THE EVOLUTION OF SOLUTE CARBON DURING

CONTINUOUS ANNEALING OF INTERSTITIAL-FREE STEEL

AND ITS EFFECTS ON BAKE HARDENABILITY

James N.A. Starling

A Thesis subrnitted to the

Department of Materials and Metallurgical Engineering

in partial fulfillment of the requirements for the

Degree of Master of Science (Engineering)

Queen' s University

Kingston, Ontario,Canada

March 1998

copyright O James Noel Andrewes Starling, 1998

 Bibliothque nationale
I*I ofCanada
National Library
du Canada
Acquisitions and Acquisitions et
Bibliographie Services services bibliographiques
395 Wellington Street 395. rue Wellington
Otbwa ON K I A ON4 Ottawa ON K IA ON4
Canada Canada

The author has granted a non-
exclusive licence allowing the
National Library of Canada to
reproduce, loan, distribute or sell
copies of this thesis in microform,
paper or electronic formats.
L'auteur a accord une licence non
exclusive permettant la
Bibliothque nationale du Canada de
reproduire, prter, distribuer ou
vendre des copies de cette thse sous
la forme de rnicrofiche/film, de
reproduction sur papier ou sur format
lectronique.

The author retains ownership of the L'auteur conserve la proprit du

copynght in this thesis. Neither the
thesis nor substantial extracts fiom it
may be printed or otherwise
reproduced without the author's
permission.
droit d'auteur qui protge cette thse.
Ni la thse ni des extraits substantiels
de celle-ci ne doivent tre imprims
ou autrement reproduits sans son
autorisation.
 ABSTRACT

Ultra-low carbon interstitial-fiee (IF) sheet steel can offer both good
formability and adequate strength for auto body paneis. These steels c m be designed
to achieve a 30-60 MPa increase in strength after press forrning and during baking.
This increase in yield strength is referred to as bake hardening, AuBH. The
strengthening mechanisms are a combination of carbon segregation to dislocations and
precipitation of coherent Fe carbides. The increase in strength occurs during the
curing of the outer body panel paint. The AQ, property is a consequence of strain
aging and has a strong dependence on the deformed microstmcture and the
concentration of solute carbon atorns.
The objective of this research is to study the evolution of solute carbon and
precipitates dunng the simulated continuous annealing of various microdloyed IF
steels. A simulated bake hardening process was developed and different continuous
annealing parameters (temperature, time, cooling rate) were studied to determine their
effects on the BoaH response. In conjunction with the bake hardening tests, a
quantitative TEM study was done to characterise the size and distribution of the
precipitates following various annealing times. These results were compared with
existing dissolution models.
The determination of the solute carbon content after various annealing cycles
was determined by measurements of the aging index and of activation volume by strain
rate change experiments. The solute carbon concentration was correiated with the
previous Ansn results and a solute carbon - AoBHrelationship was determined.
 ACKNOWLEDGMENTS

1 would first iike to thank my supervisors, Dr. J.D. Boyd and Dr. S. Saimoto,
both of whom showed great enthusiasm in my work and took the time to explain the
many theories behind much of this work.
John Thomson of Stelco Steel played an important part is setting up the initiai
study on bake hardenability with his knowledge and generous supply of materiais. He
was also very supportive in discussing my results and sharing his opinions.
1 would tike to thank the Nicol Hail technical staff: Charlie Cooney, Darryl
Dietrich, Gary Contant, Paul Nolan, and Rod Bemdt, for there expertise in their
specialized fields. They were instrumentai in offering a word of advice or lending a
hand to helping set-up an expenment.
My parents, John and Susan Starling, who have aiways encouraged me to take
on new challenges and were aiways there to support me through the difficult times
with reassuring words of advice.
Last but not least, 1 must give the biggest thanks to Donna Milligaq whose
enthusiasm and ability to listen, helped me get through the highs and lows of daily life.
1 love you.
 TABLE OF CONTENTS

ABSTRACT 1

- *
ACKNOWLEDGMENTS 11

...
TABLE OF CONTENTS III

LIST OF FIGURES vi

LIST OF TABLES

LIST OF SYMBOLS

CHAPTER 1: INTRODUCTION

CHAPTER 2: FACTORS AFFECTING THE PROCESSING OF BAKE

HARDENABLE IF STEELS
2.1 Introduction
2.2 Literature Review
2.2.1 Production of IF Sheet Steel
2.2.2 Stee1 Composition
2.2.2.1Precipitation Strengthening
2.2.2.2 Solid Solution Strengthening
2.2.2.3 Effect of C and N on Mechanical Properties
2.2.2.4Steel Composition - Nb/C Atomic Ratio
2.2.3 Processing
2.2.3.1 Batch and Continuous Annealing
2.2.3.1.1 RecrystalIization
2.2.3.1 -2 Effect of Continuous Annealing Temperature
on Mechanical Properties
2.2.3-2 Temper Rolling
2.2.3.3Press Forming
2.2.3.4Paint Baking
2.2.3.5 Hot Dip Gaivannealing Process
2.3 Experimentai
2.3.1 Nb Interstitiai-Free Steels
2.3-2 Laboratory Simulation
2.3 -2.1 Continuous Annealing
2.3 -2.2 Temper Rolling
2.3.2.3 Press-Forning
2.3 -2.4Paint-Baking
2.4 Resuits
2.4.1 Annealing Temperature
2.4.2 Annealing Holding Time

...
Ilt
 2.4.3 Cooling Rate
2.4.4 Yield Strength
2.5 Conclusions
2.6 References

CHAPTER 3 : THE EVOLUTION OF NbC PRECPITATES D W G

CONTINUOUS ANNEALING OF AN IF SHEET STEEL
3.1 Introduction
3.2 Literature Review
3.2.1 Precipitation in Nb IF Steels
3-2.2 Precipitate Solubility Product
3-2.3 Dissolution of Precipitates
3.2.4 Precipitate Coarsening
3.3 Experimental
3.3.1 Experimental Procedure
3.3.2 Experimental Apparatus
3.3.2.1 Transmission Electron Microscopy
3.3.2.2 Preparation of Carbon Replicas and Thin Foils
3.4 Results
3 -4.1 Characterization of Starting Material
3A.2 Precipitate Size and Density Evolution
3 -4.2.1 As-Rolled Condition
3 -4.2.2 Continuously Annealed Conditions
3.4.3 Surnrnary of Precipitate Results
3.5 Conclusions
3.6 References

CHAPTER 4: THE EVOLUTION OF SOLUTE CARBON D W G

CONTTNUOUS ANNEALING
4.1 Introduction
4.2 Literature Review
4.2.1 Mechanism of Bake Hardening
4.2.2 First Step - Cottrell-Effect
4.2.2.1 Strain Aging
4.2.2.2 Interstitial Atoms
4.2.2.3 Deformed Microstructure
4.2.3 Second Step - Carbide Precipitation
4.3 Experimental
4.3.1 Continuous Annealing Simulation
4.3.2 Agin Index
4.3 -3 Correlation between AGBHand Ijcr;u
4.4 Results
4.4.1 Correlation of Soiute Carbon and AaBH
4.4.1 Solute Carbon and Annealing Holding Time
4.4.2 Solute Carbon and Cooling Rate
4.5 Conclusions
4.6 References

CHAPTER 5: CORRELATION OF ACTIVATION AREA TO SOLUTE DRAG

STRESS
5.1 Introduction
5.2 Literature Review
5.2.1 Theoretical Activation Volume and Area
5.2.2 Type of Obstacles
5.2.3 Precision Strain Rate Testing
5.2.4 Cottrell-Stokes Law and the Haasen Plot
5.2.4.1 Pure Material
5.2.4.2 Obstacles in Materid
5.2.5 Determination of Solute Carbon
5.2.6 Solute Drag
5.2.6.1 Serrated Flow
5-2.7 Strain Rate Sensitivity Experiments
5.3 Experimental
5.4 Results
5.4.1 Strain Rate Sensitivity Test Results
5.4.2 Variation of Solute Carbon and Yield Stress with Annealing Time
5.4.3 Phase Diagram Construction
5.4.4 Observation of Solute Drag Effect in Haasen Plots
5.5 Conclusions
5.6 References

CHAPTER 6: SUMlMARY & CONCLUSIONS

VITA
 LIST OF FIGURES

Figure 2.1 EfTect of P, Si, and Mn on the mechanical properties of a Nb IF

sheet steel (0.005%C, 0.04%Nb)continuously anneaied at 830C for
40 seconds [7]. 7

Figure 2.2 Effect of NblC atomic percent ratio on the mechanical properties of a
Nb IF sheet steel (0.004%C) [14]. 12

Figure 2.3 A schematic diagram of a continuous annealing line [16]. 13

Figure 2.4 The effect of annealing temperature on the mechanical properties of

continuously annealed ULC-IF steei, (a) yield strength, @) elongation,
(c) n value, (d) r value [2]. 17

Figure 2.5 Effect of percent cold reduction and temper rolling of a low-carbon
(0.0074%C) steel (a) yield strength, (b) work hardenability,
(c) elongation, (d) AGBH[2 1 1. 18

Figure 2.6 A schematic diagram of a typical industrial continuous annealing-

galvannealing cycle [23, 241. 22

Figure 2.7 Effect of intemediate holding temperature (180 seconds) on AuBH

of
rapidly cooled sheet steels after annealing at 870C [1 O]. 23

Figure 2.8 A schematic diagram of the laboratory simulation of the industrial

process cycle fiom continuous annealing to paint baking. 25

Figure 2.9 Measurement of bake hardenability (AoBH)by tensile testing. 29

Figure 2.10 Effect of annealing temperature on AoBH(annealing tirne - 1 minute,

fast-air cooled). 31

Figure 2.11 Effect of cooling rate on AosH for N1 steel (annealed at 890C for 1
minute). 33

Figure 2.12 Effect of annealing time on AoeH for N 1 steel (annealed at 880C,
water quenched). 34

Figure 2.13 Effect of annealing temperature on yield strength (anneaiing time 1

minute, fast-air cooled). 36

Figure 3.1 Solubility relationships for Tic, VC., and MC. in ferrite. Hatching
indicates ranges fiom different references. Heavy lines represent
predicted relationship fiom Taylor [9].
Figure 3.2 Iso-solute carbon curves for varyng Nb/C ratio steels in femte. 46

Figure 3.3 Solute concentration profile around a precipitate particle during

dissolution [ 1O]. 47

Figure 3.4 Dissolution of sphencal cementite particle in austenite at 850C,

according to different dissolution models [ 131. 49

Figure 3.5 TEM bnght field micrograph and corresponding EDS spectrum
showing a large MnS precipitate in as-rolled condition. 55

Figure 3.6 TEM bright field micrograph and corresponding EDS spectmms
showing (A)cubodial (Nb,Ti)(C,N) precipitate and (B) globular NbC
precipitate, in as-rolled condition. 56

Figure 3.7 Precipitate size distribution of NbC in as-rolled condition. 59

Figure 3.8 Precipitate size distribution of NbC after annealing 1 minute at 880C.
60
Figure 3.9 Precipitate size distribution of NbC after annealing 3 minutes at 880C.
61
Figure 3.10 Precipitate size distribution of M C after annealing 6 minutes at 880C.
62
Figure 3.1 1 Precipitate size distribution of NbC after annealing 9 minutes at 880C.
63
Figure 3.12 EDS spectmm of a globular-shaped precipitate after annealing for 6
minutes at 880C. 64

Figure 3.13 Size evolution of NbC precipitates with various annealing times at
880C. 67

Figure 3.14 Density evolution of NbC precipitates with various anneding times at
880C. 68

Figure 3.15 Dissolution times of NbC particles at 880C in steel Nlcaiculated from
Aaron's mode1 [ 1 11. 69

Figure 4.1 A schematic diagram of the two-step increase in yield stress by bake
hardening [ 11. 74

Figure 4.2 A) plot of increase in lower yield stress vs. time after aging at different
temperatures [l], B) replot Figure 4.2a in temperature compensated
time, texp (QRT), where Q = 83736 J. 78
Figure 4.3 Effect of increased prestrain on lower yield stress, aging at 180C [ 11.
82
Figure 4.4 Relationship between aging index and solute carbon [ 141. 84

Figure 4.5 Relationship between AoBHand solute carbon content. 88

Figure 4.6 Effect of anneaihg time on solute carbon for Nt steel (annealed at
880C, water quenched). 9I

Figure 4.7 Effect of cooling rate on solute carbon for NI steel (anneded at 880C
for 1 minute). 92

Figure 5.1 A schematic diagram of the dislocation motion about a barrier. 98

Figure 5.2 Variation of the glide resistance to the dislocation motion against a
ban-ier. 98

Figure 5.3 A scbematic diagram of the effects of thermal and atherrnal obstacles
on a Haasen plot. 1O2

Figure 5.4 Solute drag profile calculated fiom James and Barnett analysis [14]. 107

Figure 5.5 Theoretical shape of Haasen plot. 108

Figure 5.6 Typical stress-strain curve at a temperature of 220C with strain

rate changes. Il5

Figure 5.7 Progression fiom an under-compensated (upper-left) to an over-

cornpeosated (lower-right) load drop, due to a strain rate change [ 191.
115

Figure 5.8 Variation in solute carbon with annealing time at 500C. 116

Figure 5.9 Variation of yield stress with annealing time for Nb-Ti (annealed at
500C) and Ti-only (annealed at 600C) IF steels. 117

Figure 5. 10 Variation in inverse activation area and yield stress with annealing
time (annealed at 660C). 119

Figure 5.1 1 a 1 a-carbide solvus line for two IF steels determined by strain rate
sensitivity measurernents and a low-carbon steel determined by internai
fiction. 120
Figure 5.12 Haasen plot showing precipitate effects and deviation from linearity due
to solute drag. Solute carbon content cdculated by back extrapolation
and solute drag methods. 123

Figure 5.13 Haasen plot showing dynamic strain agiag effects on the drag stress
profile. 124
 LIST OF TABLES

Table 2.1 Chernical composition of Nb IF steels (weight percent). 24

Table 3.1 Surnmary of precipitates present in a Nb IF steels [ I l 2, 31. 41

Table 3.2 Solubility product for Tic and NbCx in femte [ 9 ] . 44

Table 3.3 Summary of NbC precipitate size and density observed after various
annealing times at 880C. 65

Table 4.1 ~ A C J Bd~u e s (A - air, FA - fast-air, WQ - water

Correlation of A ( J to
quench). 86

Table 5.1 Chernicai Composition (weight percent) Il1

Table 5.2 Sample heat treatments, yield stress, and measured solute carbon
concentration. 114
 LIST OF SYMBOLS

activity
apparent activation area
Burgers vector
solute concentration
force
obstacle strength
Boltzmann's constant
dislocation segment length
strain rate sensitivity due to forest dislocations
strain rate sensitivity due to solute atoms
total number of atoms in solution per unit volume
strain hardening exponent
particle radius

plastic strain ratio

time
activation distance
parameter depending on the elastic constants
solute concentration at particle-rnatrix interface
solute concentration in matrix
concentration in precipitate
bulk difhsivity
Gibbs free energy of formation
AG@) Gibbs free energy of activation
AH0 standard enthalpy
K solubility product
R gas constant
AS0 standard entropy
T Temperature
V, precipitate molar volume
W work
a dimensioniess driving force

E strain rate
Y grain boundary energy
A dislocation core motion
P dislocations density
O normal stress
A aging index
A bake hardening
oo yield stress
odis dislocation-dislocation interaction stress
G solute drag stress
lower yield stress
o,k peak drag stress
a, stress due to solute atoms
WH work hardened stress
r shear stress
C structure
 CHAPTER 1

THE EVOLUTION OF SOLUTE CARBON DURLNG

CONTINUOUS ANNEALING OF LNTERSTITIALFREE
STEEL AND ITS EFFECTS ON BAKE HARDENABILITY

1 INTRODUCTION

An important objective of the automotive industry is to decrease the weight of

outer-body car panels while maintaining good formability and dent resistance. The use

of higher strength steels allows thimer gauge components and reduces the total

weight. However, conventional high strength sheet steels have insufficient formability

to meet the drawing requirements of today's more complex outer-body car panels.

