center, a small second component <strong>of</strong> the texture is still visible. At
even higher strains, the nature <strong>of</strong> the texture does not change significantly, it is just becoming
sharper. The main component <strong>of</strong> the texture can thus be described as a (110)[001] texture.
This is a typical shear texture for bcc metals. 7 From the measurements made for the steel
900A, a similar behavior can be assumed when analyzing the Ψ-scans, depicted in Figure
B.3.1. In the undeformed sample, only a weak texture is present. During deformation two
texture components became visible. At strains larger than ≈4 the main component is clearly
dominant, the second component is becoming weaker. This indicates that a (110)〈001〉 texture
evolves in the ferrite <strong>of</strong> the steel. The 2Θ scans <strong>of</strong> Armco-iron (Figure B.3.1a) and the steel
900A (Figure B.3.1b) are showing a significant peak broadening. The peaks <strong>of</strong> the cementite
in the steel begin to vanish and are no longer recognizable at higher strains. As described later
this is no pro<strong>of</strong> <strong>of</strong> a beginning dissolution <strong>of</strong> the cementite as it has been reported by various
authors, 8, 9 it is only an indication that such a dissolution may have started. The peaks were
fitted with the Voigt function, a convolution <strong>of</strong> a Gaussian and a Lorentzian function, using a
commercial fitting program. The analysis <strong>of</strong> the integral breadths <strong>of</strong> broadened peak pr<strong>of</strong>iles
is an easily applied single-line method to follow the evolution <strong>of</strong> crystallite-size and strain, a
well established procedure. 10, 11 In such a single line method it is assumed that the Cauchy
or Lorentzian component is due to the crystallite size, the Gaussian component occurs from
a strain e, which may be used as a measure <strong>of</strong> dislocation density. 12, 13 The apparent domain
size D is given by Eq. 2, the strain e can be calculated by Eq. 3.
λ
D =
βL cos Θ
e = βG
4 tan Θ
Figure B.3: psi-Scans <strong>of</strong> 900A.
(B.2)
(B.3)
B–5
B

B
B Structural Changes <strong>of</strong> Severely Plastic Deformed Rail Steel
In these equations λ is the wavelength, βL and βG are the Lorentzian and Gaussian components
<strong>of</strong> the integral breadth and Θ is the angular position <strong>of</strong> the peak. In this study, only
qualitative information will be obtained, no absolute values <strong>of</strong> D and e are calculated from
the values <strong>of</strong> βL and βG. In the pearlitic steel 900A there are also peaks from the cementite.
Nevertheless the 〈110〉 peak <strong>of</strong> the ferrite was fitted disregarding the cementite peaks, this can
be justified by the fact that the intensity <strong>of</strong> these peaks is about 100 times smaller than the
intensity <strong>of</strong> the ferrite peak (note that the y-axis in Figure 4b is logarithmic). Figures B.3.1a
and b show a normalized plot <strong>of</strong> domain size D and strain e, calculated according to Eq. 2
and Eq. 3. For Armco-iron, a fast decrease in the domain size can be observed, until at higher
strains a saturation takes place. In the rail steel, the decrease <strong>of</strong> the domain size seems to be
slower, from the present data it can not be concluded that the refinement is already finished at
ɛeq = 16. In contrast the strain e seems to increase more rapidly in the rail steel. A possible
explanation may be the concurrent deformation <strong>of</strong> the cementite and the ferrite matrix, leading
to raised accumulation <strong>of</strong> dislocations in the ferrite.
The evolution <strong>of</strong> the microstructure investigated with scanning electron microscopy (SEM)
and ion microscopy (IM) and the evolution <strong>of</strong> the mechanical strength is reported in Ref. [14].
The rapid decrease <strong>of</strong> structure size in Armco-iron and the slower decrease in the rail steel
900A as it could be followed from the x-ray investigations is also reflected in an increase in
microhardness and a decrease in the structure size observed with SEM and IM.
Figure B.4: Evolution <strong>of</strong> domain size and strain as a function <strong>of</strong> ɛeq for (a) Armco-iron (b) 900A.
B.3.2 TEM-investigations
TEM-micrographs at different strain levels were taken to study the evolution <strong>of</strong> the structure
<strong>of</strong> the cementite. The arrows indicate the shear direction. Figure 6a shows a TEM micrograph
<strong>of</strong> an undeformed sample. The lamellae are clearly visible, a sharp boundary between the
ferrite and the cementite can be observed. The lamellae spacing is ≈300 nm. After a small
deformation <strong>of</strong> ɛeq = 1 (Figure B.3.2b), there is only a small change in the microstructure,
the lamellae are well defined, the spacing does not changed significantly and the selected area
diffraction pattern (SAD) show quite sharp diffraction spots. At higher strains these sharp
spots gradually transform into concentric rings, displaying increasing missorientations even
in very small volume elements. An alignment <strong>of</strong> the lamellae has occurred after a strain <strong>of</strong>
B–6

Figure B.5: 2Θ scans <strong>of</strong> (a) Armco-iron and (b) 900A.
B.3 Results and Discussion
4, the majority <strong>of</strong> the lamellae are now parallel to the shear direction (Figure B.3.2c). The
distance between neighboring lamellae has significantly decreased, and the boarders are no
longer sharp. The dark field image reveals that missorientation is present in the cementite too.
At higher strains the lamellae are heavily fragmented, the spacing has further decreased, no
sharp boundaries are present. Small broken particles with lengths smaller than 100nm can be
observed. The change in the contrast <strong>of</strong> the cementite and the diffuse boundaries could be
explained by a beginning dissolution <strong>of</strong> the cementite lamellae.
However, a complete dissolution could not be verified, probably the deformation is still
to low. In principle, the observed changes <strong>of</strong> the microstructure are in good agreement with
previous scanning electron and ion microscope investigations 14 <strong>of</strong> the same material. The
alignment <strong>of</strong> the cementite lamellae, the decrease in lamellae spacing and the fragmentation
<strong>of</strong> the lamellae is confirmed by the TEM-investigations.
B–7
B

B
B Structural Changes <strong>of</strong> Severely Plastic Deformed Rail Steel
B–8
Figure B.6: TEM images <strong>of</strong> 900A (a) bright field image, ɛeq = 0, (b) dark field image and SAD-pattern, ɛeq = 1,
(c) bright field image, dark field image and SAD-pattern, ɛeq = 2 (d) bright field image and SAD-pattern, ɛeq = 4,
(e)bright field image and SAD pattern, ɛeq = 8, (f) dark field image <strong>of</strong> (e).

B.4 Conclusion
B.4 Conclusion
• The texture analysis during severe plastic deformation by HPT in Armco-iron and the
pearlitic rail steel 900A indicates that in both cases a (110)[001] texture develops.
• The 2Θ scans <strong>of</strong> the steel 900A reveals a disappearance <strong>of</strong> the cementite peaks at higher
strains, indicating (at least) a start <strong>of</strong> dissolution <strong>of</strong> the cementite
• Peak pr<strong>of</strong>ile analysis demonstrates the decreasing structure size and an increase in strain
(or dislocation density) in both materials, the behavior is in good agreement with previous
scanning electron microscope investigations.
• The TEM investigations confirmed the increasing missorientation even in very small
volume elements, the fragmentation <strong>of</strong> the lamellae and the alignment <strong>of</strong> the fragments
parallel to the shear plane with acceding deformation. The diffuse cementite/ferrite
boundaries at higher strains are a further indication <strong>of</strong> a beginning dissolution <strong>of</strong> the
cementite.
B–9
B

B
B–10

Bibliography to paper B
[1] G. Baumann, K. Knothe, and H. J. Fecht. Nanostructured Materials, 9:951–954, 1997.
[2] W. Oesterle, H. Rooch, A. Pyzalla, and L. Wang. Mater. Sci. Eng.A, 303:150–157, 2001.
[3] L. Wang, A. Pyzalla, W. Stadlbauer, and E. A. Werner. Mater. Sci. Eng.A, 359:31–43,
2003.
[4] H. Jiang, Y. T. Zhu, D. P. Butt, I. V. Alexandrov, and T. C. Lowe. Mater. Sci. Eng.A,
303:128–138, 2000.
[5] A. P. Zhilyaev, G. V. Nurislomova, B. G. Kim, M. D. Baro, J. A. Szpunar, and T. G.
Langdon. Acta Mater., 51:753–765, 2003.
[6] R. Z. Valiev, R. K. Islamgaliev, and I.V. Alexandrov. Prog. Mat. Sci., 45:103–189, 2000.
[7] M. Hölscher, D. Raabe, and K. Lücke. Acta metall. mater., 42:879–886, 1994.
[8] V.A. Shabashov, L.G. Korshunov, A. G. Mukoseev, V.V. Sagaradze, A.V. Makarov, V.P.
Pilyugin, S.I. Novikov, and N.F. Vildanova. Mater. Sci. Eng.A, 346:196–207, 2003.
[9] Yu. V. Ivanisenko, R.Z. Valiev, W. Lojkowski, A. Grob, and H.-J. Fecht. In Ultrafine
Grained Materials II, pages 47–54. Warrendale (USA), 2000.
[10] S. Vives, E. Gaffet, and C. Meunier. Mater. Sci. Eng. A, 2003, in press.
[11] I.V. Alexandrov, K. Zhang, A.R. Kilmametov, K.Lu, and R.Z. Valiev. Mater. Sci. Eng.A,
234-236:331–334, 1997.
[12] Th.H. De Keijser, J.I. Langford, E.J. Mittermeijer, and A.B.P. Vogels. J. Appl. Cryst.,
15:308–314, 1982.
[13] D. Louer. In Advances in X-Ray Analysis Vol. 37, volume 1994, pages 27–35. Plenum
Press, New York, 1994.
[14] F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan. Mater. Sci. Eng. A, 387-389:809–816,
2004.
B–11
B

B
B–12

C
Strain Hardening during High Pressure
Torsion <strong>Deformation</strong>
F. Wetscher 1,2 , , A. Vorhauer 1,2 and R. Pippan 1,2
1 Erich Schmid Institute <strong>of</strong> Material Sciences, Austrian Academy <strong>of</strong> Science, Jahnstraße 12,
A-8700 Leoben, Austria
2 Christian Doppler Laboratory for Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture, Jahnstraße
12, A-8700 Leoben, Austria
Abstract
Severe Plastic <strong>Deformation</strong> (SPD) has been applied to Armco-Iron and Copper by means <strong>of</strong>
High Pressure Torsion (HPT). The evolution <strong>of</strong> the shear stress during deformation was measured
under different hydrostatic pressures. By applying a simple model for a strain hardening
the shear stress - shear strain curves are fitted and the influence <strong>of</strong> the hydrostatic pressure is
studied. These results are compared to microhardness measurements, tensile strength and the
microstructural evolution.
C–1
C

C
C–2

C.1 Introduction
C.1 Introduction
Recent investigations 1–3 have shown for various materials deformed by different severe plastic
deformation (SPD) methods like high pressure torsion (HPT) or equal channel angular pressing
(ECAP) a saturation <strong>of</strong> both the microstructural refinement and the yield strength. The
question always arises, are these changes already present during the deformation or are they
the result <strong>of</strong> post-deformation recovery or recrystallisation. If there are no changes <strong>of</strong> the
developed microstructure after the deformation, then in-situ measurements <strong>of</strong> the flow stress
during the deformation should reflect the characteristics <strong>of</strong> the results from mechanical tests
after the deformation. The aim <strong>of</strong> this study is to compare the mechanical behaviour during
and after the deformation by HPT. The in-situ measurement permits, furthermore, to estimate
the effect <strong>of</strong> the hydrostatic pressure on the flow stress during HPT.
C.2 Experimental
C.2.1 Test equipment
Figure C.1 shows a sketch <strong>of</strong> our tool. The sample lies inside the cavity, in the gab between the
two anvils a burr is formed during the compressive loading. The torque is measured directly
above the sample on one <strong>of</strong> the anvils by means <strong>of</strong> strain gauges. This measured torque M
consists <strong>of</strong> the torque necessary to deform the sample Md plus a torque arising in the area <strong>of</strong>
the burr Mb.
C.2.2 Samples
To investigate the influence <strong>of</strong> the hydrostatic pressure on flow stress <strong>of</strong> severely plastic deformed
samples during and after deformation, Armco iron and pure copper have been deformed
by HPT at different hydrostatic pressures at room temperature and the torque versus
angle <strong>of</strong> twist have been measured in-situ. The hydrostatic pressure was varied between 800
MPa and 7 GPa for copper and between 1.6GPa and 7 GPa for Armco iron. The shear strain
for a number <strong>of</strong> turns n can be calculated by
γ(r) = 2πrn
t
(C.1)
The samples for HPT had a diameter <strong>of</strong> 8 mm and a thickness t <strong>of</strong> 0.8 mm, the number <strong>of</strong>
turns per minute was 0.2. Samples were deformed to strains γ <strong>of</strong> 60, 120 and 180 in the case
<strong>of</strong> Armco iron and to 220 in the case <strong>of</strong> copper at a radius r <strong>of</strong> 2mm.
From each <strong>of</strong> these samples, two sub-sized tensile test specimens were produced. The
longitudinal axis <strong>of</strong> the tensile specimens was situated at a radius <strong>of</strong> 2 mm <strong>of</strong> the HPT sample.
These specimens for tensile tests had a cross section <strong>of</strong> 0.4x0.75 mm and a gauge length <strong>of</strong>
2.5 mm. The tensile tests were performed at constant speed <strong>of</strong> cross head on a commercially
available testing device for small specimens at an initial strain rate <strong>of</strong> 10 -3 s -1 . In addition
microhardness measurements on all tensile specimens were performed; at least five indents per
sample were made. The microstructure was investigated with a scanning electron microscope
C–3
C

C
C Strain Hardening during High Pressure Torsion <strong>Deformation</strong>
Figure C.1: Design <strong>of</strong> the HPT tool.
(SEM) LEO 1525 using backscattered electrons. Special care was taken that the samples were
not exposed to elevated temperatures during the machining <strong>of</strong> the samples for tensile tests and
microscopical analyses to prevent recrystallisation.
C.2.3 Evaluation <strong>of</strong> the torque curves
To recalculate the stress-strain curves from the measured torque versus angle <strong>of</strong> twist, a simple
model for strain hardening (Equation C.2) as applied in 3, 4 is used.
τ(γ) = τS − (τS − τ0)e −Aγ
(C.2)
In this equation τS is the shear stress in the saturation area, τ0 is the shear stress for γ = 0
and A is a constant describing the shape <strong>of</strong> the curve. Such a relation can not describe the
detailed hardening behaviour at relatively small strains, however at large strains it is a useful
approximation. The torque Md for the torsinal deformation5 can be generally written as
Md =
2πτ(r)r 2 dr (C.3)
Using Equation C.1 and Equation C.2, the resulting equation for the torque is given by
Md =
2π τS − (τS − τ0)e −Aγ
r 2 dr (C.4)
C–4

C.3 Results
With this equation it is possible to fit the measured torque curves and obtain the values for
A, τ0 and the most important value for this study τS.
C.3 Results
C.3.1 Torque versus number <strong>of</strong> turns
In Figure C.2a and b, the measured torque versus angle <strong>of</strong> twist for copper and for Armco-iron
for different hydrostatic pressures are shown. In addition the fitted curves are indicated. At the
beginning a pronounced strain hardening is clearly visible, after a certain number <strong>of</strong> turns a
saturation takes place, somewhat earlier in iron than in copper. The necessary torque to deform
the sample increases with pressure. However, it has to be noted that these curves contain the
torque for the deformation plus a contribution from the burr.
Figure C.2: Measured torque versus number <strong>of</strong> turns curves and the fitted curves for (a) copper (b) Armco iron.
C–5
C

C
C Strain Hardening during High Pressure Torsion <strong>Deformation</strong>
C.3.2 Tensile tests and microhardness measurements
Figure C.3 shows the measured tensile tests for copper and Armco iron as a function <strong>of</strong> the
hydrostatic pressure for the samples deformed to γ = 220 in the case <strong>of</strong> copper and between
γ = 60 and γ = 180 in the case <strong>of</strong> Armco iron. For both Armco-iron and copper it can
be seen that there is even if only a very small effect <strong>of</strong> pressure on the tensile strength when
the samples were deformed to the saturation regime. The results from the tensile tests are in
good agreement with recently published results 6, 7 when one takes into account that in these
papers the deformation was not high enough to reach a steady state <strong>of</strong> the mechanical strength.
Additional microhardness tests also revealed no significant deviations in the microhardness <strong>of</strong>
the samples deformed at different pressures.
Figure C.3: Results <strong>of</strong> the tensile test for copper and Armco iron as a function <strong>of</strong> the hydrostatic pressure.
C.3.3 Microstructure
The micrographs <strong>of</strong> the samples were taken in radial direction <strong>of</strong> the HPT sample at a radius
<strong>of</strong> 3 mm. The shear plane is parallel to the baseline in all micrographs. It seems to be that
the size <strong>of</strong> the structural elements is somewhat smaller at higher pressures, however the effect
is even if very small (Figure C.4). The micrographs show that the structural elements are
elongated. The longitudinal axis forms an angle with the vertical that is clearly smaller than
the corresponding shear angle.
C.4 Discussion
The results from the in-situ measurement <strong>of</strong> the torque curves, tensile tests, microhardness
measurements and SEM micrographs have clearly shown that for both materials saturation in
structural size and strength takes place when they are deformed by high pressure torsion up to
very high strains. This saturation and the resulting microstructure can be explained by a steady
state deformation, where new structural elements are continuously formed (for more details
see 1 ), the current results and the micrographs <strong>of</strong> the microstructure are supporting this.
C–6

C.4 Discussion
Figure C.4: SEM micrographs (a) copper, γ = 220 hydrostatic pressure 0.95 MPa, (b) copper, γ = 220 hydrostatic
pressure 7500 MPa, (c) Armco iron, γ = 180, hydrostatic pressure 1.9 GPa, (d) Armco iron, γ = 180
hydrostatic pressure 7.5 GPa.
A surprising result is that there is almost no dependence <strong>of</strong> the mechanical strength on the
hydrostatic pressure in the investigated range when only the values obtained after the deformation
are considered. In contradiction to this, the in-situ measured torque curves apparently
show such differences that are not present in the other experiments although these differences
are not as large as one might expect. 8 An explanation for this could be that, as mentioned
before, the measured curves do not only contain the torque necessary to plastically deform
the material but also a contribution from the burr. This contribution strongly depends on the
design <strong>of</strong> the used HPT tool. We always make great efforts to ensure that the volume <strong>of</strong> the
sample lying inside the cavity is deformed by simple torsion, so Equation C.4 will describe
the fraction <strong>of</strong> the torque for the plastic deformation <strong>of</strong> this part. For the present analyses
we assume that slip between one side <strong>of</strong> the burr and the anvil takes place. In this case the
torque due to the burr Mb will be proportional to the hydrostatic pressure and the coefficient
<strong>of</strong> friction µ. Recent results from finite element calculations 9 make it reasonable to assume
C–7
C

C
C Strain Hardening during High Pressure Torsion <strong>Deformation</strong>
Armco iron copper
Hydrostatic Pressure [MPa] 1900 3800 7500 1900 3800 5700
τS [MPa] 1100 1277 1519 431 497 527
τS (corrected) [MPa] µ = 0.05 1069 1232 1405 405 448 455
τS (corrected) [MPa] µ = 0.15 1018 1107 1217 353 350 311
Table C.1: Summarisation <strong>of</strong> the calculated values for τS for different hydrostatic pressures
for our tool a linear decrease from eighty per cent <strong>of</strong> the nominal hydrostatic pressure at the
inner edge <strong>of</strong> the burr to the flow stress <strong>of</strong> the undeformed material at the outer edge. This
torque due to friction can then be subtracted from the measured torque curves and the resulting
curve can be fitted again using Equation C.4. In Table C.1 calculated τs are shown where
the coefficient <strong>of</strong> friction was 0.05 and 0.15, respectively. It can bee seen that the differences
between the shear stresses for different hydrostatic pressures are reduced more. The estimated
shear stress depends strongly on the values for the friction coefficient. Therefore it is very
difficult to obtain absolute values for the shear stress during the HPT deformation. However,
these estimations show that the effect <strong>of</strong> pressure is much smaller than one would expect from
the torque measurements.
Nevertheless, the measurement <strong>of</strong> the torque can be very useful to study changes in the
mechanical strength and to find the necessary amount <strong>of</strong> deformation to reach the steady state.
The ratios <strong>of</strong> the shear stresses in the steady state for copper and iron are similar as the ratios
<strong>of</strong> the ultimate tensile strengths. Hence it is possible to compare the hardening behaviour <strong>of</strong>
different materials when the experimental setup is the same. For example the present results
indicate that the steady state <strong>of</strong> the flow stress is reached at significantly smaller strains in the
fcc copper than in the bcc Armco iron.
C.5 Conclusion
C–8
• Depending on the material, saturation <strong>of</strong> strain hardening and grain refinement takes
place at γ <strong>of</strong> about 40 and 20 for Armco iron and copper, respectively.
• Only a very small influence <strong>of</strong> the hydrostatic pressure on the resulting microstructure
and the tensile strength on Armco-iron and copper samples deformed by high pressure
torsion could be observed.

Bibliography to paper C
[1] T. Hebesberger, H. P. Stuewe, A. Vorhauer, F. Wetscher, and R. Pippan. Acta Mater.,
53:393–402, 2005.
[2] F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan. Mater. Sci. Eng. A, 387-389:809–816,
2004.
[3] N. Q. Chinh, G. Horvath, Z. Horita, and T. G. Langdon. Acta mater., 52:3555–3563, 2004.
[4] A. Vorhauer and R. Pippan. Scripta Mater., 51:921–925, 2004.
[5] H. P. Stuewe and H. Turk. Zeitschrift f. Metallkunde, 55:699–703, 1964.
[6] M. Sus-Ryszkowska, T. Wejrzanowski, Z. Pakiela, and K. J. Kurzydlowski. Mater. Sci.
Eng.A, 369:151–156, 2004.
[7] B.Q. Han, E.J. Lavernia, and F.A.Mohamed. Met. Trans., 35A:1343–1350, 2004.
[8] M. J. Zehetbauer, J. Kohout, E. Schafler, F. Sachslehner, and A. Dubravina. J. Alloys and
Compounds, 378:329–334, 2004.
[9] W. Ecker. Master’s thesis, Leoben, 2004.
C–9
C

C
C–10

D
Formation <strong>of</strong> surface layers: the effect <strong>of</strong>
the strain path
F. Wetscher 1,2 , R. Stock 3 and R. Pippan 1,2
1 Erich Schmid Institute <strong>of</strong> Material Sciences, Austrian Academy <strong>of</strong> Science, Jahnstraße 12,
A-8700 Leoben, Austria
2 Christian Doppler Laboratory for Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture, Jahnstraße
12, A-8700 Leoben, Austria
3 voestAlpine Schienen GmbH, Leoben, Austria
Abstract
To simulate the formation <strong>of</strong> heavily deformed layers on rails in service, High Pressure Torsion
experiments as well as simple surface deformation experiments with the pearlitic rail steel
900A have been performed. The evolution <strong>of</strong> the microstructure was investigated, with close
attention to the differences between a monotonic and a cyclic deformation path. It turned out
that the resulting microstructure is markedly affected by the amount <strong>of</strong> plastic deformation per
cycle. The higher this value is, the more similar are the resulting microstructure compared to
the monotonic deformed samples. In experiments, where only the surface was deformed, the
microstructure showed the same features as in the HPT-deformed samples, thus confirming
the assumption that High Pressure Torsion is capable to simulate the processes during the surface
deformation due to overrun <strong>of</strong> the rail tracks by trains. The microhardness measurements
showed clear differences between the monotonic and cyclic deformed samples, indicating that
a monotonic deformation promotes higher dislocation densities and gives rise to the assumption
that dissolution <strong>of</strong> the cementite is taking place.
D–1
D

D
D–2

D.1 Introduction
D.1 Introduction
The formation <strong>of</strong> nanostructured layers due to friction and shear deformation is a well known
phenomenon. It can be observed in as different areas as machining like turning and grinding,
1–4 sliding 5, 6 and abrasion. 7, 8 Another industrial very important area is the behaviour <strong>of</strong>
the surface <strong>of</strong> rails. During service on the surface <strong>of</strong> rails such a nanostructured deformation
layer evolves, which is the origin <strong>of</strong> all phenomena like headcecks, ratcheting, formation <strong>of</strong>
white-etching layers and wear. 9–13 Due to an increasing exposure in respect to both load and
frequency the demands on the performance <strong>of</strong> rails steels are very high and the need for an improved
knowledge <strong>of</strong> the resulting changes in the microstructure and the mechanical properties
at the surface arises.
D.2 Experimental
To simulate the shear deformation on the surface <strong>of</strong> rail steels, the methods <strong>of</strong> severe plastic
deformation (SPD) are very suitable. Normally, these methods are used to produced nanoscale
grained materials which unique mechanical properties, 14–18 but they are also suitable to investigate
processes where both high hydrostatic pressures and high shear deformations are
combined. For this study, high pressure torsion (HPT) has been used to deform the rail steel
UIC 900A, the composition <strong>of</strong> this steel and mechanical properties are given in Table D.1.
A high pressure torsion tool consists <strong>of</strong> two anvils with a coin-shaped cavity for the sample,
see Figure D.1. While applying a pressure to supply enough friction, one anvil is turned with
respect to the other. Due to friction the sample is deformed by almost pure shear.
The equivalent von Mises strain is a function <strong>of</strong> the radius r, <strong>of</strong> the thickness t and the
number <strong>of</strong> turns n according to Equation D.1.
ɛeq = 2πrn
t √ 3
(D.1)
In this investigation, the sample had a diameter <strong>of</strong> 8 mm and a initial thickness <strong>of</strong> 0.8 mm.
The applied hydrostatic pressure was about 6.5GPa. With the method <strong>of</strong> HPT samples where
prepared with a monotonic deformation and a cyclic deformation representing a change in
the strain path. Samples with a monotonic deformation were prepared up to deformations <strong>of</strong>
ɛeq = 32 at a radius <strong>of</strong> 3 mm, for the samples with cyclic deformation, von Mises strains
<strong>of</strong> ∆ɛeq = 0.25; 0.5; 1 and 2 per cycle at a radius <strong>of</strong> 3mm and a total deformations up to
ɛeq = 64 were realised.
The third type <strong>of</strong> deformation experiment consisted <strong>of</strong> two samples <strong>of</strong> the same material
pressed together and one was turned with respect to the other. This induced a deformation <strong>of</strong>
the surface layer, which was investigated for comparison with homogeneous deformed samples.
The diameter <strong>of</strong> the sample that was shaped like a frustum <strong>of</strong> pyramid was 2 mm, the
C Si Mn P S R m A5 HB
0,76 0,35 1,0 0,017 0,04 900 10 260
Table D.1: Composition and mechanical properties <strong>of</strong> rail steel 900A
D–3
D

D
D Formation <strong>of</strong> surface layers: the effect <strong>of</strong> the strain path
Figure D.1: Principle <strong>of</strong> High Pressure Torsion
contact pressure was 1 GPa and the samples were turned 10 times and 20 times around the axis.
The changes <strong>of</strong> the microstructure was characterised by scanning electron microscopy (SEM)
using a secondary electron detector (SE) and a backscatter electron detector (BSE) depicting
the microstructure in different directions <strong>of</strong> the samples as indicated in D.2. Microhardness
measurements were performed to follow the changes <strong>of</strong> the mechanical properties.
D–4
Figure D.2: Sample preparation for microscopy

D.3 Results
D.3.1 Monotonic <strong>Deformation</strong>
D.3 Results
From the SEM-micrographs <strong>of</strong> the monotonic deformed samples taken with the SE-detector
(Figure D.3) it can be seen that at low shear strain a severe deformation <strong>of</strong> the normally brittle
lamellae takes place leading to heavily bended lamellae, sometimes the lamellae are broken.
Lamellae originally oriented roughly parallel to the shear plane try to align in this direction.
At higher strains all the lamellae and respectively fragments <strong>of</strong> lamellae are orientated parallel
to the shear plane, this can be seen both in radial and tangential direction. The pearlite
colonies are no longer recognizable. When the strain gets even higher, on the etched samples
the lamellae are hardly recognizable; it seems that dissolution <strong>of</strong> the cementite has started. The
size <strong>of</strong> the fragments is in both depicted directions smaller than 1 µm, the distance between
the lamellae respectively this fragments is decreasing from approximately 300nm in the undeformed
state to a mean value <strong>of</strong> less than 100nm. The images taken with the BSE-detector revealed
that the size <strong>of</strong> ferrite’s structural elements with the same orientation decreases quickly,
at higher shear strains the lamellae spacing is an upper bound.
D.3.2 Cyclic deformation
The SEM-images <strong>of</strong> the cyclically deformed samples reveal that the deformation per cycle
has a distinct effect on the resulting microstructure. Figure D.4 shows the development <strong>of</strong> the
microstructure after a deformation by cyclic HPT with a strain amplitude <strong>of</strong> ∆ɛeq = 0.5 per
cycle. It can be seen that even at total deformation as high as ɛeq = 16 or even ɛeq = 32 the
original cementite colonies are still recognizable. In this case <strong>of</strong> quite low amplitudes it seems
that the main feature <strong>of</strong> deformation is the fragmentation <strong>of</strong> the lamellae and it needs much
higher strains to achieve these changes at all cyclic experiments compared to a monotonic
deformation. Figure D.4d shows the microstructure after a deformation up to ɛeq = 64 (0.5
per cycle). It can be seen that now all the lamellae are broken to pieces smaller than 1 µm,
and again the distance between these pieces has been reduced compared to the undeformed
microstructure. All these pieces are more or less oriented parallel to the shear plane, just as
in Figure D.3d. Nevertheless, the microstructure <strong>of</strong> this cyclic deformed sample is different
from the monotonic deformed one. The change in the lamellae spacing is not as pronounced
as in the monotonic deformed samples and there is no indication that the etching behaviour
has changed and therefore it cannot be assumed that the carbon <strong>of</strong> the carbide has gone in
solution. When the strain per cycle becomes larger, the changes in the microstructure are
similar to those in the monotonic deformed samples. In Figure D.5 one can see that after a
total deformation <strong>of</strong> ɛeq = 8 all <strong>of</strong> the lamellae are heavily bend and fanfold. But it is still
possible to distinguish areas that belonged to the same cementite colony. In some places the
lamellae distance has been markedly decreased. At higher total deformations at first there
are no more obvious changes; in some areas a wavy structure can be observed. From these
SEM-images it cannot be definitely seen that the lamellae are heavily fragmented. Again,
after a deformation to ɛeq = 64, almost all lamellae are parallel to the shear plane, in this
case with some steps, probably resulting from the change in the strain path. If Figure D.5d is
compared to 4d (same total deformation, but different deformation per cycle) it can be seen
D–5
D

D
D Formation <strong>of</strong> surface layers: the effect <strong>of</strong> the strain path
Figure D.3: SEM micrographs <strong>of</strong> 900A deformed by monotonic HPT (a) ɛeq = 1, tangential (b) ɛeq = 2,
tangential (c) ɛeq = 8, radial (d) ɛeq = 32, tangential
that in this case the etching behaviour has changed like in the monotonic deformed sample,
but higher total deformations were needed. Figure D.6 shows SEM-images taken with the
BSE-detector (reproducing mainly orientation differences <strong>of</strong> the ferrite) <strong>of</strong> samples with the
same total deformation but a different strain per cycle. In these images two things can be
seen. First, the decrease <strong>of</strong> the lamellae spacing is larger and quicker in the samples with a
higher ∆ɛ per cycle, and second, the size <strong>of</strong> structural elements <strong>of</strong> the ferrite is limited by this
distance between the lamellae. The size <strong>of</strong> these structural elements <strong>of</strong> the ferrite decreases
quickly at early stages <strong>of</strong> deformation and is reduced then further slowly simultaneously with
the lamellae spacing.
D.3.3 <strong>Deformation</strong> <strong>of</strong> a surface due to friction
The micrographs <strong>of</strong> this experiments (Figure D.7) show all the features <strong>of</strong> the HPT deformed
samples: a severely deformation <strong>of</strong> the lamellae, fragmentation <strong>of</strong> lamellae, an alignment
D–6

D.3 Results
Figure D.4: SEM micrographs <strong>of</strong> 900A deformed by cyclic HPT with a strain amplitude <strong>of</strong> ∆ɛeq = 0.5, all
images are in radial direction (a) ɛeq = 8; (b) ɛeq = 16; (c) ɛeq = 32; (d) ɛeq = 64
parallel to the shear plane and even a change in the etching behaviour close to the surface can
be observed (Figure D.7b-d). Of course, the deformation is not uniform and there is a gradient
not only from the centre <strong>of</strong> rotation to the border but also from the surface to the bulk material
(Figure D.7a). In this case, it is not possible to define a certain strain. In some cases, cracks
were observed at the surface.
D.3.4 Microhardness measurements
Figure D.8a shows an overview <strong>of</strong> the results <strong>of</strong> the microhardness measurements. It can be
seen that for the monotonic deformed sample the microhardness almost triples after a deformation
up to ɛeq = 32, while the increase <strong>of</strong> the cyclically deformed samples is much less
pronounced. Only the sample with ∆ɛeq = 2 deformed up to ɛeq = 64 had similar hardness
like the highest monotonic sample. The lower the plastic deformation per cycle was, the lower
is the increase in microhardness at the same strain level. In Figure D.8b the microhardness
pr<strong>of</strong>ile <strong>of</strong> the sample in Figure D.7a is shown. As expected the hardness is highest on the most
D–7
D

D
D Formation <strong>of</strong> surface layers: the effect <strong>of</strong> the strain path
Figure D.5: SEM micrographs <strong>of</strong> 900A deformed by cyclic HPT with a strain amplitude <strong>of</strong> ∆ɛeq = 2, all images
are in radial direction (a) ɛeq = 8; (b) ɛeq = 16; (c) ɛeq = 32; (d) ɛeq = 64
severe deformed surface and decreases to the hardness <strong>of</strong> the undeformed bulk material.
D.4 Discussion
D.4.1 Microstructure
The development <strong>of</strong> the microstructure <strong>of</strong> this pearlitic rail steel deformed by High Pressure
Torsion has been described previously. 19, 20 The results are very similar to the results <strong>of</strong> cold
drawn wires both in respect to the microstructure and to microhardness. 21–23 Some <strong>of</strong> these
papers also show by means <strong>of</strong> tomographic atom probe or Mössbauer spectroscopy that at
least a partly dissolution <strong>of</strong> the cementite has occurred. The most common explanation for
this dissolution is that dislocations crossing the carbide drag carbon atoms out <strong>of</strong> the lamellae
due to a higher binding energy <strong>of</strong> the carbon to the dislocation compared to the binding energy
in the carbide. 23, 24 So it is reasonable to assume that such a dissolution has also taken place
D–8

D.4 Discussion
Figure D.6: SEM micrographs <strong>of</strong> 900A deformed by cyclic HPT with a total strain ɛeq = 8, all images are in
radial direction (a) ∆ɛeq = 0.25; (b) ∆ɛeq = 0.5; (c) ∆ɛeq = 0.1; (d) ∆ɛeq = 2
in the monotonic deformed samples. If a dislocation is supposed to cross a cementite lamella,
a high stress is necessary. If the strain is not high enough, the lamellae act as a barrier for
the dislocations and they pile up at the interface. So some kind <strong>of</strong> threshold value for a strain
per cycle is needed to start a dissolution <strong>of</strong> the cementite. If the strain is below this threshold,
cementite lamellae can be bent, they can break, they can align to favourable orientations,
but no dissolution will occur. If the fragments are small enough, they can be moved in the
s<strong>of</strong>t matrix without further deformation (see Figure D.4d). When the strain is high enough,
carbon atoms are dragged into the matrix and this leads to an increase <strong>of</strong> hardness due to solid
solution hardening. This further enhances the plastically deformation <strong>of</strong> the lamellae because
the difference <strong>of</strong> mechanical strength between the ferrite matrix and the carbide decreases.
The resulting microstructure (Figure D.5d) is similar to the microstructure <strong>of</strong> the monotonic
deformed sample (Figure D.3d). The comparison <strong>of</strong> the microstructure <strong>of</strong> the samples form
the friction experiments and pictures out <strong>of</strong> the literature 25, 26 with the HPT samples shows
that the use <strong>of</strong> HPT is suitable to produce samples for studying the fundamental mechanism
D–9
D