These formability requirements limit the initiai yield strength of the steel to Ievels

insufficient to provide adequate dent resistance.

The development of bake hardenable steels offers an interesting solution which

cm satisfy both the formability and strength requirements. Bake hardenable steels

have relatively low yield strength and good formability afker continuous annealing,

therefore allowing for complex press-forming operations. After press forming, the

panels go through a paint-baking cycle, between 150 to 200C for 10 to 30 minutes, to

cure the paint. During the paint-baking cycle, an increase in yield strength occurs,

which is referred to as bake hardening. The bake hardenability, AOBH, of a steel is

deterrnined by measuring the difference between the flow stress at 2% prestrain and

the lower yield stress after subsequent heating at 170C for 20 minutes.
 Bake hardening can produce a 30 to 60 MPa increase in strength d e r press

forming. The increase in strength that occurs duriog the paint baking process is a

consequence of strain agkg and is controlled primarily by the amount of interstitial

carbon or nitrogen in the femte mauix. Several studies have shown that a

concentration of interstitial atoms in the range of 5 to 15 ppm is needed to obtain an

response. Solute concentration exceeding 15 ppm results in premature

effective AoBH

manifestation of yield point elongation, while a concentration below 5 ppm wili give an

insufficient increase in the AcBHresponse.

Solute content can be controiled by varying steel composition andfor

processing parameters. The compositional variables of concem are the interstitial

elements, carbon and nitrogen and the carbide-forming elements, such as titanium or

niobium, used to stabilize the interstitial atoms as precipitates. This leads to an under-

or over-stabilized composition. Changing processing parameters c m also affect the

amount of solute available for AmH. This is especially important during continuous

annealing, where the high temperatures partially dissolve the precipitates and the rapid

cooling rates trap the interstitial atoms in the matnx. A balance between composition

and processing is needed in order to obtain sufficient levels of doBH

while maintaining

dl other mechanical properties.

The main objective of this research is to study the evolution of solute carbon

and precipitates during the simulated continuous annealing of rnicroalloyed interstitial-

free (IF) steels and relate their development to the AoBH

response. The effects of steel

composition and several continuous annealing parameters on the Aoe~response are

measured. The change in precipitate size and density, and solute carbon content as a

hnction of theses continuous annealing parameters are also studied. Functionai

relationships between the evolution of the precipitates, solute carbon content, and

AGBHresponse during annealhg are developed. The result is a better understanding of

the evolution of bake hardening in IF steels.

To sirnplie die analysis, the subject matter has been divided into individuai

chapters, each with its own Literature review and conclusions. Chapter II descnbes the

compositions and processing of IF steels and the effects of these parameters on AaBH.

Experimental results are presented on the effects of continuous annealing parameters

(temperature, tirne, and cooiing rate) on the AoBH response for a series of IF steels

having varying Nb/C ratios. In Chapter III, transmission electron microscopy (TEM)

measurements of a Nb stabilized iF steel are reported for the size and distribution of

NbC precipitates following various continuous annealing soaking times. Chapter N

discusses in detail the mechanisms controlling bake hardening. The solute carbon

content in continuously annealed sarnples is determined from measurements of the

aging index and correlated with bake hardening values. In Chapter V7the theories of

thermally activated flow are summarked and strain rate sensitivity measurements are

used to determine the solute carbon content in a flly annealed Ti-Nb IF steel. The

qualitative observations on solute drag are correlated with the solute content. Finally,

the overall conclusions from these studies are given in Chapter W.

 CHAPTER II

FACTORS AFFECTING THE PROCESSING OF BAKE

HAIiDENABLE IF STEELS

2.1 INTRODUCTION

Recent demand for reducing car body weight requires steelmakers to supply

higher strength cold rolled sheet steels without sacrifcing press formability. However,

metallurgical methods of increasing the yield strength reduce the plastic strain ratio

and total elongation. Bake-hardenable IF steels offer an interesing solution. They

provide a formable sheet dunng the stamping process but have an incremental increase

in strength during the paint-baking cycle.

The effects of both the steel composition and the processing on the amount of

carbon lefl in solution are two important aspects to understand in order to control the

A0BH response in IF steels. The niobium-to-carbon atomic ratio, as well as other soiid

solution strengthening elements can change both the formability properties and the

arnount of available solute carbon. The carbides present in the cold-rolled sheet steel

cm be partiaily dissolved during the continuous annealing cycle with its high

temperatures and fast cooling rates, therefore trapping solute carbon in solution.

The objective of this chapter is to investigate the effects of continuous

annealing parameters: temperature, time, and cooling rate on the AOBHresponse for a

senes of IF steels having varying Nb/C ratios.

 2.2 LITERATURE: REVIEW

2.2.1 Production of IF Sheet Steel

Conventional interstitial free steels (C = 50 - 100 ppm, N = 40 - 80 ppm) were

developed commercially in the 1960's following the introduction of vacuum degassing

technology. Microalloying elements such as titanium andor niobium were added to

these steels to stabilize the interstitial carbon and nitrogen atoms as carbide and nitnde

precipitates. However, excessive arnounts of Ti andlor Nb were used which did not

combine with either C or N but stayed in solid solution, making the steel expensive and

af3ecting their mechanical properties.

ltra Low Carbon WC) IF steels were developed in the 1980's as a result of

the introduction of the bottom-blown converter and a reformed RH vacuum degasser.

This technology allowed the production of steels with significantly lower levels of C

and N (C = 30 - 50 ppm, N = 20 - 30 ppm). The principle driving force of LC IF

steels was the wide spread implementation of continuous anneaiing lines in Japan [ I l .

ULC-IF cold rolled sheet steels were found to have improved formability with

higher elongation and plastic strain ratios than conventional IF steels [2, 31. The

reduction in the arnount of both interstitial and alloying elements in the matrix Iead to a

measurable difference in the texture development and irnproved deep drawability. Hot

rolling parameters, including slab reheating temperature, finishing temperature, coiling

temperature, hot rolling speed, and reduction al1 have strong effects on the deep

drawability. These effects have been related to the size and distribution of the
precipitates and grain sire of the hot band [4]. The effects of steel composition and

processing are deeply interwoven, and one parameter often cannot be isolated fiom

interaction with another element or processing step.

2.2.2 Steel Composition

2.2-2.1 Precipitarion Strengthening

The increase in the strength of a steel achieved through the precipitation of Nb,

Ti, and/or V with C and N is known as precipitation strengthening. The increase in

yield strength with increased microalloying content is a result of the increase in the

precipitate volume fraction [SI. Ti steels show poor precipitate strengthening due to

the coarser 200-1000 nrn diameter precipitates, as compared to Nb steels which tend

to have a higher density of h e , 5-30 nm diarneter precipitates [6].

2.2.2.2 Solid Solution Strengthening

The most cornmonly-used alloying elements for solid solution strengthening are

phosphorus, silicon, and manganese. Their effects on the mechanical properties of an

annealed Nb steels are s h o w in Figure 2.1. As strength is increased, formability and

elongation decrease. Phosphorus and silicon have little effect on the plastic strain

ratio, i , while manganese additions considerably decrease both ; and elongation [7].
The addition of silicon has also been found to have a detrimentai effect on coating

adhesion [SI.
 0.03 0.05 0.07 0.09
P (wt %)
( i f i i f i !
0.3 0.5 0.7 0.9
Si, Mn (wt %)

Figure 2.1 Effect of P, Si, and Mn on the mechanical properties of a Nb IF sheet

steel (O.OOS%C, 0.04%Nb) continuously annealed at 830C for 40
seconds [7].
 Phosphorus is the most desirable solid solution strengthening element in IF

steels. It can produce a yield strength increase of up to 75 MPa. However,

segregation of phosphorus to grain boundaries promotes cold-work embrittlement [ 2 ] .

This segregation is enhanced in IF steels due to the depletion of carbon atoms at the

grain boundaries.

2.2.2.3 Effect of C and N on Mechanical Properties

To improve sheet fonnability, major attention has been focused on reducing the

total amount of carbon and nitrogen dunng steelmaking and the rernoval of these

interstitial elements fiom solid solution by the addition of stabiiizing alloying additions

[ 9 ] . Nitrogen is usually considered to be effectively removed From solid solution by

the formation AIN or TiN precipitates at high temperatures. The formation of carbide

precipitates such as TiC and NbC help remove carbon from solid solution. Carbide

precipitates have higher solubilities in austenite and ferrite thaa nitride precipitates.

High levels of solute carbon are undesirable for deep drawing applications

since carbon has a deleterious effect on the preferred recrystallization [Il. With an

increasing concentration of solute carbon, the strength of the ( 1 1 1 } recrystallization

texture component decreases. This decreases the i value, which correlates with the

strength of the { 1 1 1 ) recrystallization texture and forrnability is lower.

Dunng cooling after continuous anneaiing, some solute C rnay segregate to

grain boundaries. The solute carbon atorns located at the grain boundaiies may exhibit

different strain aging behaviour than solute carbon atoms located inside the grains
[ 1O]. The segregated carbon at the grain boundaries is more stable and strain aging

will occur at much higher temperatures, compared to atoms inside the grains. It has

been reponed by S. Hanai and N. Takemoto [ t 11 that sheet steels with fine grain sizes

display higher AoBH than sheets with coarse grains at the same solute carbon

concentration. These authors conclude that solute C at the grain boundaries can

influence AcBHthrough high temperature (170C)strain aging, and that solute carbon

at the grain boundaries is not aEected by room temperature strain aging, uniike solute

atoms found within the grains.

-
2.2.2.4 Stee[ Composition No/C Atomic Ratio

Both solute interstitial and substitutional elements affect the mechanical

properties of an IF steel. Therefore, control of the alloying element-to-carbon ratio is

extremeiy important. In order to control the C content accurately in IF steel products,

strong carbide- and nitride-forming elements, such as Ti and Nb, are added to combine

with the interstitial solute C and N. The amount of Nb required to fully stabilize an IF

steel can be deterrnined fiom

= 7.75C+ 6.65(N - AV1.93)

Nbctablc

= 7.75C+ 6.65(N - Til3.42)

NbshbIe (2.2)

Where Nb,bie, Ti, Al, C, and N are concentrations in wt. %. The use of the Al or Ti

factor depends on whether the steel was designed to fully precipitate stable AIN or

TiN precipitates [ 121.

 The amount of Ti required to fully stabilize an IF steel can be expressed as

Tiaiblc= 4C + 3.42N + 1.5s (2-3)

This equation was deveioped for HSLA steels and assumes that the C is combined as

Tic, N as TiN, and S as TiS, where S is the residual amount after MnS has

precipitated. However, in ULC steels, the amount of C, N, and S are significantly

lower and their relative proportions are different. The results from recent studies

[6,13] suggest that C is principally removed fiom solid solution by the formation of

Ti&&. Therefore, when stabilization of C and N occurs by the formation of Ti4CZS2

and TiN, the amount of Ti required will be given as

Tisbbi,= 3.42N + 2(1.5S) (2.4)

This uncertainty in the precipitation types and its dependence on N and especially S,

makes it difficult to accurately control the Ti*/C ratio, where Ti* is the Ti remaining

after the formation of Ti&& and TiN. Consequently, Nb steels are preferred for

AoBHapplications.

The alloy-to-carbon ratio plays a major role in determining the overall

mechanical properties of the steel. Steels with ratios below unity will contain solute

carbon throughout hot rolling, cold rolliiig, and annealing. Excessive arnounts of

solute carbon, as discussed above, considerably deteriorate the formability of the

annealed sheet but make carbon available for bake hardening. Steels with ratios above

unity will contain solute Nb or Ti throughout processing. These steels will develop

high elongation and r values [ 7 ] . However, increased alloying solute can result in
greater surface defects [2] and insufficient carbon in solution after annealing to obtain

adequate AoBHresults.

The effect of Nb/C ratio from 0.8 - 1.7 on tensile properties of IF sheet steels

continuously annealed at 830C are illustrated in Figure 2.2 [14]. With increasing

atomic ratio, yield strength and ;values increase while elongation decreases. Even in

a sub-stoichiometnc ratio of 0.8, the aging index measured before continuous

annealing remained at zero. This suggests that even at this low ratio, no carbon is in

solid solution and that any increase in NbIC ratio increases the amount of solute Nb.

2.2.3 Processing

2.2.3.I Batch and Continuous Annealing

M e r cold rolling, sheet steel has limited formability and high hardness.

Ductility must be restored through heat treatments which results in recrystallization.

Two processes are comrnonly used: batch annealing (BA) and continuous annealing

(CA).

Batch annealing involves placing sheet steel coils in a gas fired furnace with a

controlled atmosphere. The heating rate of the coi1 is approximately 30C/hour with a

soaking temperature of approximately 700C. The coils are then slow cooled at

1O0C/hour. The process takes several days.

Continuous annealing lines combine several processes including cleaning,

annealing, overaging or galvannealing, and sometirnes temper rolling, in one

continuous operation [15,16]. A schematic diagram of a continuous annealing line is

 0.8 1.0 1.2 1.4 1.6 1.8
Nb (at %) 1C (at %)

Figure 2.2 Effect of NbIC atomic percent ratio on the mechanical properties of a
Nb IF sheet steel (0.004%C) [14].
Figure 2.3 A schematic diagram of a continuous annealing line [16].
shown in Figure 2.3. With modern low carbon IF steels, overaging is not necessary

and has been replaced by a gaivanneal step. Radiant tubes fired by natural gas are used

for heating. Heating rates of 10 - 40C/second with soaking temperatures of 800 -

900C for 40 - 80 seconds are typical. Cooling rates of 20 - 200C/second are used
depending on the gaivannealing stage.

Batch annealed steels exhibit large through-thickness anisotropy in their

mechanical properties as a result of s m d differences in the textures produced [l].

These differences lead to higher r values after batch annealing than for continuous

annealing. In batch annealing, nearly ail the carbides will precipitate during the slow

cooling cycle. In continuous annealing, some dissolution of precipitates is expected to

occur at the high soaking temperature and solute carbon becomes trapped in solid

solution dunng rapid cooling. This solute carbon makes continuous annealing

favourable for producing bake hardenable steels.

2.2.3.1. I Recrystai1ization

During high temperature continuous annealing, recrystallization of the

defomed femte occurs. Control of the recrystailized texture to the y fibre texture

with rolling plane normal parallel to < I l l > and not the unfavorable a fibre with

rolling direction in <1 IO>, is essentiai for developing fomable sheet steel. There has

been much debate on which factor holds the greater responsibility for degrading the

formation of the desired y fibre texture; solute atoms or precipitate particles. Solute

carbon at recrystailization retards the development of (1 11) fibre and carbonitride

particles interfere with the nucleation and growth of recrystailized grains [17]. There

is aiso strong evidence that solute Nb and Ti can also dictate the formation of a

desirable recrystallization texture [7]. A recent in-depth study by Hutchinson et al.