D
D Formation <strong>of</strong> surface layers: the effect <strong>of</strong> the strain path
Figure D.7: Steel 900A deformed by friction (a) light optical microscope micrograph with microhardness indents
(b) SEM-micrograph; lightly deformed area (c) SEM; severely deformed area (d) SEM; severely deformed surface
Figure D.8: Results <strong>of</strong> microhardness measurements (a) Overview <strong>of</strong> the HPT-deformed samples (b) Evolution <strong>of</strong>
the microhardness as a function <strong>of</strong> distance from the surface <strong>of</strong> a sample from the friction experiment
D–10

occurring on the surface <strong>of</strong> rails.
D.4.2 Microhardness
D.4 Discussion
The increase <strong>of</strong> hardness and therefore other mechanical properties like tensile strength can
be explained by three components: the decreasing lamellae spacing, the increasing dislocation
density and the assumed solution <strong>of</strong> carbon atoms in the ferrite. The difference <strong>of</strong> the lamellae
spacing between the samples with various deformation paths could be the major influence for
the different run <strong>of</strong> the microhardness curve. Another reason for the increased microhardness
<strong>of</strong> the monotonic deformed sample compared to the cyclically deformed ones is the higher
amount <strong>of</strong> solved carbon atoms in the ferrite matrix. But if the strain per cycle is high enough
for a start <strong>of</strong> dissolution <strong>of</strong> the cementite, the influence <strong>of</strong> solid solution hardening at a certain
total deformation should be the same. So maybe another mechanism is active. In principle,
the same amount <strong>of</strong> dislocation is needed when samples are deformed to the same total strain,
so the dislocation density should also be the same. At the very beginning <strong>of</strong> deformation, one
dislocation has to be generated from a source. If this deformation is reversed, the dislocation
could go back to the source that is acting now as a sink. This may sometimes happen for very
small deformations, but in reality, most <strong>of</strong> the generated dislocations will not be able to go back
exactly the same way and annihilate in the former source. If the strain path is reversed exactly
in the opposite direction, some dislocations can go back and annihilate, but the amount <strong>of</strong>
these dislocations is smaller when the strain gets higher. So the increase <strong>of</strong> dislocation density
is slower for samples with a smaller ∆ɛ due to an easier recombination or annihilation after
the change <strong>of</strong> the strain path.
D.4.3 Conclusions
• The monotonic deformation <strong>of</strong> pearlitic steels by HPT is dominated by an alignment and
fragmentation <strong>of</strong> the cementite lamellae. The lamellae spacing is markedly decreasing,
at the highest reached strains indications for a cementite dissolution can be observed.
• The cyclic deformation <strong>of</strong> pearlitic steel by HPT is strongly affected by the plastic strain
per cycle. When the strain per cycle is low enough, the carbides are fragmented and align
themselves in favourable directions, no indications <strong>of</strong> a dissolution can be observed. If
the strain per cycle is high enough, the resulting microstructure is similar to that <strong>of</strong> the
monotonic deformed samples, but higher total strains are needed.
• The friction experiments showed that the methods <strong>of</strong> SPD and here especially HPT are
usable to investigate the processes on the surface <strong>of</strong> rails.
• The observed changes <strong>of</strong> the microhardness can be explained when all the components
<strong>of</strong> strengthening are considered.
D–11
D

D
D–12

Bibliography to paper D
[1] Y. K. Chou and H. Song. Journal <strong>of</strong> Materials Processing Technology, 148:259–268,
2004.
[2] G. Poulachon, A. Albert, M. Schluraff, and I.S. Jawahir. International Journal <strong>of</strong> Machine
Tools and Manufacture, in press, 2005.
[3] J. Rech and A. Moisan. International Journal <strong>of</strong> Machine Tools and Manufacture,
43:543–550, 2003.
[4] J. Palmers, M. Van Stappen, J. D’Haen, and M. D’Olieslaeger. Surface and Coatings
Technology, 74-75:162–167, 1995.
[5] O. D. Tasbaz, R. J. K. Wood, M. Browne, H. E. G. Powrie, and G. Denuault. Wear,
230:86–97, 1999.
[6] E. Jisheng and D. T. Gawne. Wear, 211:1–8, 1997.
[7] L. Xu, S. Clough, P. Howard, and D. Stjohn. Wear, 181-183:112–117, 1995.
[8] O. Barrau, C. Boher, R. Gras, and F. Rezai-Aria. Wear, 255:1444–1454, 2003.
[9] U. Ol<strong>of</strong>sson and T. Telliskivi. Wear, 254:80–93, 2003.
[10] J.-K. Kim and C.-S. Kim. Mater. Sci. Eng.A, 338:191–201, 2002.
[11] A. Boehmer and T. Klimpel. Wear, 253:150–161, 2002.
[12] J. W. Ringsberg, M. Loo-Morrey, B. L. Josefson, A. Kapoor, and J. H. Beynon. Int. J.
Fatigue, 22:205–215, 2000.
[13] J. H. Beynon, A. Kapoor, and W. R. Tyfour. Wear, 197:255–265, 1996.
[14] T. Y. Zhu, T. C. Lowe, and T. G. Langdon. Scripta mater., 51:825–830, 2004.
[15] Y. H. Zhao, X. Z. Liao, Z. Jin, R. Z. Valiev, and Y. T. Zhu. Acta mater., 52:4589–4599,
2004.
[16] R. Z. Valiev, A. V. Sergueeva, and A. K. Mukherjee. Scripta mater., 49:669–674, 2003.
[17] A. Vinogradov, S. Hashimoto, and V. I. Kopylov. Mater. Sci. Eng.A, 355:277–285, 2003.
[18] M. Furukawa, Z. Horita, M., and T. G. Langdon. Mater. Sci. Eng.A, 324:82–89, 2002.
BIB–13
BIB

D
Bibliography to paper D
[19] F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan. Mater. Sci. Eng. A, 387-389:809–816,
2004.
[20] Florian Wetscher, Andreas Vorhauer, and Reinhard Pippan. Mater. Sci. Eng.A, 410-
411:213–216, 2005.
[21] M. Zelin. Acta Mater., 50(17):4431–4447, 2002.
[22] J. Languillaume, G. Kapelski, and B. Baudelet. Materials Letters, 33:241–245, 1997.
[23] F. Danoix, D. Julien, X. Sauvage, and J. Copreaux. Mater. Sci. Eng.A, 250:8–13, 1998.
[24] J. Languillaume, G. Kapelski, and B. Baudelet. Acta mater., 45:1201–1212, 1997.
[25] R. Stock. Master’s thesis, Leoben, 2001.
[26] L. Wang. PhD thesis, Berlin, 2002.
D–14

E
High Pressure Torsion <strong>of</strong> Rail Steels
F. Wetscher 1,2 , B. Tian 3 , R. Stock 4 and R. Pippan 1,2
3Erich Schmid Institute <strong>of</strong> Material Sciences, Austrian Academy <strong>of</strong> Sciences, Jahnstraße
12, A-8700 Leoben, Austria
3CD-Laboratory for Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture, Jahnstraße 12, A-8700
Leoben, Austria
3now Boehler Edelstahl
3voestalpine SCHIENEN GMBH, Kerpelystr. 199, A-8700 Leoben, Austria
Abstract
To study the influence <strong>of</strong> shear deformation on the evolution <strong>of</strong> the microstructure and the
mechanical strength in rail steels, three steels with different microstructure (two pearlitic, one
bainitic) were deformed by High Pressure Torsion (HPT). In order to evaluate in addition the
effect <strong>of</strong> the strain path, a cyclic form <strong>of</strong> HPT was applied. The mechanical strength was determined
by means <strong>of</strong> in-situ measurement <strong>of</strong> the flow stress and microhardness measurements.
The differences <strong>of</strong> the mechanical strengths between the monotonic and cyclic deformed samples
clearly indicate that a monotonic deformation promotes higher dislocation densities and
leads to the assumption that dissolution <strong>of</strong> the cementite takes place more pronounced.
E–1
E

E
E–2

E.1 Introduction
E.1 Introduction
The demands on modern rail steels are permanently increasing due to an increasing exposure
in respect to both load and frequency and so the trend goes to high-strength rail steels. It is well
known that on the surface <strong>of</strong> rails a nanostructured deformation layer evolves, 1, 2 which can
be the starting point <strong>of</strong> cracks, especially dangerous in the more brittle high-strength steels.
Hence, better knowledge <strong>of</strong> the effect <strong>of</strong> shear deformation on the microstructure and the
mechanical properties <strong>of</strong> these steels is necessary to improve, for example, service intervals
or to enhance finite element simulations for lifetime estimations or to develop design tools
for new rail steels. In order to obtain samples with a known deformation history, methods <strong>of</strong>
severe plastic deformation (SPD) are used to study these effects.
E.2 Experimental
In this study, two pearlitic rail steels with different lamellae spacing (300nm and 100nm) and
one bainitic rail steel were investigated. The compositions <strong>of</strong> these steels are shown in Table
E.1; the microstructure <strong>of</strong> the undeformed steels can be seen in Figure E.1. The samples were
deformed up to very high strains by means <strong>of</strong> HPT 3 at a hydrostatic pressure <strong>of</strong> 6.5 GPa. The
diameter <strong>of</strong> the sample was 8 mm; the initial thickness was 0.8 mm; the turning speed <strong>of</strong> the
rotating anvil was 0.2 per minute. For high pressure torsion, the equivalent strain ɛeq as a
function <strong>of</strong> the radius r, the sample thickness t and the number <strong>of</strong> turns n can be calculated by
εeq = 2πnr
t √ 3
C Si Mn Cr P S
900A 0,76 0,35 1,0 0,014 0,017 0,04
350 LHT 0,78 0,46 1,18 0,23 0,012 0,015
Dobain 430 0,70-0,82 0,4-1,0 0,7-1,1 0,4-0,7 0,02 0,02
Table E.1: Composition <strong>of</strong> the used materials
(E.1)
To study whether there is an influence <strong>of</strong> the strain path or not, also a cyclic form <strong>of</strong> HPT
for the steel 900A was used, Cyclic High Pressure Torsion, CHPT, corresponding to route C in
equal channel angular pressing. 4 To realise this, the drive mechanism was modified to allow a
reversing <strong>of</strong> the deformation. CHPT was performed with different strains per cycle ∆ɛeq ranging
from ∆ɛeq = 0.5 to ∆ɛeq = 4 at a radius <strong>of</strong> r = 3 mm up to a total deformation <strong>of</strong> ɛeq= 64 at
a radius <strong>of</strong> 3 mm. The evolution <strong>of</strong> the microstructure in radial and tangential direction <strong>of</strong> the
sample was investigated by scanning electron microscopy (SEM) and transmission electron
microscopy (TEM) using a LEO 1525 and a Philips CM12 at 120kV, respectively. The samples
for the SEM were grinded, polished and etched with Nital. The foils for the TEM were
produced using a focused ion beam workstation LEO 1540 XB. The change <strong>of</strong> the mechanical
strength was determined by microhardness measurements and in-situ measurement <strong>of</strong> the flow
stress during (C)HPT by means <strong>of</strong> strain gauges at one <strong>of</strong> the anvils, for details see. 5
E–3
E

E
E High Pressure Torsion <strong>of</strong> Rail Steels
Figure E.1: Micrographs <strong>of</strong> the undeformed material (a) SEM micrograph <strong>of</strong> 350 LHT, (b) SEM micrograph <strong>of</strong>
Dobain 430.
E.3 Results
E.3.1 Microstructure
From the series <strong>of</strong> SEM micrographs <strong>of</strong> the rail steel 350 LHT the evolution <strong>of</strong> the microstructure
for samples deformed by conventional HPT as a function <strong>of</strong> the shear strain can bee seen
in Figure E.2a - d. Figure E.2e and f show the microstructure <strong>of</strong> the pearlitic steel 900A and the
bainitic steel Dobain 430 after a deformation <strong>of</strong> ɛeq= 16. Even at low strains the relatively brittle
carbides are severely deformed, and some lamellae are broken. Favorable oriented lamellae
already begin to align parallel to the shear plane. After a strain <strong>of</strong> approximately ɛeq= 4, the
pearlite colonies are no longer recognizable. The same is true for the sheaves in case <strong>of</strong> the
bainitic steel.
When the fine pearlitic 350 LHT is compared to the coarse spaced 900A, it seems that in the
later the fragmentation at an early stage <strong>of</strong> deformation is more pronounced. Due to the small
size <strong>of</strong> the carbides in the bainitic steel, a fragmentation <strong>of</strong> these carbides is less frequent. At
higher strains, most <strong>of</strong> the lamellae are broken and the fragments are clearly oriented parallel
to the shear plane. A general decreasing <strong>of</strong> the lamellae spacing with increasing strain can be
observed. In some areas this is excessively pronounced. While in the samples deformed to low
strains the etchant has no effect on the carbides and the microstructure is easily made visible,
in the severely deformed samples also the carbides are affected by the etching.
The TEM micrographs in Figure E.3 reveal that in some areas a lamellar structure is present
with a spacing that is markedly smaller than 20 - 30 nm. In addition a pronounced thinning <strong>of</strong>
the lamellae is observed. This occurs in all investigated materials. Selected Area Diffraction
Patterns (SAD) show that the ferrite is fragmented into very small structural elements. This
is confirmed by darkfield micrographs that show that the size <strong>of</strong> the structural elements <strong>of</strong> the
ferritic matrix is determined by the spacing <strong>of</strong> the lamellar structure in one direction.
The influence <strong>of</strong> the strain path can be seen in Figure E.4. The SEM micrographs show the
E–4

E.3 Results
Figure E.2: SEM micrographs <strong>of</strong> deformed samples (a) LHT 350, ɛeq = 1, tangential direction, (b) LHT 350,
ɛeq = 2 , tangential direction, (c) LHT 350, ɛeq = 4, tangential direction, (d) LHT 350, ɛeq = 8, tangential
direction, (e) 900A,ɛeq = 16, radial direction, (f) Dobain 430, ɛeq = 16, radial direction.
microstructure <strong>of</strong> the steel 900A after a total cyclic deformation <strong>of</strong> ɛeq= 64 for a ∆ɛeq= 0.5
and 4, respectively. A significant effect <strong>of</strong> the deformation per cycle is clearly visible. The
E–5
E

E
E High Pressure Torsion <strong>of</strong> Rail Steels
Figure E.3: TEM micrographs (a) 900A, ɛeq = 8, bright field image, (b) 350 LHT ɛeq = 8 bright field image, (c)
350 LHT, ɛeq = 8, dark field image.
main feature <strong>of</strong> the samples with a small ∆ɛeq is the fragmentation <strong>of</strong> the cementite lamellae
that are more or less aligned parallel to the shear plane. The size <strong>of</strong> these fragments is in most
cases smaller than 1 µm, and the distance between the fragments has also decreased markedly
compared to the lamellae spacing <strong>of</strong> the undeformed material. Although the total deformation
in this case is larger than the largest deformation reached with monotonic deformation, no
affect <strong>of</strong> the etchant on the carbides is visible. On the other hand, the microstructure <strong>of</strong> the
sample with a high deformation per cycle is comparable to the monotonic deformed sample,
but again the total deformation is much higher. The alignment to the shear plane is not as
pronounced due to cyclic deformation. Apparently in both cases the total strain, after which
the original pearlite colonies are no longer recognizable, is by far higher than in the monotonic
deformed samples.
Figure E.4: SEM micrographs <strong>of</strong> cyclic deformed 900A in radial direction, (a) ∆ɛ = 0.5, ɛeq = 64, (b) ∆ɛ = 4,
ɛeq = 64.
E–6

E.3.2 Mechanical strengths
E.4 Discussion
The in-situ measured flow stresses for the three steels are shown in Figure E.5a. A pronounced
increase <strong>of</strong> the mechanical strengths in the steels according their initial strength is present and
for the investigated deformations no indication <strong>of</strong> a saturation <strong>of</strong> the strengthening can be
assumed. Figure E.5b and c show the in-situ measured flow stress for the steel 900A for
monotonic HPT and CHPT and the results <strong>of</strong> the microhardness measurements afterwards,
respectively, for different ∆ɛeq per cycle. It can be seen that the needed total strain to reach
the same strengths for CHPT is much higher than for HPT and there is a strong influence <strong>of</strong>
the strain per cycle.
Figure E.5: Results <strong>of</strong> the measurements <strong>of</strong> the mechanical strength (a) in-situ measurement <strong>of</strong> the flow stress in
monotonic HPT for all three steel8s, (b) comparison <strong>of</strong> the flow stress for monotonic and cyclic HPT for the steel
900A, (c) comparison <strong>of</strong> the microhardness for monotonic and cyclic HPT for the steel 900A.
E.4 Discussion
The evolution <strong>of</strong> the microstructure <strong>of</strong> the rail steel 900A deformed by High Pressure Torsion
has been described previously. 6 After very large strain, the resulting microstructure <strong>of</strong> the
rail steel 350 LHT is very similar to the rail steel 900A, although at low strains it seems
that the fragmentation <strong>of</strong> the cementite lamellae is not as pronounced, the smaller lamellae
thickness permits a better deformation <strong>of</strong> the cementite. 7 The results are very similar to those
<strong>of</strong> cold drawn pearlitic steel wires. 8–10 In a previous paper 5 it has been shown that the insitu
measured flow stress does not give absolute values due to friction on the tool, however,
E–7
E

E
E High Pressure Torsion <strong>of</strong> Rail Steels
the changes in the flow stress also reflects changes in tensile strength. The measured torque
gives approximately a triplication <strong>of</strong> the tensile strength in the investigated regime. This is
also supported by the microhardness measurement and is in agreement with values for drawn
wires. For these wires it has been shown by means <strong>of</strong> tomographic atom probe or Mössbauer
spectroscopy that at least a partly dissolution <strong>of</strong> the cementite has occurred. 9, 10 It is commonly
assumed that crossing dislocations drag carbon out <strong>of</strong> the carbide. This is possible because the
binding energy <strong>of</strong> the carbon to the dislocation is larger than the binding energy <strong>of</strong> the carbon
atom in the carbide. 10 This gives rise to the assumption that such a dissolution also takes
place in the monotonic deformed samples. For a dislocation to cross a cementite lamellae,
a high stress is necessary. Hence, the lamellae will act as a barrier for the dislocations and
they pile up at the interface, if the strain is not high enough. If the strain is below a certain
threshold to overcome this barrier, no dissolution will occur, although cementite lamellae can
be bent, they can break and align to favorable orientations. When the fragments are finally
small enough, they will be moved in the s<strong>of</strong>t matrix without much further deformation. When
the strain, i.e. the dislocation density, is large enough, carbon atoms are dragged into the
matrix which leads to an additional increase <strong>of</strong> hardness due to solid solution hardening. As a
consequence, the plastic deformation <strong>of</strong> the lamellae will be enhanced because the difference
in the mechanical strength between the ferrite matrix and the carbide decreases. Therefore
the microstructure <strong>of</strong> Figure E.2e and E.4b are quite similar. The evolution <strong>of</strong> the mechanical
strength is governed by three components: the decreasing lamellae spacing, the increasing
dislocation density and the assumed solution <strong>of</strong> carbon atoms in the ferrite. This is most
pronounced for the monotonic deformed samples and explains the increase <strong>of</strong> the strength in
HPT for the investigated steels. Of course, the difference <strong>of</strong> the lamellae spacing between the
samples with different deformation paths will have a large influence for the difference in the
mechanical strength. Another reason for different hardening in HPT and CHPT may be the
unequal amount <strong>of</strong> solved carbon atoms in the ferrite matrix <strong>of</strong> the monotonic deformed sample
compared to the cyclically deformed ones. One may assume that at the same total deformation,
the same amount <strong>of</strong> dislocations are involved; hence the dislocation density should be the
same. But if the deformation is reversed, some <strong>of</strong> the generated dislocations will be able
to annihilate in the former source. This mechanism should be more efficient at small ∆ɛeq;
therefore the increase <strong>of</strong> the dislocation density will be the larger at larger ∆ɛeq.
E.5 Conclusions
E–8
• The monotonic shear deformation <strong>of</strong> the investigated rail steels is dominated by the carbides.
Depending on the size and distribution <strong>of</strong> these carbides, fragmentation, alignment
and decrease <strong>of</strong> the spacing are the main features <strong>of</strong> the deformation.
• The CHPT experiments clearly showed a significant influence <strong>of</strong> the strain path on the
resulting microstructure and the mechanical strength.
• The trend <strong>of</strong> the mechanical strength for HPT and CHPT deformed samples can be explained
by a combination <strong>of</strong> the decreasing lamellae spacing, the increasing dislocation
density and the solution <strong>of</strong> carbon atoms.