[18] concluded that the preferred texture in IF steels is predominately a result of the

fact that the rnatrix is fiee fiom interstitial atoms, and that niobium and/or titanium

help by maintaining a fine grain size after hot rolling through the formation of

carbonitride particles.

However, precipitated Nb andor Ti particles affect the recrystallization

process in a minor rote. Nb steels usually have higher recrystallization temperatures

than Ti steels, by approximately 50C. The retarded recrystallization rate of Nb steels

is attributed to the high density of fine particles present f i e r hot rolling and coiling,

which retards dislocation motion more effectively than the coarser particles in a Ti

steel.

2.2.3.1.2 Effect of Continuous Annealing Temperature on

Mechanical Properties

During soaking at the annealing temperature, the equilibriurn solubility of

carbon can be reached. The high cooling rate prevents carbide precipitation and

produces a supersaturated carbon concentration. This solute carbon will affect the

mechanical properties such as formability and bake hardening [19]. Continuous

annealing studies [2] indicate that higher annealing temperatures promote increased

values of elongation, strain hardening exponent (n), and ;, and decreased yield and
tensile strength as shown in Figure 2.4. The high ; and n values are a result of the
texture developed during recrystallkition and are independent of the heating rate [ 1O].

This is why continuous annealed IF steels are excellent for deep drawing applications.

2.2.3.2 Temper Rolling

Temper rolling is used to meet strip shape and surface finish requirements and

eliminate any discontinuous yielding effects. It also increases the unpimed dislocation

density which affects the bake hardening mechanism [20, 211.

Expenments done by C.S. Lee and B.K. Zuidema [2 11 show that varyifig the

cold rolling reduction between 65 and 85% did not significantly influence the final

mechanical properties whereas temper rolling between 1 and 5% does influence those

properties. With increasing temper rolling, initial yield strength is increased (Figure

2.5a) but total elongation and AoBHdecrease. (Figure 2 3 and d respectively) Work

hardening (Figure 2Sb) has a maximum for temper rolling between 1.5 and 2%, and

decreases with greater percent reduction. Therefore, considering the total strength

increment due to work hardening and AOBH, (Figure 2Sd) temper rolling reduction

between 1.5 - 2 % is ideal. With greater reduction, AoBHdecreases.

Figure 2.4 The efFect of anneding temperature on the mechanical properties of
continuously annealed ULC-IFsteel, (a) yield strength, (b) elongation,
(c) n value, (d) r value [Z].
 -- CR L m
--, am
---*-* CR-
---*--
CRllP

Figure 2.5 Effect of percent cold reduction and temper rolling of a low-carbon
(0.0074%C) steel (a) yield strength, (b) work hardenability,
(c) elongation, (d) AmH[2 1 1.
 2.2.3.3 Press Forming

The sheet rnetal is stamped into the desired component shape. Dunng plastic

deformation, the density of mobile dislocations is increased and work hardening

occurs. Deformation of the sheet steel usually ranges from 1 to 5 % strain. With

larger prestrains, Aoex decreases (221. With increased prestrain, the dislocation

density becomes higher, the concentration of carbon per length of dislocation line is

less, and AoBHis decreased. However, with higher strains, more work hardening has

occurred. Therefore, the strength increment which occurs during paint baking, is most

important in the low prestrained region of the press formed part.

2.2.3.4 Paint Baking

Automotive steels, after being formed, assembled, and painted, pass through a

drying fumace at temperatures between 150 to 200C for 10 to 30 minutes to harden

the paint. The change in the properties, particularly the increase in yield stress after

the paint-baking process, is known as the bake hardening effect. Essentially, during

the baking process, strain aging occurs in steels containing small arnounts of solute

carbon in the matrix. The rnechanisms controlling bake hardening will be discussed in

more detail in Chapter 4. Previous investigations have s h o w that a range of 5 - 12

ppm of solute carbon is needed to obtain optimum bake hardenability. Above this

range, room temperature aging occurs and fonnability is decreased. Below 5 ppm, an

insufficient AcsH response is obtained.

 2.2.3.5 Hot Dip Galvannealing Process

Hot-dipped galvannealed steels are used to give corrosion resistance to

automotive outer panels. The hot-dip galvanneaiing process has become an integrated

part of the continuous anneaiing process for IF steels. M e r continuous anneaiing, a

zinc-coating is applied and subsequent annealing treatments occur prior to final

cooling [16, 23, 241. A schematic diagram of the continuous anneaiing-galvannealing

cycle is shown in Figure 2.6.

The galvannealing treatment requires an additional intermediate holding period

during the cooling stage. M e r annealing, the strip is cooled to 460C and passed

through a molten zinc bath containing 0.12 - 0.20 wt.% Ai. Excess liquid metai is
removed by a set of air nozzles called wipers. The stnp is subsequently anneaied

between 480435C for 12 seconds. Interdifision of the iron and zinc occurs, giving

rise to a coating which comprises of a senes of iron-zinc intennetallic compounds.

The strip is then cooled to ambient temperature. The galvannealing temperature and

tirne must be carefully controlled to prevent overalloying of the coating.

This intermediate holding treatment may promote the reprecipitation of

carbides resulting in an undesirable decrease in the bake hardenability of the product.

The influence of such intermediate holding tirnes on bake hardenability have been

investigated and the results are shown in Figure 2.7. A considerable drop in bake

hardening is seen in the 800-700C temperature range due to NbC precipitation, and in

the 400-300C temperature range due to precipitation of Fe3C. Little decrease in bake

hardenability is seen in the 400-600C range where the galvannealing treatment

occurs. Therefore, significant metallurgical changes affecting bake hardenability will

not occur during galvannealing. The rnetallurgical processing principles applied to

uncoated cold-rolled products can be also applied to hot-dip galvannealed products

[13, 241.

Galvannealed coatings adhere better to Ti+Nb IF steels than Ti IF steels 12,

101. In Ti IF steels, the iron/inc interface is non-uniform because zinc atoms cm

difise rapidly through the grain boundaries. This creates high concentrations of the

irodzinc alloy at the steel surface grain boundaries while depleting surrounding areas,

resulting in poor powdering resistance. This rapid diffusion of zinc does not occur in

Ti+Nb steel because of the small amounts of solute carbon that reside at the grain

boundaries. This prevention of localized difision allows uniform distribution of the

alloying layer at the steel substrate surface.

0
Soaking Time 40 - 60 sec
-
Soaking Temp 870 890'
Rapid Cooling to 550C

Galvanneal
00-520C for 12 sec

45 sec to 460C

Final
Cooling Rate

150
Time (sec)
 Intermediate holding temp. TH, t

Figure 2.7 Effect of intermediate holding temperature (180 seconds) on AoBHof

rapidly cooled sheet steels afier annealing at 870C [ l O].
2.3-1 Nb Interstitial-Free Steels

Four Nb IF sheet steels with varying Nb/C atomic ratios were supplied by

Stelco Steel. The steels were received as 3 x 10 inch (7.6 x 25 cm) sheets, cut fiom

the cold rolled coils. The pnor processing of the steel was as follows: initially the

steels slabs were reheated to 127SC and passed 9 times through a reversing roughing

mil1 were the thickness was reduced fiom 244.2 to 26.2 mm. The transfer bar was

then reduced to 2.9 mm in a 5-stand continuous finishing rnill. The sheet exited the

finishing miIl at 923OC and was cooled by water spray to 729T and coiled. The sheet

was finally cold rolled to a thickness of 0.77 mm. The chernical compositions of the

steels and their Nb/C atornic ratio were determined at Stelco and are given in Table

Table 2.1 Chernical composition of Nb IF steels (weight percent)

Code C N Ti Nb Si Mn P S Al Nb/C*
NO7 0.0040 0.0047 0.005 0.022 0.010 0.148 0.084 0.013 0.021 0.70
N1 0.0052 0.0039 0.003 0.040 0.005 0.239 0.067 0.015 0.055 0.99
NI 1 0.0034 0.0033 0.004 0.029 0.012 0.107 0.071 0.007 0.037 1.10
NI5 0.0032 0.0028 0.003 0.037 0.004 0.243 0.064 0.014 0.040 1.49

* atomic ratio
2.3.2. Laboratory Simulation

A schematic diagrarn of the laboratory simulation of the industrial processing,

from continuous annealing to paint baking, is shown in Figure 2.8.

2.3.2.I Continuous Annealing

The 3 x 10 inch (7.6 x 25 cm) sheets were degreased in an acetone solution. A

K-type thermocouple was spot welded in the center of one face. The sheet was then

mounted on its side in a jig so that both faces would be exposed in the fumace. The

sheet sample was placed in an air fmace and the temperature of the sheet was

recorded by a data acquisition cornputer. This set-up obtained a heating rate of

--12"C/second from arnbient temperature to 750C. From 750C to the desired

amealing temperature, a heating rate of -3OC/second was obtained. This allowed the

sheet to reach the annealing temperature within approxirnately 120 seconds. The

accuracy of the temperature control was approximately k C dong the length of the

sample, as determined through frnace trials, and MC fiorn the reported value.

At the end of the annealing hold time, the sheet was removed from the fmace

and cooled via one of three methods: air, fast-air, or water quench. The cooling rates

for the three methods were 1 1, 20, and 200C/second, respectively.

2.3.2.2. Temper Rolling

The annealed sheet was pickied in a 10% HCI solution to rernove any scaie.

The sheet was then ternper rolled -2% on a 10 cm Stanat rolling mill. The percent
Figure 2.8 A schematic diagrarn of the laboratory simulation of the industrial
process cycle fiom continuous annealing to paint baking.
reduction during the temper roll was caiculated by measuring the initial and final length

of the sheet.

2.3.2.3 Press- Forrning

The temper-rolled sheet was sheared into 3, 1 x 8 inch (2.5 x 20.3 cm) strips.

Tensile samples were machined fiom the strips corresponding to the ASTM E-517

Standard with a gauge section of 2 x 0.5 inch (5 x 1.2 cm). Industrial press-forming or

stamping was simulated by pre-straining the tensile samples to 2% plastic strain in a

screw-dnven Instron testing machine at a cross-head speed of 0.02 inchedminute

(0.051 cmhinute). The final load at 2% strain was recorded and the work hardened

flow stress ( at ~
0 ~ this)strain was caiculated.

2.3.2.4 Paint-Baking

The pre-strained sample was heated in an air fmace at 170C for 20 minutes

and air cooled. The accuracy of the temperature control was approximately E C .

The sample was then remounted in the testing machine and deformed at a cross-head

speed of 0.02 or 0.05 inchedminute (0.051 or 0.127 crnhinute). The paint-baked

samples usually exhibited discontinuous yielding and the lower yield stress (ON) was

recorded in each case. In cases when discontinuous yielding was not observed the

0.2% off-set method was used to determine the yield stress. The higher cross-head

speed was used when samples were taken to failure, while samples deformed at the
slower cross-head speed were deformed just past the yield point. No difference in

bake hardening behaviour was observed due to the different cross-head speeds.

The difference between the lower yield stress d e r the paint baking treatment

and the flow stress after the 2% pre-strain (aw) was used as the rneasure of
(oLY)

bake hardening, ACTeH.

A c B =~ CTLY - CTWH (2.5)

The corresponding load-elongation curves for determiring AceH f?om OLY and o~

are s h o w in Figure 2.9. A minimum of two samples were tested for each condition

and the mean value of AoeH is reported. The standard error for the mean value of each

A o s value
~ was determined and the average of al1 these values was found to be K2.4 1

MPa. This standard error is applied to al1 values and is shown as a reference in each

figure in the results section.

 17OoC,20 min
/
'
Elongot ion

Figure 2.9 Measurement of bake hardenability (AoBH) by tende testing.

2.4.1 Annealing Temperature

The variation in AGBHwith annealing temperature for the four steels annealed

for I minute and fast air cooled is shown in Figure 2.10. A C Y increases
~~ with

increasing anneding temperature for ail four steels. The arnount of NbC precipitate

dissolution during annealing increases with increasing anneaiing temperature. Also,

with increasing annealing temperature there is more time for precipitates to dissolve

during heat up. Both of these factors increase the amount of carbon present in the

matrix after annealing.

The four steels exhibit significant differences in bake hardening response which

is related to the Nb/C atomic ratio. Under-stabilized and stabilized steel (NO7 and N I )

have higher values of AceH at al1 anneaiing temperatures. The over-stabilized steels

(N 1 1 and N 15) shows very poor bake hardening response even at extremely high

temperatures. N 15 finally obtains a bake hardening value of 30 MPa with an annealing

temperature of 920C, which is well above the range of industriai annealing

ternperatures. This poor bake hardenability of over-stabilized steels is due to the

excess solute N b present which rnakes reprecipitation easier upon cooling.

 800 820 840 860 880 900 920 940

Annealing Temperature ( O C )

Figure 2.10 Effect of annealing temperature on AoeH (anneaihg time - 1 minute,

fast-air cooled).
2.4.2 Annealing Holding Time

The variation in AoBHwith annealing tirne for steel NI held at 880C and then

fast air cooled is shown in Figure 2.11. A s H increases to approximately 32 MPa in 3

minutes. m e r 3 minutes, no further change in the A O ~ His seen. This suggests that

the matrix has reached its solubility lirnit of carbon a n d o r niobium at 3 minutes. With

no frther increase in solute carbon, no funher increase in bake hardening can occur.

The high AcBH value of 27 MPa after 1 minute, compared to the maximum bake

hardening value after 3 minutes, suggests that the majority of NbC dissolution occurs

in the first minute of annealing. Therefore, holding tirnes greater than 1 minute do not

significantly affect AaBH.

2.4.3 Cooling Rate

The variation in AoeH with cooling rate is shown in Figure 2.12 for N 1 steel
annealed at 890C for 1 minute. The three cooling rates: air, fast-air, and water

quench were used. doBH

decreases with decreasing cooling rate. With slower cooling

rates, more dissolved carbon c m return to the niobium clusters and/or iron and

reprecipitate as NbC or FesC in their respective temperature ranges. Water quenching

produces the least reprecipitation of NbC or FelC upon cooling and gives the highest

value of AaBH.
 O 1 2 3 4 5 6 7

Tirne (min)

Figure 2.1 1 Effect of annealing time on AoBHfor N 1 steel (annealed at 880C,

water quenched).
 50 1 O0 150 200

Cooling Rate ('Clsec)

Figure 2.12 Effect of cooling rate on AneHfor NI steel (annealed at 890C for I
minute).
2.4.4 Yield Strength

The change in yield strength, for N07, N 1 1, and N15, after continuous

annealing for 1 minute at various temperatures and fast air cooling is given in Figure

2.13. The yield strength decreases with increasing annealing temperature with steel

NO7 showing the largest decrease in yield strength. At 820C,a value over 300 MPa

is obtained suggesting that full recrystallization has not yet occurred. With higher

annealing temperatures, full recrystallization is able to occur and yield strength

decreases rapidly. The over-stabilized NI1 and N 15 steels also decrease in strength

with increasing temperature but not to the sarne extent as the under-stabilized N07.
 800 820 840 860 880 900 920 940

Annealing Temperature (OC)

Figure 2.13 Effect of annealing temperature on yield strength (annealing time 1

minute, fast-air cooled).
 2.5 CONCLUSIONS

The effects of three continuous annealing parameters on bake hardenability

were studied. The effects of amealing temperature on yield strength and bake

hardenability for under and over-stabiiized niobium IF steels, as well as the effects of

cooling rate and holding times on a stabilized niobium IF steel, were investigated. The

following conclusions are drawn from these results:

Steels with NbIC atomic ratios above one produce poor bake hardenability values

due to the excess of solute niobium present. This excess Nb promotes

reprecipitation of NbC upon cooling.

The solubility limit of NbC is approached after one minute at the annealing

temperature and reached within three minutes.

The cooling rate and annealing temperature greatly affect bake hardenability values.

Slower cooling rates allow more time for reprecipitation, therefore resulting in less

solute carbon. Higher annealing temperatures have higher solubility limits and

allow more time for dissolution, leading to more solute carbon available for bake

hardening.
 2.6 REFERENCES

R.K. Ray, J.J. Jonas, and R E . Hook, Interrzutional Materials Review, Vol. 39,
No. 4, 1994, pp. 129- 172.

J.R Fekete, D.C. Strugala, and 2.Yao, JOM,Jannaq 1992, pp. 17-2 1.

H. Takechi, ISIJ International, Vol. 34, No. 1, 1994, pp. 1-8.

Y.S. Hwang and H.C.Chen, in International Synposium on Low-Carbon Steels

for the go's, eds. R. Asfahani and G. Tither, TMS, Pittsburgh, Pfi, 1993, pp.
475-480.

R P . Pradhan, AIME Annual Meeting, Dallas, TX, 1982.

M. Hua, C.I. Garcia, and A.J. De Ardo, in Phase Transformations During

Thermai/Mechanical Processinn of Steel, eds. E.B. Hawbolt and S. Yue, CIM,
Vancouver, BC, 1995, pp. 285-290.

G. Tither, C.I. Garcia, M. Hua, and A.J. DeArdo, in Proceedine of the

International Forum for Phvsical Metallurw of IF Steels, Iron and Steel Institute
of Japan, Tokyo, Japan, 1994, pp. 293-324.

R.C. Coleman and M. lefford, in Svmposium on Low-Carbon Steels for the 90's,
eds. R. Asfahani and G. Tither, TMS, Pittsburgh, PA, 1993, pp. 465474.

G. Krauss, D.O. Wilshynsky, and D.K. Matlock, in Interstitial-Free Sheet Sheet:

Processine. Fabrication and Properties, eds. L E . Collins and D .L Baragar, C M ,
Ottawa, ON, 1991, pp. 1-14.

K. Sakata, S. Satoh, T. Kato, and 0. Hashimoto, in Proceedine of the

International Forum for Phvsical Metallurw of IF Steels, Iron and Steel lnstitute
of Japan, Tokyo, lapan, 1994, pp. 279-288.

S. Hanai and N. Takemoto, Tetszi-to-Hagane, Vol. 68, No. 9, 1982, p. 1 169.

Y. Tokunaga and H. Kato, Nippon Steel, Intemal Report, 1986.

M. Hua, C.I. Garcia, and A.J. De Ardo, Mrtalizirgcal and Muterials

Trnnsactiorzs A , Vol. 2 8 4 Septernber 1997, pp 1769- 1780.

N. Ohashi, T. Irie, S. Satoh, O. Hashimoto, and 1. Takahashi, SAE Paper No.

810027, 1981.
M. Kannappar, Evolution of Microstmcture and Texture During Continuous
Annealing of Cold Rolled Ti-Stabilised Interstitial-Free Steel, Ph. D. Thesis,
University of British Columbia, Vancouver, BC, 1994.

V. Jagannathan, JOM, August 1993, pp. 48-5 1.

K. Tsunoyama, S. Satoh, M. hanaka, H. Abe, and K. Yamato, in Iron and Steel

in the Automotive Industry, ed. M.S. Orr, ISS, Hamilton, ON, 1990, pp. 168-
172.

W.B. Hutchinson, Intermtional Metal Reviews, Vol 29, No. 1, 1 984, p. 25

H.T. lunius, Iron and Steel E w l e e r , August 1989, pp. 45-5 1.

F.D.Bailey, RP. Foley, and D.K. Matlock, in Proceedines on: Hieh-Strenmh

Steels for the Automotive Industry, ed. R Pradhan, Iron & Steel Society,
Baltimore, MD, 1994, pp. 1 19- 133.

C.S. Lee and B.K.Zuidema, in Proceedings on: High-Strenmh Steels for the
Automotive Industry, ed. R. Pradhan, Iron & Steel Society, Baltimore, MD,
1994, pp. 103-1 10.

P. Eisen and H.P. Hougardy, Steel Research, Vol. 64, No.819, 1993, pp. 43 1-
436.

K. Taylor, Bethlehem Steel, Private Communication.

J. Thomson, Stelco Steel, Pnvate Communication.

 CHAPTER III

THE EVOLUTION OF NbC PRECIPITATES DURING

CONTINUOUS ANNEALING OF AN IF SHEET STEEL

3.1 INTRODUCTION

Microalloying elements such as Nb and Ti combine with interstitial elements to

form stable carbo-nitride precipitates. This can leave the matrix relatively free of

interstitial solute atoms which can gready effect the microstmcture and mechanical

properties. The use of continuous annealing with its high soaking temperatures and

fast cooling rates c m partially dissolve NbC precipitates and trap solute carbon in

solution. The control of the amount of solute present after continuous annealing is

critical in the production of bake hardenable steels. Severd theoretical models to

predict the dissolution rate of carbide precipitates have been developed.

A quantitative transmission electron microscopy (TEM) study was performed

to characterize the size and distribution of M C precipitates following various

continuous annealing soaking times. The results were compared to theoretical

dissolution models.
 3.2 LITERATURE REVIEW

3.2.1 Precipitation in Nb IF Steels

Four types of precipitates can form in Nb IF steels during hot rolling and

coiling [ l , 2, 31. Their size and shape are summaiized in Table 3.1. Sulfur is

scavenged by manganese to fonn large rounded precipitates, 0.5 to 1 p m in size.

Nitrogen can be stabilised with either Al or Ti. Al will usually be used to combine

with N as AIN. However, an alternative to Al is Ti which precipitates N as TiN.

When al1 the Ti is consumed, Nb will combine with the remaining N at lower

temperatures. This leads to a (Ti,Nb)N precipitate with a Ti rich center. The sizes of

Al and (Ti,Nb) nitndes both range from 40 to 100 nm but differ in shape; AIN being

rounded while (Ti,Nb)N is cubic.

At lower temperatures (below 950C) the remaining Nb and C precipitate as

NbC. These particles are formed by independent nucleation and growth, forming a

high density of srnall, 5 to 40 nm, particles.

Table 3.1 Surnmay of precipitates present in a Nb IF steels [l,2, 31

Type Size (nm) Shape

MnS 500 - 1000 rounded
(Ti,Nb)N 40 -100 square
AIN 40 -100 rounded
NbC 5 -40 globular
 In general, Nb steels show a higher density of small precipitates (10% ppt/mm2)

as compared to Ti steels (106 ppt/rnm2). The density refers to the number of

precipitates visible in a given area, as observed by electron microscopy techniques.

This difference in densities is related to the method that carbon is stabilized in the

precipitates. In Ti steels, carbon is stabilized by the transformation of TiS + T4S2C2,

between 1250 and 950C [4]. Therefore, there is relatively little carbon at 950C

when Tic is favoured to form. Hence, the nucleation of separate Tic particles is not

favoured because of the insufficient supersaturation needed to overcome the

nucleation barrier. Rather, remaining carbon will grow epitaxially on pre-existing

Ti&C2, producing a coarser dispersion of precipitates [SI.

3.2.2 Precipitate Solubility Product

At a given temperature, there is a solubility product which provides

quantitative information about the thermodynamic equilibnum between the matnx and

precipitate. This information is detennined by the temperature, the concentration of

elements, and the type of precipitate formed.

The solubility product can be derived from the analysis of the Gibbs free

energy at equilibrium.

Where AH()and dSO are the standard enthalpy and entropy, respectively, of the

reaction and a is the activity. When pure AxByexists as the standard state, the activity
can be considered equal to unity. In addition, for dilute solutions it can be

approxirnated that a, = atom Fraction of 4 that is if the activity coefficient is assumed

to be unity. The equivalent concentrations cm be expressed as the solubility product,

K,

where [A] and p]are the element concentrations in solution in wt.% and C' and D'

are constants corresponding to the entropy and enthalpy of formation of the

p recipitate, respectively.

Different types of mixed precipitates may form within a system when there are

multiple microalloying and interstitial elements. In these cases, the relationships fail

because of the rnutual solubilities of the elernents. There have been attempts made to

quanti@ the solubility of mixed precipitates [6, 7, 81, but for this study only pure

binary compounds will be considered.

The solubility data for Nb and Ti carbides and nitrides, in the austenite

temperature range have been extensively studied. However, there has been relatively

little experirnental work done on precipitate solubilities in the femte. A method of

obtaining femte solubility relationships ffom known austenite solubility products by

using information on the activity coefficients of the solutes of interest has been

developed by Taylor [9]. The resulting solubility product relationships for ferrite are

given in Table 3.2, and plotted in Figure 3.1. Equilibnum solute carbon concentrations

were calculated for varying Nb/C ratio and temperature using the NbC solubility

product given in Table 3.2. The concentrations are plotted as iso-solute carbon curves
in Figure 3.2. At 880C the solute niobium and carbon concentrations are calculated

to be 0.0 128 and 0.00 168 in W.%, respectively.

Table 3.2 Solubility product for Tic and MC, in ferrite 191.

3.2.3 Dissolution of Precipitates

Several analytical models have been developed to describe the kinetics of

particle dissolution in a metal at elevated temperatures [ 10, 1 1, 121. None of these

models are exact as rnany approximations rnust be made about the system and the

difision field around the particle. However, these approximations are suficiently

accurate to give the essential physics of the problem and to give valuable predictions

about particle dissolution.

The accepted theory of precipitate dissolution was proposed by Whelan [IO]

who considered dissolution to be approxirnately the reverse of precipitate growth. His

mode1 considers a sphericai particle embedded in an infinite rnatrix. The concentration

profile is shown schematically in Figure 3.3. The migration rate of the phase interface

is obtained from the local mass balance at the interface. By inserting the flux given by
Figure 3.1 Solubility relationships for Tic, VC,, and MC, in ferrite. Hatching
indicates ranges from different references. Heavy lines represent
predicted relationship from Taylor [9].
 700 750 800 850 900 950
Temperature (OC)

Figure 3.2 Iso-solute carbon curves for varying NblC ratio steels in ferrite.
 - - --

Distance

Figure 3.3 Soiute concentration profile around a precipitate particle during

dissolution [ 1O].
Fick's law, Whelan obtained the following dissolution rate expression

where r is initial particle radius, t, time, Dm the element bulk difisivity and a, the

dimensionless driving force given as

Cmand C,, refer to the concentrations in the matrix and precipitate respectively and C,

refers to the equilibrium concentration of solute at the particle-rnatrix interface. For a

binary intermetallic, C, is given by the solubility product.

WheIan's solution cannot be integrated anaiytically and therefore numerical

methods must be employed. From equation 3.3, the l/r term arises from the steady-

state part of the difision field and the l/& term arises fiorn the transient part. In a

later dissolution mode1 by Aaron [II], the transient term was neglected and a simpler

expression for the particle radius as a function of tirne is given

This expression is valid for short periods of time when there is no impingement of

diffusion fields from neighbouring particles.

-4 cornparison of dissolution models proposed by Whelan, Aaron, and A g e n

[12], is shown in Figure 3.4. The last mode1 uses an elaborate numencal approach that

is solved via a computer prograrn.

 W, helan

Figure 3.4 Dissolution of sphericai cementite particle in austenite at 850C,

according to different dissolution models [13].
3.2.4 Precipitate Coarsening

When precipitates become slightly superheated at an elevated temperature,

there is a tendency for smaller particles to dissolve and reprecipitate on Iarger particles

according to the dictates of the Gibbs-Thompson relationship. The classical theory for

particle coarsening was developed independently by Lifshitz and Slyovoz [14] and by

Wagner [ 151. The kinetics of particle coarsening are normally controlled by volume

diffusion through the matrix. In steady state, an expression for the particle radius as a

function of time is given as

where y is the particle-matrix interface energy and V,, the molar volume of

precipitates. This relationship has many simplifjring assumptions but fits well with

observed experimental data.

 3.3 EXPERIMENTAL

3.3.1 Experimental Procedure

Steel NI was used to examine the evolution of precipitates after continuous

annealing with various holding times. The simulated annealing cycle described in

detail in Chapter 2 was used to perform these heat treatments. Transmission electron

rnicroscopy was used to study the changes in precipitate size and density.

The N 1 steel was sectioned into smaller sheets, approximately 4 x 10 cm. The

samples were processed through the continuous annealing simulation at 880C for

hold times of 1, 3, 6, and 9 minutes and then water quenched.

3.3.2 Experimental Apparatus

3.3.2.1 Transmission Electron Microscopy

A Philips CM 20 transmission electron microscope (TEM), operating at

200 kV, was used to characterize the size and distribution of precipitates by carbon

replicas and thin foils. The precipitate size was determined by measuring the diameter

in two directions and taking the average. A minimum of 200 precipitates were

measured for each condition to determine an accurate size distribution. Precipitate

density was determined by photographing a selected region and dividing the number of

precipitates visible by the total area, giving units of ppt/prn2. A minimum of 10 fields

were used to calculate the density for each condition.

 Chernical composition of the precipitates was determined using the Voyager

energy dispersive spectroscopy (EDS) system. The EDS system uses an ultra-thin

window so the detection of elements with atomic numbers greater than 5 is possible.

3.3.2.2 Preparation of Carbon Replicas and Thin Foils

To prepare carbon replicas, the as-rolled and anneaied samples were sectioned

into 5 mm squares and mounted in bakelite. The rnounted samples were lightly ground

on silicon carbide paper and polished with 6 p and then 1 pm diamond paste.

Sarnples were then cleaned in an ultrasonic cleaning bath for 20 minutes to remove any

impunties picked-up fiom polishing. The polished and cleaned samples were etched

lightly in 2% Nital and a thin layer of carbon 150 - 200 A thick was deposited on the
surface using a vacuum evaporator. The replicas were removed by scoring the carbon

film into a grid of 2 mm squares and immersing in a 10% Nitai solution. The Nitai

over-etched the Fe-matnx and the replicas could be floated off in a dish of distilled

water. The replicas were supported on 100 or 150 mesh copper grids. New solution

was used for each sample and containers were washed between samples to prevent

cross-contamination.

To prepare thin foi1 samples, the 3- and 6-minute annealed sheet steel samples

were sectioned into 15 mm squares and ground to a thickness of about 100 p m using

silicon carbide paper. Discs 3 mm in diameter were punched out of the thimed sheet

and electropolished using a Fischione twin-jet electropolisher. The solution used was
10% perchloric acid in glacial acetic acid. The bath temperature was maintained below

-40C and a voltage of 30V was used.