Bibliography to paper E
[1] L. Wang, A. Pyzalla, W. Stadlbauer, and E. A. Werner. Mater. Sci. Eng.A, 359:31–43,
2003.
[2] G. Baumann, K. Knothe, and H. J. Fecht. Nanostructured Materials, 9:751–754, 1997.
[3] M. Furukawa, Z. Horita, M. Nemoto, and T. G. Langdon. Mater. Sci. Eng.A, 324(1-
2):82–89, 2002.
[4] V. V. Stolyarov, Y. T. Zhu, I. V. Alexandrov, T. C. Lowe, and R. Z. Valiev. Mater.Sci.
Eng.A, 299:59–67, 2001.
[5] F. Wetscher, A. Vorhauer, and R. Pippan. Mater. Sci. Eng.A, 410-411:213–216, 2005.
[6] F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan. Mater. Sci. Eng.A, 387-389:809–816,
2004.
[7] W. J. Nam, C. M. Bae, S. J. Oh, and S.-J. Kwon. Scripta Mater., 42:457–463, 2000.
[8] M. Zelin. Acta Mater., 50:4431–4447, 2002.
[9] J. Languillaume, G. Kapelski, and B. Baudelet. Materials Letters, 33:241–245, 1997.
[10] F. Danoix, D. Julien, X. Sauvage, and J. Copreaux. Mater. Sci. Eng.A, 250:8–13, 1998.
E–9
E

E
E–10

F
TEM Investigation <strong>of</strong> the Structural
Evolution in a Pearlitic Steel Deformed by
High Pressure Torsion
F. Wetscher 1,2 , R. Pippan 1,2 , S. Sturm 3 , F. Kauffmann 4 , C. Scheu 4,5 , G. Dehm 1,6
1 Erich Schmid Institute for Materials Science, Austrian Academy <strong>of</strong> Sciences, Leoben,
Austria
2 CD-Laboratory for Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture, Leoben, Austria
3 Max Planck Institute for Materials Research, Stuttgart, Germany
4 Materialprüfungsanstalt, University <strong>of</strong> Stuttgart, Germany
5 Department Physical Metallurgy and Materials Testing, University <strong>of</strong> Leoben, Austria
6 Department Materials Physics, University <strong>of</strong> Leoben, Austria
Abstract
A fully pearlitic steel was deformed by high pressure torsion up to very high strains and
the changes in the microstructure were determined by analytical and conventional transmission
electron microscopy. The imposed strain leads to a fragmentation and an alignment <strong>of</strong>
the cementite lamellae parallel to the shear plane. The electron energy-loss near-edge-fine
structures <strong>of</strong> the Fe-L2,3-edge <strong>of</strong> the iron matrix and the cementite lamellae were measured
with high spatial resolution. The results indicated that after high pressure torsion the iron
matrix contains finely dispersed carbon-rich areas that do not show the electronic fingerprint
<strong>of</strong> cementite. However, the refinement in microstructure leads to an enormous increase in
mechanical strength.
F–1
F

F
F–2

F.1 Introduction
F.1 Introduction
Cold deformation <strong>of</strong> pearlitic steels occurs in many technical applications, for instance in
wire drawing or on the surface <strong>of</strong> rails during service. 1, 2 Especially in the last example it is
important to relate the deformation and the resulting microstructure to changes <strong>of</strong> mechanical
properties and damage processes. A key feature is therefore the behaviour <strong>of</strong> the cementite
lamellae (Fe3C) under shear deformation.
In most cases, methods <strong>of</strong> severe plastic deformation are used as a new route to produce
ultrafine grained materials with improved mechanical properties. 3–5 In the present study, high
pressure torsion (HPT) is used to obtain clearly defined deformed specimens <strong>of</strong> a low alloyed
steel with a initially pearlitic microstructure. The microstructural evolution was analysed by
different transmission electron microscopy (TEM) techniques. The aim <strong>of</strong> the study is to
characterize the microstructure, especially <strong>of</strong> the cementite lamellae, resulting from the severe
plastic shear deformation.
F.2 Experimental Details and Material
A pearlitic rail steel 260 (UIC 900A) was deformed by high pressure torsion to obtain severely
deformed material. The chemical composition and the mechanical properties <strong>of</strong> the material is
given in Tab. 1, details <strong>of</strong> the HPT technique are described elsewhere. 6, 7 The samples for HPT
had a diameter <strong>of</strong> 8 mm and a thickness t <strong>of</strong> 0.8 mm (0.7 after the deformation). These discs
where deformed at room temperature under a hydrostatic pressure <strong>of</strong> 5.7 GPa. The number <strong>of</strong>
turns n was calculated according to Equation F.1 to reach equivalent von Mises strains ɛeq <strong>of</strong>
2 and 8 at a radius r <strong>of</strong> 3 mm.
ɛeq = 2πrn
t √ 3
C Si Mn Cr P S Rm [MPa]
0,76 0.35 1 0.014 0.017 0,04 900
Table F.1: Table 1 Chemical composition <strong>of</strong> the pearlitic steel, the balance is Fe [wt. %].
(F.1)
TEM specimens from the deformed samples as well as from undeformed material were
prepared by cutting several millimeter small sections <strong>of</strong> the deformed material parallel to the
torsion axis at a radius <strong>of</strong> 3 mm. The material was mechanically polished to a final thickness <strong>of</strong>
≈50 µm and further reduced in size in order to keep its magnetic volume as small as possible,
see Figure F.1. Finally, several electron transparent regions <strong>of</strong> ≈10 x 10 µm 2 were made using
a Zeiss XP 1540 focussed ion beam microscope. The TEM samples were then glued on a
copper grid and subsequently analysed with a JEOL 2000 FX equipped with a Gatan imaging
filter and a VG HB 501 UX scanning transmission electron microscope. The VG HB 501UX
has a cold-field-emission gun and is equipped with a Gatan Enfina System.
On the JEOL 2000 FX, bright field micrographs and elemental distribution maps for carbon
were taken at 200 kV. The carbon maps were measured by the three-window method. 8, 9 The
F–3
F

F
F TEM Investigation <strong>of</strong> the Structural Evolution in a Pearlitic Steel Deformed by High Pressure Torsion
Figure F.1: A section <strong>of</strong> the HPT-sample was cut out from the deformed disk for TEM investigations. Electron
transparent windows were thinned into the specimen using a focused ion beam microscope.
energy slit width was 20 eV and an acquisition time <strong>of</strong> 10 s was used for each image. With the
VG HB 501 UX electron energy-loss (EEL) spectroscopy measurements <strong>of</strong> the Fe-L2,3 edge
using the ENFINA system were performed at 100 kV. The EEL spectra were recorded with a
dispersion <strong>of</strong> 0.1 eV and an acquisition time <strong>of</strong> 20 s. The full width at half maximum (FWHM)
<strong>of</strong> the zero loss was smaller than 0.8 eV. The electron energy-loss near-edge-fine structure
(ELNES) <strong>of</strong> the Fe-L2,3 edge was evaluated by fitting the measured spectra after background
subtraction with a Gaussian function (see Appendix) and deriving values for the peak areas,
peak heights and FWHM for the Fe-L2,3 edge <strong>of</strong> the matrix and the carbide phase. For the
undeformed sample and the sample deformed up to ɛeq = 2 the beam was centred exactly on
the well defined cementite lamellae. Hence, it is possible to distinguish between measurements
in the cementite and in the ferritic phase. In the case <strong>of</strong> the specimen deformed up to ɛeq = 8,
it was not possible to determine whether the electron spot was on a cementite lamellae or
not. Therefore, linescans with a length <strong>of</strong> approximately 15 nm and a step size smaller than 1
nm where performed over a line that was crossing a deformed cementite lamellae. The beam
diameter was nominally 1 nm for these measurements.
F.3 Results
F.3.1 Microstructure and Elemental Maps
The initial structure <strong>of</strong> the material consists <strong>of</strong> 10-20 µm large pearlite colonies with ferrite
lamellae ≈300 nm and cementite lamellae ≈ 20 nm wide, as can be seen in a scanning electron
micrograph (Fig 2a). After a deformation by HPT <strong>of</strong> ɛeq = 8, pearlite colonies are no
longer visible in scanning electron micrographs and a quite uniform lamellar structure with a
decreasing lamellae spacing <strong>of</strong> deformed and broken cementite lamellae and ferrite is present,
F–4

F.4 Discussion
see Figure F.2b. The cementite fragments have predominately the form <strong>of</strong> plates with a maximum
diameter <strong>of</strong> 1 µm. 10 All the fragments are now aligned parallel to the shear plane.
This change <strong>of</strong> the microstructure does also lead to an enormous increase in the mechanical
strength as can be seen in Figure F.3 in terms <strong>of</strong> the in-situ measured torque to deform the sample
and was already reported elsewhere. 10 At high deformations, the use <strong>of</strong> TEM is necessary
to clearly resolve the microstructure. In Figure F.4 a series <strong>of</strong> bright-field TEM micrographs
and the corresponding element maps for carbon is depicted showing the development <strong>of</strong> the
microstructure as a function <strong>of</strong> the strain. In the undeformed sample the cementite lamellae
with a width <strong>of</strong> about 20 nm are straight and continuous (Figure F.4a and d). Figure F.4b and c
show a bright-field image and the corresponding carbon elemental map <strong>of</strong> a sample deformed
to ɛeq = 2. The cementite lamellae, which start to break up, are clearly resolved in the C-K
map. At this degree <strong>of</strong> deformation the crystallographic cementite structure is still preserved
as found by selected area diffraction. Also in the case <strong>of</strong> the highest deformed sample the
lamellar structure that can be seen in bright field images correspond to carbon enriched areas
in the elemental map, Figures F.4c and f. In this sample, the lamellae spacing as well as the
lamellae thickness are markedly decreased down to a minimum value <strong>of</strong> approximately 25 nm
and 2 nm, respectively.
F.3.2 ELNES Measurements
Figure F.5 shows the measured EEL spectra <strong>of</strong> ferrite and cementite <strong>of</strong> the initial microstructure.
A change in the Fe-L2,3 area-ratio from 3.24 ±0.1 for α-iron to 2.82 ±0.12 for cementite
as well as in the FWHM for both peaks is observed for the cementite phase compared to ferrite.
The values for the Fe-L2,3 peak ratio, the Fe-L2,3 area ratio, the Fe-L2 FWHM and the Fe-L3
FWHM measured from the undeformed samples where used as a fingerprint and compared to
the results <strong>of</strong> the deformed samples. In Figure F.6 all data for the area ratio and the FWHMs
are summarized. The mean values as well as the largest and smallest values obtained from
measurements <strong>of</strong> the initial microstructure are indicated by black and grey lines. It is obvious
that only the initial structure <strong>of</strong> the material and the sample deformed up to ɛeq = 2 reveal the
characteristic values for the cementite phase. The values from the samples with a deformation
<strong>of</strong> ɛeq = 8 are all in the range <strong>of</strong> the ferritic phase.
F.4 Discussion
The microstructures as can be seen in the bright field micrographs are very similar to the microstructure
resulting from wire drawing <strong>of</strong> pearlitic steels reported by Embury and Fisher 11
and Langford. 12 The lamellae spacing λ in the undeformed microstructure was about 300 nm
and decreases in the sample deformed to eq = 8 to about 25 nm. This is consistent with the
calculated value for ≈23 nm (for n = 0.55) due to the deformation according to Equation F.2
which indicates that the calculated macrostrain according to Equation F.1 is homogeneously
distributed in the microstructure. The larger spacing in some areas can be explained by inhomogeneous
or deck-<strong>of</strong>-card deformation. 13 The increase <strong>of</strong> mechanical strength in terms <strong>of</strong>
the macroscopic tensile stress resulting from the decreasing lamellae spacing can be estimated
F–5
F

F
F TEM Investigation <strong>of</strong> the Structural Evolution in a Pearlitic Steel Deformed by High Pressure Torsion
Figure F.2: Scanning electron microscopy micrographs (using secondary electrons) <strong>of</strong> (a) the undeformed material
and (b) a sample after a deformation <strong>of</strong> ɛeq = 8, in radial direction.
F–6

Figure F.3: In-situ measured torque during the HPT-deformation.
F.4 Discussion
Figure F.4: TEM bright-field micrographs <strong>of</strong> (a) the undeformed sample, ɛeq = 0, (b) the sample after a deformation
<strong>of</strong> ɛeq = 2, and (c) the sample after deformation <strong>of</strong> ɛeq = 8. Corresponding carbon maps <strong>of</strong> the samples with
(d) ɛeq = 0, (e) ɛeq = 2and (f) ɛeq = 8. See text for details.
F–7
F

F
F TEM Investigation <strong>of</strong> the Structural Evolution in a Pearlitic Steel Deformed by High Pressure Torsion
Figure F.5: Results for undeformed samples: (a) measured EEL spectra <strong>of</strong> ferrite and cementite. (b) Area ratios
and peak height ratios for several measurements <strong>of</strong> ferrite and cementite. (c) FWHM for the Fe-L3 and the Fe-L2
peaks obtained in ferrite and cementite.
according to Dollar et al. 14 by Equation F.3, where λ is the interlamellar spacing given in mm:
λnew = λt
2rπn
σE = 12.74
λ 0.5
(F.2)
(F.3)
For λ ≈= 25nm the expected tensile stress is ≈2300 MPa, which is more than twice the
initial strength <strong>of</strong> ≈900 MPa. This increase is in good agreement with the measured increase
in the microhardness [10] from 2 GPa for the undeformed sample to 4.2 GPa for a sample
deformed to ɛeq = 8 and the measured increase <strong>of</strong> the torque during the deformation (Figure
F.3).
The elemental distribution maps <strong>of</strong> carbon for the samples with different deformation clearly
show that the carbon is concentrated in the cementite lamellae. Hence, they confirm that
the observed lamellar structures in the highest deformed samples are remains <strong>of</strong> the original
pearlitic structure and the carbon is still concentrated near the remains <strong>of</strong> the cementite
lamellae.
The ELNES-measurements <strong>of</strong> the undeformed material demonstrate that the differences <strong>of</strong>
the Fe-L2,3 fine structure between the ferrite phase and the cementite phase in terms <strong>of</strong> the Fe-
L2,3 peak ratio, the Fe-L2,3 area ratio and the FWHMs are significant and allow to postulate that
a change in the chemical structure has occured. The reason for that can be seen in the different
chemical composition and consequently the different electronic structure. 15, 16 No differences
were detectable in the distance between the two peaks (measured both the peak-peak distance
as well as distance between the points <strong>of</strong> inflection) indicating a similar spin-orbit splitting <strong>of</strong>
the 2p states for ferrite and cementite. 17 The results for the sample deformed to ɛeq = 2 are
F–8

F.4 Discussion
Figure F.6: Comparison <strong>of</strong> the EEL spectroscopy measurements <strong>of</strong> the starting material and samples with ɛeq = 2
and ɛeq = 8(a) area ratios, (b) FWHM for the Fe-L3 peaks and (c) FWHM for the Fe-L2 peaks.
F–9
F

F
F TEM Investigation <strong>of</strong> the Structural Evolution in a Pearlitic Steel Deformed by High Pressure Torsion
quite similar to the undeformed material. No significant deviation for the Fe-L2,3 values from
ferrite and cementite were observed and both phases are well distinguishable. Therefore, it can
be concluded that there was no significant change <strong>of</strong> the chemical composition and structure
in the cementite due to a deformation <strong>of</strong> ɛeq = 2. In contrast, the electronic fingerprint for the
highest deformed sample (ɛeq = 8) lies in the range <strong>of</strong> the values for the ferritic phase. This
means that even in the areas where the EFTEM micrographs showed an enrichment <strong>of</strong> carbon,
this carbon is no longer present as a Fe3C carbide. Hence, at least in the measured areas, the
condition <strong>of</strong> the cementite has changed markedly due to the severe plastic deformation by HPT
at room temperature.
Similar changes <strong>of</strong> cementite are especially known for cold-drawn pearlitic wires 18–22 and
ball-milled pearlitic powders. 23 It is generally assumed that the cementite dissolves as a result
<strong>of</strong> the plastic deformation. There is less agreement about the whereabout <strong>of</strong> the carbon atoms
after the dissolution. A saturation <strong>of</strong> the carbon atom at dislocations, 22 grain boundaries 23 or
in a fine-grained martensite 20 are discussed in the literature. The methods used in the literature
comprised Mössbauer spectroscopy, 22 atom probe field ion microscopy, 19 three dimensional
atom probe methods 21 and thermomagnetic methods. 24 The most commonly used explanation
for a dissolution <strong>of</strong> cementite due to deformation is that carbon atoms are dragged out <strong>of</strong> the
carbide by crossing dislocations. This might happen because the binding energy <strong>of</strong> the carbon
atom to the dislocation is greater than the binding energy <strong>of</strong> the carbon atom in the carbide.
A more detailed explanation <strong>of</strong> this process is given by Gavriljuk, 25 for instance. Ivanisenko
et al. 24 reported a 40% decomposition <strong>of</strong> cementite due to deformation by HPT for a similar
steel as used in this study after ɛeq = 36 and a total dissolution after ɛeq = 245 measured by
a thermomagnetic method. In the present study a decomposition has been observed already at
much lower strains. The reason for this might be the different experimental setup, especially
the large shape change that is avoided in our HPT-tool which permits a defined shearing <strong>of</strong> the
samples. Another reason might be the different techniques to investigate the actual state <strong>of</strong> the
carbon. In our case we have used a technique with high spatial resolution enabling electronic
structure measurements down to the nanometer level, while the observations <strong>of</strong> 24 are mainly
based on integral techniques. The results form Sauvage 26 obtained by 3D-atom probe <strong>of</strong> HPT
deformed pearlitic steel are in good agreement with the present observations. He measured a
decrease <strong>of</strong> the initial carbon concentration <strong>of</strong> 25 at.% <strong>of</strong> the cementite down to 2 - 10 at. %
after 5 turns <strong>of</strong> HPT, observing similar carbon-rich lamellae instead <strong>of</strong> the cementite present
in the initial microstructure.
F.5 Conclusion
F–10
• In the current TEM study a markedly decrease <strong>of</strong> both the lamellae spacing and the
lamellae thickness in the pearlitic steel due to severe plastic deformation by High Pressure
Torsion was observed.
• In the bright field micrographs <strong>of</strong> the severely deformed sample a lamellar structure was
still observable and the elemental distribution maps showed an enrichment <strong>of</strong> carbon in
these thin lamellae.