 3.4 RESULTS

3.4.1 Characterization of Starting Material

TEM replicas of the as-roiled NI sheet steel revealed two types of precipitates:

MnS and NbC. The MnS precipitates were generaily ovai in shape and ranged in size

fkom 200 to 800 nm.Figure 3.5 shows a typical MnS precipitate with corresponding

EDS spectrurn. Figure 3.6 shows two differently shaped precipitates. The larger

cubodiai particle (Figure 3.6a) is a Nb-rich carbide with trace arnounts of Ti and

possibly N. A distinct nitrogen peak was never seen in any EDS scan. This could be

due to the close proxirnity of the N and C K, energy peaks and that the C peak was

usually very large resulting fiom the carbon replica film. The smaller globular particle

Figure 3.6b) is a Nb-carbide.

It was found, with furt.her study of other precipitates, that some globular

particles contained trace amounts of Ti and some cubodial particles contained no Ti.

There is no distinguishing chemical feature between the differently shaped precipitates.

Therefore, both cubic and globular precipitates will be referred to as "NbC"

throughout the rest of this work. However, it is understood that these particles are

really complex (Nb,Tii,)(CyNi,) compounds with x and y being close to 1 in value.

Some globular-shaped precipitates showed Al as well as Nb and C peaks. The

percentage of these precipitates in the matrix is insignificantly srnall and will not be

considered frther.
Figure 3.5 TEM bright field micrograph and corresponding EDS spectrum
showing a large MnS precipitate in as-rolled condition.
Figure 3.6 TEM bright field micrograph and corresponding EDS spectmms
showing (A) cubodial (Nb,Ti)(C,N) precipitate and (B) globular NbC
precipitate, in as-rolled condition.
3.4.2 Precipitate Sue and Density Evolution

3.4.2.1 As- Rolled Condition

The measured density of precipitates in NI steel in the as-rolled condition was

1 18 ppt/pm2. The average precipitate diarneter was determined to be 12.9 m. This

consisted of ail the NbC precipitates ranging from the small globular, 3 nm, to the

large cubodial, 60 m, precipitates. The histogram in Figure 3.7a shows the size

distribution of NbC precipitates in the as-rolled condition. The rnicrograph in Figure

3-7b shows a representative precipitate distribution.

3.4.2.2 Continuously Annenled Conditions

Afier annealing for 1 minute at 880C, the density of NbC precipitates was

found to be 47.6 ppt/pn2 and the average precipitate diameter was 13.4 nm. The

measured precipitate size distribution is given in Figure 3.8a and a representative

micrograph is shown in Figure 3 .ab.

After anneaiing for 3 minutes at 880C, the density of NbC precipitates was

24.7 ppt/prn2 and the average precipitate diameter was 16.6 nm. The histogram in

Figure 3.9a shows that the percentage of small precipitates ( 4 5 nm) has decreased

while the percentage of medium sized precipitates (-25 nm) has increased, cornpared

with the as-rolled condition.

M e r annealing for 6 minute at 880C, the density of NbC precipitates was

19.2 ppt/pm2 and the average precipitate diameter was 19.3 nm. On average 25 EDS

spectmms were obtained from precipitates from each anneaiing condition. It was
found that the majority of precipitates annealed for 6 minutes had EDS spectra with Ti

K, energy peaks, regardless of the precipitate shape. Figure 3.12 shows a

representative EDS spectrum with a Ti K, peak from a globular precipitate which has

been annealed for 6 minutes. These precipitates differ significantly ftom the small

globular precipitates present in the as-rolled condition (Figure 3.6b) that did not show

any Ti present.

After anneaiing for 9 minutes at 880C the density of NbC precipitates is

17.9 ppt/Clm2and the average precipitate diarneter is 17.3 nm. There is relatively little

change in the size distribution or density between holding times of 6 and 9 minutes, as

seen in Figures 3.10 and 3.1 1.

 5 10 15 20 25 30 35 40 45 50 55 60
Precipitate Diameter (nrn)

Figure 3.7 Precipitate size distribution of NbC in as-rolled condition.

 5 10 15 20 25 30 35 40 45 50 55 60
Precipitate Diameter (nm)

Figure 3.8 Precipitate size distribution of NbC after anneaiing 1 minute at 880C.
 5 10 15 20 25 30 35 40 45 50 55 60
Ptecipitate Diameter (nm)

Figure 3.9 Precipitate size distribution of NbC after annealing 3 minutes at 880C.
 5 10 15 20 25 30 35 40 45 50 55 60
Precipitate Diameter (nm)

Figure 3.LO Precipitate size distribution of M C after annealing 6 minutes at 880C.

 5 10 15 20 25 30 35 40 45 50 55 60
Precipitate Diarneter (nm)

Figure 3.1 1 Precipitate size distribution of NbC after annealing 9 minutes at 880C.
Figure 3.12 EDS spectmm of a globular-shaped precipitate after amealing for 6
minutes at 880C.
3.4.3 Surnmary of Precipitate Results

A summary of the NbC precipitate size and density observed for the various

holding times, as well as the as-rolled condition is given in Table 3.3 and graphically

represented in Figures 3.13 and 3.14. The average of the standard error for the

+
precipitate diameters was found to be f 0.68 nm and for the densities 8.16 ppt/pm2.

Table 3.3 Sumrnary of MC precipitate size and density observed

f i e r various anneaiing tirnes at 880C.

Annealing Time Mean Diameter Density

(min) (nm) (ppt/P2)

As-Rolled 12.9 -t 0.72 118.1 k 2 0

1 13.4 k 0.64 47.6 I7.3
3 16.6 i 0.64 24.7 It 3.2
6 19.3 + 0.62 19.2 k 8.5

The density of NbC precipitates decreases rapidly while the average precipitate

diameter increases during the first three minutes of annealing. With longer holding

times up to nine minutes, there is relatively little change in the precipitate density or

size. Close examination of the various histograms show a decrease in the number of

small, 5 to 10 nm, precipitates and an increase in medium-sized, 20 to 30 nm,

precipitates. The decrease in the frequency of smaller particles is due to precipitate

dissolution, which fits well with theoreticai dissolution models. Particles under 15 nm
in size could completely dissolve in under 3 minutes based on Aaron's dissolution

model illustrated in Figure 3.15. However, the increases in medium sue particles

resembles precipitate coarsening but at these annealing temperatures, the mode1 for

coarsening predicts little to no change in precipitate size over a ten minute time range.

Based on Aaron's dissolution model medium sized, 20 to 30 nrn, precipitates will only

reduce in size by 2 or 3 m in a 3 minute annealing tirne. Further dissolution will not

occur after 3 minutes as the rnatrix has become saturated. With the complete

dissolution of the finer precipitates the frequency of the medium sized precipitates will

increase, resembling precipitate coarsening.

 4 6

Holding Time (min)

Figure 3.13 Size evolution of NbC precipitates with various ameding times at
880C.
 O 2 4 6 8 1O

Holding Time (min)

Figure 3.14 Density evolution of NbC precipitates with various annealing times at
880C.
 O 200 400 600 800 1000 1200 1400 1600

Time (sec)

Figure 3.15 Dissolution tirnes of NbC pmicles at 880C in steel N 1 calculated from
Aaron's mode1 [ I l ] .
 3.5 CONCLUSIONS

The precipitate evolution of MC dunng continuous annealing was studied for

N I steel. The size and density of the precipitates were studied by transmission

electron rnicroscopy. The experimental results obtained were cornpared with

theoretical dissolution and coarsening models. The following conclusions can be

drawn from the results obtained in this study:

The overall precipitate density initiaily decreases with holding times up to 3 minutes

while precipitate dissolution is occurnng. With extended holding time, 6 minutes

and longer, there is relatively little change in density and dissolution is no longer

occurnng.

The average precipitate diameter increases with holding time and reaches a constant

value after approximately three minutes.

 3.6 REFERENCES

Wen Mu Bing, in International Svm~osiumon Low-Carbon Steel's for the 909s,

eds. R Asfahani and G. Tither, TMS,Pittsburgh, Pa., 1993, pp. 15 1-1 56.

M. Hua, C.I. Garcia, and AJ. De Ardo, in Phase Transformations Durinq

Thermal/Mechanical Processine of Steel, eds. E.B.Hawbolt and S. Yue, CIM,
Vancouver, BC, 1995, pp. 285-290.

R.E. Hook and H. Nyo, Metallurgicai Transactions A, Vol. 6 4 1975, pp. 1443-
1451.

N.Yoshinaga, K. Ushioda, S. Akamatsu, and 0-Akisue, ISIJ International, Vol.

34, 1994, pp. 24-3 2.

M. Hua, C.I. Garcia, and A.J. De Ardo, Metalizrrgrcol and Materiais

Transaclions A, Vol. 2 8 q 1997, pp. 1769- 1780.

J.G. Speer, I R . Michael, and S.S. Hansen, M e ~ a l l z i r ~ c Tramactions

al A, Vol.
1 8 4 1987, pp. 21 1-222.

R.C Sharma, V.K Lakshrnanan, and J. S. Kirkaidy, Metaliwgzcal Tra~~suction

A,
Vol. 1 5 4 1984, pp. 545-553.

P.R. Rios, Material Science and Engineering, Vol. A 142, 199 1, pp. 87-94.

K.A. Taylor, Scripta Metallzirgica, Vol. 32, No. 1, 1995, pp. 7-12.

M.J. Whelan, Metal Science Journal, Vol. 3, 1969, pp. 95-97.

H.B. Aaron, MefalScienceJoiirnal, Vol. 2, 1968, pp. 192-193.

J. Agren, J. Phys. Chern. Solids, Vol. 43, No. 4, 1982, pp. 385-391.

J. Agren, Scat~dinavianJournal of MetalIurgy, Vol. 19, 1 990, pp. 2-8.

LM. Lifshitz and V.V. Slyozov, J. Phys. C'hem. Solids, Vol. 19, No. 1, 1961, pp.
35-50.

C. Wagner. Z. Elektrochern.. Vol. 65. 196 1. o. 58 1.

 CHAPTER IV

THE EVOLUTION OF SOLUTE CARBON DURING

CONTINUOUS ANNEALING

4.1 INTRODUCTION

Bake hardenable steel derives its designation from the strengthening which

occurs during the cunng of the outer-body panel paint, typically done at 150 to 200C

for 10 to 30 minutes. This results in a finished part with an increased yield strength of

approximately 30 to 60 MPa. The AcBHproperty is a consequence of strain aging and

has a strong dependence on the dislocation microstmcture and the concentration of

interstitid solute atoms.

A general understanding of the effects of dislocation distribution and solute

carbon content is needed in order to oljtimally control the AGBH response. This

chapter will summarize the rnechanisms of AoBHand the factors affecthg thern. The

aging index will be used to determine the solute carbon content in the matnx after

various continuous anneding cycles. These results \vil1 be correlated with A o e data.
~
 4.2 LITERATURE REVIEW

4.2.1 Mechanism of Bake Hardening

The strengthening mechanism in bake hardening is based on strain aging by

interstitial solute atorns. In modern Nb IF steels, nearly al1 of the nitrogen is combined

as duminum or titanium nitrides and only small amounts of carbon in solid solution

control the ABH response. There are still uncertainties about the mechanisms of bake

hardening. Current theory proposes that the increase in yield strength during paint-

baking occurs by a two step process: step one is the segregation of carbon to pin

dislocations by the Cottrell-effect and step two, the strengthening by precipitation of

fine carbides [l]. The first step is generally completed before the second step begins.

A schematic illustration showing the contribution of the two step mechanism for

increasing yield stress by bake hardening is s h o w in Figure 4.1.

4.2.2 First Step - Cottrell-Effect

Dunng the paint-baking cycle, the first step is initiated by the segregation of

carbon atoms to dislocations. Cottrell-atmospheres then form, which effectively

anchor the dislocations. The carbon atorns produce tetragonal distortion as well as a

dilation. The former interacts with screw dislocations and the latter with edge

dislocations, giving nse to an increase in strength [2]. This is a form of static strain

aging.
 1 . step 2. step
Cot trell-effect CottreII-effect
and pariicle hardeniris

log. a g i n g i i m e

Figure 4.1 A schematic diagrarn of the two-step increase in yield stress by bake
hardening [ i l .
4.2-2. I Struin Aging

Strain aging is the change in properties of a metal that occurs by the

interaction, during or d e r plastic deformation, of point defects and dislocations.

These point defects are usually interstitial solute atoms. When they occur during

plastic deformation, the process is termed dynamic strain aging. When the property

changes occur afker plastic deformation, the process is termed static strain aging.

Strain aging can be a valuable and economical means of strengthening steels.

The rnost common rneasure of property changes during strain aging is the effects

produced on the ioad-elongation curves in a uniaxial tensile test. The effects on the

mechanical properties can be seen as an increase in the yield stress, a return of yield

point elongation with an increase in Luders strain, an increase in the ultimate tensile

stress, and a decrease in total elongation [3, 41.

The theory of strain aging proposed by Cottrell and Bilby in 1949 [SI is the

generally accepted theory. In general, for strain aging to occur, three criteria must be

met: 1. unpinned dislocations must be present, 2. interstitial solute atoms must be able

to d i a s e to the dislocations, and 3. the interstitial solute atoms must then be able to

pin the dislocations. Strain aging can be divided into two stages. Initially, there is a

rapid increase in the upper yield point by strain induced short range ordering of

interstitial solutes in the stress fieid of a dislocation. This is known as the Snoek-effect

[6]. It is complete in approxirnately the tirne required for a single jump of an

interstitial atom in the stress field of a dislocation. The second and slower stage is

caused by long range diffusion of the interstitial solute atoms towards the dislocation
core. These atoms concentrate in the stress field of a dislocation and form a Cottrell-

atmosphere. The incurred interaction energy between the solute and the dislocation's

strain field anchor the dislocations effectively and gives rise to the increase in strength.

The yield strength obtained afker straining and aging is hence dependent on the local

dislocation density and the solute concentration. These parameters are infiuenced by:

the amount of strain and the work-hardening rate of the steel during prestraining, the

initial level and location of dissolved interstitial atoms, and the femte grain size [7].

4.2.2.2 Interstitial A toms

Interstitial carbon and nitrogen atoms are very mobile in femte iron at ambient

temperatures. In mild steeis, strain aging is largely attnbuted to nitrogen with a

smaller contribution from carbon. The reason for this is that nitrogen is much more

soluble than carbon at low temperatures. For temperatures less than 100C, the

solubility of nitrogen is approxirnately 100 times greater than that of carbon [3, 81.

It is expected that the magnitude of strain aging wilt be proportional to the

concentration of interstitial solute atoms in the matrix. The size of the upper yield

point is controlled by the formation of Cottrell atmospheres that lock the dislocations

during aging. At higher interstitial solute contents, complete locking of mobile

dislocations is expected to occur at lower aging temperatures due to the reduced

distance between the solute atoms and distocations. If the initial level of interstitial

solute atoms in the matnx is decreased, only minimal locking will occur. However, it

has been found that the quantity needed to cornpletely lock dislocations is quite srnall
[7, 91. A nitrogen content of 1 ppm can produce a detectable change in the

mechanical properties of a steel and 10 to 20 ppm can completely lock ail dislocations

[8]. The degree of atrnosphere formation at tirne I has been theoretically expressed by

Cottrell and Bilby [SI as

where no is the total number of atoms in solution per unit volume, A is the permitted

dislocation core motion of locked dislocations in the order of the core radius, and A is

a parameter that depends on the elastic constants.