F.6 Appendix
• Electron energy-loss near-edge-fine structure measurements were successfully performed
at the samples and analysed by using exponentially modified Gaussian functions. With
this method reliable ratios and peak width were obtained which act as fingerprints <strong>of</strong> the
iron atoms.
• The electron energy-loss near-edge-fine structure measurements <strong>of</strong> the carbon-rich areas
in the most severe deformed sample did not show the characteristics <strong>of</strong> cementite indicating
that cementite is no longer in these areas. This leads to the conclusion that due
to the severe plastic deformation by high pressure torsion at least a partial dissolution <strong>of</strong>
the carbon from the cementite has occurred.
F.6 Appendix
F.6.1 Fitting <strong>of</strong> the EELS-spectra
The measured EELS-spectra consist <strong>of</strong> element specific ionisation edges superimposed on a
background which is mainly arising from plural-scattering events <strong>of</strong> outer shell electrons. 8, 17
This background has to be removed to analyse the edges. For background subtraction, a powerlaw
approximation:
Ibackground = AE −r
(F.4)
was used. 17 The remaining signal was then analyzed with the s<strong>of</strong>tware Mathematica from
Wolfram Research. For the investigated Fe-L2,3 edge transitions occur from the 2p states into
unoccupied 3d and 4s states. 17 The transition into the narrow unfilled 3d band leads to white
lines. These structures are superimposed on a smooth background which is due to transitions
into the extended 4s bands. To describe the white lines as a function <strong>of</strong> the energy E, two
exponentially modified Gaussian functions:
Iwhiteline = B0
e
2B3
B 2 2
2B 2 3
were used. The parameters <strong>of</strong> this function are:
B0 = Area <strong>of</strong> the deconvolved Gaussian
B1 = Centre <strong>of</strong> the deconvolved Gaussian
B2 = Width
B3 = Distortion
+ B1−E
B E − B1
3 Erf √ −
B2 2 B2
√ +
B3 2
B3
|B3|
(F.5)
This function is a mathematical convolution <strong>of</strong> an Gaussian and a exponential function. Due
to its capability to describe asymmetric peaks it is widely used in the field <strong>of</strong> chromatography.
An overview is given by Jeansonne and Foley. 27 The reason why this function was used in our
study is that it gave much better agreement with the measured white lines than the normally
used Lorentz-function. 28 The smooth background intensity in the threshold region results
from the transitions into the 4s band which is free-electron like. 17 Therefore, the density <strong>of</strong>
F–11
F

F
F TEM Investigation <strong>of</strong> the Structural Evolution in a Pearlitic Steel Deformed by High Pressure Torsion
Figure F.7: (a) Comparison between a measured spectrum (ferrite, ɛeq = 0) and the fit. (b) Plot <strong>of</strong> the individual
functions used for the fit, see Appendix for further details.
states is proportional to the square root <strong>of</strong> energy and the intensity was fitted with C and E0 as
parameters:
I4s−states = C E − E0
(F.6)
The actual fitting was done using a least-square method. After choosing reasonable starting
parameters, pairs <strong>of</strong> parameters (for example B0 for the Fe-L2 peak and for Fe-L3) were
calculated separately and used as constants for the next fitting step. After calculating all the
parameters, the whole procedure was repeated. A third repetition <strong>of</strong> this gave no significant
changes <strong>of</strong> the values for the parameters. In Figure F.6.1a a measured spectra is compared with
the final fit. The separate graphs for the fitted functions are displayed in Figure F.6.1b. From
the obtained functions, all other values like the peak height, the peak area and the FWHM
for both peaks as well as the corresponding ratios (Fe-L2,3 peak ratio, Fe-L2,3 area ratio) were
calculated; see Equation F.7 and Equation F.8, respectively.
peak high ratio = Maximum (IW hiteline,F eL3 )
Maximum (IW hiteline,F eL2 )
area ratio =
B1L3+10
IW hiteline,F eL3
B1L3−10
dE
B1L2+10
IW hiteline,F eL2
B1L2−10
dE
F.7 Acknowledgements
(F.7)
(F.8)
The authors want to thank Dr. P. Pointer and Mr. R. Stock from voestAlpine Schienen GmbH
for providing the material and their support <strong>of</strong> the CD-laboratory. The help <strong>of</strong> Mr. J. Thomas
with the STEM measurements is gratefully acknowledged.
F–12

[1] M. Zelin. Acta Mater., 50(17):4431–4447, 2002.
Bibliography to paper F
[2] G. Baumann, H. J. Fecht, and S. Liebelt. Wear, 191:133–140, 1996.
[3] R. Z. Valiev, R. K. Islamgaliev, and I.V. Alexandrov. Prog. Mat. Sci., 45:103–189, 2000.
[4] R. Z. Valiev and I.V. Alexandrov. Ann. Chim. - Sci. Mat., 27:3–14, 2002.
[5] V. Y. Gertsman, R. Birringer, R. Z. Valiev, and H. Gleiter. Scripta Metall., 30:229–234,
1994.
[6] T. Hebesberger, H. P. Stüwe, A. Vorhauer, F. Wetscher, and R. Pippan. Acta Mater.,
53:393–402, 2005.
[7] O. Dimitrov. Ann. Chim. -Sci. Mat., 27:15–24, 2002.
[8] D. B. Williams and C. B. Carter. Transmission Electron Microscopy: A Textbook for
Material Science. Plenum Press, New York, 1996.
[9] F. H<strong>of</strong>er, W. Grogger, G. Kothleitner, and P. Wasbichler. Ultramicroscopy, 67:83–103,
1997.
[10] F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan. Mater. Sci. Eng. A, 387-389:809–816,
2004.
[11] J. D. Embury and R. M. Fisher. Acta Met., 14:147–159, 1966.
[12] G. Langford. Met. Trans., 1:465–477, 1970.
[13] G. Langford. Met. Trans., 8A:861–875, 1977.
[14] M. Dollar, I. M. Bernstein, and A. W. Thompson. Acta Metall., 36:311–320, 1988.
[15] R. D. Leapman and L. A. Grunes. Phys. Rev. Lett., 45:397–401, 1980.
[16] R. D. Leapman, L. A. Grunes, and P. L. Fejes. Phys. Rev. B., 26:614, 1982.
[17] R. F. Egerton. Electron Energy Loss Spectroscopy in the Electron Microscop, 2 nd ed.
Plenum Press, New York, 1996.
[18] J. Languillaume, G. Kapelski, and B. Baudelet. Acta Mater., 45:1201–1212, 1997.
BIB–13
BIB

F
Bibliography to paper F
[19] H. G. Read, W. T. Reynolds Jr., K. Hono, and T. Tarui. Scripta Mater., 37:1221–1230,
1997.
[20] K. Hono, M. Ohnuma, M. Murayama, S. Nishida, A. Yoshie, and T. Takahashi. Scripta
Mater., 44:977–983, 2001.
[21] F. Danoix, D. Julien, X. Sauvage, and J. Copreaux. Mater. Sci. Eng.A, 250:8–13, 1998.
[22] W. J. Nam, C. M. Bae, S. J. Oh, and S.-J. Kwon. Scripta Mater., 42:457–463, 2000.
[23] M. Umemoto. Mater. Trans., JIM, 44:1900–1911, 2003.
[24] Y. Ivanisenko, W. Lojkowski, R. Z. Valiev, and H.-J. Fecht. Acta Mater., 51:5555–5570,
2003.
[25] V. G. Gavriljuk. Mater. Sci. Eng.A, 345:81–89, 2002.
[26] X. Sauvage. Mater. Sci. Forum, 503-504:433–438, 2006.
[27] M. S. Jeansonne and J. P. Foley. J. Chromatogr. A, 594:1–8, 1992.
[28] T. Manoubi, M. Tence, M. G. Walls, and C. Colliex. Microsc. Microanal. Microstruct.,
1:23–39, 1990.
F–14

G
Fracture Processes in Severe Plastic
Deformed Rail Steels
F. Wetscher 1,2 , R. Stock 3 and R. Pippan 1,2
1 Erich Schmid Institute for Materials Science, Austrian Academy <strong>of</strong> Sciences, Leoben,
Austria
2 CD-Laboratory for Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture, Leoben, Austria
G.1 Abstract
3 voestAlpine Schienen GmbH, Leoben, Austria
A fully pearlitic steel used as a material for railway tracks was deformed by equal channel angular
pressing. The shear deformation results in an alignment <strong>of</strong> the pearlite colonies and the
cementite lamellae. This formation <strong>of</strong> a preferred direction leads to an anisotropy in mechanical
properties, especially in the fracture toughness. To investigate this, two critical directions
were defined and compact tension samples <strong>of</strong> undeformed and deformed materials were tested.
With increasing strain, the fracture toughness decreases for samples where the crack can easily
cross the deformed colonies.
G–1
G

G
G–2

G.2 Introduction
G.2 Introduction
The investigation <strong>of</strong> the deformation and the resulting changes in mechanical properties <strong>of</strong>
pearlitic steels has a long history. 1–3 This is mainly because cold deformation <strong>of</strong> pearlitic
steels is a process that happens in many technical applications. For instance, in wire drawing
this leads to an enormous increase in mechanical strength. Another field is the formation <strong>of</strong> a
deformation layer on the surface <strong>of</strong> rails during service. For the development <strong>of</strong> new materials,
for a better scheduling <strong>of</strong> service intervals and for realistic simulations <strong>of</strong> wheel-rail contacts a
better knowledge <strong>of</strong> the influence <strong>of</strong> shear deformation on the mechanical properties <strong>of</strong> these
steels is necessary. Since this deformation layer is the origin <strong>of</strong> many defects like headchecks
and ratcheting, 4, 5 especially the fracture properties are <strong>of</strong> great importance. In the last year,
many techniques were developed to deform materials up to very high strains under controlled
conditions. The most useful <strong>of</strong> these so-called severe plastic deformation (SPD) methods are
equal channel angular pressing (ECAP) 6, 7 and high pressure torsion. 8, 9 Especially with ECAP
it is possible to obtain samples with a known deformation history and precisely known strains
that are also large enough to determine fracture mechanics properties. The aim <strong>of</strong> the present
study is to determine the fracture toughness <strong>of</strong> a deformed pearlitic steel as a function <strong>of</strong> the
deformation in different directions and to investigate the influence <strong>of</strong> the microstructure as a
function <strong>of</strong> the applied shear strain.
G.3 Experimental
The material used in this study is the pearlitic rail steel 900 UIC, the composition <strong>of</strong> this steel
is 0.76 wt.% C, 0.35 wt.% Si, 1 wt.% Mn, 0.14 wt.% S and 0.17 wt.% P, the rest is Fe. To
obtain material with a defined pre-deformation, ECAP was applied. The samples for ECAP
had a size <strong>of</strong> 10 x 10 x 50 mm and were pressed through the die at room temperature. The
pressing speed was approximately 0.5 mm/s. The used tool has an intersection angle Φ <strong>of</strong>
120°; therefore the strain per path according to Equations G.1 10 was 0.67.
ɛ = 2
√ 3 cot Φ
2
(G.1)
Samples were pressed 1, 2 and 3 times through the tool using route A 7 resulting in strains
<strong>of</strong> 0.67, 1.34 and 2.01. From the deformed as well as the undeformed samples, compacttension
(CT) specimens were machined. Two different geometries were chosen in respect
<strong>of</strong> the ECAP-sample geometry, see Figure G.3b. In the following, samples after N passes
<strong>of</strong> ECAP will be referred as NA or NB. The samples were cut with a diamond-wire saw to
prevent a rise in temperature and therefore recovery in the material. The samples were grinded
and polished to the final thickness. The CT-specimens had a width W = 8 mm and a thickness
B = 2 mm.
To create a sharp crack, a notch was cut into the samples with a diamond-wire saw and
further sharpened by polishing the notch ground with a razor blade. A fatigue crack was
produced under compression-compression loading with a ∆K <strong>of</strong> 30 MPa m and 10.000 to
20.000 load cycles.
G–3
G

G
G Fracture Processes in Severe Plastic Deformed Rail Steels
Figure G.1: (a) Sketch <strong>of</strong> the ECAP tool: The sample is pressed through intersecting channels by a plunger;
(b) Definition <strong>of</strong> the two sample orientations in respect to the ECAP sample and CT-specimens after the fracture
toughness test.
Fracture toughness tests were carried out on a testing machine from Kammrath and Weiss at
a constant cross-head speed <strong>of</strong> 2.5 µm/s. For each deformation grade and sample orientation,
three specimens were tested. The analysis <strong>of</strong> the fracture toughness was done according to the
standard E 399.
The microstructure as well as the fracture surface was investigated using a Zeiss 1425 scanning
electron microscope at 5 kV. For microstructural investigations, the samples were grinded,
polished and finally etched using pikrin acid. Micrographs were taken using secondary electrons.
G.4 Results
G.4.1 Microstructure and crack path
Figure G.4.1 is a series <strong>of</strong> micrographs showing the microstructure <strong>of</strong> the material and the
crack path for the different geometries. The as-received material consists <strong>of</strong> pearlite colonies
with a size <strong>of</strong> approximately 10µm, the average lamellae spacing is 300 nm. The colonies are
more or less randomly orientated and are almost equiaxed. After three passes <strong>of</strong> ECAP, most
<strong>of</strong> the pearlite colonies are severely deformed and favorable orientated lamellae are aligned
parallel to the pressing direction. Other lamellae are severely deformed and a fragmentation
starts. Figure G.4.1c depicts the crack path through an undeformed sample after a fracture
G–4

G.4 Results
toughness test. It can be seen that the crack propagates through colony boundaries as well as
directly through colonies, breaking up the lamellas.
In Figure G.4.1d, a part <strong>of</strong> the crack through a sample 3A can be seen. In all these micrographs,
the original notch is parallel to the baseline <strong>of</strong> the image. In this sample, the crack
runs along the elongated pearlite colonies and through the aligned lamellae even if this means
that the direction <strong>of</strong> the crack propagation is perpendicular to the original notch. The crack
propagation in a sample 3B is depicted in Figure G.4.1e and f. Also in this case, the crack
follows the deformed cementite colonies and the aligned lamellae.
G.4.2 Fracture toughness
The fracture toughness was evaluated from force-displacement curves. In Figure G.4.2a, typical
force-displacement curves for samples after 3 passes are shown. From these curves, PQ was
determined; in most cases PQ was equal to Pmax. The stress intensity factor KQ was calculated
using Equation 2 (the function f(a/W) was used according to the standard E399):
KQ = PQ
a
f
BW 0.5 W
(G.2)
The results for KQ can be seen in Figure G.4.2b. The highest fracture toughness was measured
for the undeformed material. The fracture toughness <strong>of</strong> the ECAP-deformed samples
with orientation B markedly decreases with the number <strong>of</strong> passes. Reasonable values <strong>of</strong> KQ
for deformed samples with orientation A could only be obtained after the first step. Samples
with two or three passes did not fracture in the direction <strong>of</strong> the notch. In Figure G.3b, samples
after the fracture test are compared. In the case <strong>of</strong> a sample 3B, the crack propagates straight
trough the CT-specimen. The crack in the sample 3A propagated perpendicular to this direction,
i.e. it propagates parallel to the loading direction. Therefore, a value <strong>of</strong> KQ perpendicular
to the aligned pearlite colonies could not be obtained. Due to the small sample size and the
quite high fracture toughness, the calculated KQ values are not valid KIC values except for the
samples 3B.
G.4.3 Fracture surface
The fracture surface <strong>of</strong> the undeformed sample is depicted in Figure G.4.3a. It can be seen
that the surface is very rough on a macroscopic level and shows a typical appearance <strong>of</strong> a
cleavage-like fracture. In contradiction to this, the fracture surface <strong>of</strong> the samples after three
ECAP passes is quite smooth as it is shown in Figure G.4.3b for the sample 3A. Figure G.4.3c
and d compare the fracture surface at a much higher magnification. Here, the differences
are not so obvious. On this level, the surface is dominated by large areas <strong>of</strong> cleavage-like
fracture surface, surrounded by a more ductile dimple fracture <strong>of</strong> the remaining section. The
occurrence <strong>of</strong> these areas with a dimple fracture is much more pronounced in the case <strong>of</strong> the
undeformed specimens. In all these micrographs, the macroscopic crack propagation direction
was from the left to the right side <strong>of</strong> the image.
G–5
G

G
G Fracture Processes in Severe Plastic Deformed Rail Steels
Figure G.2: SEM micrographs <strong>of</strong> (a) the as-received material, (b) sample after 3 ECAP passes, (c) as-received
sample after fracture toughness testing, (d) sample 3A after fracture toughness testing and (e) and (f) sample 3B
after fracture toughness testing.
G–6

G.5 Discussion
Figure G.3: (a) Load-Displacement curves for samples 3A and 3B, (b) Summary <strong>of</strong> the results from the fracture
toughness tests.
G.5 Discussion
G.5.1 Microstructure
The present results <strong>of</strong> the development <strong>of</strong> the microstructure due to the shear deformation
are in good agreement with previous studies <strong>of</strong> the same material deformed by high pressure
torsion. 11, 12 The cementite colonies become elongated; favorable orientated lamellae begin
to align parallel to the shear plane and others are severely deformed, see Figure G.4.1b. Although
by using Route A in ECAP the shear plane is changed between the passes, 6 the overall
effect is a simple shear in the pressing direction, leading to a similar deformation as in high
pressure torsion. ECAP <strong>of</strong> low-carbon steels was performed, for instance, by Fukuda et al. 13
and Kim et al. 14 In these studies, the influence <strong>of</strong> the shear deformation on the microstructure
and mechanical properties was investigated. Although the microstructure in these steels was
ferritic-pearlitic, the results concerning the pearlitic phase are comparable to the present findings.
A fully pearlitic steel deformed by ECAP was investigated by Wang et al. 15 They also
found a severe deformation <strong>of</strong> the lamellae, together with a spheroidization <strong>of</strong> the cementite
lamellae. This occurrence <strong>of</strong> globular cementite, which was not found in this investigation,
was obviously an effect <strong>of</strong> the elevated temperature <strong>of</strong> 923 K during ECAP. Together with the
change in the microstructure, also mechanical properties are changed. Due to the severe cold
working and a decrease in the lamellae spacing, an increase in the mechanical strength occurs.
From measurements during and after HPT an increase <strong>of</strong> the tensile strength from ≈ 900 MPa
to ≈ 1500 MPa can be expected. 16 Similar changes <strong>of</strong> the mechanical strength due to ECAP
are reported in other studies for low-carbon steels, see for example. 17 What is missing in all
these studies is that a strong anisotropy due to the alignment <strong>of</strong> the lamellae has to evolve,
especially for fracture properties.
G–7
G

G
G Fracture Processes in Severe Plastic Deformed Rail Steels
Figure G.4: SEM micrograph <strong>of</strong> the fracture surface <strong>of</strong> (a) and (c) as-received material and (b) and (d) sample
3A.
G–8

G.5.2 Fracture
G.5 Discussion
Fracture <strong>of</strong> pearlitic microstructures was investigated in detail in many studies, see, for instance.
18–21 Today, the model <strong>of</strong> shear cracking suggested by Miller and Smith 21 is widely
accepted. In this model, shear stresses cause a cracking <strong>of</strong> a cementite plates. This local
weakening promotes further shear deformation <strong>of</strong> the ferrite and the cracking <strong>of</strong> neighboring
plates. Pores grow at the cracks in the cementite and coalescent finally. This is exactly what
can be seen in Figure G.4.1c and especially in Figure G.5.2a for the undeformed sample after
the fracture toughness tests. In Figure G.5.2a additional cracks in the cementite are visible.
Above and below the actual crack, the initial state <strong>of</strong> the model for shear cracking can be observed.
A few neighboring lamellae are broken along a line and pores have formed. For the
undeformed sample, this seems to be the main mechanism <strong>of</strong> fracture. Due to the fact that
the pearlite colonies are more or less randomly orientated, the propagating crack locally <strong>of</strong>ten
changes the direction while retaining a global mode I propagation. This results in the highest
measured fracture toughness, see Figure G.4.2b, as well as in the observed very rough fracture
surface.
Figure G.5: SEM micrograph <strong>of</strong> the crack in (a) as-received sample and (b) sample 3A.
The fracture toughness experiments <strong>of</strong> the samples after three passes showed a distinctive
anisotropy. In the samples with orientation A, the crack was deflected by 90°as can be seen
in Figure G.3b. Therefore, the crack propagates in the same direction as in the sample with
orientation B. This explains why the fracture surface <strong>of</strong> both samples look very similar and
there are no obvious differences, as can be seen in Figure G.4.1d-f, although it has to noted
the fracture mode in both samples is different. In the sample 3B, the crack propagated due to a
pure mode I load while in the sample with orientation A a mixed mode I - mode II propagation
was present due to the crack deflection. A similar crack deflection was observed by Toribio
et al. 22–24 for notched tensile specimen <strong>of</strong> drawn pearlitic wires. They found that pearlitic
pseudocolonies may trigger the deflection <strong>of</strong> the crack. The occurrence <strong>of</strong> such pseudocolonies
is a function <strong>of</strong> the drawing strain. Such extreme slender colonies are also observed after
G–9
G

G
G Fracture Processes in Severe Plastic Deformed Rail Steels
the pure shear deformation by ECAP. However, in Figures 2c-2f and 5b, almost no evidence
<strong>of</strong> shear cracking for the deformed samples is present. It seems that the crack runs mainly
along the boundaries <strong>of</strong> the deformed lamellae or parallel to the aligned lamellae, avoiding
crossing cementite lamellae. After three passes <strong>of</strong> ECAP, the alignment <strong>of</strong> the microstructure
is pronounced enough that even a 90°deflection <strong>of</strong> the crack is more favorable than to follow
the highest stress in mode I propagation. The microstructure and not the stress field is now the
determining factor for crack propagation.
Due to the appearance <strong>of</strong> possible fracture passes with very low resistance against crack
propagation, the fracture toughness decreases markedly in the case <strong>of</strong> sample 3B.
G.6 Conclusions
G–10
• <strong>Deformation</strong> <strong>of</strong> a fully pearlitic rail steel by equal channel angular pressing at room temperature
leads to an alignment <strong>of</strong> the cementite colonies and <strong>of</strong> the cementite lamellae.
• This alignment leads to a markedly anisotropy in fracture properties. In the direction
parallel to the elongated colonies, the fracture toughness decreases with the increasing
number <strong>of</strong> ECAP passes.
• In the samples without pre-deformation, mainly shear cracking was observed. In the
samples after three ECAP passes, the crack propagated most notably along the borders
<strong>of</strong> the aligned colonies and between aligned cementite lamellae.

Bibliography to paper G
[1] J. D. Embury and R. M. Fisher. Acta Met., 14:147–159, 1966.
[2] G. Langford. Met. Trans., 1:465–477, 1970.
[3] G. Langford. Met. Trans., 8A:861–875, 1977.
[4] J. H. Beynon, A. Kapoor, and W. R. Tyfour. Wear, 197:255–265, 1996.
[5] U. Ol<strong>of</strong>sson and T. Telliskivi. Wear, 254:80–93, 2003.
[6] V. M. Segal. Mater. Sci. Eng.A, 197:154–157, 1995.
[7] M. Furukawa, Y. Iwahashi, Z. Horita, M. Nomoto, and T. G. Langdon. Mater. Sci. Eng.A,
257:328–332, 1998.
[8] H. Jiang, Y. T. Zhu, D. P. Butt, I. V. Alexandrov, and T. C. Lowe. Mater. Sci. Eng.A,
290:128–138, 2000.
[9] V. V. Stolyarov, Y. T. Zhu, T. C. Lowe, R. K. Islamgaliev, and R. Z. Valiev. Mater. Sci.
Eng.A, 282:78–85, 2000.
[10] O. Dimitrov. Annales de Chimie Science des Materiaux, 27:15–24, 2002.
[11] F. Wetscher, A. Vorhauer, R. Stock, and R. Pippan. Mater. Sci. Eng. A, 387-389:809–816,
2004.
[12] F. Wetscher, R.Stock, B.Tian, and R. Pippan. In Ultrafine Grained Materials III, pages
673–678. TMS, Warrendale (USA), 2004.
[13] Y. Fukuda, K. Oh-ishi, Z. Horita, and T. G. Langdon. Acta Mater., 50:1359–1368, 2002.
[14] J. Kim, I. Kim, and D. H. Shin. Scripta Mater., 45:421–426, 2001.
[15] J. T. Wang, J. Huang, Z. De, Z. Zhang, and X. Zhao. In Ultrafine Grained Materials III,
pages 673–678. TMS, Warrendale (USA), 2004.
[16] F. Wetscher, B. Tian, R. Stock, and R. Pippan. Mater. Sci. Forum, 503-504:455–460,
2006.
[17] J. T. Wang, C. Xu, Z. Z. Du, G. Z. Qu, and T. G. Langdon. Mater. Sci. Eng.A, 410-
411:312–315, 2005.
BIB–11
BIB

G
Bibliography to paper G
[18] J. M. Hyzak and I. M. Bernstein. Met. Trans., 7A:1217–1224, 1976.
[19] Y. J. Park and I. M. Bernstein. Met. Trans., 10A:1653–1664, 1979.
[20] J. J. Lewandowski and A. W. Thompson. Met. Trans., 17A:1769–1786, 1986.
[21] L. E. Miller and G. C. Smith. J.I.S.I., 208:998–1005, 1970.
[22] J. Toribio, E. Ovejero, and M. Toledano. Int. J. Fracture, 87:L83–L88, 1997.
[23] J. Toribio and J. Ayaso. Int. J. Fracture, 115:L29–L34, 2002.
[24] J. Toribio and J. Ayaso. Mater. Sci. Eng.A, 343:265–272, 2003.
G–12

H
Cyclic High Pressure Torsion <strong>of</strong>
Nickel and Armco Iron
F. Wetscher 1,2 and R. Pippan 1,2
1 Erich Schmid Institute for Materials Science, Austrian Academy <strong>of</strong> Sciences, Leoben,
Austria
2 CD-Laboratory for Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture, Leoben, Austria
Abstract
Cyclic high pressure torsion, a modified version <strong>of</strong> high pressure torsion, is applied to
Armco-iron and nickel. The results in terms <strong>of</strong> microstructure and flow stress are compared
to samples deformed by conventional high pressure torsion. For both processes and both materials,
a saturation in the decrease <strong>of</strong> the structure size and the increase in the flow stress is
observed. The minimum size <strong>of</strong> the structural elements which is obtainable is smallest for the
conventionally high pressure torsion deformed samples and increases with decreasing strain
per cycle in cyclic high pressure torsion.
H–1
H

H
H–2

H.1 Introduction
H.1 Introduction
In the last years numerous papers have proven the capability <strong>of</strong> the methods <strong>of</strong> severe plastic
deformation (SPD) to produce ultra fine grained and nanograined materials for new applications,
see for example. 1–3 Especially by using equal channel angular pressing (ECAP) with
the different routes it is possible to determine the influence <strong>of</strong> the strain path on grain refinement
and the mechanical properties. 4, 5 Nevertheless, the possibilities to vary the strain per
pass are very limited and do in most cases comprise a change in the geometry <strong>of</strong> the tool or
the application <strong>of</strong> composed passes as used in. 4 Furthermore, the total strain reachable by
ECAP is limited by practical considerations. The applying <strong>of</strong> extremely high strains is very
time-consuming for the incremental process <strong>of</strong> ECAP. The simplest method to reach extremely
high strains is high pressure torsion (HPT), 6–8 with the restriction that no change in the strain
path can be achieved easily.
Therefore, in this paper a cyclic form <strong>of</strong> HPT is introduced, cyclic high pressure torsion
(CHPT). Due to modifications <strong>of</strong> our HPT tool it is now possible to cyclically reverse the
deformation after a chosen time. The aim <strong>of</strong> this study is to evaluate the differences between
samples deformed by HPT and CHPT in terms <strong>of</strong> microstructure and mechanical strengths and
the influence <strong>of</strong> the strain per cycle over a wide range <strong>of</strong> this value.
H.2 Experimental
The materials used in this study are the bcc Armco iron and the fcc nickel (99.99%). Disks <strong>of</strong>
the materials with an original grain size <strong>of</strong> approximately 50µm were produced with a radius r
<strong>of</strong> 4 mm and a thickness t <strong>of</strong> 0.8 mm. These disks were deformed by HPT to a total equivalent
strain <strong>of</strong> 64 and CHTP to total equivalent shear strains ɛeq,tot <strong>of</strong> 4, 8, 32 and 64 at a radius r =
3 mm. The shear strain per cycle ∆ɛ was choosen to be 0.5, 1, 2 and 4 at a radius <strong>of</strong> r = 3 mm
for this investigation. For comparision, samples <strong>of</strong> nickel with ∆ɛ = 0.5 and 4 and ɛeq,tot=256
were also examined. The total equivalent shear strain and the strain per cycle are calculated
according to Equation H.1 and Equation H.2 where ϕ is the rotation angle and m is the number
<strong>of</strong> cycles ∗ .
∆ɛeq = ϕr
t √ 3
ɛeq,tot = m∆ɛeq
(H.1)
(H.2)
The hydrostatic pressure was kept constant for all experiments at a value <strong>of</strong> 5.7 GPa, the
number <strong>of</strong> turns per minute was 0.2. All deformation experiments were carried out at room
temperature, no significant heating <strong>of</strong> the samples due to the deformation did occur. To evaluate
the changes <strong>of</strong> the flow stress, the torque was measured in-situ by means <strong>of</strong> strain gauges.
From this measured torque, an upper value for the flow stress can be calculated. 9, 10 A sketch
<strong>of</strong> our HPT-tool can be seen in Figure H.1. The microstructure was investigated with a Zeiss
∗ one cycle means a deformation in one direction, similar to one pass in ECAP, it corresponds to 1
2
conventional fatigue
cycle in
H–3
H

H
H Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
Figure H.1: Sketch <strong>of</strong> the HPT-tool.
1525 scanning electron microscope (SEM) at 20kV using backscattered electrons (BSE). Orientation
image maps were measured for nickel samples deformed with a ∆ɛ <strong>of</strong> 0.5 and 4 by
analyzing the Kikuchi patterns caused by back scattered electrons (EBSD). The sample preparation
comprised consecutive grinding, polishing, etching and a final polishing step. Special
care was taken that the samples were not exposed to elevated temperatures during this process
to prevent recrystallisation. All micrographs were taken in radial direction as can be seen in
Figure H.2.
H.3 Results
H.3.1 Flow stress
A typical measurement <strong>of</strong> the torque which can be related to the flow stress during CHPT with
a ∆ɛ = 1 <strong>of</strong> Armco iron as a function <strong>of</strong> time which is proportional to the accumulated strain
is presented in Figure H.3a. The negative values <strong>of</strong> the flow stress correspond to an arbitrary
definition <strong>of</strong> a positive and negative direction <strong>of</strong> the deformation. At the first few cycles, the
increase <strong>of</strong> the flow stress would be almost identical to the flow curve <strong>of</strong> monotonic deformed
samples when absolute values are taken. In both materials a pronounced work hardening at
the first few cycles is present, with an increasing number <strong>of</strong> cycles the increase <strong>of</strong> the absolute
value <strong>of</strong> the maximum <strong>of</strong> the flow stress becomes very low. With higher total strains the
H–4

Figure H.2: Sample preparation for microscopy.
H.3 Results
cyclic flow curve shows another feature. After changing the strain path, the shear stress at the
beginning <strong>of</strong> the new cycle is markedly lower as the maximum shear stress at the end <strong>of</strong> the
previous cycle and a period with a slow increase <strong>of</strong> the flow stress is present. As can be seen
in Figure H.3b for nickel, this behavior depends on the strain per cycle. It is most pronounced
at ∆ɛ = 1 and 2. After this, a further work hardening occures with a steep increase <strong>of</strong> the flow
stress until it reachs approximately the same absolute value at the end <strong>of</strong> the cycle as in the
previous one. (Figure H.4). It can be seen in Figure H.3b that the slope <strong>of</strong> this hardening stage
decreases with increasing ∆ɛ. Figure H.4a and H.4b show the maximum <strong>of</strong> the flow stresses
at each cycle as a function <strong>of</strong> the total deformation for all applied ∆ɛ for Armco iron and
nickel, respectively. In both materials the same features are visible. Just like in the monotonic
deformed samples, the maximum torque saturates after a certain saturation strain. The lower
the ∆ɛ is, the lower is also the strain necesarry for the onset <strong>of</strong> this steady state deformation.
The quickest saturation can be observed at a ∆ɛ = 0.5, the highest strain to reach the saturation
is necessary for the monotonic deformation. It can also be noted for both materials that the
higher ∆ɛ was, the closer are the curves to the curve for the monotonic deformed samples after
the onset <strong>of</strong> saturation. A difference between the two materials is also the onset strain for the
saturation. This strain is higher for the bcc iron compared to the fcc nickel. The maximum
torque and therefore the maximum flow stress is slightly higher for Armco iron than for nickel.
H.3.2 Microstructure
The resulting microstructures after a deformation to a total strain <strong>of</strong> ɛeq,tot = 64 for different
∆ɛ as can be seen by using backscattered electrons are depicted in Figure H.5 and Figure H.6
for Armco iron and nickel. In all presented micrographs the baseline <strong>of</strong> the picture is parallel
to the shear plane. For both materials the same features are recognizable. The microstructure
for the samples with ∆ɛ = 0.5 consists <strong>of</strong> quite equiaxed elements, the structure size is
markedly smaller than 1 µm, only a weak preferred direction <strong>of</strong> the elements is detectable.
H–5
H

H
H Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
Figure H.3: (a) Measured torque during CHPT experiment, Armco-iron, ∆ɛ = 1 and during a HPT experiment
(b) Details <strong>of</strong> measured torque curves during CHPT <strong>of</strong> nickel at different ∆ɛ, the decrease in the torque after
changing the shear direction (+ and -) is indicated.
Figure H.4: Measured torque during HPT and the maximum torques in the CHPT experiments (a) for Armco-iron
and (b) for nickel.
H–6

H.3 Results
With increasing ∆ɛ a decreasing structure size and an aspect ratio <strong>of</strong> the elements markedly
larger than 1 can be observed. Furthermore, a preferred direction <strong>of</strong> the structural elements is
now present. The microstructure <strong>of</strong> the nickel samples deformed monotonically and the sample
deformed with ∆ɛ = 4 are almost identical. The alignment <strong>of</strong> the structural elements is
naturally in the opposite direction, because the last cycle in the CHPT experiments is always
in the ’negative’ direction.
Figure H.5: SEM micrographs <strong>of</strong> Armco-iron taken in radial direction for ɛeq,tot = 64 at (a) ∆ɛ = 0.5, (b)∆ɛ =
1, (c)∆ɛ = 2, (d)∆ɛ = 4.
In Figure H.7, the microstructure <strong>of</strong> Nickel after monotonic deformation to strains <strong>of</strong> ɛeq=0.5,
1, 2 and 4 (as large as the strain increments in the CHPT experiments) is depicted. After a strain
<strong>of</strong> ɛeq = 0.5 the beginning <strong>of</strong> the formation <strong>of</strong> a substructure within a grain can be seen. When
the strain gets larger, more and more well defined elements form, until the whole microstructure
is very homogeneous, compare Figure H.6e and Figure H.7d. During further deformation,
the microstructure is still somewhat refined, until this process reachs a steady state.
Figure H.8 shows the inverse pole figures calculated from the EBSD measurements for
different stages <strong>of</strong> the development <strong>of</strong> the microstructure for nickel samples deformed with
∆ɛ = 0.5 and ∆ɛ = 4. The white lines correspond to boundaries with a missorientation
H–7
H

H
H Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
Figure H.6: SEM micrographs <strong>of</strong> nickel taken in radial direction for ɛeq,tot = 64 at (a) ∆ɛ = 0.5, (b)∆ɛ = 1,
(c)∆ɛ = 2, (d)∆ɛ = 4, (e) monotonically deformed.
H–8

H.3 Results
Figure H.7: SEM micrographs <strong>of</strong> nickel deformed by HPT in radial direction for (a) ɛeq = 0.5, (b) ɛeq = 1, (c)
ɛeq = 2 and (d) ɛeq = 4.
H–9
H

H
H Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
Figure H.8: Orientation maps <strong>of</strong> nickel in axial direction and the standard triangle, (a)∆ɛ = 0.5;ɛeq,tot = 4,
(b)∆ɛ = 0.5; ɛeq,tot = 16, (c)∆ɛ = 0.5;ɛeq,tot = 64, (d)∆ɛ = 4;ɛeq,tot = 4, (e)∆ɛ = 4; ɛeq,tot = 16,
(f)∆ɛ = 4;ɛeq,tot = 64, (g) ɛeq,tot = 64, monotonically deformed.
H–10