In a low-carbon steel prestrained 2% and aged at 170C, the first step is

completed almost immediately as shown in Figure 4-23. When the same steel is aged

at 50C, completion of the first step takes over 2000 minutes. These results agree well

with equation 4.1, assuming 7 atoms per atom plane of dislocation [IO]. Plotting the

data fiom Figure 4.2a as increase in yield stress vs. 1 - exp (2)
- , wiih Q = 83736 1,

gives a straight line as shown in Figure 4.2b. The 83736 J corresponds to the

activation e n e r c for carbon difision in f e ~ t e2nd shows that the strengthening

mechanisrns are controlled by the rate of carbon difision.

 i

: aging after
- 2% prestrain
b

10 1O0 1000 10000

Aging time i n minutes

Figure 4.2 A) plot of increase in Iower yield stress vs. time afier aging at different
temperatures [Il, B) reptot of Figure 4.2a in temperature compensated
time, fexp(-Q/RT), where Q = 83736 J.
 4.2.2.3 Deformed Microstructure

The density of dislocations will affect the number of solute atoms needed to

lock al1 the dislocations in a given time. The fraction of solutes that have segregated

to dislocations can be expressed as

where p is the density of dislocations per unit area. Using the data fiom Figure 4.2,

equation 4.2 predicts a dislocation density of 10"/cm2. It is difficult to relate this

dislocation density to the amount of prestraining as both heterogeneous and

homogeneous deformation has taken place dunng prestraining.

Temper rolling is a heterogeneous deformation process that develops a non-

uniform microstructure of dislocation tangles. The density of the dislocations is

greater near the surface than at the half thickness of the sheet. The extent of the non-

uniformity of the dislocation rnicrostmcture depends on the tensiIe/cornpression

balance of the temper rolling process.

The formation of a non-uniform dislocation density leads to a cornplex

situation dunng strain aging because the kinetics and magnitude of strain aging

depends on the dislocation density. With a non-uniform dislocation microstructure, a

carbon concentration gradient may be developed dunng aging between regions with

different dislocation densities. This may lead to a situation where the carbon in

solution may be too srnall in some places to saturate and lock al1 the dislocations

during aging, which could affect the overall mechanical properties of the steel [ I LI.
 During starnping the sheet steel is not uniforrnly deformed as in AoBHtesting.

The 2% prestrain is used as an average strain for the whole auto-body sheet. In lower

prestrained areas the dislocation fiee space is increased, making the tirne and

temperatures important factors in allowing for the difision of carbon to the

dislocations. As prestrain increases, the dislocation density increases and the amount

of carbon per dislocation unit decreases, therefore decreasing the AoSH.

The stamping process typically varies between plane-strain tension and biaxial

stretching but is sirnulated experirnentally by uniaxial tensile deformation. The aging

behaviour of specimens prestrained by biaxial stretching are qualitatively similar to

samples prestrained in simple tension. However, biaxial prestraining produces a higher

yield stress as compared to uniaxial prestrained specimens. These differences in yield

stress are related to the relative severity of the change in strain paths.

4.2.3 Second Step - Carbide Precipitation

Precipitation hardening of fine carbides occurs after the formation of the

Cottrell atmospheres of step one is complete. The structure and rnorphology of these

carbides has not yet been determined. However, from experiments performed by Elsen

and Hougardy [ I l , it is believed that the carbides form coherent particles in the matnx.

The second stage particles form at dislocations and increasing prestraining can

decrease the total strengthening increment. Figure 4.3 shows the effect of increased

prestrain on the increase in lower yield stress. At these baking temperatures (180C)

the increase in prestraining, and dislocation density, does not affect the tirne necessary
for the initial increase in yield stress due to the Cottrell effect (20 MPa increase).

Assuming that at 5% prestrain ail the dislocations are locked by the solute atoms, the

formation of second phase particles strip the Cottrell atmosphere unless new solute

atoms are able to difise to the dislocation to take their place. Therefore, with a lower

dislocation density (1% prestrain) new precipitates will form but the Cottrell

atmosphere will be replenished, leading to an increase in strength due to both

precipitate hardening and the Cottreil atmospheres. However, with a higher

dislocation density al1 the solute atoms are used to lock the dislocations and when the

precipitate forms, the stripped atmospheres will not be able to reforrn. Therefore, the

increase in precipitate hardening will be offset by the decrease in the Cottrell effect,

leading to a lower overall strengthening increase.

 1 Aging at 180C

1 1O 100 1O00 10000

A g i n g time in minutes

Figure 4.3 Effect of increased prestrain on lower yield stress, aging at 180C [Il.
 4.3 EXPERIMENTAL

43.1 Continuous Annealing Simulation

Steel N1 (composition given in Table 2.1) was used to examine the evolution

of solute carbon during a continuous annealing cycle with various holding tirnes and

cooling rates. A simulated industriai continuous annealing cycle, discussed in Chapter

2, was used to perform the heat treatments. The solute carbon content was

determined by the aging index (AoAi) method [13, 141, and the AOM values were

corretated with AOBHvalues.

4.3.2 Aging Index

The same procedure for determining the AcBHresponse was used to determine

the AoAI response. However, the annealed sheet was ternper rolled 0.8% instead of

2%, the pre-straining was 7.5% instead of 2%, and the aging was done for 60 minutes

at 100C instead of the 20 minutes at 170C. The difference between the lower yield

stress afler aging (ory)and the flow stress &er the 7.5% pre-strain ( o ~was
) used to

determine the aging index. The solute carbon content was determined by using the

correlation between the aging index and the solute carbon content shown in Figure 4.4

[14]. The accuracy of the aging index correlation is k ppm. Each data point in the

results section represents an individual tensile test.

Figure 4.4 Relationship between aging index and solute carbon [14].
4.3.3 Correlation between AcAIand AGBH

was determined by annealing a 3 x 10

The relationship between AcAi and daBH

inch (7.6 x 25 cm) sheet and sectioning it into 3, 1 x 8 inch ( 2.5 x 20.3 cm) strips.

The aging index was measured on one stnp and the solute carbon content was

determined. The AoBHvalue was then determined by the same procedure outlined in

Chapter 2, on the remaining two strips. Sirice the same initial sheet was used to

determine the AcBH and AcAI values, a correlation between the AoBK value and the

amount of solute carbon found by the AoAi method can be determined.

 4.4 RESULTS
4.4.1 Correlation o f Solute Carbon and AcBH

The correlation between solute carbon and A o B ~response is shown in

Figure 4.5. Each point represents an individual bake hardening simulation. The solute
carbon content was determined by the aging index and related to the AOBHresponse as
explained in the expex-imental procedure. Table 4.1 gives the AoM, [Clppm, and AGBH
values for N 1 steel with varying heat treatments.

Table 4.1 Correlation of AoM to AoBHvalues (A - air, FA - fast-air, WQ -

water quench).
I Annealing Treatment
l AG, @Pa)

1 min. @ 883C - WQ 57.0

51.8

3 min. @ 878OC - WQ 67.8

54.8

A 6 min. @ 875C - WQ

6 min. @ 884OC - WQ 57.0

51.8

r ra7
1 rnin. @ 879OC - FA

1 min. @ 926C - FA 70.1

1 min. @ 855C - A

1 min. @ 881C - A 33.4

1 min. @ 882C - A
A linear regression was performed on the data and plotted on Figure 4.5. The best-fit

equation is

AoBH= 1.74[C] + 11-63 (4-3)

where [Cl is the solute carbon concentration in ppmw and AoBHis the bake hardening

response in MPa. The variation of the data with respect to the best-fit equation is

known as the adjusted squared multiple residual, R*. As R* approaches 1 the better

the fit of the model. R* for equation 4.3 was detennined to be 0.77. Equation 4.3

shows that there is a direct relationship between solute carbon content and the AGBH

response. This relationship can now be used to determine the amount of solute carbon

needed afier processing in order to obtain a certain level of AcBH. Equation 4.3 will be

used to compare measured solute carbon results from this Chapter with AoBti results

obtained from Chapter 2. In general, automotive manufactures are demanding a AGBH

value of 30 MPa for outer-body components [13]. From Figure 4.5, a AGBHvalue of

30 MPa corresponds to 10.5 ppm of solute carbon.

Cornparhg the AoBH versus [Cl curve in Figure 4.5 to the AGAI versus [Cl

curve in Figure 4.4, shows that for equal arnounts of solute carbon, the A ~ e r r

procedure produces a smaller increase in strength than the AoM procedure. This

difference in strength is due to the effect of the deformed microstmcture on the

mechanism of strain aging. The effect of the deformed microstructure on both the

Cottrell and precipitate hardening effect was discussed in sections 1.2.2.3 and 4 - 2 3 .

The deformed microstmcture developed for AaoHand AoAi responses are significantly
Figure 4.5 and solute carbon content.
Relationship between doBH
different due to the amount of ternper r o h g and prestraining imparted to the samples.

The difference in dislocation densities could affect the degree of pinning for the

amount of carbon available during the first step of aging.

The procedure that develops the greater density of dislocation would hinder

the second step by removing all the solute atoms fiom the matrix. However, the

second step is not thought to contribute significantly to the strengthening of the steel

because of the time allowed. A detailed TEM study of tested samples to see if any

precipitation of fine carbides has occurred would confirm this hypothesis.

4.4.2 Solute Carbon and Annealing Holding Time

The variation in solute carbon content, with annealing time for steel N1 held at

880C and then water quenched, is shown in Figure 4.6. The solute content increases

to approximately 16.5 ppm with holding time of 3 minutes. M e r 3 minutes, there is

no further change in the solute content. Equation 4.3 predicts that the A o s ~values

should range fiom 36.5 MPa after I minute to 40 MPa d e r 6 minutes, when the

solubility limit of 17 pprn has been reached. Figure 2.11, presents the same heat

treatment schedule with experimentally determined AoBH values. Results range fiom

27 to 33 MPa over the same tirne period. The predicted A o e ~values are greater then

the experimentally determined values but show the same trend. This is consistent with

the observation that the solubility limit of the carbon has been reached after 3 minutes

and that the equilibnum carbon content is -17 pprn as calculated theoretically in

Chapter 3, Figure 3 -2.

 Steel NI has a Nb/C atomic ratio of 1, suggesting that, theoretically, there

should be no excess Nb or C in the matrix pnor to continuous annealing. However,

this is not usually the case and some carbon is left in the matrix. Through the scope of

these expenments, it is impossible to determine the starting solute carbon

concentration in the as-rolled material. The data suggest that the majonty of

dissolution must occur in the first minute of annealing. This is also supported by the

AaBHand precipitate data presented in Chapters 2 and 3 .

4.4.3 Solute Carbon and Cooling Rate

The variation in solute carbon with cooling rate is s h o w in Figure 4.7 for N1

samples annealed at 880C for 1 minute. The three cooling rates; air, fast-air, and

water quench, discussed in the experirnental procedure of Chapter 2, were used. The

amount of solute carbon decreases with decreasing cooling rate. The decreasing trend

in solute carbon in Figure 4.7 corresponds to the decreasing trend in A O B ~in Figure

2.11. with decreasing cooling rates. A cornparison between the experimentaily

detennined AoeH values in Figure 2.1 1 and the predicted values from Equation 4.3 is

not possible due to the different annealing temperatures used for the two experiments.

These results confirm the hypothesis proposed in Chapter 2 that the A o B response
~

decreases with decreasing cooling rate, due to reprecipitation of NbC andlor FeiC

precipitates. This is an example of autotempering of steel [IO].

 O 1 2 3 4 5 6 7
Holding Time (min)

Figure 4.6 Effect of annealing tirne on solute carbon for N1 steel (annealed at
880C,water quenched).
 50 1O0 150 200
Cooling Rate ('Clsec)

Figure 4.7 Effect of cooling rate on solute carbon for N 1 steel (annealed at 880C
for 1 minute).
 4.5 CONCLUSIONS

The effect of continuous annealing holding time and cooling rate on the

evolution of solute carbon was studied for steel NI. The solute carbon content was

measured by the aging index method and results were correlated with the bake

hardening response. The following conclusions are drawn:

The theoretical solubility limit of 17 ppm of carbon is approached afler one minute

of annealing and reached within three minutes.

Slower cooling rates dlow more time for the reprecipitation of NbC and/or FejC,

therefore leaving less solute carbon in the matnx.

A relationship between arnount of solute carbon and the bake hardening response

has been developed and the theoreticai and experirnental results were compared.
 4.6 REFERENCES

P. Elsen and H.P.Hougardy, Sleel Research, Vol. 64, No.8/9, 1993, pp. 43 1-
436.

M. E . Fine, Metallurgical and Mirterials Transaction A, Vol. 27, Septemb er

1996, pp. 23 97-24 i 8.

W.C. Leslie, The Phvsical Metallurw of SteeIs, McGraw-Hill, New York, NY,
198 1 pp. 79-90.

J.R. Low and M. Gensamer, T m s . AIME, Vol. 158, 1944, pp. 207-249.

A.H. Cottrell and B . A Bilby, froc. Phys. Soc., 1949, pp. 49-62.

J.L. Snoek, Physzca, Vol. 8, 1941, p. 71 1.

F.D. Bailey, R.P. Foley, and D.K. Matlock, in Proceedines on: High-Streneth
Steels for the Automotive Industry, ed. R Pradhan, Iron & Steel Society,
Baltimore, MD, 1994, pp. 1 19-133.

Effects of Strain Aging and Related Phenomena upon the Toughness of Carbon
and Carbon-Manganese Weld Metals, Ontario Hydro Research Division, Report
NO. 86-323-K, 1987.

D.V. Wilson and B. Russell. Acta Metczllz~r~ca,

Vol. 7, September 1959, pp.
628-63 1.

G.R. Speich, Trans. AIME, Vol. 245, December 1969, pp. 2553-2563.

J.M. Rubianes and P. Zimmer, in Proceedinss on: Hkh-Stren-gth Steels for the
Automotive Industry, ed. R. Pradhan, Iron & Steel Society, Baltimore, MD,
1994, pp. 1 1 1-1 18.

D.V. Wilson and B. Russell, Acta MetaZIurgica, Vol. 8, July 1960, pp. 468-479.

K. Ushioda, O. Akisue, K. Koyama, and T. Hayashida, in Development in the

Annealins of Sheet Steels, eds. R. Pradhan and 1. Gupta, TMS/AIME, Cincinnati,
OH, 199 1, pp. 26 1-286.

H. Katoh, H. Takechi, N. Takahashi, and M. Abe, in Technolog of

Continuouslv Annealed Cold-Rolled Sheet Steel, ed. R. Pradhan, TMSIAIME,
Detroit, MI, 1984, pp. 37-58.

B.K. Zuidema, National Steel, Private Communication-

 CHAPTER V

CORRELATION OF ACTIVATION AREA TO THE SOLUTE

DRAG STRESS

5.1 INTRODUCTION

With the existing theoretical knowledge of dislocation-defect interactions a

microstructural parameter obtained fiom strain rate sensitivity measurements can be

used to quantitatively assess the defect population. The thermodynarnic response

measured under tensile conditions is the change in Gibbs fiee energy, G, as a function

of stress. This partial differential = A V is referred to as the activation volume

and is measured using strain rate change experiments. The activation volume

theoretically reveals the interaction of a dislocation segment with obstacles to its

motion

Precision strain rate sensitivity measurements have been used to determine the

solute carbon concentration in a Nb-Ti IF steel. The linear superposition principle is

employed to separate out the contributions from solute hardening and forest

hardening. A plot of the inverse activation area versus applied stress is initially linear

and its intercept at the 0.02% proof stress can be correlated to the carbon

concentration in the rnatrix.