H.3 Results
between 2°and 15°, the black lines show boundaries with a missorientation greater than 15°.
It can be seen that after a deformation to total strains higher than 16 no significant changes in
the structure size occurs. The differences between the samples deformed with ∆ɛ = 0.5 and
∆ɛ = 4 are similar to those seen in the BSE-micrographs.
The evolution <strong>of</strong> the structural size as well as the ratio <strong>of</strong> the length <strong>of</strong> high angle boundaries
(HAB) (≥15°) to low angel boundaries (LAB) (2°-15°) for nickel is presented in Figure H.9.
The grain size presented here was evaluated from the EBSD measurements using a missorientation
<strong>of</strong> ≥ 5°as a criterion to define a grain boundary. Again, a saturation in the decrease
<strong>of</strong> the structural size can be seen. With increasing total strain, the fraction <strong>of</strong> the high angle
boundaries increases, somewhat faster for the samples with the higher ∆ɛ. It can be seen
that the higher the ∆ɛ was, the higher was the fraction <strong>of</strong> the HAG. The highest value for the
fraction <strong>of</strong> the HAG as well as the smallest structure size was measured for the monotonic
deformed sample.
In Figure H.10, the missorientation angle distributions <strong>of</strong> the nickel samples are compared
to the random Mackenzie distribution. Only missorientations larger than 5°were taken into
account for the calculation <strong>of</strong> these graphs. Of course, a lot <strong>of</strong> boundaries with smaller missorientations
exists, but they are mainly within the grains. For the samples deformed with
∆ɛ = 0.5 almost no correlation between the missorientation distribution and the random distribution
can be seen. For higher total deformations, a slightly higher amount <strong>of</strong> high angle
boundaries is present, but the fraction <strong>of</strong> small missorientations is still very high. In contradiction
to this, the missorientation distribution <strong>of</strong> samples deformed with ∆ɛ = 4 converge to
the random distribution with increasing total strain. The samples deformed up to ɛeq,tot = 256
as well as the missorientation distribution <strong>of</strong> the monotonic deformed sample with ɛeq = 64
show no further shift to higher missorientation angles.
Figure H.9: (a) Size <strong>of</strong> the structural elements <strong>of</strong> nickel as a function <strong>of</strong> the total equivalent strain (b) ratio between
the length <strong>of</strong> the HAB and the LAB.
H–11
H

H
H Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
Figure H.10: Distribution <strong>of</strong> the missorientation angle in deformed nickel samples (a) Samples deformed with
∆ɛ = 0.5 and monotonically deformed sample, ɛeq = 64 (b) Samples deformed with ∆ɛ = 4.
H.4 Discussion
The cyclic high pressure torsion can be seen in two contexts. Firstly, as an advancement <strong>of</strong>
the conventional high pressure torsion and also as a similar process like ECAP Route C, and
secondly, as a borderline case <strong>of</strong> fatigue.
H.4.1 Severe Plastic <strong>Deformation</strong>
SPD <strong>of</strong> pure metals is quite common and therefore a number <strong>of</strong> papers deal with the evolution
<strong>of</strong> the microstructure and mechanical properties <strong>of</strong> nickel and pure iron. After HPT, grain
sizes in the range <strong>of</strong> 100 nm to 400nm for nickel are reported in the literature. 11–14 The grain
sizes reported for nickel after ECAP are quite higher, namely between 300nm and 450nm. 14–17
When comparing such results, one has always to take care that different methods are used to
evaluate the grain size. Another important factor is the level <strong>of</strong> impurities and alloying elements
<strong>of</strong> the used materials. 18 Similar results are reported for SPD <strong>of</strong> pure iron. 19–21 In the
present investigation, the structure size at saturation <strong>of</strong> nickel was determined from 370 nm
for the monotonic deformed samples up to 850 nm for the samples deformed with ∆ɛ = 0.5.
The determination <strong>of</strong> the structure size as well as the measurement <strong>of</strong> the flow curve both for
nickel and Armco-iron clearly show that in HPT experiments after a total equivalent strain
larger than approximately 20 no further refinement <strong>of</strong> the structure occures. This behavior was
also shown for pure copper in HPT experiments. 9, 22 A similar saturation in the structure size
and in the mechanical strength can also be observed in CHPT experiments, as can be seen in
Figure H.3 and Figure H.8. It can be seen that the strain necessary for the onset <strong>of</strong> saturation
depends on the strain increment ∆ɛ. The larger ∆ɛ is, the larger has to be that onset strain
and the more similar is the microstructure <strong>of</strong> CHPT deformed samples compared to the microstructure
<strong>of</strong> monotonic HPT deformed samples. Hence, an obvious result is that the strain
per cycle clearly determines the resulting structure size. Conventional HPT could be compared
to ECAP Route A, where there is no rotation <strong>of</strong> the billet between the passes. Hence,
elements are continiously sheared in one direction during multiple passes. 23 CHPT is more
H–12

H.4 Discussion
comparable to ECAP Route C, where the billet is rotated by 180°along the billet axis. This
leads to a foreward and backward shearing <strong>of</strong> an element during successive passes. 23 Therefore,
it can be concluded that ECAP Route A will be more effective than Route C in terms
<strong>of</strong> grain refinement, although not necessarily in terms <strong>of</strong> producing an equiaxed microstructure.
This behavior should be well pronounced in most ECAP processes, where a tool angle
<strong>of</strong> 90°is used and the resulting strain per pass is equal to 1. But as can be seen in Figure H.4,
the differences at the beginning are not so large as in the steady state regime. This effect <strong>of</strong>
the strain path at ECAP was also found in other studies, for instance in. 24, 25 The conflictive
results from other studies26 may be contributed to the smaller differences at low total strains
where the differences are not as significant.
In contrast to the saturation in grain size and the mechanical strength, it seems that the ratio
between the length <strong>of</strong> the HAB and LAB as well as the increase in missorientation between
neighbouring elements does not saturate as fast as the strength. Hence, one can assume that
the mechanical strength is mainly determined by the size <strong>of</strong> the structural elements and not by
the missorientation between these elements.
Why does the structure size saturate in HPT and CHPT at different levels?
The saturation <strong>of</strong> the structure size in single phase materials due to severe plastic deformation
and its reasons is a topic that is not widely discussed in the literature until recently. 22, 27, 28
It is assumed that processes similar to dynamic recrystallisation or grain boundary sliding are
taking place. For a more detailed discussion <strong>of</strong> this, see.
Richert et al. 29 reported that similare stages <strong>of</strong> deformation are observed in cyclic experiments
as in experiments with monotonic deformation, although the accumulated strains have to be
much higher. The reason for this are annihilation processes that are induced by the reversal
<strong>of</strong> the shear strain. Stüwe 30 described this behaviour mathematical by an efficiency factor η.
This factor η describes the efficiency to create new dislocations and store them in the structure.
Only η times the shear strain increment contributes to the increase in dislocation density. There
are two limiting cases for η. If η = 0, no further strain hardening can occure. All dislocations
that are created during the deformation in one direction run backwards and recombine in the
former sources during the reversal <strong>of</strong> the deformation or the number <strong>of</strong> generated dislocations
is equal to the number <strong>of</strong> annihilated dislocations in one cycle. The other case is that for
both directions new dislocations are generated and are stored in the microstructure, therefore
η would be equal to 1. If this is true there should be no difference between the cyclic and the
monotonic deformation. In Figure H.3a it can be seen that this may be the case for the very
first few cycles. After a certain number <strong>of</strong> cycles, the efficiency factor decreases, until after
the onset <strong>of</strong> saturation η is obviously zero over an even number <strong>of</strong> cycles and no further work
hardening occurs. Hence, in this description <strong>of</strong> the process, the efficiency factor η is itself a
function <strong>of</strong> the total strain and also differs for different ∆ɛ. It is near one at low total strain
and approachs zero as the number <strong>of</strong> cycles increases. This efficiency factor can be a useful
tool to describe cyclic deformation modes, although it has to be noted that no structural model
is combined with it at the moment.
But the detailed measurement <strong>of</strong> the torque show that this description with η is only true
when one just looks at the reached maximum values for the torque at the end <strong>of</strong> the cycle.
During the deformation, the torque and therefore the flow stress developes in a different way
as described in section 3.1. As can be seen in Figure H.3, after a reversal <strong>of</strong> the strain path, the
22, 27
H–13
H

H
H Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
torque at the beginning <strong>of</strong> the cycle is significantly lower than the reached maximum torque
after the previous cycle and increases only slightly for a significant fraction <strong>of</strong> the whole cycle.
This region may be the consequence <strong>of</strong> dislocations running backwards toward their sources
or newly generated dislocations that annihilate with dislocation present on the same slip plane
but with different sign from the previous cycle. Following this, the flow stress increases with
a rate that is decreasing with increasing ∆ɛ. This means that after the process <strong>of</strong> a reversal
<strong>of</strong> dislocation movement or annihilation <strong>of</strong> dislocations is exhausted, new dislocations are
created to further plastically deform the sample. For samples with a larger structure size (at
the beginning <strong>of</strong> deformation or small ∆ɛ) a high work harding capability exists, therefore
the rate <strong>of</strong> work hardening is large. For the samples with a structure size near the minimum
structure size, almost no work hardening capability is left, leading to a slow increase <strong>of</strong> the
flow stress. From the measurement it can be assumed that the flow stress and the structure size
<strong>of</strong> cyclically deformed samples would reach the level <strong>of</strong> the monotonic deformed samples if
the deformation would be carried on monotonic. This behavior can be seen for both materials
with the before mentioned difference in the onset <strong>of</strong> the steady state region.
H.4.2 Fatigue
Of course, it has to be emphazised that the results from CHPT and HPT cannot directly be
compared to results from fatigue tests. Nevertheless, many features that are well known in
fatigue have similarities with the behaviour <strong>of</strong> CHPT deformed materials.
Fatigue <strong>of</strong> nickel and Armco-iron was investigated in many studies, see for example. 31–35
The presented cyclic hardening curves in these studies show similar characteristics like the
measured curves in the present investigation. After an intense hardening stage at the beginning,
the stress amplitude saturates in strain-controlled experiments. It is commonly reported
34, 36–39
that the saturation stress is a function <strong>of</strong> the plastic strain amplitude for symmetric tensioncompression
tests both for various single- and polycrystalline materials. The saturation stress
increases with the plastic strain amplitude, in some cases a plateau region exists as can be seen
in Figure H.11a (from 38 ). A question which was not solved till now is: Exists a limit for the
cyclic hardening and, if it exists, where is it?
Of course, this can not be easily investigated due to the limit <strong>of</strong> the plastic strain amplitude in a
compression-tension experiment. In the present study, these limits do not exist, although it has
to be minded that the deformation mode is different. The increase <strong>of</strong> the saturation stress (here
in terms <strong>of</strong> the torque) is compared to the monotonic flow curve in Figure H.11b and H.11c for
Armco-iron and nickel, respectively. For both materials it is obvious that the saturation stress
increases with the plastic strain amplitude but the slope <strong>of</strong> this increase decreases markedly.
Finally, for a ∆ɛ = 4, the saturation stress has about reached the saturation stress for the
monotonic deformation mode. The data from conventional fatigue tests would just cover the
very left side <strong>of</strong> the diagrams.
The occurrence <strong>of</strong> a saturation <strong>of</strong> the stress in fatigue is generally explained by the development
<strong>of</strong> a dislocation network that allows for a further deformation without an increase in the
dislocation density. The cell size <strong>of</strong> this dislocation network decreases with increasing plastic
strain amplitude and leads therefore to a higher saturation stress according to the mesh-length
theory <strong>of</strong> work hardening. 40 Davidson and Lankford 41 measured the subgrain size in low car-
H–14

H.5 Conclusions
bon steels as a function <strong>of</strong> the cyclic stress range and compared them with results from other
studies. A main result is the decrease <strong>of</strong> the subgrain size with the increasing cyclic stress,
the smallest subgrains where found for the monotonic deformation in a tensile test. This is
in good agreement with the present results as can be seen in Figure H.9a. The higher the
∆ɛ was, the smaller was the structure size in the saturation regime. The smallest structure
size was measured for the monotonic deformed sample. The decrease <strong>of</strong> the structure size for
HPT-deformed copper was investigated by Hebesberger et al. 22 They observed a saturation in
the decrease <strong>of</strong> the structure size at higher strains similar as in the presented experiments in
Armco iron and nickel.
Another feature <strong>of</strong> conventional fatigue tests is the onset <strong>of</strong> the saturation in mechanical
strength. Generally it is reported that saturation is reached earlyer in terms <strong>of</strong> the number <strong>of</strong>
cycles for larger plastic strain amplitudes. When for example the results from 39 are expressed
in terms <strong>of</strong> the product <strong>of</strong> number <strong>of</strong> cycles for saturation and the plastic strain amplitude
(a value comparable to the equivalent strain as calculated in our study), the same behavior
for fatigue and CHPT is present: The higher the strain per cycle, the higher has to be the
accumulated strain until saturation occurs. The highest total strain to reach the saturation is
needed for the monotonic deformation in HPT.
In fatigue experiments, the typical plastic strain amplitudes are quite low (in the order <strong>of</strong>
10 −3 ) compared to the plastic strains in ECAP Route C or this investigation. Nevertheless,
the cell size is comparable or only somewhat larger than the structure size after SPD and the
laws govering this size seem to be similar. The same characteristics both in terms <strong>of</strong> structure
size and strain hardening are present in both processes. But the much higher stresses that
are measured for SPD-deformed materials and now also for the CHPT deformed nickel and
Armco iron show that especially the boundaries cannot be directly compared to the structures
in fatigued materials. One difference is <strong>of</strong> course the higher missorientation between the elements
in the SPD-deformed materials. But as could be seen in Figure H.10 and Figure H.4, the
flow stress is mainly influenced by the structure size and not by the missorientation. Another
difference is the high hydrostatic pressure under which SPD takes place. It is assumed 42 that
the vacancy concentration is extremely high in these conditions and these vacancies may influence
the formation <strong>of</strong> the boundaries. The boundaries (which are no real grain boundaries)
in SPD-deformed materials are sometimes referred as non-equilibrium grain boundaries 6 and
seem therefore be a special feature <strong>of</strong> the very large plastic strains only reached by applying
the methods <strong>of</strong> SPD.
H.5 Conclusions
• Cyclic High Pressure Torsion, a modified version <strong>of</strong> High Pressure Torsion has been
introduced, which permits to easily study many parameters <strong>of</strong> cyclic severe plastic deformation.
The method has been applied to Armco iron and nickel.
• Monoton deformation by HPT <strong>of</strong> both material results in a saturation <strong>of</strong> strain hardening.
A similar saturation occures due to deformation by cyclic high pressure torsion, but the
accumulated strain to reach saturation decreases with decreasing ∆ɛ.
H–15
H

H
H Cyclic High Pressure Torsion <strong>of</strong> Nickel and Armco Iron
Figure H.11: (a) Cyclic stress strain curve in fatigued polycrystalline copper, 38 (b) Cyclic torque strain curve and
the monoton flow curve for iron, (c) Cyclic torque strain curve and the monoton flow curve for nickel (∆ɛ,ɛ at
r = 3mm).
H–16

H.5 Conclusions
• The steady state that is reached in terms <strong>of</strong> mechanical strength is also reached in structure
size, the differences in the saturation strength is also reflected in differences <strong>of</strong> the
structure size. The only value that still changes somewhat is the missorientation distribution.
The higher the total deformation, the more similar is the orientation distribution
to the Mackenzie distribution.
• The structure size strongly depends on the strain increment. The larger the strain increment
is, the smaller gets the structure size after the onset <strong>of</strong> saturation. The smallest
structure size was reached for the monotonic deformation by High Pressure Torsion.
H–17
H

H
H–18

Bibliography to paper H
[1] Y. T. Zhu, T. G. Langdon, R. S. Mishra, S. L. Semiatin, M. J. Saran, and T. C. Lowe,
editors. Ultrafine Grained Materials II. TMS Publications, Warrendale, Pennsylvania,
2002.
[2] Y. T. Zhu, T. G. Langdon, R. Z. Valiev, S. L. Semiatin, D. H. Shin, and T. C. Lowe,
editors. Ultrafine Grained Materials III. TMS Publications, Warrendale, Pennsylvania,
2004.
[3] M. J. Zehetbauer and R. Z. Valiev, editors. Nanomaterials by Severe Plastic <strong>Deformation</strong>.
J. Wiley, VCH Weinheim (Germany), 2002.
[4] A. Vinogradov, T. Ishida, K. Kitagawa, and V.I. Kopylov. Acta Mater., 53:2181–2192,
2005.
[5] P. L. Sun, P. W. Kao, and C. P. Chang. In Y. T. Zhu, T. G. Langdon, R. S. Mishra, S. L.
Semiatin, M. J. Saran, and T. C. Lowe, editors, Ultrafine Grained Materials II, pages
35–42. TMS Publications, Warrendale, Pennsylvania, 2002.
[6] R. Z. Valiev, R. K. Islamgaliev, and I. V. Alexandrov. Progr. Mat. Sci., 45:103–189, 2000.
[7] T. Hebesberger, A. Vorhauer, H. P. Stüwe, and R. Pippan. In M. J. Zehetbauer and R. Z.
Valiev, editors, Nanomaterials by Severe Plastic <strong>Deformation</strong>, pages 447–452. J. Wiley,
VCH Weinheim (Germany), 2002.
[8] A. Vorhauer and R. Pippan. Scripta Mater., 51:921–925, 2004.
[9] F. Wetscher, A. Vorhauer, and R. Pippan. Mat. Sci. Eng. A, in press, 2005.
[10] A. Vorhauer and R. Pippan. submitted to Acta Mater.
[11] R. K. Islamgaliev, F. Chmelik, and R. Kuzel. Mater. Sci. Eng., A, 237:43–51, 1997.
[12] F. Dalla Torre, P. Spätig, R. Schäublin, and M. Victoria. Acta Mat, 53:2337–2349, 2005.
[13] E. Schaffler and R. Pippan. Mat. Sci. Eng.A, 387-389:799–804, 2004.
[14] A. P. Zhilyaev, B.-K. Kim, J. A. Szpunar, M. D. Baró, and T. G. Langdon. Mat. Sci.
Eng.A, 391:377–389, 2004.
[15] K. Neishi, Z. Horita, and T. G. Langdon. Mat. Sci. Eng.A, 325:54–58, 2002.
BIB–19
BIB