Solute drag can manifest itself during strain rate sensitivity experiments.

Assuming the additivity of the solute drag stress to the dislocation interaction stress,
the influence of solute drag c m be portrayed on the Haasen plot as the solute drag

profile giving rise to a second method to determine solute carbon concentration in

solid solution.

5.2.1 Theoretical Activation Volume and Area

From the theory of thermal activation of dislocation motion and the interaction

of a dislocation with an obstacle to its motion, the relationship

can be formulated, where E, represents total attempt rate spectrum of participating

strain centers for the activation process and AG(@, the Gibb's free energy of

activation. Rearranging for AG and differentiating with respect to stress at constant

temperature and interna1 structure, X, gives

1 .i

The intemal structure refers to precipitate size, shape, volume fraction, solute

distribution, grain size, dislocation density, and dislocation distribution.

The apparent activation volume, AV, is the volume that a dislocation must

overcome by thermal activation and is defined as the negative stress derivative of the

activation energy [ I l . The apparent activation area is given as Aa' = AVib, where the
apparent activation area represents the area of the slip plane swept by the dislocation

dunng activation. Therefore, equation 5.2 can be written as

5.2.2 Type of Obstacles

A simple physical mode1 of a dislocation-obstacle interaction is schematicaily

shown in Figure 5.1. With an extemally applied stress, 7 , the work done in moving a

dislocation segment of length, 1, through a distance, d, is

AW = fx = ~ b l d (5 -4)

where f is the force per unit length acting on the dislocation. The tem bld has units

of volume and is generaily referred to as the activation volume.

A force-distance diagram for a given obstacle is shown in Figure 5.2. For a

given applied stress, a dislocation will glide. When the dislocation interacts with an

obstacle, the obstacle will provide a resistive force. The work done to overcome this

obstacle is shown by the shaded area between points 1 and 2. These points represent

stable and unstable equilibrium positions respectively. The area above the shaded

region represents the free energy of thermal activation, AG. This is the energy above

the applied stress necessary for the dislocation to pass an obstacle. The shape of the

force-distance cuve changes with the type of obstacle to be overcome. These

obstacles or bamiers to dislocation motion can be classed into two categories, thermal

and athermal.
Figure 5.1 A schematic diagram of the dislocation motion about a barrer.

+d< Distance, x

Figure 5.2 Variation of the glide resistance to the dislocation motion against a
barrier.
 Thermal obstacles, such as solute atoms or clusters, are generaily discrete and

widely spaced acting at a point on a dislocation. They generally have small stress

fields and will impart a strong local force on a dislocation's motion. Therefore, they

are also referred to as short range obstacles. The distance to overcorne these obstacles

is small and a thermal fluctuation can usuaily provide the energy needed to overcome

these obstacles. An athemd obstacle has a high interaction energy and cannot be

overcome by thermal activation. Athermal obstacles occupy many atomic sites, such

as precipitates or particles, and produce large stress fields which can extend over the

length of a dislocation. They are also referred to as long range obstacles.

5.2.3 Precision Strain Rate Testing

In order to determine the activation volume, the right-hand side of equation 5.3

must be solved for expenmentally. This c m be achieved by differential strain rate

change tests, done at constant temperature, pressure, where atmospheric pressure is

assumed, and structure. The macroscopic aspect of a materiai does not normally

change during a typicai strain rate change, but the dislocation density and distribution

may change. The largest inhibitor to maintaining a constant dislocation structure

during a strain rate change is the interaction between the testing machine and the

sample. When a strain rate change occurs, a matenal's load-bearing capacity changes.

The sudden rate change causes the load fiame of the machine to elastically compensate

to maintain equilibrium. This elastic compensation of the load frarne interferes with
the measurement of the change in the load bearng capacity of the material, which from

equation 5.3 must be in a state of constant structure.

To overcome this difficulty, Champion et al. [2], and Saimoto and Basinski [3]

developed a tensometer with an interactive magnetic device which nulIlfies the elastic

effects of the load frame when a strain rate change occurs. At the instant a strain rate

change occurs, an electromagnet inserted in the compression part of the test frarne is

activated which contracts the outer test fiame. If the amount the load frame contracts,

controlled by the operator through a voltage input, is equal to its elastic effects, a

perfect compensation can be obtained [4]. A constant structure is maintained and the

response of the abrupt rate change is only due to the true elastic response of the

material [SI. The expenmentd measurement simulates as close as possible the

theoretical partial differentiation given in equation 5.3.

5.2.4 Cottrell-Stokes Law and the Haasen Plot

5.2.4.1 Pure Material

The Haasen plot [6] is produced by plotting the inverse apparent activation

area against fiow stress. In a pure material, the Haasen plot is linear and passes

through the origin. The slope is proportional to the strain rate sensitivity, m, and

represents the dislocation-dislocation interactions. This linearity represents the

Cottrell-Stokes law [7]. According to the conditions outlined above, the Cottrell-

Stokes law States that the strain rate sensitivity is independent of work hardening and

that the stress is proportional to I/Aa' [8].

 5.2.4.2 Obstacles in Material

In real materials, dislocation motion is hindered by a variety of obstacles.

Glide resistance depends on how the effects of the different kinds of obstacles

superimpose on each other.

Superposition of the glide resistance can be linear, square, or of a general

power form. Consider the principles of linear superposition developed by Kocks,

Argon, and Ashby [ 9 ] . They considered two types of obstacles, one strong and the

other weak. When the density of the weak obstacles is an order of magnitude greater

than the strong obstacles, the glide dislocation bows out between the two strong

obstacles and the bowed-out segment moves against the resistance of the weak

obstacles. Together their effective glide resistance is a linear sum of the two kinds of

obstacles. These principles imply that the reciprocals of the activation areas are also

linearly superimposed. Therefore, superpositioning of solute hardening on forest

hardening is expected to follow the linear superposition Iaw. Where the forest

dislocations are strong, obstacles and solute atoms are weaker but have much higher

density.

A material with obstacles other than dislocations will still obey the Cottrell-

Stokes law but the line will not pass through the origin on the Haasen plot. The

influence of the obstacles will shift the Haasen plot line vertically up or down as shown

in Figure 5.3. When the second component or obstacle is more athermal than the

dislocations, the line will shift down and intercept the positive stress a i s . A more

thermal obstacle leads to an upward shift of the line and intercepts the positive inverse
 1 ore Thermal

1 / o or est Dislocations

Flow Stress

Figure 5.3 A schernatic diagrarn of the effects of thermal and athermal obstacles
on a Haasen plot.
activation area axis. Accurate rneasurement of the apparent activation area through

strain rate sensitivity measurements wiil form the basis for determinin the solute

carbon content in the rnatrix.

5.2.5 Determination of Solute Carbon

The theory of linear superposition forms the basis of separating out the

contributions from individual processes of plastic deformation [ 1O]. Separation of the

individual effects of microstructural features such as solute atoms, ciusters,

precipitates, and dislocations is possible with accurate rneasurement of the activation

volume. The effect of the thermally activated obstacles on the activation volume cm

be interpreted as corresponding to the square root of the solute concentration for

randornly dispersed obstacles, as suggested by MacEwen [ I l ] .

The flow stress is the summation of, firstly the thermal stress which is due to

short range obstacles and, secondly, the athermal stress which is due to long range

obstacles [7]. In the case of work hardening the change in thermal stress is

proportional to the flow stress and this proportionality is constant throughout

deformation. Considering solute atoms and forest dislocations as the two obstacle

types, the flow stress can be written as

~=O,+cT,-+bo (5.5)

where a, and a
, are the stresses due to solute atoms and forest dislocations,

o a friction stress. Equation 5.3 can now be rewritten as

respectively and o
 where (cr-aJ represents the flow stress contribution solely from newly generat

forest dislocations and m p n d m, are experirnental strain rate sensitivities defined as

h ( a - a,)
m= (5 - 7)
An E

During testing, it is assumed that forest dislocations and solute atoms are the

only two obstacles that are thermally activatable. Since the solute concentration does

not change dunng deformation, then the influence of solutes is seen as the positive

intercept on the Haasen plot at yielding or when the flow stress equals zero.

For random obstacles obeying Friedel statistics, the apparent activation area is

given as [ 1 1]

k'=
-2 -bAy
,lx (5.8)

where Ay is the distance traveled across a barrier upon activation which is equai to 26,

and f br> is the strengthening factor by interstitial atoms in b.c.c. materials. Typically,

fb. is of the order of 0.1 [Il]. Substituting the known constants into equation 5.8

leads to

c = 17.8b4
(Ad)-' (5-9)

where c is deterrnined to be the soIute concentration in atomic fraction.

5.2.6 Solute Drag

Besides dislocation-dislocation interaction, dislocation-solute interaction must

be considered. A solute can slow a moving dislocation if the thermal fluctuation of the

solute atom can resonate with the dislocation motion. This results in a dynamic state

of Cottrell atmospheres. Energy minimization of a system can occur if solute atoms

jump to sites where they increase the solute-dislocation interaction. This results in a

dnving force for solute atoms to diffise towards dislocations and f o m an atmosphere

around the dislocation. Cottrell and Bilby [12] considered that, with an applied stress,

atmospheres could dynamically form and dissolve. The formation of an atmosphere

around a moving dislocation and the resulting drag on the dislocation was also studied

by Cottrell and Iawson [13]. For a moving dislocation, an asyrnrnetq in the solute

distribution is developed around the dislocation which leads to a drag force opposing

the applied force. The solute drag stress is dependent on dislocation velocity and

solute concentration.

A recent mode1 of solute drag was proposed by James and Bamett [14]. They

used statistics to derive the solute distribution around the dislocation and dso ta

derive the resulting drag force profile on the dislocation. The results of their mode1

are shown in Figure 5.4 for a specirnen with 1 18 ppm (by weight) of carbon in solution

at 220C. The large curve represents the drag stress profile, while the smaller curves

represent the stress changes when a 1/4 or 1/10 strain rate change occurs. The drag

stress for the 1/10 rate change is normalized to that of 1/4 for easy cornparison with

the Haasen plot.

 Since solute drag stress, oh,, is very srnall in cornparison to the dislocation-

dislocation interaction stress, QL, and is dependent only on dislocation velocity, a

linear superposition of the two can be considered [ I 51.

Gtotai= c d i s + c h g (S.10)

The dislocation interaction stress obeys the Cottrell-Strokes law and corresponds to

the linear component of the Haasen plot. Depending on the temperature and strain

rate of the tensile test, the drag stress component could be observed in the Haasen plot

and will manifest itself as a deviation from Iinearity. A theoretical prediction of this is

shown in Figure 5.5.

The initial slope of the Haasen plot suggests that dislocation velocity lies to the

right of the drag peak where negative strain rate sensitivity is seen. As the velocity

decreases through the drag profile, the solute drag component of rate sensitivity

increases giving a positive deviation from linearity at higher stresses.

Through mathematical manipulation of the James & Barnett model, Upadhyaya

1151 found that solute content, colcould be related to the drag stress peak, o,,k,

c. =87.886~~~ (5.1 1)

where c, is in atomic ppm andk,a is in MPa. it was found by this relationship that

predicted results were approxirnately an order of magnitude larger than experimental

results determined on a Ti IF sheet steel [ 151.

 Temperature 220C

[Cl= 118 ppm

\/ (normalized)

Log Velocity (mls)

Figure 5.4 Solute drag profile calculated frorn lames and Barnett analysis [14]
 Drag Profile

Max A ~ O , - ~

Corrected Stress (MPa)

Figure 5.5 Theoretical shape of Haasen plot.

 S. 2-6.1 Semated Flow

Serrated flow or jerky fiow is a discontinuous flow behavior which can occur

during mechanical working. During uniaxiai tensile testing, serrated flow manifests

itself as load jerks. Often there is a criticai strain at which serrated flow initiates.

Hayes and Hayes [16] showed that there are two manifestations of serrated flow;

normal and anomalous behaviour. Below a criticai testing temperature, normal

behaviour is observed. As the strain rate is decreased, the critical strain is decreased,

which is the classical dynamic strain aging case. Above a critical testing temperature,

anomalous behaviour is observed where the critical strain increases with decreasing

strain rate.

Mulford and Kocks [17] have shown a relationship benveen strain rate

sensitivity and the onset and disappearance of serrated flow. When strain rate

sensitivity changes from a positive to a negative value, senations cm appear. Two

mechanisms for the disappearance of serrated flow are possible (161. In the first, the

carbon diffuses down the dislocation line to a precipitate sink and in the second, the

carbon reacts with carbide-forrning species on the dislocation line.

5.2.7 Strain Rate Sensitivity Experiments

Strain rate sensitivity measurements were used to determine the solute carbon

content and the a / a-carbide soIvus line for an anneded Ti-Nb IF steel. The Nb steels

previously used for the AOBH experiments could not be used. The Nb steels have a
high phosphoms content and phosphoms is therrnally activatable in the testing

temperature range. This Ieads to an increase in the apparent inverse activation area

rneasured. The effects of the phosphorus and carbon could not be separated.

Therefore, the method for determuiing the solute carboa content discussed in section

5.2.5 is invalid for the Nb steels and a low phosphorus steel is needed.

The solute carbon content for the Ti-Nb steel was determined fiom the

intercept of the inverse activation area axis on the Haasen plot. The effect of solute

drag on the Haasen plot was also studied by strain rate change tests, assuming its

linear additivity to the dislocation-dislocation interaction.

 5.3 EXPERIMENTAL

The steel used was a comrnercially produced Nb-Ti IF transfer bar. The

chernical composition is given in Table 5.1 .

Table 5.1 Chemical Composition (weight percent)

A stnp 2.5 cm wide was cut parallel to the rolling direction from the 2.5 cm

thick transfer bar. The bar was turned down on a lathe to a diarneter of 1.25 cm. The

rod was then double bagged in Sedpak stainless steel bags, filled with an 0 2 - Ar

mixture. These gases were chosen as the best compromise between an inert

atmosphere and an atmosphere to prevent decarburization and nitriding of the

specimens. The rod was solutionized in an air furnace at 950C for one hour and

furnace cooled. The rod was then swaged to a final diameter of 2 mm, with twelve

decreasing swage blocks, and sectioned into 50 mm lengths.

The sarnple rods were degreased in acetone. For each heat treatment two rods

were double bagged in Sedpaks, with an O2 - Ar mixture and solutionized in a high

temperature salt bath at 850C for 10 minutes to ensure an equilibrium composition

has been reached. The specimens were then step annealed to 660, 500, or 400C for

various times and finally water quenched. Two rods were in each bag for validation of
reproducibility. The specimens were kept in a freezer at -20C between heat treatrnent

and testing tirne, which was necessary to prevent aging of the specimens at roorn

temperature.

Strain rate sensitivity measurements on the specimens were performed on a

special screw driven tende apparatus 12, 31. The gauge length of the specimens was

approximately 35 mm. Nominal strain rates between 2.83~10"s-' and 7 . 0 9 ~ 1 0S-'~

were used with intermittent 114 andor 1/10 rate changes. The compensation applied

for the machine cornpliance was chosen such that the initial load drop was about 95%

of the estimated load change upon rate change. This slight under-compensation leads

to consistent results of &5% in the measured activation area [ 5 ] . Tests were done at

either 220 or 240C in a silicone oil bath with a temperature control of M.1C. These

test temperatures were used to rninimize the effect of Peierls friction stress [15]. The

total carbon content for two specimens (5 min. @ 500C and 240 min. @ 500C) was

detennined in a Leco CS 444 at Dofasco Inc. The specimens were tested to determine

if any carburization had occurred during heat treating or subsequent testing. Both

samples were found to be identical in their total carbon content (0.0045wt%).