BIB
Bibliography to paper H
[16] N. Hansen, X. Huang, and D. A. Hughes. In Y. T. Zhu, T. G. Langdon, R. S. Mishra,
S. L. Semiatin, M. J. Saran, and T. C. Lowe, editors, Ultrafine Grained Materials II,
pages 35–42. TMS Publications, Warrendale, Pennsylvania, 2002.
[17] N. Krasilnikov, W. Lojkowski, Z. Pakiela, and R. Valiev. Mat. Sci. Eng.A, 397:330–337,
2005.
[18] R. Pippan, A. Vorhauer, F. Wetscher, M. Faleschini, M. Hafok, and I. Sabirov. Mater.
Sci. Forum, 503-504:407–412, 2006.
[19] R. Z. Valiev, Yu. V. Ivanisenko, E. F. Rauch, and B. Baudelet. Acta Mat., 44:4705–4712,
1996.
[20] Yu. Ivanisenko, A. V. Sergueeva, A. Minkow, R. Z. Valiev, and H.-J. Fecht. In M. J. Zehetbauer
and R. Z. Valiev, editors, Nanomaterials by Severe Plastic <strong>Deformation</strong>, pages
453–458. J. Wiley, VCH Weinheim (Germany), 2002.
[21] Florian Wetscher, Andreas Vorhauer, Richard Stock, and Reinhard Pippan. Materials
Science and Engineering A, 387-389:809–816, 2004.
[22] T. Hebesberger, H. P. Stüwe, A. Vorhauer, F. Wetscher, and R. Pippan. Acta Mater.,
53:393–402, 2005.
[23] M. Furukawa, Y. Iwahashi, Z. Horita, M. Nomoto, and T. G. Langdon. Mater. Sci. Eng.A,
257:328–332, 1998.
[24] A. Gholinia, P. B. Prangnell, and M. V. Markushev. Acta Mater., 48:1115–1130, 2000.
[25] L. Dupuy and E. F. Rauch. Mat. Sci. Eng.A, 337:241–247, 2002.
[26] T. C. Lowe, Y. T. Zhu, S. I. Semjatin, and D. R. Berg. In T. C. Lowe and R. Z. Valiev,
editors, Investigations and Applications <strong>of</strong> SPD, page 237. Kluwer Academic Publishers,
Norwell, 2000.
[27] H. P. Stüwe. Mater. Sci. Forum, 503-504:175–178, 2006.
[28] K. Nakashima, M. Suzuki, Y. Futamura, T. Tsuchiyama, and S. Takaki. Mater. Sci.
Forum, 503-504:627–632, 2006.
[29] M. Richert, H. P. Stüwe, M. J. Zehetbauer, J. Richert, R. Pippan, Ch. Motz, and
E. Schafler. Mater. Sci. Eng., A, 355:180–185, 2003.
[30] H. P. Stüwe. In M. J. Zehetbauer and R. Z. Valiev, editors, Nanomaterials by Severe
Plastic <strong>Deformation</strong>, pages 55–64. J. Wiley, VCH Weinheim (Germany), 2002.
[31] D.J. Morrison and V. Chopra. Mater. Sci. Eng., A, 177:29–42, 1994.
[32] P. Lukas and L. Kunz. Mater. Sci. Eng., A, 189:1–7, 1994.
BIB–20

Bibliography to paper H
[33] C. Buque, J. Bretschneider, A. Schwab, and C. Holste. Mater. Sci. Eng., A, 300:254–262,
2001.
[34] Y. El-Madhoun, A. Mohamed, and M. N. Bassim. Mater. Sci. Eng., A, 385:140–147,
2004.
[35] H. Haddou, M. Risbet, G. Marichal, and X. Feaugas. Mater. Sci. Eng., A, 379:102–111,
2004.
[36] H. Mughrabi. Mat. Sci. Eng., 33:207–223, 1978.
[37] A. Giese, A. Styczynski, and Y. Estrin. Mat. Sci. Eng.A, 124:L11–L13, 1990.
[38] C. D. Liu, M. N. Bassim, and D. X. You. Acta metall. mater., 42:3695–3704, 1994.
[39] Y. El-Madhoun, A. Mohamed, and M. N. Bassim. Mat. Sci. Eng.A, 359:220–227, 2003.
[40] D. Kuhlmann-Wilsdorf. Metall. Trans., 1:3173–3179, 1970.
[41] D. L. Davidson and J. Lankford. Int. J. Fracture, 17:257–275, 1981.
[42] M. Zehetbauer, H. P. Stüwe, A. Vorhauer, E. Schafler, and J. Kohout. In M. J. Zehetbauer
and R. Z. Valiev, editors, Nanomaterials by Severe Plastic <strong>Deformation</strong>, pages 435–446.
J. Wiley, VCH Weinheim (Germany), 2002.
H–21
H

H
H–22

I
Structural Evolution during Cyclic Severe
Plastic <strong>Deformation</strong>
F. Wetscher 1,2 and R. Pippan 1,2
1 Erich Schmid Institute for Materials Science, Austrian Academy <strong>of</strong> Sciences, Leoben,
Austria
2 CD-Laboratory for Local Analysis <strong>of</strong> <strong>Deformation</strong> and Fracture, Leoben, Austria
Abstract
To investigate the influence <strong>of</strong> cyclically large plastic deformations, a new form <strong>of</strong> high
pressure torsion, cyclic high pressure torsion, is presented. With this method experiments
where carried out with plastic shear strains (equivalent von Mises strains) per cycle ∆ɛeq between
0.5 and 4 up to total strains <strong>of</strong> ɛ = 64. The materials used in this investigation are the
bcc Armco-iron and the fcc pure nickel. For both materials, the initial grain size <strong>of</strong> some ten
micrometer decreased markedly while the mechanical strength, measured in-situ during the
deformation experiment, increased. The evolution <strong>of</strong> the microstructure was investigated by
means <strong>of</strong> scanning electron microscopy using backscattered electrons and electron backscatter
diffraction. At higher ∆ɛeq the resulting microstructure is much finer, similar like in conventional
fatigue experiments. This difference in the microstructure results also in a difference in
the mechanical strengths.
I–1
I

I
I–2

I.1 Introduction
I.1 Introduction
The knowledge <strong>of</strong> the microstructural evolution during fatigue is <strong>of</strong> great technological importance,
hence a large number <strong>of</strong> investigations deal with this problem, see for example. 1–3 If
the plastic strain amplitude and/or the number <strong>of</strong> cycles is very large, the formation <strong>of</strong> fatigue
cracks is inevitable. Therefore, the range to investigate cyclic plastic deformations is quite
small and constricted to quite low strain amplitudes. To overcome these limits and to expand
the investigable range, new methods have to be applied. High pressure torsion (HPT) 4, 5 is a
method to severely plastically deform materials by simple torsion. Due to the high hydrostatic
pressure, very large strains can be reached without crack initiation. To investigate the influence
<strong>of</strong> very high cyclic deformations, a cyclic form <strong>of</strong> HPT, cyclic high pressure torsion (CHPT)
was developed.
I.2 Experimental Procedure
Samples <strong>of</strong> pure nickel and Armco iron (grain size ≈ 50µm with a diameter <strong>of</strong> 8 mm and
a thickness <strong>of</strong> 0.8 mm were deformed by CHPT at room temperature under a hydrostatic
pressure <strong>of</strong> 5.7 GPa with 0.2 turns per minute. The equivalent von Mises strain ɛeq for a
deformation in one direction was calculated according to equation 1, the total deformation
according to equation 2:
∆ɛeq = 2πrn
t √ 3
ɛeq,total = N∆ɛeq
For this investigation, the number <strong>of</strong> turns n was chosen to reach a ∆ɛ at a radius r = 3 mm
<strong>of</strong> 0.5, 1, 2 and 4. The number <strong>of</strong> cycles N for each experiment was chosen to obtain samples
with ɛeq,total between 4 and 64 (one cycle in the present study is corresponding to 1
2 cycle in
conventional fatigue). Figure I.1 shows a sketch <strong>of</strong> the used HPT-tool, the torque during the
experiment was measured in-situ using strain gauges.
Microstructural investigations in the indicated directions were performed on a Zeiss 1525
scanning electron microscope (SEM) at 20 kV using backscattered electrons. For determing
information about missorientation, orientation image maps (OIM) were measured using back
scattered Kikuchi lines.
I.3 Results
In Figure I.2 the microstructure <strong>of</strong> nickel after a total deformation <strong>of</strong> ɛeq,total = 64 with
different ∆ɛ is compared. In both samples, the microstructure in tangential direction consists
<strong>of</strong> almost equiaxed structural elements; the size <strong>of</strong> the elements is significantly smaller for
the samples cycled with a higher ∆ɛ. The same relation in terms <strong>of</strong> the size <strong>of</strong> the elements
is present in micrographs depicting the radial direction. In the sample deformed with ∆ɛ =
0.5, again the microstructure is quite equiaxed, and no preferred direction is recognisable. In
(I.1)
(I.2)
I–3
I

I
I Structural Evolution during Cyclic Severe Plastic <strong>Deformation</strong>
Figure I.1: Principle <strong>of</strong> high pressure torsion and the sample directions for microstructural investigations
contradiction to this, in the sample with ∆ɛ = 4, many <strong>of</strong> the elements are elongated and
aligned along a certain direction.
Figure I.3 shows the corresponding micrographs <strong>of</strong> Armco iron. The microstructural elements
are somewhat coarser compared to nickel. While in nickel the elements viewed in
tangential direction are equiaxed, the elements in Armco iron in all micrographs are somewhat
elongated and a preferred direction is present.
The OIM images <strong>of</strong> deformed Armco iron are presented in Figure I.4. Figure I.4a - c depict
the microstructural evolution during a deformation with ∆ɛ = 0.5 for different ɛeq,total. After
a deformation <strong>of</strong> ɛeq,total = 4, remains <strong>of</strong> the original grains or grain boundaries are still
recognisable, these vanish after higher total deformations. The decreasing structure size can
be seen in Figure I.4c - e, where OIM images <strong>of</strong> Armco iron after a total deformation <strong>of</strong> 64 with
different ∆ɛ are compared. In all these images, boundaries with missorientations between 2
and 15 degrees are indicated with white lines, missorientations larger than 15°are marked with
black lines. From these OIM images, structural sizes (elements with a missorientation larger
than 5°) and missorientation distributions where calculated. Figure I.5a shows the size <strong>of</strong> the
structural elements <strong>of</strong> Armco iron and nickel as a function <strong>of</strong> the total deformation. It can be
seen for all materials that a quick decrease <strong>of</strong> the structure size from the original grain size <strong>of</strong>
some ten micrometer has occurred. At higher total strains, no further refinement occurs, the
structure size remains constant. The onset <strong>of</strong> this saturation in structure refinement seems to
occur at lower total strains for the fcc nickel compared to the bcc Armco iron. As seen in the
I–4

I.3 Results
Figure I.2: SEM micrographs <strong>of</strong> CHPT deformed nickel in different directions with a ɛeq,total = 64; (a) ∆ɛ =
0.5, tangential direction, (b)∆ɛ = 4 , tangential direction, (c) ∆ɛ = 0.5, radial direction and (d), ∆ɛ = 4, radial
direction.
I–5
I

I
I Structural Evolution during Cyclic Severe Plastic <strong>Deformation</strong>
micrographs, the structure size decreases with increasing ∆ɛ. Generally, the structure size <strong>of</strong>
nickel is somewhat smaller than the size <strong>of</strong> Armco iron.
Figure I.3: SEM micrographs <strong>of</strong> CHPT deformed Armco iron in different directions with a ɛeq,total = 64; (a)
∆ɛ = 0.5, tangential direction, (b)∆ɛ = 4 , tangential direction, (c) ∆ɛ = 0.5, radial direction and (d), ∆ɛ = 4,
radial direction.
The missorientation angle distribution <strong>of</strong> different samples and the random Mackenzie distribution
are compared in Figure I.5b. It can be seen that the ratio <strong>of</strong> low angle boundaries for
similar samples is much higher for Armco iron than for nickel, especially for low ∆ɛ. For the
highest ∆ɛ, the missorientation angle distribution is very similar to the Mackenzie distribution.
The development <strong>of</strong> the shear stress (here in terms <strong>of</strong> the measured torque) as a function <strong>of</strong>
the total equivalent strain for nickel can be seen in Figure I.6a. For monotonic deformation,
a stage <strong>of</strong> intensive strain hardening is present that finally saturates after a strain <strong>of</strong> approximately
10. In the cyclic experiments, strain hardening is also present, especially in the first
few cycles. There, the curves are almost identical. But the lower the ∆ɛ is, the faster the
saturation in strain hardening occurs and the lower is the torque and therefore the mechanical
strength after saturation. Figure I.6b compares the torque for certain ∆ɛ in saturation with the
monotonic curve, showing the increase <strong>of</strong> the mechanical strength with increasing ∆ɛ. It can
I–6

I.4 Discussion
be seen that saturation torque almost reaches the saturation torque for monotonic deformation
for the highest applied ∆ɛ.
Figure I.4: Orientation image maps <strong>of</strong> CHPT deformed iron in radial direction with (a) ∆ɛ = 0.5, ɛeq,total = 4,
(b) ∆ɛ = 0.5, ɛeq,total = 16, (c) ∆ɛ = 0.5, ɛeq,total = 64, (d) ∆ɛ = 1, ɛeq,total = 64 and (e) ∆ɛ = 2,
ɛeq,total = 64.
I.4 Discussion
In many studies, 6–8 the microstructural evolution <strong>of</strong> nickel and Armco iron during fatigue was
investigated. These studies were limited to small plastic amplitudes due to practical reasons.
Therefore, it is not clear whether or not a limit for the cyclic hardening exists. In most cases
the cyclic saturation stress increases with an increasing plastic amplitude, sometimes, a plateau
region is reported. 9, 10 The increase <strong>of</strong> the mechanical strength is explained according to the
mesh-length theory 11 by a formation <strong>of</strong> a dislocation network, the size <strong>of</strong> the dislocation cells
decreases with an increasing plastic amplitude. These dislocation networks also allow for a
deformation without a change in the dislocation density and therefore without further work
hardening.
The same features that are reported in conventional fatigue are also present in this study.
The higher the plastic amplitude is, the smaller is the observed structure size. This decreasing
I–7
I

I
I Structural Evolution during Cyclic Severe Plastic <strong>Deformation</strong>
Figure I.5: (a) Size <strong>of</strong> the structural elements for Armco iron and nickel. (b) Missorientation angle distribution
for Armco iron and nickel, all samples with ɛeq,total = 64.
Figure I.6: (a) maxima <strong>of</strong> the in-situ measured torque during the CHPT deformation <strong>of</strong> nickel for different plastic
strain amplitudes. (b) Comparison <strong>of</strong> the torque after the onset <strong>of</strong> saturation with the torque during monotonic
deformation for Armco iron and nickel.
I–8

I.5 Conclusions
structure size is also visible in an increase in the measured torque and therefore in the mechanical
strength, 12 see Figure I.6a. Figure I.6b is similar to a conventional cyclic stress strain
curve and show the increase <strong>of</strong> the saturation stress with increasing plastic strain amplitude. In
this diagram, the results from conventional fatigue tests would only lie in the very left corner.
It is obvious that the limit for the plastic strain hardening is the curve for monotonic deformation;
it is not possible to reach a smaller structure size and therefore a higher mechanical
strength by applying a cyclic deformation. Generally it is reported that the onset <strong>of</strong> saturation
occurs after a smaller number <strong>of</strong> cycles for higher plastic strains. 13, 14 Nevertheless, converted
to total equivalent strains, this means that the onset <strong>of</strong> saturation occurs faster for smaller plastic
strain amplitudes. This is exactly what can be seen in Figure I.6a. The larger the ∆ɛ, the
larger is the total strain necessary to reach saturation. Nevertheless, despite all the similarities
between CHPT and conventional fatigue, some things are quite different. The structure sizes
determined in the present study is in the same range than the cell size in fatigue. However,
the mechanical strength measured here is much higher than in fatigue. Another difference
is the missorientation between the elements. While in fatigue mainly low angel boundaries
are present, after CHPT most boundaries are large angle boundaries. In fatigued material,
the boundaries consist <strong>of</strong> dislocations, the exact nature <strong>of</strong> the boundaries in SPD material is
presently not known. Sometimes it is assumed that also these boundaries are made up <strong>of</strong> dislocations,
but with a much higher density so that the dislocations loose their identity and they
are only referred as non-equilibrium boundaries. 15 A recent study 16 suggests that mainly the
state <strong>of</strong> the boundary and not the larger missorientation causes the higher strains.
I.5 Conclusions
• Cyclic high pressure torsion <strong>of</strong> Armco iron and nickel leads to a decreasing structure
size and an increase in the mechanical strength.
• This change <strong>of</strong> properties saturates after a certain number <strong>of</strong> cycles. The total strains for
the onset <strong>of</strong> saturation depends on the plastic amplitude.
• The larger the plastic amplitude, the smaller is the structure size and the higher the
mechanical strength after the onset <strong>of</strong> saturation.
• The results <strong>of</strong> the performed experiments are similar to results from conventional fatigue
tests and can be considered as a limiting chase for very large plastic amplitudes.
I–9
I

I
I–10

[1] H. Mughrabi. Mater. Sci. Eng., 33:207–223, 1978.
Bibliography to paper I
[2] P. Haehner, B. Tippelt, and C. Holste. Acta mater., 46:5073–5084, 1998.
[3] J. Zhang and Y. Jiang. Int. J. Plast., 21:2191–2211, 2005.
[4] H. Jiang, Y. T. Zhu, D. P. Butt, I. V. Alexandrov, and T. C. Lowe. Mater. Sci. Eng.A,
290:128–138, 2000.
[5] A. Vorhauer and R. Pippan. Scripta Mater., 51:921–925, 2004.
[6] D.J. Morrison and V. Chopra. Mater. Sci. Eng., A, 177:29–42, 1994.
[7] P. Lukas and L. Kunz. Mater. Sci. Eng., A, 189:1–7, 1994.
[8] Y. El-Madhoun, A. Mohamed, and M. N. Bassim. Mater. Sci. Eng., A, 385:140–147,
2004.
[9] C. D. Liu, M. N. Bassim, and D. X. You. Acta metall. mater., 42:3695–3704, 1994.
[10] C. Buque, J. Bretschneider, A. Schwab, and C. Holste. Mater. Sci. Eng., A, 300:254–262,
2001.
[11] D. Kuhlmann-Wilsdorf. Metall. Trans., 1:3173–3179, 1970.
[12] F. Wetscher, A. Vorhauer, and R. Pippan. Mater. Sci. Eng. A, 410-411:213–216, 2005.
[13] Y. El-Madhoun, A. Mohamed, and M. N. Bassim. Mat. Sci. Eng.A, 359:220–227, 2003.
[14] H. Haddou, M. Risbet, G. Marichal, and X. Feaugas. Mater. Sci. Eng., A, 379:102–111,
2004.
[15] R. Z. Valiev, R. K. Islamgaliev, and I. V. Alexandrov. Progr. Mat. Sci., 45:103–189, 2000.
[16] F. Wetscher and R. Pippan. submitted to Phil. Mag., 2006.
I–11
I