Tests were analyzed and plotted as Haasen plots. Mamyama et al. [18]

calculated that under Orowan bowing of a dislocation between two obstacles the

elastic prestrain was 0.02%. This defines the microscopie point of yielding which is

necessary for the determination of activation volume of solutes without the

complications of forest dislocations. Hence, the 0.02% proof stress was used instead

of the engineering value of 0.2% as the flow stress.

 4.4 RESULTS

5.4.1 Strain Rate Sensitivity Test Results

A typical stress-strain curve with strain rate changes is s h o w in Figure 5.6.

The sarnples were pulled until necking occurred at which tirne there is non-unifom

deformation and accurate strain rate change data is unattainable. Figure 5.7, shows

the progression from an under-compensated (upper-lefi) through to an over-

compensated (lower-right) load drop, due to a strain rate change. Suain rate

sensitivity test results are presented in Table 5.2, giving heat treatment, yield strength,

and measured solute carbon concentration.

5.4.2 Variation of Solute Carbon and Yield Stress with Annealing Time

Figure 5.8 shows the decreasing concentration of solute carbon with time,

calculated from strain rate sensitivity tests. During anneaiing at a lower temperature,

after equilibrating at 850C for 10 minutes, precipitation of a second phase is

occumng which decreases the solute carbon concentration. Equilibrium will be

reached if sufficient time is given at the annealing temperature and a a / a-carbide

solvus point can be determined.

The variation in yield stress with annealing time is plotted for two different

steels in Figure 5.9. The Ti steel, studied by Upadhyaya [15], was annealed at 600C

and the present Ti-% steel at 500C. The Ti steel shows constant values f i e r one

hour while the present Ti-Nb steel shows a dramatic increase in strength. During
annealing and the precipitation of a second phase, the loss of yield stress due to

decreased solute content is compensated for by an increase in precipitation hardening.

Therefore, since the Ti-Nb steel shows a significant increase in strength over time, the

results suggest that Nb is a stronger precipitate hardening element than Ti.

Table 5.2 Sample heat treatments, yield stress, and measured solute
carbon concentration.

Apparent Inverse Solute Carbon

Activation Area Concentration
Ippml

60 min @ 500C 1

120 min @ 500C

220
WQ
240 min @ 500C
220
WQ
10 min @ 660C
240
WQ

1 30 min 6fJ 660C 1 refer to text

60 min @ 660C
240 refer to tex?
WQ
 0.00 0-01 0.02 0.03 0.04 0.05 0.06
True Strain
Figure 5.6 Typicd stress-strain curve at a temperature of 220C with strain rate
changes.

Under Compensated

Used Cnterion

: \

'.

Over Compensated

Figure 5.7 Progression from an under-compensated (upper-lefi) to an over-

compensated (lower-right) load drop, due to a strain rate change [tg].
 O 50 100 150 200 250

Time (min)

Figure 5.8 Variation in solute carbon with annealing time at 500C.

 O 50 1 O0 150 200 250 300
Time (min)

Figure 5.9 Variation of yield stress with annealing time for Nb-Ti (anneaied at
500C) and Ti-only (annealed at 600C)IF steels.
5.4.3 Phase Diagram Construction

The solute content was determined at 850, 660, and 500C. The specimens

were step annealed and held at their respective temperatures until equilibrium was

reached. The exceptions are specimens annealed at 660C which showed an initial

decrease in the inverse activation area but increased d e r an annealing time of 15

minutes, as shown in Figure 5.10. This apparent anomaly of the increase in the

apparent inverse activation area is attributed to the fact that at 660C the Tic is more

supersaturated then the NbC (Figure 5.1 1). Hence Tic precipitates but its hardening

effect is less than the solid solution effect of the carbon being removed. However, at

longer times NbC starts to precipitate leading not only to an increase in strength but an

increase in strain rate sensitivity due to the thermal activation of the disc like

precipitates which initially manifests the phenomenon of chernical strengthening. The

collaborating evidence for this precipitate effect can be seen in Figure 5.12. At yield

when the dislocation population is small the solute content in the inter particle regions

c m be detected. However, as deformation progresses the cutting of precipitates leads

to increased strain rate sensitivity.

The resulting a / a-carbide solvus line for the Nb-Ti IF steel is shown in Figure

5.1 1. For cornparison, interna1 friction data for a low-carbon steel from Wert [20] and

strain rate sensitivity data for a Ti IF steel by Upadhyaya [7] are given. The difference

between equilibrium solute contents of the Nb-Ti and Ti steels shows that Ti is a better

carbide former then Nb. Both microalloying elements are better carbide formers than

iron.
 tInverse Activation
Area
! 43- Yield Stress
I

O 10 20 30 40 50 60 70
Time (min)

Figure 5.10 Variation in inverse activation area and yield stress with annealing time
(annealed at 660C).
 iooo

O
-O

2
Y
3
2
a
e2
Upadhyaya [15] - Ti-only IF
steel
-
a Present work Nb+Ti IF
steel
A Wert [20]- Low-carbon steel

100

Carbon Concentration (ppm by wt)

Figure 5.1 1 a / a-carbide solvus line for two IF steels determined by strain rate
sensitivity measurements and a Iow-carbon steel determined by interna1
fiction.
5.4.4 Observation of Solute Drag Effect in Eiaasen Plots

The predicted deviation of the Haasen plot from linearity was observed as

straining of the sarnples progressed. The Haasen plot for a specimen annealed after 4

hours at 500C and tested at 220C is shown in Figure 5.12. The solute carbon in

solution using the 0.02% proof stress intercept method can be determined precisely to

be 3.9 ppm by. However, determining the solute in solution dynamically by the James

and Barnett drag stress method, leads to a concentration value 30 times greater.

Three factors have hindered our abiiity to relate the James and Barnett solute

drag mode1 to our results. Firstly, the theoretical James and Bamett mode1 assumes

that the interactions are only hydrostatic and hence the calibration factor is smdler

than the actual case. Carbon is in interstitial sites with both hydrostatic and

distortional strains around it. Nandedkar and Johnson [2 11 have considered the

distortion around the carbon atom and shown that the interaction potential could be up

to 5 times higher than that due to hydrostatic interaction alone. Furthemore, Nabarro

[22] has evaiuated Fleischer's analysis and concluded that the interstitial piming effect

could be IO times larger.

A funher complication in detemiining a correlation factor is that the NbC

precipitates become obstacles that can be thermally activated at the test temperature

and strain rate ranges used. This interaction is more thermally activatable than the

dislocation-dislocation interaction and shows itself as an increased plateau on the

Haasen plot. Thus a Ti-only stabilized IF steel would be better to more precisely

determine the calibration factor.

 Finally, the concentration range in which the solute drag mode1 can be used is

restricted. The Haasen plot, Figure 5.13, for a specimen annealed at 850C and water

quenched, shows an initial deviation from linearity due to solute drag but then

decreases. This decrease in strain rate sensitivity is attributed to dynamic strain aging

if sufficent amounts of solute carbon are present. As the dislocation velocity decreases

with increasing density the normal Portevin-Le Chatelier effect takes place [16].

Therefore, with solute carbon concentrations greater than -30 ppm the linear

additivity of equation 5.10 is no longer true and the maximum drag stress peak due to

solute carbon is unattainable.

 10 min. @ 850C -
240 min. @ 500C - WQ

Test Temperature 220C

AH = 0.004631
Max Goras = 7.92 MPa
(Cl= 118 ppm
1 Ppt Effects

0.002

3.9 ppm l
O

Corrected Stress (MPa)

---
'
---
+ 114-
-
Rate Changes
- --
a 1110 Rate Changes

Figure 5-12 Haasen plot showing precipitate effects and deviation from linearity due
to solute drag. Solute carbon content calculated by back extrapolation
and solute drag methods.
 Test Temperature 220C

Effect of dynamic
O strain aging

u + +

Corrected Stress (MPa)

.- --A-- - - -- - - - - - - . -

+ 114 Rate Changes o-1/10

- - - -- -
. -.- .- . - --
Rate Changes
. -- -

Figure 5.13 Haasen plot showing dynamic strain aging effects on the drag stress
profile.
 5.5 CONCLUSIONS

The kinetics of carbide precipitation in a niobium-titanium stabilized interstitial-

fiee steel were studied. Solute composition was determined by precision strain rate

sensitivity using inverse activation area versus flow stress at yielding on the Haasen

plot. A correlation between the composition and solute drag stress profile was

perfomed. The following conclusions c m be made:

. Niobium is a more effective precipitate hardening element than titanium.

The determination of the equilibrium carbon content d e r annealing at different

temperatures fomed the basis for an estirnate of the a / a-NbC solvus line. The

higher solubility of NbC than Tic make Nb-stabilized steels more attractive for use

in bake hardening applications.

A correlation factor between composition and the solute drag profile could not be

accurateiy obtained since the precipitates became thermally activatable and thus it

was not possible to separate out the precipitate and solute drag contributions.
 5.6 REFERENCES

J. W. Christian and B.C.Masters, Proc. Roy. Soc. London A, Vol. 28 1, 1964, pp.
223 -240.

H.G. Champion, M.S. Duesbery, and S. Saimoto, Scripta Meialf., Vol. 17, 1983,
pp. 135-140.

S. Saimoto and Z.S. Basinski, Method and Apparatus for Measunng

Thermodynamics Response, U.S. Patent No. 4, 572001, 1986.

M.J. Bull, Development of a Constitutive Relationship for an Aluminum

Chromium Alloy, M. Sc. Thesis, Queen's University, Ontario, Canada, 1981.

S. Saimoto, Journal of Eng. Mater. and Tech., Vol. 109, 1987, pp. 230-235.

P. Haasen, Phii. Mag., Ser. 8, Vol. 3, 1958, pp. 384-418.

A.H. Cotre11 and R J . Stokes, Proc. Roy. Suc. London A, Vol. 233, 1955, pp. 17-
34.

S. Saimoto and H. Sang, Acta Metafl., Vol. 3 1, 1983, pp 1873-1881.

U.F. Kocks, A S . Argon, M.F. Ashby, in Thermodvnamics and Kinetics of Slip,

Progress in Materials Science, Vol. 19, I 975.

R.A. Mulford, Acta. Metall., Vol. 27, 1979, pp. 1 1 15- 1124.

S.R. MacEwen, ActaMetall., Vol. 30, 1982, pp. 1431-1442.

A.H. Cottrell and B.A Bilby, Proc. Phys. Soc. Lo~zdon,Vol. 62, 1949, pp. 49-62

A.H. Cottrell and M.A. Jaswon, Proc. Roy. Soc. London A, Vol. 199, 1949, pp.
104-1 14.

W. James and D.M Bamett, in Solute Defect Interaction: Theory and

Ex~eriment,eds. S. Saimoto, G.R Purdy, and G.V. Kidson, Pergamon Press,
1986, pp. 136-142.

K. Upadhyaya, Analysis of Thennodynamic Response for Materials

Characterization Using Tensile and Nanoindentation Testing, Ph.D. Thesis,
Queen's University, Ontario, Canada, 1994.

R.W. Hayes and W.C. Hayes, Acta Metali., Vol. 32, 1984, pp. 259-267.
17. R A Mulford and V.F. Kocks, Acta Metall., Vol. 27, 1979, p. 1125.

18. K. Maruyama, T, Inui, and H. Oikawa, in Aspects of Hieh Temperature

Deformation and Fracture, eds. Y. Hosoi, H. Yoshinaga, H. Oikawa, and K.
Mamyama, Japan Institute of Metais, 1993.

19. D. Diak, Queen's University, Pnvate Communication.

20. C . A Wen, Trans. MME: JoumalofMetaLr, Vol. 188, 1950, pp. 1242-1244.

2 1 . A. S. Nandedkar and R A Johnson, Acta Metall., Vol. 48, 1982, pp. 2055-2059.

22. F.R.N. Nabarro, in Solution Hardenine. Dislocations and Properties of Real

Matenals, The Metals Society, 1984, pp. 1 52- 169.
 C W T E R VI

SUMMARY & CONCLUSIONS

Quantitative results have been obtained for the effects of different annealing

process variables (temperature, time, and cooling rate) on the AoeH response for

various Nb steels. The chemical composition of the precipitates in the flly stabilized

Nb steel was determined to be Nb-rich. The evolution of the precipitate size and

density after annealing for various times was measured by TEM. The solute carbon

content was determined by the ageing index rnethod for the same stabilized Nb steel

and by strain rate sensitivity measurements for a Ti-Nb steel. Both methods produced

quantitative results, from various annealing cycles, on the kinetics of carbide

dissolution and precipitation. Precipitate hardening effects were also observed. The

dissolution of the precipitates and evolution of the solute carbon content has been

related to the AoSH response. From these results the following conclusions can be

made:

The increase in doBHis directly related tu the amount of solute carbon present in

the matrix after processing and a mathematical correlation between the two has

been developed.

The solute carbon content available for bake hardening is determined by the

dissolution of NbC precipitates during continuous annealing. Initially, precipitate

density decreases and average precipitate diameter increases with holding times up

to 3 minutes. With eaended holding times, there is little change in precipitate

density and average diameter.

The theoretical solubility limit for NbC in the fully stabilized steel is approached

after one minute of annealing and reached within three minutes.

Dissolution of fine NbC precipitates is very rapid at 880C, while medium-sized

and larger precipitates do not dissolve significantly, in accordance with Aaron's

theoretical dissolution model.

The cooling rate greatly affects th response. With slower cooling rates,

more time is allowed for the reprecipitation of NbC and/or FelC, therefore leaving

less solute carbon in the matrix for bake hardening.

Higher annealing temperatures produce higher AoBH values. The higher the

annealing temperature the greater the solubility limit. The solubility limit is closely

approached within the first minute of annealing.

Over-stabilized steels produce poor A ( J values

~ ~ due to the excess solute niobium

present in the rnatrix. This excess makes reprecipitation of NbC easier dunng
 cooling. Under-stabilized steels produce higher AceH responses with lower

annealing temperatures and are less sensitive to cooling rates.

The increase in yield stress and decrease in solute carbon concentration with

increased annealing time showed Nb to be a stronger agent for precipitation

hardening than Ti.

NbC is more soluble then Tic. Therefore, Nb steels ailow more carbon to be

present in solution d e r annealing for bake hardening.

For the testing temperature and strain rate used, solute drag effects manifest a

deviation from linearity in the Haasen plot. A correlation between this

experimentally measured drag peak and the solute carbon concentration was

hindered by the formation of thermally activatable precipitates, which result in an

increased plateau in the Haasen plot.

Functional relationships between the evolution of the precipitates, solute

carbon content, and AG^^ response have been determined. These relationships result

in a better understanding of how the processing variables affect the evolution of Aoeli

in IF steels. Further work is needed in studying the effects of the deformed

microstructure on the aging kinetics and the interaction of different species of solute

atorns with disIocations.

 lMAGE EVALUATION
TEST TARGET (QA-3)

APPLIED I W G t . lnc
-
=
-. --
1653 East Main Street
--
Rochester, NY 14609 USA
--
--
- -
=A - - Phone: 716/482-0300
F a : 7161288-5989

Q 1993. Applied Image. lm..All Rights R e s e t v e d